Fix: The minimum size of the historical message window for the classification operator is 1. #12778 (#12779)

### What problem does this PR solve?

Fix: The minimum size of the historical message window for the
classification operator is 1. #12778

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)

.github/copilot-instructions.md

@@ -0,0 +1 @@
+Refer to [AGENTS.MD](../AGENTS.md) for all repo instructions.

.github/workflows/release.yml

@@ -3,11 +3,18 @@ name: release
 on:
   schedule:
     - cron: '0 13 * * *'  # This schedule runs every 13:00:00Z(21:00:00+08:00)
+  # https://github.com/orgs/community/discussions/26286?utm_source=chatgpt.com#discussioncomment-3251208
+  # "The create event does not support branch filter and tag filter."
   # The "create tags" trigger is specifically focused on the creation of new tags, while the "push tags" trigger is activated when tags are pushed, including both new tag creations and updates to existing tags.
-  create:
+  push:
     tags:
       - "v*.*.*"                  # normal release
-      - "nightly"                 # the only one mutable tag
+
+permissions:
+  contents: write
+  actions: read
+  checks: read
+  statuses: read
 
 # https://docs.github.com/en/actions/using-jobs/using-concurrency
 concurrency:
@@ -21,9 +28,9 @@ jobs:
       - name: Ensure workspace ownership
         run: echo "chown -R ${USER} ${GITHUB_WORKSPACE}" && sudo chown -R ${USER} ${GITHUB_WORKSPACE}
 
-      # https://github.com/actions/checkout/blob/v3/README.md
+      # https://github.com/actions/checkout/blob/v6/README.md
       - name: Check out code
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           token: ${{ secrets.GITHUB_TOKEN }}  # Use the secret as an environment variable
           fetch-depth: 0
@@ -31,12 +38,12 @@ jobs:
 
       - name: Prepare release body
         run: |
-          if [[ ${GITHUB_EVENT_NAME} == "create" ]]; then
+          if [[ ${GITHUB_EVENT_NAME} != "schedule" ]]; then
             RELEASE_TAG=${GITHUB_REF#refs/tags/}
-            if [[ ${RELEASE_TAG} == "nightly" ]]; then
-              PRERELEASE=true
-            else
+            if [[ ${RELEASE_TAG} == v* ]]; then
               PRERELEASE=false
+            else
+              PRERELEASE=true
             fi
             echo "Workflow triggered by create tag: ${RELEASE_TAG}"
           else
@@ -55,7 +62,7 @@ jobs:
           git fetch --tags
           if [[ ${GITHUB_EVENT_NAME} == "schedule" ]]; then
             # Determine if a given tag exists and matches a specific Git commit.
-            # actions/checkout@v4 fetch-tags doesn't work when triggered by schedule
+            # actions/checkout@v6 fetch-tags doesn't work when triggered by schedule
             if [ "$(git rev-parse -q --verify "refs/tags/${RELEASE_TAG}")" = "${GITHUB_SHA}" ]; then
               echo "mutable tag ${RELEASE_TAG} exists and matches ${GITHUB_SHA}"
             else
@@ -75,6 +82,14 @@ jobs:
           # The body field does not support environment variable substitution directly.
           body_path: release_body.md
 
+      - name: Build and push image
+        run: |
+          sudo docker login --username infiniflow --password-stdin <<< ${{ secrets.DOCKERHUB_TOKEN }}
+          sudo docker build --build-arg NEED_MIRROR=1 --build-arg HTTPS_PROXY=${HTTPS_PROXY} --build-arg HTTP_PROXY=${HTTP_PROXY} -t infiniflow/ragflow:${RELEASE_TAG} -f Dockerfile .
+          sudo docker tag infiniflow/ragflow:${RELEASE_TAG} infiniflow/ragflow:latest
+          sudo docker push infiniflow/ragflow:${RELEASE_TAG}
+          sudo docker push infiniflow/ragflow:latest
+
       - name: Build and push ragflow-sdk
         if: startsWith(github.ref, 'refs/tags/v')
         run: |
@@ -84,11 +99,3 @@ jobs:
         if: startsWith(github.ref, 'refs/tags/v')
         run: |
           cd admin/client && uv build && uv publish --token ${{ secrets.PYPI_API_TOKEN }}
-
-      - name: Build and push image
-        run: |
-          sudo docker login --username infiniflow --password-stdin <<< ${{ secrets.DOCKERHUB_TOKEN }}
-          sudo docker build --build-arg NEED_MIRROR=1 -t infiniflow/ragflow:${RELEASE_TAG} -f Dockerfile .
-          sudo docker tag infiniflow/ragflow:${RELEASE_TAG} infiniflow/ragflow:latest
-          sudo docker push infiniflow/ragflow:${RELEASE_TAG}
-          sudo docker push infiniflow/ragflow:latest

.github/workflows/tests.yml

@@ -1,4 +1,6 @@
 name: tests
+permissions:
+  contents: read
 
 on:
   push:
@@ -12,7 +14,7 @@ on:
   # The only difference between pull_request and pull_request_target is the context in which the workflow runs:
   # — pull_request_target workflows use the workflow files from the default branch, and secrets are available.
   # — pull_request workflows use the workflow files from the pull request branch, and secrets are unavailable.
-  pull_request_target:
+  pull_request:
     types: [ synchronize, ready_for_review ]
     paths-ignore:
       - 'docs/**'
@@ -31,12 +33,9 @@ jobs:
     name: ragflow_tests
     # https://docs.github.com/en/actions/using-jobs/using-conditions-to-control-job-execution
     # https://github.com/orgs/community/discussions/26261
-    if: ${{ github.event_name != 'pull_request_target' || contains(github.event.pull_request.labels.*.name, 'ci') }}
+    if: ${{ github.event_name != 'pull_request' || (github.event.pull_request.draft == false && contains(github.event.pull_request.labels.*.name, 'ci')) }}
     runs-on: [ "self-hosted", "ragflow-test" ]
     steps:
-      # https://github.com/hmarr/debug-action
-      #- uses: hmarr/debug-action@v2
-
       - name: Ensure workspace ownership
         run: |
           echo "Workflow triggered by ${{ github.event_name }}"
@@ -44,7 +43,7 @@ jobs:
 
       # https://github.com/actions/checkout/issues/1781
       - name: Check out code
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           ref: ${{ (github.event_name == 'pull_request' || github.event_name == 'pull_request_target') && format('refs/pull/{0}/merge', github.event.pull_request.number) || github.sha }}
           fetch-depth: 0
@@ -53,7 +52,7 @@ jobs:
       - name: Check workflow duplication
         if: ${{ !cancelled() && !failure() }}
         run: |
-          if [[ ${GITHUB_EVENT_NAME} != "pull_request_target" && ${GITHUB_EVENT_NAME} != "schedule" ]]; then
+          if [[ ${GITHUB_EVENT_NAME} != "pull_request" && ${GITHUB_EVENT_NAME} != "schedule" ]]; then
             HEAD=$(git rev-parse HEAD)
             # Find a PR that introduced a given commit
             gh auth login --with-token <<< "${{ secrets.GITHUB_TOKEN }}"
@@ -78,7 +77,7 @@ jobs:
                 fi
               fi
             fi
-          elif [[ ${GITHUB_EVENT_NAME} == "pull_request_target" ]]; then
+          elif [[ ${GITHUB_EVENT_NAME} == "pull_request" ]]; then
             PR_NUMBER=${{ github.event.pull_request.number }}
             PR_SHA_FP=${RUNNER_WORKSPACE_PREFIX}/artifacts/${GITHUB_REPOSITORY}/PR_${PR_NUMBER}
             # Calculate the hash of the current workspace content
@@ -87,6 +86,9 @@ jobs:
             mkdir -p ${RUNNER_WORKSPACE_PREFIX}/artifacts/${GITHUB_REPOSITORY}
             echo "${PR_SHA} ${GITHUB_RUN_ID}" > ${PR_SHA_FP}
           fi
+          ARTIFACTS_DIR=${RUNNER_WORKSPACE_PREFIX}/artifacts/${GITHUB_REPOSITORY}/${GITHUB_RUN_ID}
+          echo "ARTIFACTS_DIR=${ARTIFACTS_DIR}" >> ${GITHUB_ENV}
+          rm -rf ${ARTIFACTS_DIR} && mkdir -p ${ARTIFACTS_DIR}
 
       # https://github.com/astral-sh/ruff-action
       - name: Static check with Ruff
@@ -98,7 +100,7 @@ jobs:
       - name: Check comments of changed Python files
         if: ${{ false }}
         run: |
-          if [[ ${{ github.event_name }} == 'pull_request_target' ]]; then
+          if [[ ${{ github.event_name }} == 'pull_request' || ${{ github.event_name }} == 'pull_request_target' ]]; then
             CHANGED_FILES=$(git diff --name-only ${{ github.event.pull_request.base.sha }}...${{ github.event.pull_request.head.sha }} \
               | grep -E '\.(py)$' || true)
 
@@ -127,13 +129,21 @@ jobs:
             fi
           fi
 
+      - name: Run unit test
+        run: |
+          uv sync --python 3.12 --group test --frozen
+          source .venv/bin/activate
+          which pytest || echo "pytest not in PATH"
+          echo "Start to run unit test"
+          python3 run_tests.py
+
       - name: Build ragflow:nightly
         run: |
           RUNNER_WORKSPACE_PREFIX=${RUNNER_WORKSPACE_PREFIX:-${HOME}}
           RAGFLOW_IMAGE=infiniflow/ragflow:${GITHUB_RUN_ID}
           echo "RAGFLOW_IMAGE=${RAGFLOW_IMAGE}" >> ${GITHUB_ENV}
           sudo docker pull ubuntu:22.04
-          sudo DOCKER_BUILDKIT=1 docker build --build-arg NEED_MIRROR=1 -f Dockerfile -t ${RAGFLOW_IMAGE} .
+          sudo DOCKER_BUILDKIT=1 docker build --build-arg NEED_MIRROR=1 --build-arg HTTPS_PROXY=${HTTPS_PROXY} --build-arg HTTP_PROXY=${HTTP_PROXY} -f Dockerfile -t ${RAGFLOW_IMAGE} .
           if [[ ${GITHUB_EVENT_NAME} == "schedule" ]]; then
             export HTTP_API_TEST_LEVEL=p3
           else
@@ -154,7 +164,7 @@ jobs:
           INFINITY_THRIFT_PORT=$((23817 + RUNNER_NUM * 10))
           INFINITY_HTTP_PORT=$((23820 + RUNNER_NUM * 10))
           INFINITY_PSQL_PORT=$((5432 + RUNNER_NUM * 10))
-          MYSQL_PORT=$((5455 + RUNNER_NUM * 10))
+          EXPOSE_MYSQL_PORT=$((5455 + RUNNER_NUM * 10))
           MINIO_PORT=$((9000 + RUNNER_NUM * 10))
           MINIO_CONSOLE_PORT=$((9001 + RUNNER_NUM * 10))
           REDIS_PORT=$((6379 + RUNNER_NUM * 10))
@@ -174,7 +184,7 @@ jobs:
           echo -e "INFINITY_THRIFT_PORT=${INFINITY_THRIFT_PORT}" >> docker/.env
           echo -e "INFINITY_HTTP_PORT=${INFINITY_HTTP_PORT}" >> docker/.env
           echo -e "INFINITY_PSQL_PORT=${INFINITY_PSQL_PORT}" >> docker/.env
-          echo -e "MYSQL_PORT=${MYSQL_PORT}" >> docker/.env
+          echo -e "EXPOSE_MYSQL_PORT=${EXPOSE_MYSQL_PORT}" >> docker/.env
           echo -e "MINIO_PORT=${MINIO_PORT}" >> docker/.env
           echo -e "MINIO_CONSOLE_PORT=${MINIO_CONSOLE_PORT}" >> docker/.env
           echo -e "REDIS_PORT=${REDIS_PORT}" >> docker/.env
@@ -193,34 +203,41 @@ jobs:
           echo "HOST_ADDRESS=http://host.docker.internal:${SVR_HTTP_PORT}" >> ${GITHUB_ENV}
 
           sudo docker compose -f docker/docker-compose.yml -p ${GITHUB_RUN_ID} up -d
-          uv sync --python 3.10 --only-group test --no-default-groups --frozen && uv pip install sdk/python
+          uv sync --python 3.12 --only-group test --no-default-groups --frozen && uv pip install sdk/python --group test
 
       - name: Run sdk tests against Elasticsearch
         run: |
           export http_proxy=""; export https_proxy=""; export no_proxy=""; export HTTP_PROXY=""; export HTTPS_PROXY=""; export NO_PROXY=""
-          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS}/v1/system/ping > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          source .venv/bin/activate && pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api
+          source .venv/bin/activate && set -o pipefail; pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api 2>&1 | tee es_sdk_test.log
 
-      - name: Run frontend api tests against Elasticsearch
+      - name: Run web api tests against Elasticsearch
         run: |
           export http_proxy=""; export https_proxy=""; export no_proxy=""; export HTTP_PROXY=""; export HTTPS_PROXY=""; export NO_PROXY=""
-          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS}/v1/system/ping > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          source .venv/bin/activate && pytest -s --tb=short sdk/python/test/test_frontend_api/get_email.py sdk/python/test/test_frontend_api/test_dataset.py
+          source .venv/bin/activate && set -o pipefail; pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_web_api/ 2>&1 | tee es_web_api_test.log
           
       - name: Run http api tests against Elasticsearch
         run: |
           export http_proxy=""; export https_proxy=""; export no_proxy=""; export HTTP_PROXY=""; export HTTPS_PROXY=""; export NO_PROXY=""
-          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS}/v1/system/ping > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          source .venv/bin/activate && pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api
+          source .venv/bin/activate && set -o pipefail; pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api 2>&1 | tee es_http_api_test.log
+
+      - name: Collect ragflow log
+        if: ${{ !cancelled() }}
+        run: |
+          cp -r docker/ragflow-logs ${ARTIFACTS_DIR}/ragflow-logs-es
+          echo "ragflow log" && tail -n 200 docker/ragflow-logs/ragflow_server.log
+          sudo rm -rf docker/ragflow-logs
 
       - name: Stop ragflow:nightly
         if: always()  # always run this step even if previous steps failed
@@ -236,30 +253,36 @@ jobs:
       - name: Run sdk tests against Infinity
         run: |
           export http_proxy=""; export https_proxy=""; export no_proxy=""; export HTTP_PROXY=""; export HTTPS_PROXY=""; export NO_PROXY=""
-          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS}/v1/system/ping > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api
+          source .venv/bin/activate && set -o pipefail; DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api 2>&1 | tee infinity_sdk_test.log
 
-      - name: Run frontend api tests against Infinity
+      - name: Run web api tests against Infinity
         run: |
           export http_proxy=""; export https_proxy=""; export no_proxy=""; export HTTP_PROXY=""; export HTTPS_PROXY=""; export NO_PROXY=""
-          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS}/v1/system/ping > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short sdk/python/test/test_frontend_api/get_email.py sdk/python/test/test_frontend_api/test_dataset.py
+          source .venv/bin/activate && set -o pipefail; DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_web_api/ 2>&1 | tee infinity_web_api_test.log
 
       - name: Run http api tests against Infinity
         run: |
           export http_proxy=""; export https_proxy=""; export no_proxy=""; export HTTP_PROXY=""; export HTTPS_PROXY=""; export NO_PROXY=""
-          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS}/v1/system/ping > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api
+          source .venv/bin/activate && set -o pipefail; DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api 2>&1 | tee infinity_http_api_test.log
 
+      - name: Collect ragflow log
+        if: ${{ !cancelled() }}
+        run: |
+          cp -r docker/ragflow-logs ${ARTIFACTS_DIR}/ragflow-logs-infinity
+          echo "ragflow log" && tail -n 200 docker/ragflow-logs/ragflow_server.log
+          sudo rm -rf docker/ragflow-logs
       - name: Stop ragflow:nightly
         if: always()  # always run this step even if previous steps failed
         run: |

.gitignore

@@ -44,6 +44,7 @@ cl100k_base.tiktoken
 chrome*
 huggingface.co/
 nltk_data/
+uv-x86_64*.tar.gz
 
 # Exclude hash-like temporary files like 9b5ad71b2ce5302211f9c61530b329a4922fc6a4
 *[0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f]*
@@ -51,6 +52,13 @@ nltk_data/
 .venv
 docker/data
 
+# OceanBase data and conf
+docker/oceanbase/conf
+docker/oceanbase/data
+
+# SeekDB data and conf
+docker/seekdb
+
 
 #--------------------------------------------------#
 # The following was generated with gitignore.nvim: #
@@ -195,3 +203,11 @@ ragflow_cli.egg-info
 
 # Default backup dir
 backup
+
+
+.hypothesis
+
+
+# Added by cargo
+
+/target

AGENTS.md

@@ -0,0 +1,110 @@
+# RAGFlow Project Instructions for GitHub Copilot
+
+This file provides context, build instructions, and coding standards for the RAGFlow project.
+It is structured to follow GitHub Copilot's [customization guidelines](https://docs.github.com/en/copilot/concepts/prompting/response-customization).
+
+## 1. Project Overview
+RAGFlow is an open-source RAG (Retrieval-Augmented Generation) engine based on deep document understanding. It is a full-stack application with a Python backend and a React/TypeScript frontend.
+
+- **Backend**: Python 3.10+ (Flask/Quart)
+- **Frontend**: TypeScript, React, UmiJS
+- **Architecture**: Microservices based on Docker.
+  - `api/`: Backend API server.
+  - `rag/`: Core RAG logic (indexing, retrieval).
+  - `deepdoc/`: Document parsing and OCR.
+  - `web/`: Frontend application.
+
+## 2. Directory Structure
+- `api/`: Backend API server (Flask/Quart).
+  - `apps/`: API Blueprints (Knowledge Base, Chat, etc.).
+  - `db/`: Database models and services.
+- `rag/`: Core RAG logic.
+  - `llm/`: LLM, Embedding, and Rerank model abstractions.
+- `deepdoc/`: Document parsing and OCR modules.
+- `agent/`: Agentic reasoning components.
+- `web/`: Frontend application (React + UmiJS).
+- `docker/`: Docker deployment configurations.
+- `sdk/`: Python SDK.
+- `test/`: Backend tests.
+
+## 3. Build Instructions
+
+### Backend (Python)
+The project uses **uv** for dependency management.
+
+1. **Setup Environment**:
+   ```bash
+   uv sync --python 3.12 --all-extras
+   uv run download_deps.py
+   ```
+
+2. **Run Server**:
+   - **Pre-requisite**: Start dependent services (MySQL, ES/Infinity, Redis, MinIO).
+     ```bash
+     docker compose -f docker/docker-compose-base.yml up -d
+     ```
+   - **Launch**:
+     ```bash
+     source .venv/bin/activate
+     export PYTHONPATH=$(pwd)
+     bash docker/launch_backend_service.sh
+     ```
+
+### Frontend (TypeScript/React)
+Located in `web/`.
+
+1. **Install Dependencies**:
+   ```bash
+   cd web
+   npm install
+   ```
+
+2. **Run Dev Server**:
+   ```bash
+   npm run dev
+   ```
+   Runs on port 8000 by default.
+
+### Docker Deployment
+To run the full stack using Docker:
+```bash
+cd docker
+docker compose -f docker-compose.yml up -d
+```
+
+## 4. Testing Instructions
+
+### Backend Tests
+- **Run All Tests**:
+  ```bash
+  uv run pytest
+  ```
+- **Run Specific Test**:
+  ```bash
+  uv run pytest test/test_api.py
+  ```
+
+### Frontend Tests
+- **Run Tests**:
+  ```bash
+  cd web
+  npm run test
+  ```
+
+## 5. Coding Standards & Guidelines
+- **Python Formatting**: Use `ruff` for linting and formatting.
+  ```bash
+  ruff check
+  ruff format
+  ```
+- **Frontend Linting**:
+  ```bash
+  cd web
+  npm run lint
+  ```
+- **Pre-commit**: Ensure pre-commit hooks are installed.
+  ```bash
+  pre-commit install
+  pre-commit run --all-files
+  ```
+

CLAUDE.md

@@ -45,7 +45,7 @@ RAGFlow is an open-source RAG (Retrieval-Augmented Generation) engine based on d
 ### Backend Development
 ```bash
 # Install Python dependencies
-uv sync --python 3.10 --all-extras
+uv sync --python 3.12 --all-extras
 uv run download_deps.py
 pre-commit install
 

Dockerfile

@@ -1,5 +1,5 @@
 # base stage
-FROM ubuntu:22.04 AS base
+FROM ubuntu:24.04 AS base
 USER root
 SHELL ["/bin/bash", "-c"]
 
@@ -10,60 +10,67 @@ WORKDIR /ragflow
 # Copy models downloaded via download_deps.py
 RUN mkdir -p /ragflow/rag/res/deepdoc /root/.ragflow
 RUN --mount=type=bind,from=infiniflow/ragflow_deps:latest,source=/huggingface.co,target=/huggingface.co \
-    cp /huggingface.co/InfiniFlow/huqie/huqie.txt.trie /ragflow/rag/res/ && \
     tar --exclude='.*' -cf - \
         /huggingface.co/InfiniFlow/text_concat_xgb_v1.0 \
         /huggingface.co/InfiniFlow/deepdoc \
-        | tar -xf - --strip-components=3 -C /ragflow/rag/res/deepdoc 
+        | tar -xf - --strip-components=3 -C /ragflow/rag/res/deepdoc
 
 # https://github.com/chrismattmann/tika-python
 # This is the only way to run python-tika without internet access. Without this set, the default is to check the tika version and pull latest every time from Apache.
 RUN --mount=type=bind,from=infiniflow/ragflow_deps:latest,source=/,target=/deps \
     cp -r /deps/nltk_data /root/ && \
-    cp /deps/tika-server-standard-3.0.0.jar /deps/tika-server-standard-3.0.0.jar.md5 /ragflow/ && \
+    cp /deps/tika-server-standard-3.2.3.jar /deps/tika-server-standard-3.2.3.jar.md5 /ragflow/ && \
     cp /deps/cl100k_base.tiktoken /ragflow/9b5ad71b2ce5302211f9c61530b329a4922fc6a4
 
-ENV TIKA_SERVER_JAR="file:///ragflow/tika-server-standard-3.0.0.jar"
+ENV TIKA_SERVER_JAR="file:///ragflow/tika-server-standard-3.2.3.jar"
 ENV DEBIAN_FRONTEND=noninteractive
 
 # Setup apt
 # Python package and implicit dependencies:
 # opencv-python: libglib2.0-0 libglx-mesa0 libgl1
 # aspose-slides: pkg-config libicu-dev libgdiplus         libssl1.1_1.1.1f-1ubuntu2_amd64.deb
-# python-pptx:   default-jdk                              tika-server-standard-3.0.0.jar
+# python-pptx:   default-jdk                              tika-server-standard-3.2.3.jar
 # selenium:      libatk-bridge2.0-0                       chrome-linux64-121-0-6167-85
 # Building C extensions: libpython3-dev libgtk-4-1 libnss3 xdg-utils libgbm-dev
 RUN --mount=type=cache,id=ragflow_apt,target=/var/cache/apt,sharing=locked \
+    apt update && \
+    apt --no-install-recommends install -y ca-certificates; \
     if [ "$NEED_MIRROR" == "1" ]; then \
-        sed -i 's|http://ports.ubuntu.com|http://mirrors.tuna.tsinghua.edu.cn|g' /etc/apt/sources.list; \
-        sed -i 's|http://archive.ubuntu.com|http://mirrors.tuna.tsinghua.edu.cn|g' /etc/apt/sources.list; \
+        sed -i 's|http://archive.ubuntu.com/ubuntu|https://mirrors.tuna.tsinghua.edu.cn/ubuntu|g' /etc/apt/sources.list.d/ubuntu.sources; \
+        sed -i 's|http://security.ubuntu.com/ubuntu|https://mirrors.tuna.tsinghua.edu.cn/ubuntu|g' /etc/apt/sources.list.d/ubuntu.sources; \
     fi; \
     rm -f /etc/apt/apt.conf.d/docker-clean && \
     echo 'Binary::apt::APT::Keep-Downloaded-Packages "true";' > /etc/apt/apt.conf.d/keep-cache && \
     chmod 1777 /tmp && \
     apt update && \
-    apt --no-install-recommends install -y ca-certificates && \
-    apt update && \
     apt install -y libglib2.0-0 libglx-mesa0 libgl1 && \
     apt install -y pkg-config libicu-dev libgdiplus && \
     apt install -y default-jdk && \
     apt install -y libatk-bridge2.0-0 && \
     apt install -y libpython3-dev libgtk-4-1 libnss3 xdg-utils libgbm-dev && \
     apt install -y libjemalloc-dev && \
-    apt install -y python3-pip pipx nginx unzip curl wget git vim less && \
+    apt install -y nginx unzip curl wget git vim less && \
     apt install -y ghostscript && \
     apt install -y pandoc && \
-    apt install -y texlive
+    apt install -y texlive && \
+    apt install -y fonts-freefont-ttf fonts-noto-cjk && \
+    apt install -y postgresql-client
 
-RUN if [ "$NEED_MIRROR" == "1" ]; then \
-        pip3 config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple && \
-        pip3 config set global.trusted-host pypi.tuna.tsinghua.edu.cn; \
+# Install uv
+RUN --mount=type=bind,from=infiniflow/ragflow_deps:latest,source=/,target=/deps \
+    if [ "$NEED_MIRROR" == "1" ]; then \
         mkdir -p /etc/uv && \
-        echo "[[index]]" > /etc/uv/uv.toml && \
+        echo 'python-install-mirror = "https://registry.npmmirror.com/-/binary/python-build-standalone/"' > /etc/uv/uv.toml && \
+        echo '[[index]]' >> /etc/uv/uv.toml && \
         echo 'url = "https://pypi.tuna.tsinghua.edu.cn/simple"' >> /etc/uv/uv.toml && \
-        echo "default = true" >> /etc/uv/uv.toml; \
+        echo 'default = true' >> /etc/uv/uv.toml; \
     fi; \
-    pipx install uv
+    arch="$(uname -m)"; \
+    if [ "$arch" = "x86_64" ]; then uv_arch="x86_64"; else uv_arch="aarch64"; fi; \
+    tar xzf "/deps/uv-${uv_arch}-unknown-linux-gnu.tar.gz" \
+    && cp "uv-${uv_arch}-unknown-linux-gnu/"* /usr/local/bin/ \
+    && rm -rf "uv-${uv_arch}-unknown-linux-gnu" \
+    && uv python install 3.12
 
 ENV PYTHONDONTWRITEBYTECODE=1 DOTNET_SYSTEM_GLOBALIZATION_INVARIANT=1
 ENV PATH=/root/.local/bin:$PATH
@@ -79,12 +86,12 @@ RUN --mount=type=cache,id=ragflow_apt,target=/var/cache/apt,sharing=locked \
 # A modern version of cargo is needed for the latest version of the Rust compiler.
 RUN apt update && apt install -y curl build-essential \
     && if [ "$NEED_MIRROR" == "1" ]; then \
-         # Use TUNA mirrors for rustup/rust dist files
+         # Use TUNA mirrors for rustup/rust dist files \
          export RUSTUP_DIST_SERVER="https://mirrors.tuna.tsinghua.edu.cn/rustup"; \
          export RUSTUP_UPDATE_ROOT="https://mirrors.tuna.tsinghua.edu.cn/rustup/rustup"; \
          echo "Using TUNA mirrors for Rustup."; \
        fi; \
-    # Force curl to use HTTP/1.1
+    # Force curl to use HTTP/1.1 \
     curl --proto '=https' --tlsv1.2 --http1.1 -sSf https://sh.rustup.rs | bash -s -- -y --profile minimal \
     && echo 'export PATH="/root/.cargo/bin:${PATH}"' >> /root/.bashrc
 
@@ -101,10 +108,10 @@ RUN --mount=type=cache,id=ragflow_apt,target=/var/cache/apt,sharing=locked \
     apt update && \
     arch="$(uname -m)"; \
     if [ "$arch" = "arm64" ] || [ "$arch" = "aarch64" ]; then \
-        # ARM64 (macOS/Apple Silicon or Linux aarch64)
+        # ARM64 (macOS/Apple Silicon or Linux aarch64) \
         ACCEPT_EULA=Y apt install -y unixodbc-dev msodbcsql18; \
     else \
-        # x86_64 or others
+        # x86_64 or others \
         ACCEPT_EULA=Y apt install -y unixodbc-dev msodbcsql17; \
     fi || \
     { echo "Failed to install ODBC driver"; exit 1; }
@@ -148,11 +155,14 @@ RUN --mount=type=cache,id=ragflow_uv,target=/root/.cache/uv,sharing=locked \
     else \
         sed -i 's|pypi.tuna.tsinghua.edu.cn|pypi.org|g' uv.lock; \
     fi; \
-    uv sync --python 3.10 --frozen
+    uv sync --python 3.12 --frozen && \
+    # Ensure pip is available in the venv for runtime package installation (fixes #12651)
+    .venv/bin/python3 -m ensurepip --upgrade
 
 COPY web web
 COPY docs docs
 RUN --mount=type=cache,id=ragflow_npm,target=/root/.npm,sharing=locked \
+    export NODE_OPTIONS="--max-old-space-size=4096" && \
     cd web && npm install && npm run build
 
 COPY .git /ragflow/.git
@@ -183,11 +193,11 @@ COPY deepdoc deepdoc
 COPY rag rag
 COPY agent agent
 COPY graphrag graphrag
-COPY agentic_reasoning agentic_reasoning
 COPY pyproject.toml uv.lock ./
 COPY mcp mcp
 COPY plugin plugin
 COPY common common
+COPY memory memory
 
 COPY docker/service_conf.yaml.template ./conf/service_conf.yaml.template
 COPY docker/entrypoint.sh ./

Dockerfile.deps

@@ -3,7 +3,7 @@
 FROM scratch
 
 # Copy resources downloaded via download_deps.py
-COPY chromedriver-linux64-121-0-6167-85 chrome-linux64-121-0-6167-85 cl100k_base.tiktoken libssl1.1_1.1.1f-1ubuntu2_amd64.deb libssl1.1_1.1.1f-1ubuntu2_arm64.deb tika-server-standard-3.0.0.jar tika-server-standard-3.0.0.jar.md5 libssl*.deb /
+COPY chromedriver-linux64-121-0-6167-85 chrome-linux64-121-0-6167-85 cl100k_base.tiktoken libssl1.1_1.1.1f-1ubuntu2_amd64.deb libssl1.1_1.1.1f-1ubuntu2_arm64.deb tika-server-standard-3.2.3.jar tika-server-standard-3.2.3.jar.md5 libssl*.deb uv-x86_64-unknown-linux-gnu.tar.gz uv-aarch64-unknown-linux-gnu.tar.gz /
 
 COPY nltk_data /nltk_data
 

README.md

@@ -22,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.22.1">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.23.1">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -37,7 +37,7 @@
 
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Document</a> |
-  <a href="https://github.com/infiniflow/ragflow/issues/4214">Roadmap</a> |
+  <a href="https://github.com/infiniflow/ragflow/issues/12241">Roadmap</a> |
   <a href="https://twitter.com/infiniflowai">Twitter</a> |
   <a href="https://discord.gg/NjYzJD3GM3">Discord</a> |
   <a href="https://demo.ragflow.io">Demo</a>
@@ -72,7 +72,7 @@
 
 ## 💡 What is RAGFlow?
 
-[RAGFlow](https://ragflow.io/) is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context layer for LLMs. It offers a streamlined RAG workflow adaptable to enterprises of any scale. Powered by a converged context engine and pre-built agent templates, RAGFlow enables developers to transform complex data into high-fidelity, production-ready AI systems with exceptional efficiency and precision.
+[RAGFlow](https://ragflow.io/) is a leading open-source Retrieval-Augmented Generation ([RAG](https://ragflow.io/basics/what-is-rag)) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context layer for LLMs. It offers a streamlined RAG workflow adaptable to enterprises of any scale. Powered by a converged [context engine](https://ragflow.io/basics/what-is-agent-context-engine) and pre-built agent templates, RAGFlow enables developers to transform complex data into high-fidelity, production-ready AI systems with exceptional efficiency and precision.
 
 ## 🎮 Demo
 
@@ -85,6 +85,7 @@ Try our demo at [https://demo.ragflow.io](https://demo.ragflow.io).
 
 ## 🔥 Latest Updates
 
+- 2025-12-26 Supports 'Memory' for AI agent.
 - 2025-11-19 Supports Gemini 3 Pro.
 - 2025-11-12 Supports data synchronization from Confluence, S3, Notion, Discord, Google Drive.
 - 2025-10-23 Supports MinerU & Docling as document parsing methods.
@@ -187,15 +188,15 @@ releases! 🌟
 > All Docker images are built for x86 platforms. We don't currently offer Docker images for ARM64.
 > If you are on an ARM64 platform, follow [this guide](https://ragflow.io/docs/dev/build_docker_image) to build a Docker image compatible with your system.
 
-> The command below downloads the `v0.22.1` edition of the RAGFlow Docker image. See the following table for descriptions of different RAGFlow editions. To download a RAGFlow edition different from `v0.22.1`, update the `RAGFLOW_IMAGE` variable accordingly in **docker/.env** before using `docker compose` to start the server.
+> The command below downloads the `v0.23.1` edition of the RAGFlow Docker image. See the following table for descriptions of different RAGFlow editions. To download a RAGFlow edition different from `v0.23.1`, update the `RAGFLOW_IMAGE` variable accordingly in **docker/.env** before using `docker compose` to start the server.
 
 ```bash
    $ cd ragflow/docker
-  
-   # git checkout v0.22.1
+
+   # git checkout v0.23.1
    # Optional: use a stable tag (see releases: https://github.com/infiniflow/ragflow/releases)
-   # This steps ensures the **entrypoint.sh** file in the code matches the Docker image version.
-   
+   # This step ensures the **entrypoint.sh** file in the code matches the Docker image version.
+
    # Use CPU for DeepDoc tasks:
    $ docker compose -f docker-compose.yml up -d
 
@@ -206,10 +207,10 @@ releases! 🌟
 
 > Note: Prior to `v0.22.0`, we provided both images with embedding models and slim images without embedding models. Details as follows:
 
-| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-| ----------------- | --------------- | --------------------- | ------------------------ |
-| v0.21.1           | &approx;9       | ✔️                    | Stable release           |
-| v0.21.1-slim      | &approx;2       | ❌                    | Stable release           |
+| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?        |
+|-------------------|-----------------|-----------------------|----------------|
+| v0.21.1           | &approx;9       | ✔️                    | Stable release |
+| v0.21.1-slim      | &approx;2       | ❌                     | Stable release |
 
 > Starting with `v0.22.0`, we ship only the slim edition and no longer append the **-slim** suffix to the image tag.
 
@@ -232,7 +233,7 @@ releases! 🌟
     * Running on all addresses (0.0.0.0)
    ```
 
-   > If you skip this confirmation step and directly log in to RAGFlow, your browser may prompt a `network anormal`
+   > If you skip this confirmation step and directly log in to RAGFlow, your browser may prompt a `network abnormal`
    > error because, at that moment, your RAGFlow may not be fully initialized.
    >
 5. In your web browser, enter the IP address of your server and log in to RAGFlow.
@@ -302,6 +303,15 @@ cd ragflow/
 docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly .
 ```
 
+Or if you are behind a proxy, you can pass proxy arguments:
+
+```bash
+docker build --platform linux/amd64 \
+  --build-arg http_proxy=http://YOUR_PROXY:PORT \
+  --build-arg https_proxy=http://YOUR_PROXY:PORT \
+  -f Dockerfile -t infiniflow/ragflow:nightly .
+```
+
 ## 🔨 Launch service from source for development
 
 1. Install `uv` and `pre-commit`, or skip this step if they are already installed:
@@ -314,7 +324,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 # install RAGFlow dependent python modules
+   uv sync --python 3.12 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
@@ -386,7 +396,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
 
 ## 📜 Roadmap
 
-See the [RAGFlow Roadmap 2025](https://github.com/infiniflow/ragflow/issues/4214)
+See the [RAGFlow Roadmap 2026](https://github.com/infiniflow/ragflow/issues/12241)
 
 ## 🏄 Community
 

README_id.md

@@ -22,7 +22,7 @@
         <img alt="Lencana Daring" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.22.1">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.23.1">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Rilis%20Terbaru" alt="Rilis Terbaru">
@@ -37,7 +37,7 @@
 
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Dokumentasi</a> |
-  <a href="https://github.com/infiniflow/ragflow/issues/4214">Peta Jalan</a> |
+  <a href="https://github.com/infiniflow/ragflow/issues/12241">Peta Jalan</a> |
   <a href="https://twitter.com/infiniflowai">Twitter</a> |
   <a href="https://discord.gg/NjYzJD3GM3">Discord</a> |
   <a href="https://demo.ragflow.io">Demo</a>
@@ -72,7 +72,7 @@
 
 ## 💡 Apa Itu RAGFlow?
 
-[RAGFlow](https://ragflow.io/) adalah mesin RAG (Retrieval-Augmented Generation) open-source terkemuka yang mengintegrasikan teknologi RAG mutakhir dengan kemampuan Agent untuk menciptakan lapisan kontekstual superior bagi LLM. Menyediakan alur kerja RAG yang efisien dan dapat diadaptasi untuk perusahaan segala skala. Didukung oleh mesin konteks terkonvergensi dan template Agent yang telah dipra-bangun, RAGFlow memungkinkan pengembang mengubah data kompleks menjadi sistem AI kesetiaan-tinggi dan siap-produksi dengan efisiensi dan presisi yang luar biasa.
+[RAGFlow](https://ragflow.io/) adalah mesin [RAG](https://ragflow.io/basics/what-is-rag) (Retrieval-Augmented Generation) open-source terkemuka yang mengintegrasikan teknologi RAG mutakhir dengan kemampuan Agent untuk menciptakan lapisan kontekstual superior bagi LLM. Menyediakan alur kerja RAG yang efisien dan dapat diadaptasi untuk perusahaan segala skala. Didukung oleh mesin konteks terkonvergensi dan template Agent yang telah dipra-bangun, RAGFlow memungkinkan pengembang mengubah data kompleks menjadi sistem AI kesetiaan-tinggi dan siap-produksi dengan efisiensi dan presisi yang luar biasa.
 
 ## 🎮 Demo
 
@@ -85,6 +85,7 @@ Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
 
 ## 🔥 Pembaruan Terbaru
 
+- 2025-12-26 Mendukung 'Memori' untuk agen AI.
 - 2025-11-19 Mendukung Gemini 3 Pro.
 - 2025-11-12 Mendukung sinkronisasi data dari Confluence, S3, Notion, Discord, Google Drive.
 - 2025-10-23 Mendukung MinerU & Docling sebagai metode penguraian dokumen.
@@ -187,12 +188,12 @@ Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
 > Semua gambar Docker dibangun untuk platform x86. Saat ini, kami tidak menawarkan gambar Docker untuk ARM64.
 > Jika Anda menggunakan platform ARM64, [silakan gunakan panduan ini untuk membangun gambar Docker yang kompatibel dengan sistem Anda](https://ragflow.io/docs/dev/build_docker_image).
 
-> Perintah di bawah ini mengunduh edisi v0.22.1 dari gambar Docker RAGFlow. Silakan merujuk ke tabel berikut untuk deskripsi berbagai edisi RAGFlow. Untuk mengunduh edisi RAGFlow yang berbeda dari v0.22.1, perbarui variabel RAGFLOW_IMAGE di docker/.env sebelum menggunakan docker compose untuk memulai server.
+> Perintah di bawah ini mengunduh edisi v0.23.1 dari gambar Docker RAGFlow. Silakan merujuk ke tabel berikut untuk deskripsi berbagai edisi RAGFlow. Untuk mengunduh edisi RAGFlow yang berbeda dari v0.23.1, perbarui variabel RAGFLOW_IMAGE di docker/.env sebelum menggunakan docker compose untuk memulai server.
 
 ```bash
    $ cd ragflow/docker
-   
-   # git checkout v0.22.1
+
+   # git checkout v0.23.1
    # Opsional: gunakan tag stabil (lihat releases: https://github.com/infiniflow/ragflow/releases)
    # This steps ensures the **entrypoint.sh** file in the code matches the Docker image version.
 
@@ -206,10 +207,10 @@ Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
 
 > Catatan: Sebelum `v0.22.0`, kami menyediakan image dengan model embedding dan image slim tanpa model embedding. Detailnya sebagai berikut:
 
-| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-| ----------------- | --------------- | --------------------- | ------------------------ |
-| v0.21.1           | &approx;9       | ✔️                    | Stable release           |
-| v0.21.1-slim      | &approx;2       | ❌                    | Stable release           |
+| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?        |
+|-------------------|-----------------|-----------------------|----------------|
+| v0.21.1           | &approx;9       | ✔️                    | Stable release |
+| v0.21.1-slim      | &approx;2       | ❌                     | Stable release |
 
 > Mulai dari `v0.22.0`, kami hanya menyediakan edisi slim dan tidak lagi menambahkan akhiran **-slim** pada tag image.
 
@@ -232,7 +233,7 @@ Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
     * Running on all addresses (0.0.0.0)
    ```
 
-   > Jika Anda melewatkan langkah ini dan langsung login ke RAGFlow, browser Anda mungkin menampilkan error `network anormal`
+   > Jika Anda melewatkan langkah ini dan langsung login ke RAGFlow, browser Anda mungkin menampilkan error `network abnormal`
    > karena RAGFlow mungkin belum sepenuhnya siap.
    >
 2. Buka browser web Anda, masukkan alamat IP server Anda, dan login ke RAGFlow.
@@ -276,6 +277,15 @@ cd ragflow/
 docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly .
 ```
 
+Jika berada di belakang proxy, Anda dapat melewatkan argumen proxy:
+
+```bash
+docker build --platform linux/amd64 \
+  --build-arg http_proxy=http://YOUR_PROXY:PORT \
+  --build-arg https_proxy=http://YOUR_PROXY:PORT \
+  -f Dockerfile -t infiniflow/ragflow:nightly .
+```
+
 ## 🔨 Menjalankan Aplikasi dari untuk Pengembangan
 
 1. Instal `uv` dan `pre-commit`, atau lewati langkah ini jika sudah terinstal:
@@ -288,7 +298,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 # install RAGFlow dependent python modules
+   uv sync --python 3.12 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
@@ -358,7 +368,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
 
 ## 📜 Roadmap
 
-Lihat [Roadmap RAGFlow 2025](https://github.com/infiniflow/ragflow/issues/4214)
+Lihat [Roadmap RAGFlow 2026](https://github.com/infiniflow/ragflow/issues/12241)
 
 ## 🏄 Komunitas
 

README_ja.md

@@ -22,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.22.1">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.23.1">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -37,7 +37,7 @@
 
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Document</a> |
-  <a href="https://github.com/infiniflow/ragflow/issues/4214">Roadmap</a> |
+  <a href="https://github.com/infiniflow/ragflow/issues/12241">Roadmap</a> |
   <a href="https://twitter.com/infiniflowai">Twitter</a> |
   <a href="https://discord.gg/NjYzJD3GM3">Discord</a> |
   <a href="https://demo.ragflow.io">Demo</a>
@@ -53,7 +53,7 @@
 
 ## 💡 RAGFlow とは？
 
-[RAGFlow](https://ragflow.io/) は、先進的なRAG（Retrieval-Augmented Generation）技術と Agent 機能を融合し、大規模言語モデル（LLM）に優れたコンテキスト層を構築する最先端のオープンソース RAG エンジンです。あらゆる規模の企業に対応可能な合理化された RAG ワークフローを提供し、統合型コンテキストエンジンと事前構築されたAgentテンプレートにより、開発者が複雑なデータを驚異的な効率性と精度で高精細なプロダクションレディAIシステムへ変換することを可能にします。
+[RAGFlow](https://ragflow.io/) は、先進的な[RAG](https://ragflow.io/basics/what-is-rag)（Retrieval-Augmented Generation）技術と Agent 機能を融合し、大規模言語モデル（LLM）に優れたコンテキスト層を構築する最先端のオープンソース RAG エンジンです。あらゆる規模の企業に対応可能な合理化された RAG ワークフローを提供し、統合型[コンテキストエンジン](https://ragflow.io/basics/what-is-agent-context-engine)と事前構築されたAgentテンプレートにより、開発者が複雑なデータを驚異的な効率性と精度で高精細なプロダクションレディAIシステムへ変換することを可能にします。
 
 ## 🎮 Demo
 
@@ -66,7 +66,8 @@
 
 ## 🔥 最新情報
 
-- 2025-11-19 Gemini 3 Proをサポートしています
+- 2025-12-26 AIエージェントの「メモリ」機能をサポート。
+- 2025-11-19 Gemini 3 Proをサポートしています。
 - 2025-11-12 Confluence、S3、Notion、Discord、Google Drive からのデータ同期をサポートします。
 - 2025-10-23 ドキュメント解析方法として MinerU と Docling をサポートします。
 - 2025-10-15 オーケストレーションされたデータパイプラインのサポート。
@@ -167,12 +168,12 @@
 > 現在、公式に提供されているすべての Docker イメージは x86 アーキテクチャ向けにビルドされており、ARM64 用の Docker イメージは提供されていません。
 > ARM64 アーキテクチャのオペレーティングシステムを使用している場合は、[このドキュメント](https://ragflow.io/docs/dev/build_docker_image)を参照して Docker イメージを自分でビルドしてください。
 
-> 以下のコマンドは、RAGFlow Docker イメージの v0.22.1 エディションをダウンロードします。異なる RAGFlow エディションの説明については、以下の表を参照してください。v0.22.1 とは異なるエディションをダウンロードするには、docker/.env ファイルの RAGFLOW_IMAGE 変数を適宜更新し、docker compose を使用してサーバーを起動してください。
+> 以下のコマンドは、RAGFlow Docker イメージの v0.23.1 エディションをダウンロードします。異なる RAGFlow エディションの説明については、以下の表を参照してください。v0.23.1 とは異なるエディションをダウンロードするには、docker/.env ファイルの RAGFLOW_IMAGE 変数を適宜更新し、docker compose を使用してサーバーを起動してください。
 
 ```bash
    $ cd ragflow/docker
 
-   # git checkout v0.22.1
+   # git checkout v0.23.1
    # 任意: 安定版タグを利用 (一覧: https://github.com/infiniflow/ragflow/releases)
    # この手順は、コード内の entrypoint.sh ファイルが Docker イメージのバージョンと一致していることを確認します。
 
@@ -186,15 +187,15 @@
 
 > 注意：`v0.22.0` より前のバージョンでは、embedding モデルを含むイメージと、embedding モデルを含まない slim イメージの両方を提供していました。詳細は以下の通りです：
 
-| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-| ----------------- | --------------- | --------------------- | ------------------------ |
-| v0.21.1           | &approx;9       | ✔️                    | Stable release           |
-| v0.21.1-slim      | &approx;2       | ❌                    | Stable release           |
+| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?        |
+|-------------------|-----------------|-----------------------|----------------|
+| v0.21.1           | &approx;9       | ✔️                    | Stable release |
+| v0.21.1-slim      | &approx;2       | ❌                     | Stable release |
 
 > `v0.22.0` 以降、当プロジェクトでは slim エディションのみを提供し、イメージタグに **-slim** サフィックスを付けなくなりました。
 
-   1. サーバーを立ち上げた後、サーバーの状態を確認する:   
-   
+   1. サーバーを立ち上げた後、サーバーの状態を確認する:
+
    ```bash
    $ docker logs -f docker-ragflow-cpu-1
    ```
@@ -276,6 +277,15 @@ cd ragflow/
 docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly .
 ```
 
+プロキシ環境下にいる場合は、プロキシ引数を指定できます：
+
+```bash
+docker build --platform linux/amd64 \
+  --build-arg http_proxy=http://YOUR_PROXY:PORT \
+  --build-arg https_proxy=http://YOUR_PROXY:PORT \
+  -f Dockerfile -t infiniflow/ragflow:nightly .
+```
+
 ## 🔨 ソースコードからサービスを起動する方法
 
 1. `uv` と `pre-commit` をインストールする。すでにインストールされている場合は、このステップをスキップしてください:
@@ -288,7 +298,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 # install RAGFlow dependent python modules
+   uv sync --python 3.12 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
@@ -358,7 +368,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
 
 ## 📜 ロードマップ
 
-[RAGFlow ロードマップ 2025](https://github.com/infiniflow/ragflow/issues/4214) を参照
+[RAGFlow ロードマップ 2026](https://github.com/infiniflow/ragflow/issues/12241) を参照
 
 ## 🏄 コミュニティ
 

README_ko.md

@@ -22,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.22.1">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.23.1">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -37,7 +37,7 @@
 
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Document</a> |
-  <a href="https://github.com/infiniflow/ragflow/issues/4214">Roadmap</a> |
+  <a href="https://github.com/infiniflow/ragflow/issues/12241">Roadmap</a> |
   <a href="https://twitter.com/infiniflowai">Twitter</a> |
   <a href="https://discord.gg/NjYzJD3GM3">Discord</a> |
   <a href="https://demo.ragflow.io">Demo</a>
@@ -54,7 +54,7 @@
 
 ## 💡 RAGFlow란?
 
-[RAGFlow](https://ragflow.io/) 는 최첨단 RAG(Retrieval-Augmented Generation)와 Agent 기능을 융합하여 대규모 언어 모델(LLM)을 위한 우수한 컨텍스트 계층을 생성하는 선도적인 오픈소스 RAG 엔진입니다. 모든 규모의 기업에 적용 가능한 효율적인 RAG 워크플로를 제공하며, 통합 컨텍스트 엔진과 사전 구축된 Agent 템플릿을 통해 개발자들이 복잡한 데이터를 예외적인 효율성과 정밀도로 고급 구현도의 프로덕션 준비 완료 AI 시스템으로 변환할 수 있도록 지원합니다.
+[RAGFlow](https://ragflow.io/) 는 최첨단 [RAG](https://ragflow.io/basics/what-is-rag)(Retrieval-Augmented Generation)와 Agent 기능을 융합하여 대규모 언어 모델(LLM)을 위한 우수한 컨텍스트 계층을 생성하는 선도적인 오픈소스 RAG 엔진입니다. 모든 규모의 기업에 적용 가능한 효율적인 RAG 워크플로를 제공하며, 통합 [컨텍스트 엔진](https://ragflow.io/basics/what-is-agent-context-engine)과 사전 구축된 Agent 템플릿을 통해 개발자들이 복잡한 데이터를 예외적인 효율성과 정밀도로 고급 구현도의 프로덕션 준비 완료 AI 시스템으로 변환할 수 있도록 지원합니다.
 
 ## 🎮 데모
 
@@ -67,6 +67,7 @@
 
 ## 🔥 업데이트
 
+- 2025-12-26 AI 에이전트의 '메모리' 기능 지원.
 - 2025-11-19 Gemini 3 Pro를 지원합니다.
 - 2025-11-12 Confluence, S3, Notion, Discord, Google Drive에서 데이터 동기화를 지원합니다.
 - 2025-10-23 문서 파싱 방법으로 MinerU 및 Docling을 지원합니다.
@@ -169,12 +170,12 @@
 > 모든 Docker 이미지는 x86 플랫폼을 위해 빌드되었습니다. 우리는 현재 ARM64 플랫폼을 위한 Docker 이미지를 제공하지 않습니다.
 > ARM64 플랫폼을 사용 중이라면, [시스템과 호환되는 Docker 이미지를 빌드하려면 이 가이드를 사용해 주세요](https://ragflow.io/docs/dev/build_docker_image).
 
-   > 아래 명령어는 RAGFlow Docker 이미지의 v0.22.1 버전을 다운로드합니다. 다양한 RAGFlow 버전에 대한 설명은 다음 표를 참조하십시오. v0.22.1과 다른 RAGFlow 버전을 다운로드하려면, docker/.env 파일에서 RAGFLOW_IMAGE 변수를 적절히 업데이트한 후 docker compose를 사용하여 서버를 시작하십시오.
+   > 아래 명령어는 RAGFlow Docker 이미지의 v0.23.1 버전을 다운로드합니다. 다양한 RAGFlow 버전에 대한 설명은 다음 표를 참조하십시오. v0.23.1과 다른 RAGFlow 버전을 다운로드하려면, docker/.env 파일에서 RAGFLOW_IMAGE 변수를 적절히 업데이트한 후 docker compose를 사용하여 서버를 시작하십시오.
 
    ```bash
    $ cd ragflow/docker
-   
-   # git checkout v0.22.1
+
+   # git checkout v0.23.1
    # Optional: use a stable tag (see releases: https://github.com/infiniflow/ragflow/releases)
    # 이 단계는 코드의 entrypoint.sh 파일이 Docker 이미지 버전과 일치하도록 보장합니다.
 
@@ -188,10 +189,10 @@
 
 > 참고: `v0.22.0` 이전 버전에서는 embedding 모델이 포함된 이미지와 embedding 모델이 포함되지 않은 slim 이미지를 모두 제공했습니다. 자세한 내용은 다음과 같습니다:
 
-| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-| ----------------- | --------------- | --------------------- | ------------------------ |
-| v0.21.1           | &approx;9       | ✔️                    | Stable release           |
-| v0.21.1-slim      | &approx;2       | ❌                    | Stable release           |
+| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?        |
+|-------------------|-----------------|-----------------------|----------------|
+| v0.21.1           | &approx;9       | ✔️                    | Stable release |
+| v0.21.1-slim      | &approx;2       | ❌                     | Stable release |
 
 > `v0.22.0`부터는 slim 에디션만 배포하며 이미지 태그에 **-slim** 접미사를 더 이상 붙이지 않습니다.
 
@@ -213,7 +214,7 @@
     * Running on all addresses (0.0.0.0)
    ```
 
-   > 만약 확인 단계를 건너뛰고 바로 RAGFlow에 로그인하면, RAGFlow가 완전히 초기화되지 않았기 때문에 브라우저에서 `network anormal` 오류가 발생할 수 있습니다.
+   > 만약 확인 단계를 건너뛰고 바로 RAGFlow에 로그인하면, RAGFlow가 완전히 초기화되지 않았기 때문에 브라우저에서 `network abnormal` 오류가 발생할 수 있습니다.
 
 2. 웹 브라우저에 서버의 IP 주소를 입력하고 RAGFlow에 로그인하세요.
    > 기본 설정을 사용할 경우, `http://IP_OF_YOUR_MACHINE`만 입력하면 됩니다 (포트 번호는 제외). 기본 HTTP 서비스 포트 `80`은 기본 구성으로 사용할 때 생략할 수 있습니다.
@@ -270,6 +271,15 @@ cd ragflow/
 docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly .
 ```
 
+프록시 환경인 경우, 프록시 인수를 전달할 수 있습니다：
+
+```bash
+docker build --platform linux/amd64 \
+  --build-arg http_proxy=http://YOUR_PROXY:PORT \
+  --build-arg https_proxy=http://YOUR_PROXY:PORT \
+  -f Dockerfile -t infiniflow/ragflow:nightly .
+```
+
 ## 🔨 소스 코드로 서비스를 시작합니다.
 
 1. `uv` 와 `pre-commit` 을 설치하거나, 이미 설치된 경우 이 단계를 건너뜁니다:
@@ -283,7 +293,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 # install RAGFlow dependent python modules
+   uv sync --python 3.12 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
@@ -362,7 +372,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
 
 ## 📜 로드맵
 
-[RAGFlow 로드맵 2025](https://github.com/infiniflow/ragflow/issues/4214)을 확인하세요.
+[RAGFlow 로드맵 2026](https://github.com/infiniflow/ragflow/issues/12241)을 확인하세요.
 
 ## 🏄 커뮤니티
 

README_pt_br.md

@@ -22,7 +22,7 @@
         <img alt="Badge Estático" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.22.1">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.23.1">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Última%20Relese" alt="Última Versão">
@@ -37,7 +37,7 @@
 
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Documentação</a> |
-  <a href="https://github.com/infiniflow/ragflow/issues/4214">Roadmap</a> |
+  <a href="https://github.com/infiniflow/ragflow/issues/12241">Roadmap</a> |
   <a href="https://twitter.com/infiniflowai">Twitter</a> |
   <a href="https://discord.gg/NjYzJD3GM3">Discord</a> |
   <a href="https://demo.ragflow.io">Demo</a>
@@ -73,7 +73,7 @@
 
 ## 💡 O que é o RAGFlow?
 
-[RAGFlow](https://ragflow.io/) é um mecanismo de RAG (Retrieval-Augmented Generation) open-source líder que fusiona tecnologias RAG de ponta com funcionalidades Agent para criar uma camada contextual superior para LLMs. Oferece um fluxo de trabalho RAG otimizado adaptável a empresas de qualquer escala. Alimentado por um motor de contexto convergente e modelos Agent pré-construídos, o RAGFlow permite que desenvolvedores transformem dados complexos em sistemas de IA de alta fidelidade e pronto para produção com excepcional eficiência e precisão.
+[RAGFlow](https://ragflow.io/) é um mecanismo de [RAG](https://ragflow.io/basics/what-is-rag) (Retrieval-Augmented Generation) open-source líder que fusiona tecnologias RAG de ponta com funcionalidades Agent para criar uma camada contextual superior para LLMs. Oferece um fluxo de trabalho RAG otimizado adaptável a empresas de qualquer escala. Alimentado por [um motor de contexto](https://ragflow.io/basics/what-is-agent-context-engine) convergente e modelos Agent pré-construídos, o RAGFlow permite que desenvolvedores transformem dados complexos em sistemas de IA de alta fidelidade e pronto para produção com excepcional eficiência e precisão.
 
 ## 🎮 Demo
 
@@ -86,6 +86,7 @@ Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
 
 ## 🔥 Últimas Atualizações
 
+- 26-12-2025 Suporte à função 'Memória' para agentes de IA.
 - 19-11-2025 Suporta Gemini 3 Pro.
 - 12-11-2025 Suporta a sincronização de dados do Confluence, S3, Notion, Discord e Google Drive.
 - 23-10-2025 Suporta MinerU e Docling como métodos de análise de documentos.
@@ -187,12 +188,12 @@ Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
 > Todas as imagens Docker são construídas para plataformas x86. Atualmente, não oferecemos imagens Docker para ARM64.
 > Se você estiver usando uma plataforma ARM64, por favor, utilize [este guia](https://ragflow.io/docs/dev/build_docker_image) para construir uma imagem Docker compatível com o seu sistema.
 
-    > O comando abaixo baixa a edição`v0.22.1` da imagem Docker do RAGFlow. Consulte a tabela a seguir para descrições de diferentes edições do RAGFlow. Para baixar uma edição do RAGFlow diferente da `v0.22.1`, atualize a variável `RAGFLOW_IMAGE` conforme necessário no **docker/.env** antes de usar `docker compose` para iniciar o servidor.
+    > O comando abaixo baixa a edição`v0.23.1` da imagem Docker do RAGFlow. Consulte a tabela a seguir para descrições de diferentes edições do RAGFlow. Para baixar uma edição do RAGFlow diferente da `v0.23.1`, atualize a variável `RAGFLOW_IMAGE` conforme necessário no **docker/.env** antes de usar `docker compose` para iniciar o servidor.
 
 ```bash
    $ cd ragflow/docker
-   
-   # git checkout v0.22.1
+
+   # git checkout v0.23.1
    # Opcional: use uma tag estável (veja releases: https://github.com/infiniflow/ragflow/releases)
    # Esta etapa garante que o arquivo entrypoint.sh no código corresponda à versão da imagem do Docker.
 
@@ -206,10 +207,10 @@ Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
 
 > Nota: Antes da `v0.22.0`, fornecíamos imagens com modelos de embedding e imagens slim sem modelos de embedding. Detalhes a seguir:
 
-| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-| ----------------- | --------------- | --------------------- | ------------------------ |
-| v0.21.1           | &approx;9       | ✔️                    | Stable release           |
-| v0.21.1-slim      | &approx;2       | ❌                    | Stable release           |
+| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?        |
+|-------------------|-----------------|-----------------------|----------------|
+| v0.21.1           | &approx;9       | ✔️                    | Stable release |
+| v0.21.1-slim      | &approx;2       | ❌                     | Stable release |
 
 > A partir da `v0.22.0`, distribuímos apenas a edição slim e não adicionamos mais o sufixo **-slim** às tags das imagens.
 
@@ -231,7 +232,7 @@ Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
     * Rodando em todos os endereços (0.0.0.0)
    ```
 
-   > Se você pular essa etapa de confirmação e acessar diretamente o RAGFlow, seu navegador pode exibir um erro `network anormal`, pois, nesse momento, seu RAGFlow pode não estar totalmente inicializado.
+   > Se você pular essa etapa de confirmação e acessar diretamente o RAGFlow, seu navegador pode exibir um erro `network abnormal`, pois, nesse momento, seu RAGFlow pode não estar totalmente inicializado.
    >
 5. No seu navegador, insira o endereço IP do seu servidor e faça login no RAGFlow.
 
@@ -293,6 +294,15 @@ cd ragflow/
 docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly .
 ```
 
+Se você estiver atrás de um proxy, pode passar argumentos de proxy:
+
+```bash
+docker build --platform linux/amd64 \
+  --build-arg http_proxy=http://YOUR_PROXY:PORT \
+  --build-arg https_proxy=http://YOUR_PROXY:PORT \
+  -f Dockerfile -t infiniflow/ragflow:nightly .
+```
+
 ## 🔨 Lançar o serviço a partir do código-fonte para desenvolvimento
 
 1. Instale o `uv` e o `pre-commit`, ou pule esta etapa se eles já estiverem instalados:
@@ -305,7 +315,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 # instala os módulos Python dependentes do RAGFlow
+   uv sync --python 3.12 # instala os módulos Python dependentes do RAGFlow
    uv run download_deps.py
    pre-commit install
    ```
@@ -375,7 +385,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
 
 ## 📜 Roadmap
 
-Veja o [RAGFlow Roadmap 2025](https://github.com/infiniflow/ragflow/issues/4214)
+Veja o [RAGFlow Roadmap 2026](https://github.com/infiniflow/ragflow/issues/12241)
 
 ## 🏄 Comunidade
 

README_tzh.md

@@ -22,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.22.1">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.23.1">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -37,7 +37,7 @@
 
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Document</a> |
-  <a href="https://github.com/infiniflow/ragflow/issues/4214">Roadmap</a> |
+  <a href="https://github.com/infiniflow/ragflow/issues/12241">Roadmap</a> |
   <a href="https://twitter.com/infiniflowai">Twitter</a> |
   <a href="https://discord.gg/NjYzJD3GM3">Discord</a> |
   <a href="https://demo.ragflow.io">Demo</a>
@@ -72,7 +72,7 @@
 
 ## 💡 RAGFlow 是什麼？
 
-[RAGFlow](https://ragflow.io/) 是一款領先的開源 RAG（Retrieval-Augmented Generation）引擎，通過融合前沿的 RAG 技術與 Agent 能力，為大型語言模型提供卓越的上下文層。它提供可適配任意規模企業的端到端 RAG 工作流，憑藉融合式上下文引擎與預置的 Agent 模板，助力開發者以極致效率與精度將複雜數據轉化為高可信、生產級的人工智能系統。
+[RAGFlow](https://ragflow.io/) 是一款領先的開源 [RAG](https://ragflow.io/basics/what-is-rag)（Retrieval-Augmented Generation）引擎，通過融合前沿的 RAG 技術與 Agent 能力，為大型語言模型提供卓越的上下文層。它提供可適配任意規模企業的端到端 RAG 工作流，憑藉融合式[上下文引擎](https://ragflow.io/basics/what-is-agent-context-engine)與預置的 Agent 模板，助力開發者以極致效率與精度將複雜數據轉化為高可信、生產級的人工智能系統。
 
 ## 🎮 Demo 試用
 
@@ -85,15 +85,16 @@
 
 ## 🔥 近期更新
 
-- 2025-11-19 支援 Gemini 3 Pro.
+- 2025-12-26 支援AI代理的「記憶」功能。
+- 2025-11-19 支援 Gemini 3 Pro。
 - 2025-11-12 支援從 Confluence、S3、Notion、Discord、Google Drive 進行資料同步。
 - 2025-10-23 支援 MinerU 和 Docling 作為文件解析方法。
 - 2025-10-15 支援可編排的資料管道。
 - 2025-08-08 支援 OpenAI 最新的 GPT-5 系列模型。
-- 2025-08-01 支援 agentic workflow 和 MCP
+- 2025-08-01 支援 agentic workflow 和 MCP。
 - 2025-05-23 為 Agent 新增 Python/JS 程式碼執行器元件。
 - 2025-05-05 支援跨語言查詢。
-- 2025-03-19 PDF和DOCX中的圖支持用多模態大模型去解析得到描述.
+- 2025-03-19 PDF和DOCX中的圖支持用多模態大模型去解析得到描述。
 - 2024-12-18 升級了 DeepDoc 的文檔佈局分析模型。
 - 2024-08-22 支援用 RAG 技術實現從自然語言到 SQL 語句的轉換。
 
@@ -124,7 +125,7 @@
 
 ### 🍔 **相容各類異質資料來源**
 
-- 支援豐富的文件類型，包括 Word 文件、PPT、excel 表格、txt 檔案、圖片、PDF、影印件、影印件、結構化資料、網頁等。
+- 支援豐富的文件類型，包括 Word 文件、PPT、excel 表格、txt 檔案、圖片、PDF、影印件、複印件、結構化資料、網頁等。
 
 ### 🛀 **全程無憂、自動化的 RAG 工作流程**
 
@@ -186,12 +187,12 @@
 > 所有 Docker 映像檔都是為 x86 平台建置的。目前，我們不提供 ARM64 平台的 Docker 映像檔。
 > 如果您使用的是 ARM64 平台，請使用 [這份指南](https://ragflow.io/docs/dev/build_docker_image) 來建置適合您系統的 Docker 映像檔。
 
-> 執行以下指令會自動下載 RAGFlow Docker 映像 `v0.22.1`。請參考下表查看不同 Docker 發行版的說明。如需下載不同於 `v0.22.1` 的 Docker 映像，請在執行 `docker compose` 啟動服務之前先更新 **docker/.env** 檔案內的 `RAGFLOW_IMAGE` 變數。
+> 執行以下指令會自動下載 RAGFlow Docker 映像 `v0.23.1`。請參考下表查看不同 Docker 發行版的說明。如需下載不同於 `v0.23.1` 的 Docker 映像，請在執行 `docker compose` 啟動服務之前先更新 **docker/.env** 檔案內的 `RAGFLOW_IMAGE` 變數。
 
 ```bash
    $ cd ragflow/docker
-   
-   # git checkout v0.22.1
+
+   # git checkout v0.23.1
    # 可選：使用穩定版標籤（查看發佈：https://github.com/infiniflow/ragflow/releases）
    # 此步驟確保程式碼中的 entrypoint.sh 檔案與 Docker 映像版本一致。
 
@@ -205,10 +206,10 @@
 
 > 注意：在 `v0.22.0` 之前的版本，我們會同時提供包含 embedding 模型的映像和不含 embedding 模型的 slim 映像。具體如下：
 
-| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-| ----------------- | --------------- | --------------------- | ------------------------ |
-| v0.21.1           | &approx;9       | ✔️                    | Stable release           |
-| v0.21.1-slim      | &approx;2       | ❌                    | Stable release           |
+| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?        |
+|-------------------|-----------------|-----------------------|----------------|
+| v0.21.1           | &approx;9       | ✔️                    | Stable release |
+| v0.21.1-slim      | &approx;2       | ❌                     | Stable release |
 
 > 從 `v0.22.0` 開始，我們只發佈 slim 版本，並且不再在映像標籤後附加 **-slim** 後綴。
 
@@ -236,7 +237,7 @@
     * Running on all addresses (0.0.0.0)
    ```
 
-   > 如果您跳過這一步驟系統確認步驟就登入 RAGFlow，你的瀏覽器有可能會提示 `network anormal` 或 `網路異常`，因為 RAGFlow 可能並未完全啟動成功。
+   > 如果您跳過這一步驟系統確認步驟就登入 RAGFlow，你的瀏覽器有可能會提示 `network abnormal` 或 `網路異常`，因為 RAGFlow 可能並未完全啟動成功。
    >
 5. 在你的瀏覽器中輸入你的伺服器對應的 IP 位址並登入 RAGFlow。
 
@@ -302,6 +303,15 @@ cd ragflow/
 docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly .
 ```
 
+若您位於代理環境，可傳遞代理參數：
+
+```bash
+docker build --platform linux/amd64 \
+  --build-arg http_proxy=http://YOUR_PROXY:PORT \
+  --build-arg https_proxy=http://YOUR_PROXY:PORT \
+  -f Dockerfile -t infiniflow/ragflow:nightly .
+```
+
 ## 🔨 以原始碼啟動服務
 
 1. 安裝 `uv` 和 `pre-commit`。如已安裝，可跳過此步驟：
@@ -315,7 +325,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 # install RAGFlow dependent python modules
+   uv sync --python 3.12 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
@@ -389,7 +399,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
 
 ## 📜 路線圖
 
-詳見 [RAGFlow Roadmap 2025](https://github.com/infiniflow/ragflow/issues/4214) 。
+詳見 [RAGFlow Roadmap 2026](https://github.com/infiniflow/ragflow/issues/12241) 。
 
 ## 🏄 開源社群
 

README_zh.md

@@ -22,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.22.1">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.23.1">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -37,7 +37,7 @@
 
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Document</a> |
-  <a href="https://github.com/infiniflow/ragflow/issues/4214">Roadmap</a> |
+  <a href="https://github.com/infiniflow/ragflow/issues/12241">Roadmap</a> |
   <a href="https://twitter.com/infiniflowai">Twitter</a> |
   <a href="https://discord.gg/NjYzJD3GM3">Discord</a> |
   <a href="https://demo.ragflow.io">Demo</a>
@@ -72,7 +72,7 @@
 
 ## 💡 RAGFlow 是什么？
 
-[RAGFlow](https://ragflow.io/) 是一款领先的开源检索增强生成（RAG）引擎，通过融合前沿的 RAG 技术与 Agent 能力，为大型语言模型提供卓越的上下文层。它提供可适配任意规模企业的端到端 RAG 工作流，凭借融合式上下文引擎与预置的 Agent 模板，助力开发者以极致效率与精度将复杂数据转化为高可信、生产级的人工智能系统。
+[RAGFlow](https://ragflow.io/) 是一款领先的开源检索增强生成（[RAG](https://ragflow.io/basics/what-is-rag)）引擎，通过融合前沿的 RAG 技术与 Agent 能力，为大型语言模型提供卓越的上下文层。它提供可适配任意规模企业的端到端 RAG 工作流，凭借融合式[上下文引擎](https://ragflow.io/basics/what-is-agent-context-engine)与预置的 Agent 模板，助力开发者以极致效率与精度将复杂数据转化为高可信、生产级的人工智能系统。
 
 ## 🎮 Demo 试用
 
@@ -85,7 +85,8 @@
 
 ## 🔥 近期更新
 
-- 2025-11-19 支持 Gemini 3 Pro.
+- 2025-12-26 支持AI代理的“记忆”功能。
+- 2025-11-19 支持 Gemini 3 Pro。
 - 2025-11-12 支持从 Confluence、S3、Notion、Discord、Google Drive 进行数据同步。
 - 2025-10-23 支持 MinerU 和 Docling 作为文档解析方法。
 - 2025-10-15 支持可编排的数据管道。
@@ -93,7 +94,7 @@
 - 2025-08-01 支持 agentic workflow 和 MCP。
 - 2025-05-23 Agent 新增 Python/JS 代码执行器组件。
 - 2025-05-05 支持跨语言查询。
-- 2025-03-19 PDF 和 DOCX 中的图支持用多模态大模型去解析得到描述.
+- 2025-03-19 PDF 和 DOCX 中的图支持用多模态大模型去解析得到描述。
 - 2024-12-18 升级了 DeepDoc 的文档布局分析模型。
 - 2024-08-22 支持用 RAG 技术实现从自然语言到 SQL 语句的转换。
 
@@ -187,12 +188,12 @@
 > 请注意，目前官方提供的所有 Docker 镜像均基于 x86 架构构建，并不提供基于 ARM64 的 Docker 镜像。
 > 如果你的操作系统是 ARM64 架构，请参考[这篇文档](https://ragflow.io/docs/dev/build_docker_image)自行构建 Docker 镜像。
 
-   > 运行以下命令会自动下载 RAGFlow Docker 镜像 `v0.22.1`。请参考下表查看不同 Docker 发行版的描述。如需下载不同于 `v0.22.1` 的 Docker 镜像，请在运行 `docker compose` 启动服务之前先更新 **docker/.env** 文件内的 `RAGFLOW_IMAGE` 变量。
+   > 运行以下命令会自动下载 RAGFlow Docker 镜像 `v0.23.1`。请参考下表查看不同 Docker 发行版的描述。如需下载不同于 `v0.23.1` 的 Docker 镜像，请在运行 `docker compose` 启动服务之前先更新 **docker/.env** 文件内的 `RAGFLOW_IMAGE` 变量。
 
    ```bash
    $ cd ragflow/docker
-   
-   # git checkout v0.22.1
+
+   # git checkout v0.23.1
    # 可选：使用稳定版本标签（查看发布：https://github.com/infiniflow/ragflow/releases）
    # 这一步确保代码中的 entrypoint.sh 文件与 Docker 镜像的版本保持一致。
 
@@ -203,13 +204,13 @@
    # sed -i '1i DEVICE=gpu' .env
    # docker compose -f docker-compose.yml up -d
    ```
-   
+
    > 注意：在 `v0.22.0` 之前的版本，我们会同时提供包含 embedding 模型的镜像和不含 embedding 模型的 slim 镜像。具体如下：
 
-   | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-   | ----------------- | --------------- | --------------------- | ------------------------ |
-   | v0.21.1           | &approx;9       | ✔️                    | Stable release           |
-   | v0.21.1-slim      | &approx;2       | ❌                    | Stable release           |
+   | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?        |
+   |-------------------|-----------------|-----------------------|----------------|
+   | v0.21.1           | &approx;9       | ✔️                    | Stable release |
+   | v0.21.1-slim      | &approx;2       | ❌                     | Stable release |
 
    > 从 `v0.22.0` 开始，我们只发布 slim 版本，并且不再在镜像标签后附加 **-slim** 后缀。
 
@@ -237,7 +238,7 @@
     * Running on all addresses (0.0.0.0)
    ```
 
-   > 如果您在没有看到上面的提示信息出来之前，就尝试登录 RAGFlow，你的浏览器有可能会提示 `network anormal` 或 `网络异常`。
+   > 如果您在没有看到上面的提示信息出来之前，就尝试登录 RAGFlow，你的浏览器有可能会提示 `network abnormal` 或 `网络异常`。
 
 5. 在你的浏览器中输入你的服务器对应的 IP 地址并登录 RAGFlow。
    > 上面这个例子中，您只需输入 http://IP_OF_YOUR_MACHINE 即可：未改动过配置则无需输入端口（默认的 HTTP 服务端口 80）。
@@ -301,6 +302,15 @@ cd ragflow/
 docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly .
 ```
 
+如果您处在代理环境下，可以传递代理参数：
+
+```bash
+docker build --platform linux/amd64 \
+  --build-arg http_proxy=http://YOUR_PROXY:PORT \
+  --build-arg https_proxy=http://YOUR_PROXY:PORT \
+  -f Dockerfile -t infiniflow/ragflow:nightly .
+```
+
 ## 🔨 以源代码启动服务
 
 1. 安装 `uv` 和 `pre-commit`。如已经安装，可跳过本步骤：
@@ -315,7 +325,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 # install RAGFlow dependent python modules
+   uv sync --python 3.12 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
@@ -392,7 +402,7 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
 
 ## 📜 路线图
 
-详见 [RAGFlow Roadmap 2025](https://github.com/infiniflow/ragflow/issues/4214) 。
+详见 [RAGFlow Roadmap 2026](https://github.com/infiniflow/ragflow/issues/12241) 。
 
 ## 🏄 开源社区
 

SECURITY.md

@@ -6,8 +6,8 @@ Use this section to tell people about which versions of your project are
 currently being supported with security updates.
 
 | Version | Supported          |
-| ------- | ------------------ |
-| <=0.7.0   | :white_check_mark: |
+|---------|--------------------|
+| <=0.7.0 | :white_check_mark: |
 
 ## Reporting a Vulnerability
 

admin/build_cli_release.sh

@@ -21,7 +21,7 @@ cp pyproject.toml release/$PROJECT_NAME/pyproject.toml
 cp README.md release/$PROJECT_NAME/README.md
 
 mkdir release/$PROJECT_NAME/$SOURCE_DIR/$PACKAGE_DIR -p
-cp admin_client.py release/$PROJECT_NAME/$SOURCE_DIR/$PACKAGE_DIR/admin_client.py
+cp ragflow_cli.py release/$PROJECT_NAME/$SOURCE_DIR/$PACKAGE_DIR/ragflow_cli.py
 
 if [ -d "release/$PROJECT_NAME/$SOURCE_DIR" ]; then
     echo "✅ source dir: release/$PROJECT_NAME/$SOURCE_DIR"

admin/client/README.md

@@ -48,7 +48,7 @@ It consists of a server-side Service and a command-line client (CLI), both imple
 1.  Ensure the Admin Service is running.
 2.  Install ragflow-cli.
     ```bash
-    pip install ragflow-cli==0.22.1
+    pip install ragflow-cli==0.23.1
     ```
 3.  Launch the CLI client:
     ```bash

admin/client/admin_client.py

@@ -1,978 +0,0 @@
-#
-#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-
-import argparse
-import base64
-from cmd import Cmd
-
-from Cryptodome.PublicKey import RSA
-from Cryptodome.Cipher import PKCS1_v1_5 as Cipher_pkcs1_v1_5
-from typing import Dict, List, Any
-from lark import Lark, Transformer, Tree
-import requests
-import getpass
-
-GRAMMAR = r"""
-start: command
-
-command: sql_command | meta_command
-
-sql_command: list_services
-           | show_service
-           | startup_service
-           | shutdown_service
-           | restart_service
-           | list_users
-           | show_user
-           | drop_user
-           | alter_user
-           | create_user
-           | activate_user
-           | list_datasets
-           | list_agents
-           | create_role
-           | drop_role
-           | alter_role
-           | list_roles
-           | show_role
-           | grant_permission
-           | revoke_permission
-           | alter_user_role
-           | show_user_permission
-           | show_version
-
-// meta command definition
-meta_command: "\\" meta_command_name [meta_args]
-
-meta_command_name: /[a-zA-Z?]+/
-meta_args: (meta_arg)+
-
-meta_arg: /[^\\s"']+/ | quoted_string
-
-// command definition
-
-LIST: "LIST"i
-SERVICES: "SERVICES"i
-SHOW: "SHOW"i
-CREATE: "CREATE"i
-SERVICE: "SERVICE"i
-SHUTDOWN: "SHUTDOWN"i
-STARTUP: "STARTUP"i
-RESTART: "RESTART"i
-USERS: "USERS"i
-DROP: "DROP"i
-USER: "USER"i
-ALTER: "ALTER"i
-ACTIVE: "ACTIVE"i
-PASSWORD: "PASSWORD"i
-DATASETS: "DATASETS"i
-OF: "OF"i
-AGENTS: "AGENTS"i
-ROLE: "ROLE"i
-ROLES: "ROLES"i
-DESCRIPTION: "DESCRIPTION"i
-GRANT: "GRANT"i
-REVOKE: "REVOKE"i
-ALL: "ALL"i
-PERMISSION: "PERMISSION"i
-TO: "TO"i
-FROM: "FROM"i
-FOR: "FOR"i
-RESOURCES: "RESOURCES"i
-ON: "ON"i
-SET: "SET"i
-VERSION: "VERSION"i
-
-list_services: LIST SERVICES ";"
-show_service: SHOW SERVICE NUMBER ";"
-startup_service: STARTUP SERVICE NUMBER ";"
-shutdown_service: SHUTDOWN SERVICE NUMBER ";"
-restart_service: RESTART SERVICE NUMBER ";"
-
-list_users: LIST USERS ";"
-drop_user: DROP USER quoted_string ";"
-alter_user: ALTER USER PASSWORD quoted_string quoted_string ";"
-show_user: SHOW USER quoted_string ";"
-create_user: CREATE USER quoted_string quoted_string ";"
-activate_user: ALTER USER ACTIVE quoted_string status ";"
-
-list_datasets: LIST DATASETS OF quoted_string ";"
-list_agents: LIST AGENTS OF quoted_string ";"
-
-create_role: CREATE ROLE identifier [DESCRIPTION quoted_string] ";"
-drop_role: DROP ROLE identifier ";"
-alter_role: ALTER ROLE identifier SET DESCRIPTION quoted_string ";"
-list_roles: LIST ROLES ";"
-show_role: SHOW ROLE identifier ";"
-
-grant_permission: GRANT action_list ON identifier TO ROLE identifier ";"
-revoke_permission: REVOKE action_list ON identifier FROM ROLE identifier ";"
-alter_user_role: ALTER USER quoted_string SET ROLE identifier ";"
-show_user_permission: SHOW USER PERMISSION quoted_string ";"
-
-show_version: SHOW VERSION ";"
-
-action_list: identifier ("," identifier)*
-
-identifier: WORD
-quoted_string: QUOTED_STRING
-status: WORD
-
-QUOTED_STRING: /'[^']+'/ | /"[^"]+"/
-WORD: /[a-zA-Z0-9_\-\.]+/
-NUMBER: /[0-9]+/
-
-%import common.WS
-%ignore WS
-"""
-
-
-class AdminTransformer(Transformer):
-
-    def start(self, items):
-        return items[0]
-
-    def command(self, items):
-        return items[0]
-
-    def list_services(self, items):
-        result = {'type': 'list_services'}
-        return result
-
-    def show_service(self, items):
-        service_id = int(items[2])
-        return {"type": "show_service", "number": service_id}
-
-    def startup_service(self, items):
-        service_id = int(items[2])
-        return {"type": "startup_service", "number": service_id}
-
-    def shutdown_service(self, items):
-        service_id = int(items[2])
-        return {"type": "shutdown_service", "number": service_id}
-
-    def restart_service(self, items):
-        service_id = int(items[2])
-        return {"type": "restart_service", "number": service_id}
-
-    def list_users(self, items):
-        return {"type": "list_users"}
-
-    def show_user(self, items):
-        user_name = items[2]
-        return {"type": "show_user", "user_name": user_name}
-
-    def drop_user(self, items):
-        user_name = items[2]
-        return {"type": "drop_user", "user_name": user_name}
-
-    def alter_user(self, items):
-        user_name = items[3]
-        new_password = items[4]
-        return {"type": "alter_user", "user_name": user_name, "password": new_password}
-
-    def create_user(self, items):
-        user_name = items[2]
-        password = items[3]
-        return {"type": "create_user", "user_name": user_name, "password": password, "role": "user"}
-
-    def activate_user(self, items):
-        user_name = items[3]
-        activate_status = items[4]
-        return {"type": "activate_user", "activate_status": activate_status, "user_name": user_name}
-
-    def list_datasets(self, items):
-        user_name = items[3]
-        return {"type": "list_datasets", "user_name": user_name}
-
-    def list_agents(self, items):
-        user_name = items[3]
-        return {"type": "list_agents", "user_name": user_name}
-
-    def create_role(self, items):
-        role_name = items[2]
-        if len(items) > 4:
-            description = items[4]
-            return {"type": "create_role", "role_name": role_name, "description": description}
-        else:
-            return {"type": "create_role", "role_name": role_name}
-
-    def drop_role(self, items):
-        role_name = items[2]
-        return {"type": "drop_role", "role_name": role_name}
-
-    def alter_role(self, items):
-        role_name = items[2]
-        description = items[5]
-        return {"type": "alter_role", "role_name": role_name, "description": description}
-
-    def list_roles(self, items):
-        return {"type": "list_roles"}
-
-    def show_role(self, items):
-        role_name = items[2]
-        return {"type": "show_role", "role_name": role_name}
-
-    def grant_permission(self, items):
-        action_list = items[1]
-        resource = items[3]
-        role_name = items[6]
-        return {"type": "grant_permission", "role_name": role_name, "resource": resource, "actions": action_list}
-
-    def revoke_permission(self, items):
-        action_list = items[1]
-        resource = items[3]
-        role_name = items[6]
-        return {
-            "type": "revoke_permission",
-            "role_name": role_name,
-            "resource": resource, "actions": action_list
-        }
-
-    def alter_user_role(self, items):
-        user_name = items[2]
-        role_name = items[5]
-        return {"type": "alter_user_role", "user_name": user_name, "role_name": role_name}
-
-    def show_user_permission(self, items):
-        user_name = items[3]
-        return {"type": "show_user_permission", "user_name": user_name}
-
-    def show_version(self, items):
-        return {"type": "show_version"}
-
-    def action_list(self, items):
-        return items
-
-    def meta_command(self, items):
-        command_name = str(items[0]).lower()
-        args = items[1:] if len(items) > 1 else []
-
-        # handle quoted parameter
-        parsed_args = []
-        for arg in args:
-            if hasattr(arg, 'value'):
-                parsed_args.append(arg.value)
-            else:
-                parsed_args.append(str(arg))
-
-        return {'type': 'meta', 'command': command_name, 'args': parsed_args}
-
-    def meta_command_name(self, items):
-        return items[0]
-
-    def meta_args(self, items):
-        return items
-
-
-def encrypt(input_string):
-    pub = '-----BEGIN PUBLIC KEY-----\nMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEArq9XTUSeYr2+N1h3Afl/z8Dse/2yD0ZGrKwx+EEEcdsBLca9Ynmx3nIB5obmLlSfmskLpBo0UACBmB5rEjBp2Q2f3AG3Hjd4B+gNCG6BDaawuDlgANIhGnaTLrIqWrrcm4EMzJOnAOI1fgzJRsOOUEfaS318Eq9OVO3apEyCCt0lOQK6PuksduOjVxtltDav+guVAA068NrPYmRNabVKRNLJpL8w4D44sfth5RvZ3q9t+6RTArpEtc5sh5ChzvqPOzKGMXW83C95TxmXqpbK6olN4RevSfVjEAgCydH6HN6OhtOQEcnrU97r9H0iZOWwbw3pVrZiUkuRD1R56Wzs2wIDAQAB\n-----END PUBLIC KEY-----'
-    pub_key = RSA.importKey(pub)
-    cipher = Cipher_pkcs1_v1_5.new(pub_key)
-    cipher_text = cipher.encrypt(base64.b64encode(input_string.encode('utf-8')))
-    return base64.b64encode(cipher_text).decode("utf-8")
-
-
-def encode_to_base64(input_string):
-    base64_encoded = base64.b64encode(input_string.encode('utf-8'))
-    return base64_encoded.decode('utf-8')
-
-
-class AdminCLI(Cmd):
-    def __init__(self):
-        super().__init__()
-        self.parser = Lark(GRAMMAR, start='start', parser='lalr', transformer=AdminTransformer())
-        self.command_history = []
-        self.is_interactive = False
-        self.admin_account = "admin@ragflow.io"
-        self.admin_password: str = "admin"
-        self.session = requests.Session()
-        self.access_token: str = ""
-        self.host: str = ""
-        self.port: int = 0
-
-    intro = r"""Type "\h" for help."""
-    prompt = "admin> "
-
-    def onecmd(self, command: str) -> bool:
-        try:
-            result = self.parse_command(command)
-
-            if isinstance(result, dict):
-                if 'type' in result and result.get('type') == 'empty':
-                    return False
-
-            self.execute_command(result)
-
-            if isinstance(result, Tree):
-                return False
-
-            if result.get('type') == 'meta' and result.get('command') in ['q', 'quit', 'exit']:
-                return True
-
-        except KeyboardInterrupt:
-            print("\nUse '\\q' to quit")
-        except EOFError:
-            print("\nGoodbye!")
-            return True
-        return False
-
-    def emptyline(self) -> bool:
-        return False
-
-    def default(self, line: str) -> bool:
-        return self.onecmd(line)
-
-    def parse_command(self, command_str: str) -> dict[str, str]:
-        if not command_str.strip():
-            return {'type': 'empty'}
-
-        self.command_history.append(command_str)
-
-        try:
-            result = self.parser.parse(command_str)
-            return result
-        except Exception as e:
-            return {'type': 'error', 'message': f'Parse error: {str(e)}'}
-
-    def verify_admin(self, arguments: dict, single_command: bool):
-        self.host = arguments['host']
-        self.port = arguments['port']
-        print(f"Attempt to access ip: {self.host}, port: {self.port}")
-        url = f"http://{self.host}:{self.port}/api/v1/admin/login"
-
-        attempt_count = 3
-        if single_command:
-            attempt_count = 1
-
-        try_count = 0
-        while True:
-            try_count += 1
-            if try_count > attempt_count:
-                return False
-
-            if single_command:
-                admin_passwd = arguments['password']
-            else:
-                admin_passwd = getpass.getpass(f"password for {self.admin_account}: ").strip()
-            try:
-                self.admin_password = encrypt(admin_passwd)
-                response = self.session.post(url, json={'email': self.admin_account, 'password': self.admin_password})
-                if response.status_code == 200:
-                    res_json = response.json()
-                    error_code = res_json.get('code', -1)
-                    if error_code == 0:
-                        self.session.headers.update({
-                            'Content-Type': 'application/json',
-                            'Authorization': response.headers['Authorization'],
-                            'User-Agent': 'RAGFlow-CLI/0.22.1'
-                        })
-                        print("Authentication successful.")
-                        return True
-                    else:
-                        error_message = res_json.get('message', 'Unknown error')
-                        print(f"Authentication failed: {error_message}, try again")
-                        continue
-                else:
-                    print(f"Bad response，status: {response.status_code}, password is wrong")
-            except Exception as e:
-                print(str(e))
-                print(f"Can't access {self.host}, port: {self.port}")
-
-    def _format_service_detail_table(self, data):
-        if isinstance(data, list):
-            return data
-        if not all([isinstance(v, list) for v in data.values()]):
-            # normal table
-            return data
-        # handle task_executor heartbeats map, for example {'name': [{'done': 2, 'now': timestamp1}, {'done': 3, 'now': timestamp2}]
-        task_executor_list = []
-        for k, v in data.items():
-            # display latest status
-            heartbeats = sorted(v, key=lambda x: x["now"], reverse=True)
-            task_executor_list.append({
-                "task_executor_name": k,
-                **heartbeats[0],
-            } if heartbeats else {"task_executor_name": k})
-        return task_executor_list
-
-    def _print_table_simple(self, data):
-        if not data:
-            print("No data to print")
-            return
-        if isinstance(data, dict):
-            # handle single row data
-            data = [data]
-
-        columns = list(set().union(*(d.keys() for d in data)))
-        columns.sort()
-        col_widths = {}
-
-        def get_string_width(text):
-            half_width_chars = (
-                " !\"#$%&'()*+,-./0123456789:;<=>?@"
-                "ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`"
-                "abcdefghijklmnopqrstuvwxyz{|}~"
-                "\t\n\r"
-            )
-            width = 0
-            for char in text:
-                if char in half_width_chars:
-                    width += 1
-                else:
-                    width += 2
-            return width
-
-        for col in columns:
-            max_width = get_string_width(str(col))
-            for item in data:
-                value_len = get_string_width(str(item.get(col, '')))
-                if value_len > max_width:
-                    max_width = value_len
-            col_widths[col] = max(2, max_width)
-
-        # Generate delimiter
-        separator = "+" + "+".join(["-" * (col_widths[col] + 2) for col in columns]) + "+"
-
-        # Print header
-        print(separator)
-        header = "|" + "|".join([f" {col:<{col_widths[col]}} " for col in columns]) + "|"
-        print(header)
-        print(separator)
-
-        # Print data
-        for item in data:
-            row = "|"
-            for col in columns:
-                value = str(item.get(col, ''))
-                if get_string_width(value) > col_widths[col]:
-                    value = value[:col_widths[col] - 3] + "..."
-                row += f" {value:<{col_widths[col] - (get_string_width(value) - len(value))}} |"
-            print(row)
-
-        print(separator)
-
-    def run_interactive(self):
-
-        self.is_interactive = True
-        print("RAGFlow Admin command line interface - Type '\\?' for help, '\\q' to quit")
-
-        while True:
-            try:
-                command = input("admin> ").strip()
-                if not command:
-                    continue
-
-                print(f"command: {command}")
-                result = self.parse_command(command)
-                self.execute_command(result)
-
-                if isinstance(result, Tree):
-                    continue
-
-                if result.get('type') == 'meta' and result.get('command') in ['q', 'quit', 'exit']:
-                    break
-
-            except KeyboardInterrupt:
-                print("\nUse '\\q' to quit")
-            except EOFError:
-                print("\nGoodbye!")
-                break
-
-    def run_single_command(self, command: str):
-        result = self.parse_command(command)
-        self.execute_command(result)
-
-    def parse_connection_args(self, args: List[str]) -> Dict[str, Any]:
-        parser = argparse.ArgumentParser(description='Admin CLI Client', add_help=False)
-        parser.add_argument('-h', '--host', default='localhost', help='Admin service host')
-        parser.add_argument('-p', '--port', type=int, default=9381, help='Admin service port')
-        parser.add_argument('-w', '--password', default='admin', type=str, help='Superuser password')
-        parser.add_argument('command', nargs='?', help='Single command')
-        try:
-            parsed_args, remaining_args = parser.parse_known_args(args)
-            if remaining_args:
-                command = remaining_args[0]
-                return {
-                    'host': parsed_args.host,
-                    'port': parsed_args.port,
-                    'password': parsed_args.password,
-                    'command': command
-                }
-            else:
-                return {
-                    'host': parsed_args.host,
-                    'port': parsed_args.port,
-                }
-        except SystemExit:
-            return {'error': 'Invalid connection arguments'}
-
-    def execute_command(self, parsed_command: Dict[str, Any]):
-
-        command_dict: dict
-        if isinstance(parsed_command, Tree):
-            command_dict = parsed_command.children[0]
-        else:
-            if parsed_command['type'] == 'error':
-                print(f"Error: {parsed_command['message']}")
-                return
-            else:
-                command_dict = parsed_command
-
-        # print(f"Parsed command: {command_dict}")
-
-        command_type = command_dict['type']
-
-        match command_type:
-            case 'list_services':
-                self._handle_list_services(command_dict)
-            case 'show_service':
-                self._handle_show_service(command_dict)
-            case 'restart_service':
-                self._handle_restart_service(command_dict)
-            case 'shutdown_service':
-                self._handle_shutdown_service(command_dict)
-            case 'startup_service':
-                self._handle_startup_service(command_dict)
-            case 'list_users':
-                self._handle_list_users(command_dict)
-            case 'show_user':
-                self._handle_show_user(command_dict)
-            case 'drop_user':
-                self._handle_drop_user(command_dict)
-            case 'alter_user':
-                self._handle_alter_user(command_dict)
-            case 'create_user':
-                self._handle_create_user(command_dict)
-            case 'activate_user':
-                self._handle_activate_user(command_dict)
-            case 'list_datasets':
-                self._handle_list_datasets(command_dict)
-            case 'list_agents':
-                self._handle_list_agents(command_dict)
-            case 'create_role':
-                self._create_role(command_dict)
-            case 'drop_role':
-                self._drop_role(command_dict)
-            case 'alter_role':
-                self._alter_role(command_dict)
-            case 'list_roles':
-                self._list_roles(command_dict)
-            case 'show_role':
-                self._show_role(command_dict)
-            case 'grant_permission':
-                self._grant_permission(command_dict)
-            case 'revoke_permission':
-                self._revoke_permission(command_dict)
-            case 'alter_user_role':
-                self._alter_user_role(command_dict)
-            case 'show_user_permission':
-                self._show_user_permission(command_dict)
-            case 'show_version':
-                self._show_version(command_dict)
-            case 'meta':
-                self._handle_meta_command(command_dict)
-            case _:
-                print(f"Command '{command_type}' would be executed with API")
-
-    def _handle_list_services(self, command):
-        print("Listing all services")
-
-        url = f'http://{self.host}:{self.port}/api/v1/admin/services'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(f"Fail to get all services, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _handle_show_service(self, command):
-        service_id: int = command['number']
-        print(f"Showing service: {service_id}")
-
-        url = f'http://{self.host}:{self.port}/api/v1/admin/services/{service_id}'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            res_data = res_json['data']
-            if 'status' in res_data and res_data['status'] == 'alive':
-                print(f"Service {res_data['service_name']} is alive, ")
-                if isinstance(res_data['message'], str):
-                    print(res_data['message'])
-                else:
-                    data = self._format_service_detail_table(res_data['message'])
-                    self._print_table_simple(data)
-            else:
-                print(f"Service {res_data['service_name']} is down, {res_data['message']}")
-        else:
-            print(f"Fail to show service, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _handle_restart_service(self, command):
-        service_id: int = command['number']
-        print(f"Restart service {service_id}")
-
-    def _handle_shutdown_service(self, command):
-        service_id: int = command['number']
-        print(f"Shutdown service {service_id}")
-
-    def _handle_startup_service(self, command):
-        service_id: int = command['number']
-        print(f"Startup service {service_id}")
-
-    def _handle_list_users(self, command):
-        print("Listing all users")
-
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(f"Fail to get all users, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _handle_show_user(self, command):
-        username_tree: Tree = command['user_name']
-        user_name: str = username_tree.children[0].strip("'\"")
-        print(f"Showing user: {user_name}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{user_name}'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            table_data = res_json['data']
-            table_data.pop('avatar')
-            self._print_table_simple(table_data)
-        else:
-            print(f"Fail to get user {user_name}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _handle_drop_user(self, command):
-        username_tree: Tree = command['user_name']
-        user_name: str = username_tree.children[0].strip("'\"")
-        print(f"Drop user: {user_name}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{user_name}'
-        response = self.session.delete(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            print(res_json["message"])
-        else:
-            print(f"Fail to drop user, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _handle_alter_user(self, command):
-        user_name_tree: Tree = command['user_name']
-        user_name: str = user_name_tree.children[0].strip("'\"")
-        password_tree: Tree = command['password']
-        password: str = password_tree.children[0].strip("'\"")
-        print(f"Alter user: {user_name}, password: {password}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{user_name}/password'
-        response = self.session.put(url, json={'new_password': encrypt(password)})
-        res_json = response.json()
-        if response.status_code == 200:
-            print(res_json["message"])
-        else:
-            print(f"Fail to alter password, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _handle_create_user(self, command):
-        user_name_tree: Tree = command['user_name']
-        user_name: str = user_name_tree.children[0].strip("'\"")
-        password_tree: Tree = command['password']
-        password: str = password_tree.children[0].strip("'\"")
-        role: str = command['role']
-        print(f"Create user: {user_name}, password: {password}, role: {role}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users'
-        response = self.session.post(
-            url,
-            json={'user_name': user_name, 'password': encrypt(password), 'role': role}
-        )
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(f"Fail to create user {user_name}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _handle_activate_user(self, command):
-        user_name_tree: Tree = command['user_name']
-        user_name: str = user_name_tree.children[0].strip("'\"")
-        activate_tree: Tree = command['activate_status']
-        activate_status: str = activate_tree.children[0].strip("'\"")
-        if activate_status.lower() in ['on', 'off']:
-            print(f"Alter user {user_name} activate status, turn {activate_status.lower()}.")
-            url = f'http://{self.host}:{self.port}/api/v1/admin/users/{user_name}/activate'
-            response = self.session.put(url, json={'activate_status': activate_status})
-            res_json = response.json()
-            if response.status_code == 200:
-                print(res_json["message"])
-            else:
-                print(f"Fail to alter activate status, code: {res_json['code']}, message: {res_json['message']}")
-        else:
-            print(f"Unknown activate status: {activate_status}.")
-
-    def _handle_list_datasets(self, command):
-        username_tree: Tree = command['user_name']
-        user_name: str = username_tree.children[0].strip("'\"")
-        print(f"Listing all datasets of user: {user_name}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{user_name}/datasets'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            table_data = res_json['data']
-            for t in table_data:
-                t.pop('avatar')
-            self._print_table_simple(table_data)
-        else:
-            print(f"Fail to get all datasets of {user_name}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _handle_list_agents(self, command):
-        username_tree: Tree = command['user_name']
-        user_name: str = username_tree.children[0].strip("'\"")
-        print(f"Listing all agents of user: {user_name}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{user_name}/agents'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            table_data = res_json['data']
-            for t in table_data:
-                t.pop('avatar')
-            self._print_table_simple(table_data)
-        else:
-            print(f"Fail to get all agents of {user_name}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _create_role(self, command):
-        role_name_tree: Tree = command['role_name']
-        role_name: str = role_name_tree.children[0].strip("'\"")
-        desc_str: str = ''
-        if 'description' in command:
-            desc_tree: Tree = command['description']
-            desc_str = desc_tree.children[0].strip("'\"")
-
-        print(f"create role name: {role_name}, description: {desc_str}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/roles'
-        response = self.session.post(
-            url,
-            json={'role_name': role_name, 'description': desc_str}
-        )
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(f"Fail to create role {role_name}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _drop_role(self, command):
-        role_name_tree: Tree = command['role_name']
-        role_name: str = role_name_tree.children[0].strip("'\"")
-        print(f"drop role name: {role_name}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/roles/{role_name}'
-        response = self.session.delete(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(f"Fail to drop role {role_name}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _alter_role(self, command):
-        role_name_tree: Tree = command['role_name']
-        role_name: str = role_name_tree.children[0].strip("'\"")
-        desc_tree: Tree = command['description']
-        desc_str: str = desc_tree.children[0].strip("'\"")
-
-        print(f"alter role name: {role_name}, description: {desc_str}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/roles/{role_name}'
-        response = self.session.put(
-            url,
-            json={'description': desc_str}
-        )
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(
-                f"Fail to update role {role_name} with description: {desc_str}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _list_roles(self, command):
-        print("Listing all roles")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/roles'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(f"Fail to list roles, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _show_role(self, command):
-        role_name_tree: Tree = command['role_name']
-        role_name: str = role_name_tree.children[0].strip("'\"")
-        print(f"show role: {role_name}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/roles/{role_name}/permission'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(f"Fail to list roles, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _grant_permission(self, command):
-        role_name_tree: Tree = command['role_name']
-        role_name_str: str = role_name_tree.children[0].strip("'\"")
-        resource_tree: Tree = command['resource']
-        resource_str: str = resource_tree.children[0].strip("'\"")
-        action_tree_list: list = command['actions']
-        actions: list = []
-        for action_tree in action_tree_list:
-            action_str: str = action_tree.children[0].strip("'\"")
-            actions.append(action_str)
-        print(f"grant role_name: {role_name_str}, resource: {resource_str}, actions: {actions}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/roles/{role_name_str}/permission'
-        response = self.session.post(
-            url,
-            json={'actions': actions, 'resource': resource_str}
-        )
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(
-                f"Fail to grant role {role_name_str} with {actions} on {resource_str}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _revoke_permission(self, command):
-        role_name_tree: Tree = command['role_name']
-        role_name_str: str = role_name_tree.children[0].strip("'\"")
-        resource_tree: Tree = command['resource']
-        resource_str: str = resource_tree.children[0].strip("'\"")
-        action_tree_list: list = command['actions']
-        actions: list = []
-        for action_tree in action_tree_list:
-            action_str: str = action_tree.children[0].strip("'\"")
-            actions.append(action_str)
-        print(f"revoke role_name: {role_name_str}, resource: {resource_str}, actions: {actions}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/roles/{role_name_str}/permission'
-        response = self.session.delete(
-            url,
-            json={'actions': actions, 'resource': resource_str}
-        )
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(
-                f"Fail to revoke role {role_name_str} with {actions} on {resource_str}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _alter_user_role(self, command):
-        role_name_tree: Tree = command['role_name']
-        role_name_str: str = role_name_tree.children[0].strip("'\"")
-        user_name_tree: Tree = command['user_name']
-        user_name_str: str = user_name_tree.children[0].strip("'\"")
-        print(f"alter_user_role user_name: {user_name_str}, role_name: {role_name_str}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{user_name_str}/role'
-        response = self.session.put(
-            url,
-            json={'role_name': role_name_str}
-        )
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(
-                f"Fail to alter user: {user_name_str} to role {role_name_str}, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _show_user_permission(self, command):
-        user_name_tree: Tree = command['user_name']
-        user_name_str: str = user_name_tree.children[0].strip("'\"")
-        print(f"show_user_permission user_name: {user_name_str}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{user_name_str}/permission'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(
-                f"Fail to show user: {user_name_str} permission, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _show_version(self, command):
-        print("show_version")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/version'
-        response = self.session.get(url)
-        res_json = response.json()
-        if response.status_code == 200:
-            self._print_table_simple(res_json['data'])
-        else:
-            print(f"Fail to show version, code: {res_json['code']}, message: {res_json['message']}")
-
-    def _handle_meta_command(self, command):
-        meta_command = command['command']
-        args = command.get('args', [])
-
-        if meta_command in ['?', 'h', 'help']:
-            self.show_help()
-        elif meta_command in ['q', 'quit', 'exit']:
-            print("Goodbye!")
-        else:
-            print(f"Meta command '{meta_command}' with args {args}")
-
-    def show_help(self):
-        """Help info"""
-        help_text = """
-Commands:
-  LIST SERVICES
-  SHOW SERVICE <service>
-  STARTUP SERVICE <service>
-  SHUTDOWN SERVICE <service>
-  RESTART SERVICE <service>
-  LIST USERS
-  SHOW USER <user>
-  DROP USER <user>
-  CREATE USER <user> <password>
-  ALTER USER PASSWORD <user> <new_password>
-  ALTER USER ACTIVE <user> <on/off>
-  LIST DATASETS OF <user>
-  LIST AGENTS OF <user>
-
-Meta Commands:
-  \\?, \\h, \\help     Show this help
-  \\q, \\quit, \\exit   Quit the CLI
-        """
-        print(help_text)
-
-
-def main():
-    import sys
-
-    cli = AdminCLI()
-
-    args = cli.parse_connection_args(sys.argv)
-    if 'error' in args:
-        print(f"Error: {args['error']}")
-        return
-
-    if 'command' in args:
-        if 'password' not in args:
-            print("Error: password is missing")
-            return
-        if cli.verify_admin(args, single_command=True):
-            command: str = args['command']
-            print(f"Run single command: {command}")
-            cli.run_single_command(command)
-    else:
-        if cli.verify_admin(args, single_command=False):
-            print(r"""
-                ____  ___   ______________                 ___       __          _     
-               / __ \/   | / ____/ ____/ /___ _      __   /   | ____/ /___ ___  (_)___ 
-              / /_/ / /| |/ / __/ /_  / / __ \ | /| / /  / /| |/ __  / __ `__ \/ / __ \
-             / _, _/ ___ / /_/ / __/ / / /_/ / |/ |/ /  / ___ / /_/ / / / / / / / / / /
-            /_/ |_/_/  |_\____/_/   /_/\____/|__/|__/  /_/  |_\__,_/_/ /_/ /_/_/_/ /_/ 
-            """)
-            cli.cmdloop()
-
-
-if __name__ == '__main__':
-    main()

admin/client/http_client.py

@@ -0,0 +1,166 @@
+#
+#  Copyright 2026 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+
+import time
+import json
+from typing import Any, Dict, Optional
+
+import requests
+# from requests.sessions import HTTPAdapter
+
+
+class HttpClient:
+    def __init__(
+            self,
+            host: str = "127.0.0.1",
+            port: int = 9381,
+            api_version: str = "v1",
+            api_key: Optional[str] = None,
+            connect_timeout: float = 5.0,
+            read_timeout: float = 60.0,
+            verify_ssl: bool = False,
+    ) -> None:
+        self.host = host
+        self.port = port
+        self.api_version = api_version
+        self.api_key = api_key
+        self.login_token: str | None = None
+        self.connect_timeout = connect_timeout
+        self.read_timeout = read_timeout
+        self.verify_ssl = verify_ssl
+
+    def api_base(self) -> str:
+        return f"{self.host}:{self.port}/api/{self.api_version}"
+
+    def non_api_base(self) -> str:
+        return f"{self.host}:{self.port}/{self.api_version}"
+
+    def build_url(self, path: str, use_api_base: bool = True) -> str:
+        base = self.api_base() if use_api_base else self.non_api_base()
+        if self.verify_ssl:
+            return f"https://{base}/{path.lstrip('/')}"
+        else:
+            return f"http://{base}/{path.lstrip('/')}"
+
+    def _headers(self, auth_kind: Optional[str], extra: Optional[Dict[str, str]]) -> Dict[str, str]:
+        headers = {}
+        if auth_kind == "api" and self.api_key:
+            headers["Authorization"] = f"Bearer {self.api_key}"
+        elif auth_kind == "web" and self.login_token:
+            headers["Authorization"] = self.login_token
+        elif auth_kind == "admin" and self.login_token:
+            headers["Authorization"] = self.login_token
+        else:
+            pass
+        if extra:
+            headers.update(extra)
+        return headers
+
+    def request(
+            self,
+            method: str,
+            path: str,
+            *,
+            use_api_base: bool = True,
+            auth_kind: Optional[str] = "api",
+            headers: Optional[Dict[str, str]] = None,
+            json_body: Optional[Dict[str, Any]] = None,
+            data: Any = None,
+            files: Any = None,
+            params: Optional[Dict[str, Any]] = None,
+            stream: bool = False,
+            iterations: int = 1,
+    ) -> requests.Response | dict:
+        url = self.build_url(path, use_api_base=use_api_base)
+        merged_headers = self._headers(auth_kind, headers)
+        # timeout: Tuple[float, float] = (self.connect_timeout, self.read_timeout)
+        session = requests.Session()
+        # adapter = HTTPAdapter(pool_connections=100, pool_maxsize=100)
+        # session.mount("http://", adapter)
+        if iterations > 1:
+            response_list = []
+            total_duration = 0.0
+            for _ in range(iterations):
+                start_time = time.perf_counter()
+                response = session.get(url, headers=merged_headers, json=json_body, data=data, stream=stream)
+                # response = requests.request(
+                #     method=method,
+                #     url=url,
+                #     headers=merged_headers,
+                #     json=json_body,
+                #     data=data,
+                #     files=files,
+                #     params=params,
+                #     timeout=timeout,
+                #     stream=stream,
+                #     verify=self.verify_ssl,
+                # )
+                end_time = time.perf_counter()
+                total_duration += end_time - start_time
+                response_list.append(response)
+            return {"duration": total_duration, "response_list": response_list}
+        else:
+            return session.get(url, headers=merged_headers, json=json_body, data=data, stream=stream)
+            # return requests.request(
+            #     method=method,
+            #     url=url,
+            #     headers=merged_headers,
+            #     json=json_body,
+            #     data=data,
+            #     files=files,
+            #     params=params,
+            #     timeout=timeout,
+            #     stream=stream,
+            #     verify=self.verify_ssl,
+            # )
+
+    def request_json(
+            self,
+            method: str,
+            path: str,
+            *,
+            use_api_base: bool = True,
+            auth_kind: Optional[str] = "api",
+            headers: Optional[Dict[str, str]] = None,
+            json_body: Optional[Dict[str, Any]] = None,
+            data: Any = None,
+            files: Any = None,
+            params: Optional[Dict[str, Any]] = None,
+            stream: bool = False,
+    ) -> Dict[str, Any]:
+        response = self.request(
+            method,
+            path,
+            use_api_base=use_api_base,
+            auth_kind=auth_kind,
+            headers=headers,
+            json_body=json_body,
+            data=data,
+            files=files,
+            params=params,
+            stream=stream,
+        )
+        try:
+            return response.json()
+        except Exception as exc:
+            raise ValueError(f"Non-JSON response from {path}: {exc}") from exc
+
+    @staticmethod
+    def parse_json_bytes(raw: bytes) -> Dict[str, Any]:
+        try:
+            return json.loads(raw.decode("utf-8"))
+        except Exception as exc:
+            raise ValueError(f"Invalid JSON payload: {exc}") from exc

admin/client/parser.py

@@ -0,0 +1,623 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+
+from lark import Transformer
+
+GRAMMAR = r"""
+start: command
+
+command: sql_command | meta_command
+
+sql_command: login_user
+           | ping_server
+           | list_services
+           | show_service
+           | startup_service
+           | shutdown_service
+           | restart_service
+           | register_user
+           | list_users
+           | show_user
+           | drop_user
+           | alter_user
+           | create_user
+           | activate_user
+           | list_datasets
+           | list_agents
+           | create_role
+           | drop_role
+           | alter_role
+           | list_roles
+           | show_role
+           | grant_permission
+           | revoke_permission
+           | alter_user_role
+           | show_user_permission
+           | show_version
+           | grant_admin
+           | revoke_admin
+           | set_variable
+           | show_variable
+           | list_variables
+           | list_configs
+           | list_environments
+           | generate_key
+           | list_keys
+           | drop_key
+           | show_current_user
+           | set_default_llm
+           | set_default_vlm
+           | set_default_embedding
+           | set_default_reranker
+           | set_default_asr
+           | set_default_tts
+           | reset_default_llm
+           | reset_default_vlm
+           | reset_default_embedding
+           | reset_default_reranker
+           | reset_default_asr
+           | reset_default_tts
+           | create_model_provider
+           | drop_model_provider
+           | create_user_dataset_with_parser
+           | create_user_dataset_with_pipeline
+           | drop_user_dataset
+           | list_user_datasets
+           | list_user_dataset_files
+           | list_user_agents
+           | list_user_chats
+           | create_user_chat
+           | drop_user_chat
+           | list_user_model_providers
+           | list_user_default_models
+           | parse_dataset_docs
+           | parse_dataset_sync
+           | parse_dataset_async
+           | import_docs_into_dataset
+           | search_on_datasets
+           | benchmark
+
+// meta command definition
+meta_command: "\\" meta_command_name [meta_args]
+
+meta_command_name: /[a-zA-Z?]+/
+meta_args: (meta_arg)+
+
+meta_arg: /[^\\s"']+/ | quoted_string
+
+// command definition
+
+LOGIN: "LOGIN"i
+REGISTER: "REGISTER"i
+LIST: "LIST"i
+SERVICES: "SERVICES"i
+SHOW: "SHOW"i
+CREATE: "CREATE"i
+SERVICE: "SERVICE"i
+SHUTDOWN: "SHUTDOWN"i
+STARTUP: "STARTUP"i
+RESTART: "RESTART"i
+USERS: "USERS"i
+DROP: "DROP"i
+USER: "USER"i
+ALTER: "ALTER"i
+ACTIVE: "ACTIVE"i
+ADMIN: "ADMIN"i
+PASSWORD: "PASSWORD"i
+DATASET: "DATASET"i
+DATASETS: "DATASETS"i
+OF: "OF"i
+AGENTS: "AGENTS"i
+ROLE: "ROLE"i
+ROLES: "ROLES"i
+DESCRIPTION: "DESCRIPTION"i
+GRANT: "GRANT"i
+REVOKE: "REVOKE"i
+ALL: "ALL"i
+PERMISSION: "PERMISSION"i
+TO: "TO"i
+FROM: "FROM"i
+FOR: "FOR"i
+RESOURCES: "RESOURCES"i
+ON: "ON"i
+SET: "SET"i
+RESET: "RESET"i
+VERSION: "VERSION"i
+VAR: "VAR"i
+VARS: "VARS"i
+CONFIGS: "CONFIGS"i
+ENVS: "ENVS"i
+KEY: "KEY"i
+KEYS: "KEYS"i
+GENERATE: "GENERATE"i
+MODEL: "MODEL"i
+MODELS: "MODELS"i
+PROVIDER: "PROVIDER"i
+PROVIDERS: "PROVIDERS"i
+DEFAULT: "DEFAULT"i
+CHATS: "CHATS"i
+CHAT: "CHAT"i
+FILES: "FILES"i
+AS: "AS"i
+PARSE: "PARSE"i
+IMPORT: "IMPORT"i
+INTO: "INTO"i
+WITH: "WITH"i
+PARSER: "PARSER"i
+PIPELINE: "PIPELINE"i
+SEARCH: "SEARCH"i
+CURRENT: "CURRENT"i
+LLM: "LLM"i
+VLM: "VLM"i
+EMBEDDING: "EMBEDDING"i
+RERANKER: "RERANKER"i
+ASR: "ASR"i
+TTS: "TTS"i
+ASYNC: "ASYNC"i
+SYNC: "SYNC"i
+BENCHMARK: "BENCHMARK"i
+PING: "PING"i
+
+login_user: LOGIN USER quoted_string ";"
+list_services: LIST SERVICES ";"
+show_service: SHOW SERVICE NUMBER ";"
+startup_service: STARTUP SERVICE NUMBER ";"
+shutdown_service: SHUTDOWN SERVICE NUMBER ";"
+restart_service: RESTART SERVICE NUMBER ";"
+
+register_user: REGISTER USER quoted_string AS quoted_string PASSWORD quoted_string ";"
+list_users: LIST USERS ";"
+drop_user: DROP USER quoted_string ";"
+alter_user: ALTER USER PASSWORD quoted_string quoted_string ";"
+show_user: SHOW USER quoted_string ";"
+create_user: CREATE USER quoted_string quoted_string ";"
+activate_user: ALTER USER ACTIVE quoted_string status ";"
+
+list_datasets: LIST DATASETS OF quoted_string ";"
+list_agents: LIST AGENTS OF quoted_string ";"
+
+create_role: CREATE ROLE identifier [DESCRIPTION quoted_string] ";"
+drop_role: DROP ROLE identifier ";"
+alter_role: ALTER ROLE identifier SET DESCRIPTION quoted_string ";"
+list_roles: LIST ROLES ";"
+show_role: SHOW ROLE identifier ";"
+
+grant_permission: GRANT identifier_list ON identifier TO ROLE identifier ";"
+revoke_permission: REVOKE identifier_list ON identifier FROM ROLE identifier ";"
+alter_user_role: ALTER USER quoted_string SET ROLE identifier ";"
+show_user_permission: SHOW USER PERMISSION quoted_string ";"
+
+show_version: SHOW VERSION ";"
+
+grant_admin: GRANT ADMIN quoted_string ";"
+revoke_admin: REVOKE ADMIN quoted_string ";"
+
+generate_key: GENERATE KEY FOR USER quoted_string ";"
+list_keys: LIST KEYS OF quoted_string ";"
+drop_key: DROP KEY quoted_string OF quoted_string ";"
+
+set_variable: SET VAR identifier identifier ";"
+show_variable: SHOW VAR identifier ";"
+list_variables: LIST VARS ";"
+list_configs: LIST CONFIGS ";"
+list_environments: LIST ENVS ";"
+
+benchmark: BENCHMARK NUMBER NUMBER user_statement
+
+user_statement: ping_server
+                | show_current_user
+                | create_model_provider
+                | drop_model_provider
+                | set_default_llm
+                | set_default_vlm
+                | set_default_embedding
+                | set_default_reranker
+                | set_default_asr
+                | set_default_tts
+                | reset_default_llm
+                | reset_default_vlm
+                | reset_default_embedding
+                | reset_default_reranker
+                | reset_default_asr
+                | reset_default_tts
+                | create_user_dataset_with_parser
+                | create_user_dataset_with_pipeline
+                | drop_user_dataset
+                | list_user_datasets
+                | list_user_dataset_files
+                | list_user_agents
+                | list_user_chats
+                | create_user_chat
+                | drop_user_chat
+                | list_user_model_providers
+                | list_user_default_models
+                | import_docs_into_dataset
+                | search_on_datasets
+
+ping_server: PING ";"
+show_current_user: SHOW CURRENT USER ";"
+create_model_provider: CREATE MODEL PROVIDER quoted_string quoted_string ";"
+drop_model_provider: DROP MODEL PROVIDER quoted_string ";"
+set_default_llm: SET DEFAULT LLM quoted_string ";"
+set_default_vlm: SET DEFAULT VLM quoted_string ";"
+set_default_embedding: SET DEFAULT EMBEDDING quoted_string ";"
+set_default_reranker: SET DEFAULT RERANKER quoted_string ";"
+set_default_asr: SET DEFAULT ASR quoted_string ";"
+set_default_tts: SET DEFAULT TTS quoted_string ";"
+
+reset_default_llm: RESET DEFAULT LLM ";"
+reset_default_vlm: RESET DEFAULT VLM ";"
+reset_default_embedding: RESET DEFAULT EMBEDDING ";"
+reset_default_reranker: RESET DEFAULT RERANKER ";"
+reset_default_asr: RESET DEFAULT ASR ";"
+reset_default_tts: RESET DEFAULT TTS ";"
+
+list_user_datasets: LIST DATASETS ";"
+create_user_dataset_with_parser: CREATE DATASET quoted_string WITH EMBEDDING quoted_string PARSER quoted_string ";" 
+create_user_dataset_with_pipeline: CREATE DATASET quoted_string WITH EMBEDDING quoted_string PIPELINE quoted_string ";" 
+drop_user_dataset: DROP DATASET quoted_string ";"
+list_user_dataset_files: LIST FILES OF DATASET quoted_string ";"
+list_user_agents: LIST AGENTS ";"
+list_user_chats: LIST CHATS ";"
+create_user_chat: CREATE CHAT quoted_string ";"
+drop_user_chat: DROP CHAT quoted_string ";"
+list_user_model_providers: LIST MODEL PROVIDERS ";"
+list_user_default_models: LIST DEFAULT MODELS ";"
+import_docs_into_dataset: IMPORT quoted_string INTO DATASET quoted_string ";"
+search_on_datasets: SEARCH quoted_string ON DATASETS quoted_string ";"
+
+parse_dataset_docs: PARSE quoted_string OF DATASET quoted_string ";"
+parse_dataset_sync: PARSE DATASET quoted_string SYNC ";"
+parse_dataset_async: PARSE DATASET quoted_string ASYNC ";"
+
+identifier_list: identifier ("," identifier)*
+
+identifier: WORD
+quoted_string: QUOTED_STRING
+status: WORD
+
+QUOTED_STRING: /'[^']+'/ | /"[^"]+"/
+WORD: /[a-zA-Z0-9_\-\.]+/
+NUMBER: /[0-9]+/
+
+%import common.WS
+%ignore WS
+"""
+
+
+class RAGFlowCLITransformer(Transformer):
+    def start(self, items):
+        return items[0]
+
+    def command(self, items):
+        return items[0]
+
+    def login_user(self, items):
+        email = items[2].children[0].strip("'\"")
+        return {"type": "login_user", "email": email}
+
+    def ping_server(self, items):
+        return {"type": "ping_server"}
+
+    def list_services(self, items):
+        result = {"type": "list_services"}
+        return result
+
+    def show_service(self, items):
+        service_id = int(items[2])
+        return {"type": "show_service", "number": service_id}
+
+    def startup_service(self, items):
+        service_id = int(items[2])
+        return {"type": "startup_service", "number": service_id}
+
+    def shutdown_service(self, items):
+        service_id = int(items[2])
+        return {"type": "shutdown_service", "number": service_id}
+
+    def restart_service(self, items):
+        service_id = int(items[2])
+        return {"type": "restart_service", "number": service_id}
+
+    def register_user(self, items):
+        user_name: str = items[2].children[0].strip("'\"")
+        nickname: str = items[4].children[0].strip("'\"")
+        password: str = items[6].children[0].strip("'\"")
+        return {"type": "register_user", "user_name": user_name, "nickname": nickname, "password": password}
+
+    def list_users(self, items):
+        return {"type": "list_users"}
+
+    def show_user(self, items):
+        user_name = items[2]
+        return {"type": "show_user", "user_name": user_name}
+
+    def drop_user(self, items):
+        user_name = items[2]
+        return {"type": "drop_user", "user_name": user_name}
+
+    def alter_user(self, items):
+        user_name = items[3]
+        new_password = items[4]
+        return {"type": "alter_user", "user_name": user_name, "password": new_password}
+
+    def create_user(self, items):
+        user_name = items[2]
+        password = items[3]
+        return {"type": "create_user", "user_name": user_name, "password": password, "role": "user"}
+
+    def activate_user(self, items):
+        user_name = items[3]
+        activate_status = items[4]
+        return {"type": "activate_user", "activate_status": activate_status, "user_name": user_name}
+
+    def list_datasets(self, items):
+        user_name = items[3]
+        return {"type": "list_datasets", "user_name": user_name}
+
+    def list_agents(self, items):
+        user_name = items[3]
+        return {"type": "list_agents", "user_name": user_name}
+
+    def create_role(self, items):
+        role_name = items[2]
+        if len(items) > 4:
+            description = items[4]
+            return {"type": "create_role", "role_name": role_name, "description": description}
+        else:
+            return {"type": "create_role", "role_name": role_name}
+
+    def drop_role(self, items):
+        role_name = items[2]
+        return {"type": "drop_role", "role_name": role_name}
+
+    def alter_role(self, items):
+        role_name = items[2]
+        description = items[5]
+        return {"type": "alter_role", "role_name": role_name, "description": description}
+
+    def list_roles(self, items):
+        return {"type": "list_roles"}
+
+    def show_role(self, items):
+        role_name = items[2]
+        return {"type": "show_role", "role_name": role_name}
+
+    def grant_permission(self, items):
+        action_list = items[1]
+        resource = items[3]
+        role_name = items[6]
+        return {"type": "grant_permission", "role_name": role_name, "resource": resource, "actions": action_list}
+
+    def revoke_permission(self, items):
+        action_list = items[1]
+        resource = items[3]
+        role_name = items[6]
+        return {"type": "revoke_permission", "role_name": role_name, "resource": resource, "actions": action_list}
+
+    def alter_user_role(self, items):
+        user_name = items[2]
+        role_name = items[5]
+        return {"type": "alter_user_role", "user_name": user_name, "role_name": role_name}
+
+    def show_user_permission(self, items):
+        user_name = items[3]
+        return {"type": "show_user_permission", "user_name": user_name}
+
+    def show_version(self, items):
+        return {"type": "show_version"}
+
+    def grant_admin(self, items):
+        user_name = items[2]
+        return {"type": "grant_admin", "user_name": user_name}
+
+    def revoke_admin(self, items):
+        user_name = items[2]
+        return {"type": "revoke_admin", "user_name": user_name}
+
+    def generate_key(self, items):
+        user_name = items[4]
+        return {"type": "generate_key", "user_name": user_name}
+
+    def list_keys(self, items):
+        user_name = items[3]
+        return {"type": "list_keys", "user_name": user_name}
+
+    def drop_key(self, items):
+        key = items[2]
+        user_name = items[4]
+        return {"type": "drop_key", "key": key, "user_name": user_name}
+
+    def set_variable(self, items):
+        var_name = items[2]
+        var_value = items[3]
+        return {"type": "set_variable", "var_name": var_name, "var_value": var_value}
+
+    def show_variable(self, items):
+        var_name = items[2]
+        return {"type": "show_variable", "var_name": var_name}
+
+    def list_variables(self, items):
+        return {"type": "list_variables"}
+
+    def list_configs(self, items):
+        return {"type": "list_configs"}
+
+    def list_environments(self, items):
+        return {"type": "list_environments"}
+
+    def create_model_provider(self, items):
+        provider_name = items[3].children[0].strip("'\"")
+        provider_key = items[4].children[0].strip("'\"")
+        return {"type": "create_model_provider", "provider_name": provider_name, "provider_key": provider_key}
+
+    def drop_model_provider(self, items):
+        provider_name = items[3].children[0].strip("'\"")
+        return {"type": "drop_model_provider", "provider_name": provider_name}
+
+    def show_current_user(self, items):
+        return {"type": "show_current_user"}
+
+    def set_default_llm(self, items):
+        llm_id = items[3].children[0].strip("'\"")
+        return {"type": "set_default_model", "model_type": "llm_id", "model_id": llm_id}
+
+    def set_default_vlm(self, items):
+        vlm_id = items[3].children[0].strip("'\"")
+        return {"type": "set_default_model", "model_type": "img2txt_id", "model_id": vlm_id}
+
+    def set_default_embedding(self, items):
+        embedding_id = items[3].children[0].strip("'\"")
+        return {"type": "set_default_model", "model_type": "embd_id", "model_id": embedding_id}
+
+    def set_default_reranker(self, items):
+        reranker_id = items[3].children[0].strip("'\"")
+        return {"type": "set_default_model", "model_type": "reranker_id", "model_id": reranker_id}
+
+    def set_default_asr(self, items):
+        asr_id = items[3].children[0].strip("'\"")
+        return {"type": "set_default_model", "model_type": "asr_id", "model_id": asr_id}
+
+    def set_default_tts(self, items):
+        tts_id = items[3].children[0].strip("'\"")
+        return {"type": "set_default_model", "model_type": "tts_id", "model_id": tts_id}
+
+    def reset_default_llm(self, items):
+        return {"type": "reset_default_model", "model_type": "llm_id"}
+
+    def reset_default_vlm(self, items):
+        return {"type": "reset_default_model", "model_type": "img2txt_id"}
+
+    def reset_default_embedding(self, items):
+        return {"type": "reset_default_model", "model_type": "embd_id"}
+
+    def reset_default_reranker(self, items):
+        return {"type": "reset_default_model", "model_type": "reranker_id"}
+
+    def reset_default_asr(self, items):
+        return {"type": "reset_default_model", "model_type": "asr_id"}
+
+    def reset_default_tts(self, items):
+        return {"type": "reset_default_model", "model_type": "tts_id"}
+
+    def list_user_datasets(self, items):
+        return {"type": "list_user_datasets"}
+
+    def create_user_dataset_with_parser(self, items):
+        dataset_name = items[2].children[0].strip("'\"")
+        embedding = items[5].children[0].strip("'\"")
+        parser_type = items[7].children[0].strip("'\"")
+        return {"type": "create_user_dataset", "dataset_name": dataset_name, "embedding": embedding,
+                "parser_type": parser_type}
+
+    def create_user_dataset_with_pipeline(self, items):
+        dataset_name = items[2].children[0].strip("'\"")
+        embedding = items[5].children[0].strip("'\"")
+        pipeline = items[7].children[0].strip("'\"")
+        return {"type": "create_user_dataset", "dataset_name": dataset_name, "embedding": embedding,
+                "pipeline": pipeline}
+
+    def drop_user_dataset(self, items):
+        dataset_name = items[2].children[0].strip("'\"")
+        return {"type": "drop_user_dataset", "dataset_name": dataset_name}
+
+    def list_user_dataset_files(self, items):
+        dataset_name = items[4].children[0].strip("'\"")
+        return {"type": "list_user_dataset_files", "dataset_name": dataset_name}
+
+    def list_user_agents(self, items):
+        return {"type": "list_user_agents"}
+
+    def list_user_chats(self, items):
+        return {"type": "list_user_chats"}
+
+    def create_user_chat(self, items):
+        chat_name = items[2].children[0].strip("'\"")
+        return {"type": "create_user_chat", "chat_name": chat_name}
+
+    def drop_user_chat(self, items):
+        chat_name = items[2].children[0].strip("'\"")
+        return {"type": "drop_user_chat", "chat_name": chat_name}
+
+    def list_user_model_providers(self, items):
+        return {"type": "list_user_model_providers"}
+
+    def list_user_default_models(self, items):
+        return {"type": "list_user_default_models"}
+
+    def parse_dataset_docs(self, items):
+        document_list_str = items[1].children[0].strip("'\"")
+        document_names = document_list_str.split(",")
+        if len(document_names) == 1:
+            document_names = document_names[0]
+            document_names = document_names.split(" ")
+        dataset_name = items[4].children[0].strip("'\"")
+        return {"type": "parse_dataset_docs", "dataset_name": dataset_name, "document_names": document_names}
+
+    def parse_dataset_sync(self, items):
+        dataset_name = items[2].children[0].strip("'\"")
+        return {"type": "parse_dataset", "dataset_name": dataset_name, "method": "sync"}
+
+    def parse_dataset_async(self, items):
+        dataset_name = items[2].children[0].strip("'\"")
+        return {"type": "parse_dataset", "dataset_name": dataset_name, "method": "async"}
+
+    def import_docs_into_dataset(self, items):
+        document_list_str = items[1].children[0].strip("'\"")
+        document_paths = document_list_str.split(",")
+        if len(document_paths) == 1:
+            document_paths = document_paths[0]
+            document_paths = document_paths.split(" ")
+        dataset_name = items[4].children[0].strip("'\"")
+        return {"type": "import_docs_into_dataset", "dataset_name": dataset_name, "document_paths": document_paths}
+
+    def search_on_datasets(self, items):
+        question = items[1].children[0].strip("'\"")
+        datasets_str = items[4].children[0].strip("'\"")
+        datasets = datasets_str.split(",")
+        if len(datasets) == 1:
+            datasets = datasets[0]
+            datasets = datasets.split(" ")
+        return {"type": "search_on_datasets", "datasets": datasets, "question": question}
+
+    def benchmark(self, items):
+        concurrency: int = int(items[1])
+        iterations: int = int(items[2])
+        command = items[3].children[0]
+        return {"type": "benchmark", "concurrency": concurrency, "iterations": iterations, "command": command}
+
+    def action_list(self, items):
+        return items
+
+    def meta_command(self, items):
+        command_name = str(items[0]).lower()
+        args = items[1:] if len(items) > 1 else []
+
+        # handle quoted parameter
+        parsed_args = []
+        for arg in args:
+            if hasattr(arg, "value"):
+                parsed_args.append(arg.value)
+            else:
+                parsed_args.append(str(arg))
+
+        return {"type": "meta", "command": command_name, "args": parsed_args}
+
+    def meta_command_name(self, items):
+        return items[0]
+
+    def meta_args(self, items):
+        return items

admin/client/pyproject.toml

@@ -1,14 +1,14 @@
 [project]
 name = "ragflow-cli"
-version = "0.22.1"
+version = "0.23.1"
 description = "Admin Service's client of [RAGFlow](https://github.com/infiniflow/ragflow). The Admin Service provides user management and system monitoring. "
 authors = [{ name = "Lynn", email = "lynn_inf@hotmail.com" }]
 license = { text = "Apache License, Version 2.0" }
 readme = "README.md"
-requires-python = ">=3.10,<3.13"
+requires-python = ">=3.12,<3.15"
 dependencies = [
     "requests>=2.30.0,<3.0.0",
-    "beartype>=0.18.5,<0.19.0",
+    "beartype>=0.20.0,<1.0.0",
     "pycryptodomex>=3.10.0",
     "lark>=1.1.0",
 ]
@@ -20,5 +20,8 @@ test = [
     "requests-toolbelt>=1.0.0",
 ]
 
+[tool.setuptools]
+py-modules = ["ragflow_cli", "parser"]
+
 [project.scripts]
-ragflow-cli = "admin_client:main"
+ragflow-cli = "ragflow_cli:main"

admin/client/ragflow_cli.py

@@ -0,0 +1,322 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+
+import sys
+import argparse
+import base64
+import getpass
+from cmd import Cmd
+from typing import Any, Dict, List
+
+import requests
+import warnings
+from Cryptodome.Cipher import PKCS1_v1_5 as Cipher_pkcs1_v1_5
+from Cryptodome.PublicKey import RSA
+from lark import Lark, Tree
+from parser import GRAMMAR, RAGFlowCLITransformer
+from http_client import HttpClient
+from ragflow_client import RAGFlowClient, run_command
+from user import login_user
+
+warnings.filterwarnings("ignore", category=getpass.GetPassWarning)
+
+def encrypt(input_string):
+    pub = "-----BEGIN PUBLIC KEY-----\nMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEArq9XTUSeYr2+N1h3Afl/z8Dse/2yD0ZGrKwx+EEEcdsBLca9Ynmx3nIB5obmLlSfmskLpBo0UACBmB5rEjBp2Q2f3AG3Hjd4B+gNCG6BDaawuDlgANIhGnaTLrIqWrrcm4EMzJOnAOI1fgzJRsOOUEfaS318Eq9OVO3apEyCCt0lOQK6PuksduOjVxtltDav+guVAA068NrPYmRNabVKRNLJpL8w4D44sfth5RvZ3q9t+6RTArpEtc5sh5ChzvqPOzKGMXW83C95TxmXqpbK6olN4RevSfVjEAgCydH6HN6OhtOQEcnrU97r9H0iZOWwbw3pVrZiUkuRD1R56Wzs2wIDAQAB\n-----END PUBLIC KEY-----"
+    pub_key = RSA.importKey(pub)
+    cipher = Cipher_pkcs1_v1_5.new(pub_key)
+    cipher_text = cipher.encrypt(base64.b64encode(input_string.encode("utf-8")))
+    return base64.b64encode(cipher_text).decode("utf-8")
+
+
+def encode_to_base64(input_string):
+    base64_encoded = base64.b64encode(input_string.encode("utf-8"))
+    return base64_encoded.decode("utf-8")
+
+
+
+
+
+class RAGFlowCLI(Cmd):
+    def __init__(self):
+        super().__init__()
+        self.parser = Lark(GRAMMAR, start="start", parser="lalr", transformer=RAGFlowCLITransformer())
+        self.command_history = []
+        self.account = "admin@ragflow.io"
+        self.account_password: str = "admin"
+        self.session = requests.Session()
+        self.host: str = ""
+        self.port: int = 0
+        self.mode: str = "admin"
+        self.ragflow_client = None
+
+    intro = r"""Type "\h" for help."""
+    prompt = "ragflow> "
+
+    def onecmd(self, command: str) -> bool:
+        try:
+            result = self.parse_command(command)
+
+            if isinstance(result, dict):
+                if "type" in result and result.get("type") == "empty":
+                    return False
+
+            self.execute_command(result)
+
+            if isinstance(result, Tree):
+                return False
+
+            if result.get("type") == "meta" and result.get("command") in ["q", "quit", "exit"]:
+                return True
+
+        except KeyboardInterrupt:
+            print("\nUse '\\q' to quit")
+        except EOFError:
+            print("\nGoodbye!")
+            return True
+        return False
+
+    def emptyline(self) -> bool:
+        return False
+
+    def default(self, line: str) -> bool:
+        return self.onecmd(line)
+
+    def parse_command(self, command_str: str) -> dict[str, str]:
+        if not command_str.strip():
+            return {"type": "empty"}
+
+        self.command_history.append(command_str)
+
+        try:
+            result = self.parser.parse(command_str)
+            return result
+        except Exception as e:
+            return {"type": "error", "message": f"Parse error: {str(e)}"}
+
+    def verify_auth(self, arguments: dict, single_command: bool, auth: bool):
+        server_type = arguments.get("type", "admin")
+        http_client = HttpClient(arguments["host"], arguments["port"])
+        if not auth:
+            self.ragflow_client = RAGFlowClient(http_client, server_type)
+            return True
+
+        user_name = arguments["username"]
+        attempt_count = 3
+        if single_command:
+            attempt_count = 1
+
+        try_count = 0
+        while True:
+            try_count += 1
+            if try_count > attempt_count:
+                return False
+
+            if single_command:
+                user_password = arguments["password"]
+            else:
+                user_password = getpass.getpass(f"password for {user_name}: ").strip()
+
+            try:
+                token = login_user(http_client, server_type, user_name, user_password)
+                http_client.login_token = token
+                self.ragflow_client = RAGFlowClient(http_client, server_type)
+                return True
+            except Exception as e:
+                print(str(e))
+                print("Can't access server for login (connection failed)")
+
+    def _format_service_detail_table(self, data):
+        if isinstance(data, list):
+            return data
+        if not all([isinstance(v, list) for v in data.values()]):
+            # normal table
+            return data
+        # handle task_executor heartbeats map, for example {'name': [{'done': 2, 'now': timestamp1}, {'done': 3, 'now': timestamp2}]
+        task_executor_list = []
+        for k, v in data.items():
+            # display latest status
+            heartbeats = sorted(v, key=lambda x: x["now"], reverse=True)
+            task_executor_list.append(
+                {
+                    "task_executor_name": k,
+                    **heartbeats[0],
+                }
+                if heartbeats
+                else {"task_executor_name": k}
+            )
+        return task_executor_list
+
+    def _print_table_simple(self, data):
+        if not data:
+            print("No data to print")
+            return
+        if isinstance(data, dict):
+            # handle single row data
+            data = [data]
+
+        columns = list(set().union(*(d.keys() for d in data)))
+        columns.sort()
+        col_widths = {}
+
+        def get_string_width(text):
+            half_width_chars = " !\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~\t\n\r"
+            width = 0
+            for char in text:
+                if char in half_width_chars:
+                    width += 1
+                else:
+                    width += 2
+            return width
+
+        for col in columns:
+            max_width = get_string_width(str(col))
+            for item in data:
+                value_len = get_string_width(str(item.get(col, "")))
+                if value_len > max_width:
+                    max_width = value_len
+            col_widths[col] = max(2, max_width)
+
+        # Generate delimiter
+        separator = "+" + "+".join(["-" * (col_widths[col] + 2) for col in columns]) + "+"
+
+        # Print header
+        print(separator)
+        header = "|" + "|".join([f" {col:<{col_widths[col]}} " for col in columns]) + "|"
+        print(header)
+        print(separator)
+
+        # Print data
+        for item in data:
+            row = "|"
+            for col in columns:
+                value = str(item.get(col, ""))
+                if get_string_width(value) > col_widths[col]:
+                    value = value[: col_widths[col] - 3] + "..."
+                row += f" {value:<{col_widths[col] - (get_string_width(value) - len(value))}} |"
+            print(row)
+
+        print(separator)
+
+    def run_interactive(self, args):
+        if self.verify_auth(args, single_command=False, auth=args["auth"]):
+            print(r"""
+                ____  ___   ______________                 ________    ____
+               / __ \/   | / ____/ ____/ /___ _      __   / ____/ /   /  _/
+              / /_/ / /| |/ / __/ /_  / / __ \ | /| / /  / /   / /    / /  
+             / _, _/ ___ / /_/ / __/ / / /_/ / |/ |/ /  / /___/ /____/ /   
+            /_/ |_/_/  |_\____/_/   /_/\____/|__/|__/   \____/_____/___/   
+            """)
+            self.cmdloop()
+
+        print("RAGFlow command line interface - Type '\\?' for help, '\\q' to quit")
+
+    def run_single_command(self, args):
+        if self.verify_auth(args, single_command=True, auth=args["auth"]):
+            command = args["command"]
+            result = self.parse_command(command)
+            self.execute_command(result)
+
+
+    def parse_connection_args(self, args: List[str]) -> Dict[str, Any]:
+        parser = argparse.ArgumentParser(description="RAGFlow CLI Client", add_help=False)
+        parser.add_argument("-h", "--host", default="127.0.0.1", help="Admin or RAGFlow service host")
+        parser.add_argument("-p", "--port", type=int, default=9381, help="Admin or RAGFlow service port")
+        parser.add_argument("-w", "--password", default="admin", type=str, help="Superuser password")
+        parser.add_argument("-t", "--type", default="admin", type=str, help="CLI mode, admin or user")
+        parser.add_argument("-u", "--username", default=None,
+                            help="Username (email). In admin mode defaults to admin@ragflow.io, in user mode required.")
+        parser.add_argument("command", nargs="?", help="Single command")
+        try:
+            parsed_args, remaining_args = parser.parse_known_args(args)
+            # Determine username based on mode
+            username = parsed_args.username
+            if parsed_args.type == "admin":
+                if username is None:
+                    username = "admin@ragflow.io"
+
+            if remaining_args:
+                if remaining_args[0] == "command":
+                    command_str = ' '.join(remaining_args[1:]) + ';'
+                    auth = True
+                    if remaining_args[1] == "register":
+                        auth = False
+                    else:
+                        if username is None:
+                            print("Error: username (-u) is required in user mode")
+                            return {"error": "Username required"}
+                    return {
+                        "host": parsed_args.host,
+                        "port": parsed_args.port,
+                        "password": parsed_args.password,
+                        "type": parsed_args.type,
+                        "username": username,
+                        "command": command_str,
+                        "auth": auth
+                    }
+                else:
+                    return {"error": "Invalid command"}
+            else:
+                auth = True
+                if username is None:
+                    auth = False
+                return {
+                    "host": parsed_args.host,
+                    "port": parsed_args.port,
+                    "type": parsed_args.type,
+                    "username": username,
+                    "auth": auth
+                }
+        except SystemExit:
+            return {"error": "Invalid connection arguments"}
+
+    def execute_command(self, parsed_command: Dict[str, Any]):
+        command_dict: dict
+        if isinstance(parsed_command, Tree):
+            command_dict = parsed_command.children[0]
+        else:
+            if parsed_command["type"] == "error":
+                print(f"Error: {parsed_command['message']}")
+                return
+            else:
+                command_dict = parsed_command
+
+        # print(f"Parsed command: {command_dict}")
+        run_command(self.ragflow_client, command_dict)
+
+def main():
+
+    cli = RAGFlowCLI()
+
+    args = cli.parse_connection_args(sys.argv)
+    if "error" in args:
+        print("Error: Invalid connection arguments")
+        return
+
+    if "command" in args:
+        # single command mode
+        # for user mode, api key or password is ok
+        # for admin mode, only password
+        if "password" not in args:
+            print("Error: password is missing")
+            return
+
+        cli.run_single_command(args)
+    else:
+        cli.run_interactive(args)
+
+
+if __name__ == "__main__":
+    main()

admin/client/user.py

@@ -0,0 +1,65 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+
+from http_client import HttpClient
+
+
+class AuthException(Exception):
+    def __init__(self, message, code=401):
+        super().__init__(message)
+        self.code = code
+        self.message = message
+
+
+def encrypt_password(password_plain: str) -> str:
+    try:
+        from api.utils.crypt import crypt
+    except Exception as exc:
+        raise AuthException(
+            "Password encryption unavailable; install pycryptodomex (uv sync --python 3.12 --group test)."
+        ) from exc
+    return crypt(password_plain)
+
+
+def register_user(client: HttpClient, email: str, nickname: str, password: str) -> None:
+    password_enc = encrypt_password(password)
+    payload = {"email": email, "nickname": nickname, "password": password_enc}
+    res = client.request_json("POST", "/user/register", use_api_base=False, auth_kind=None, json_body=payload)
+    if res.get("code") == 0:
+        return
+    msg = res.get("message", "")
+    if "has already registered" in msg:
+        return
+    raise AuthException(f"Register failed: {msg}")
+
+
+def login_user(client: HttpClient, server_type: str, email: str, password: str) -> str:
+    password_enc = encrypt_password(password)
+    payload = {"email": email, "password": password_enc}
+    if server_type == "admin":
+        response = client.request("POST", "/admin/login", use_api_base=True, auth_kind=None, json_body=payload)
+    else:
+        response = client.request("POST", "/user/login", use_api_base=False, auth_kind=None, json_body=payload)
+    try:
+        res = response.json()
+    except Exception as exc:
+        raise AuthException(f"Login failed: invalid JSON response ({exc})") from exc
+    if res.get("code") != 0:
+        raise AuthException(f"Login failed: {res.get('message')}")
+    token = response.headers.get("Authorization")
+    if not token:
+        raise AuthException("Login failed: missing Authorization header")
+    return token

admin/client/uv.lock

@@ -0,0 +1,298 @@
+version = 1
+revision = 3
+requires-python = ">=3.10, <3.13"
+
+[[package]]
+name = "beartype"
+version = "0.22.6"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/88/e2/105ceb1704cb80fe4ab3872529ab7b6f365cf7c74f725e6132d0efcf1560/beartype-0.22.6.tar.gz", hash = "sha256:97fbda69c20b48c5780ac2ca60ce3c1bb9af29b3a1a0216898ffabdd523e48f4", size = 1588975, upload-time = "2025-11-20T04:47:14.736Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/98/c9/ceecc71fe2c9495a1d8e08d44f5f31f5bca1350d5b2e27a4b6265424f59e/beartype-0.22.6-py3-none-any.whl", hash = "sha256:0584bc46a2ea2a871509679278cda992eadde676c01356ab0ac77421f3c9a093", size = 1324807, upload-time = "2025-11-20T04:47:11.837Z" },
+]
+
+[[package]]
+name = "certifi"
+version = "2025.11.12"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/a2/8c/58f469717fa48465e4a50c014a0400602d3c437d7c0c468e17ada824da3a/certifi-2025.11.12.tar.gz", hash = "sha256:d8ab5478f2ecd78af242878415affce761ca6bc54a22a27e026d7c25357c3316", size = 160538, upload-time = "2025-11-12T02:54:51.517Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/70/7d/9bc192684cea499815ff478dfcdc13835ddf401365057044fb721ec6bddb/certifi-2025.11.12-py3-none-any.whl", hash = "sha256:97de8790030bbd5c2d96b7ec782fc2f7820ef8dba6db909ccf95449f2d062d4b", size = 159438, upload-time = "2025-11-12T02:54:49.735Z" },
+]
+
+[[package]]
+name = "charset-normalizer"
+version = "3.4.4"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/13/69/33ddede1939fdd074bce5434295f38fae7136463422fe4fd3e0e89b98062/charset_normalizer-3.4.4.tar.gz", hash = "sha256:94537985111c35f28720e43603b8e7b43a6ecfb2ce1d3058bbe955b73404e21a", size = 129418, upload-time = "2025-10-14T04:42:32.879Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/1f/b8/6d51fc1d52cbd52cd4ccedd5b5b2f0f6a11bbf6765c782298b0f3e808541/charset_normalizer-3.4.4-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:e824f1492727fa856dd6eda4f7cee25f8518a12f3c4a56a74e8095695089cf6d", size = 209709, upload-time = "2025-10-14T04:40:11.385Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/5c/af/1f9d7f7faafe2ddfb6f72a2e07a548a629c61ad510fe60f9630309908fef/charset_normalizer-3.4.4-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4bd5d4137d500351a30687c2d3971758aac9a19208fc110ccb9d7188fbe709e8", size = 148814, upload-time = "2025-10-14T04:40:13.135Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/79/3d/f2e3ac2bbc056ca0c204298ea4e3d9db9b4afe437812638759db2c976b5f/charset_normalizer-3.4.4-cp310-cp310-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:027f6de494925c0ab2a55eab46ae5129951638a49a34d87f4c3eda90f696b4ad", size = 144467, upload-time = "2025-10-14T04:40:14.728Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/ec/85/1bf997003815e60d57de7bd972c57dc6950446a3e4ccac43bc3070721856/charset_normalizer-3.4.4-cp310-cp310-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:f820802628d2694cb7e56db99213f930856014862f3fd943d290ea8438d07ca8", size = 162280, upload-time = "2025-10-14T04:40:16.14Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/3e/8e/6aa1952f56b192f54921c436b87f2aaf7c7a7c3d0d1a765547d64fd83c13/charset_normalizer-3.4.4-cp310-cp310-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:798d75d81754988d2565bff1b97ba5a44411867c0cf32b77a7e8f8d84796b10d", size = 159454, upload-time = "2025-10-14T04:40:17.567Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/36/3b/60cbd1f8e93aa25d1c669c649b7a655b0b5fb4c571858910ea9332678558/charset_normalizer-3.4.4-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:9d1bb833febdff5c8927f922386db610b49db6e0d4f4ee29601d71e7c2694313", size = 153609, upload-time = "2025-10-14T04:40:19.08Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/64/91/6a13396948b8fd3c4b4fd5bc74d045f5637d78c9675585e8e9fbe5636554/charset_normalizer-3.4.4-cp310-cp310-manylinux_2_31_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:9cd98cdc06614a2f768d2b7286d66805f94c48cde050acdbbb7db2600ab3197e", size = 151849, upload-time = "2025-10-14T04:40:20.607Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/b7/7a/59482e28b9981d105691e968c544cc0df3b7d6133152fb3dcdc8f135da7a/charset_normalizer-3.4.4-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:077fbb858e903c73f6c9db43374fd213b0b6a778106bc7032446a8e8b5b38b93", size = 151586, upload-time = "2025-10-14T04:40:21.719Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/92/59/f64ef6a1c4bdd2baf892b04cd78792ed8684fbc48d4c2afe467d96b4df57/charset_normalizer-3.4.4-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:244bfb999c71b35de57821b8ea746b24e863398194a4014e4c76adc2bbdfeff0", size = 145290, upload-time = "2025-10-14T04:40:23.069Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/6b/63/3bf9f279ddfa641ffa1962b0db6a57a9c294361cc2f5fcac997049a00e9c/charset_normalizer-3.4.4-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:64b55f9dce520635f018f907ff1b0df1fdc31f2795a922fb49dd14fbcdf48c84", size = 163663, upload-time = "2025-10-14T04:40:24.17Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/ed/09/c9e38fc8fa9e0849b172b581fd9803bdf6e694041127933934184e19f8c3/charset_normalizer-3.4.4-cp310-cp310-musllinux_1_2_riscv64.whl", hash = "sha256:faa3a41b2b66b6e50f84ae4a68c64fcd0c44355741c6374813a800cd6695db9e", size = 151964, upload-time = "2025-10-14T04:40:25.368Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/d2/d1/d28b747e512d0da79d8b6a1ac18b7ab2ecfd81b2944c4c710e166d8dd09c/charset_normalizer-3.4.4-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:6515f3182dbe4ea06ced2d9e8666d97b46ef4c75e326b79bb624110f122551db", size = 161064, upload-time = "2025-10-14T04:40:26.806Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/bb/9a/31d62b611d901c3b9e5500c36aab0ff5eb442043fb3a1c254200d3d397d9/charset_normalizer-3.4.4-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:cc00f04ed596e9dc0da42ed17ac5e596c6ccba999ba6bd92b0e0aef2f170f2d6", size = 155015, upload-time = "2025-10-14T04:40:28.284Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/1f/f3/107e008fa2bff0c8b9319584174418e5e5285fef32f79d8ee6a430d0039c/charset_normalizer-3.4.4-cp310-cp310-win32.whl", hash = "sha256:f34be2938726fc13801220747472850852fe6b1ea75869a048d6f896838c896f", size = 99792, upload-time = "2025-10-14T04:40:29.613Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/eb/66/e396e8a408843337d7315bab30dbf106c38966f1819f123257f5520f8a96/charset_normalizer-3.4.4-cp310-cp310-win_amd64.whl", hash = "sha256:a61900df84c667873b292c3de315a786dd8dac506704dea57bc957bd31e22c7d", size = 107198, upload-time = "2025-10-14T04:40:30.644Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/b5/58/01b4f815bf0312704c267f2ccb6e5d42bcc7752340cd487bc9f8c3710597/charset_normalizer-3.4.4-cp310-cp310-win_arm64.whl", hash = "sha256:cead0978fc57397645f12578bfd2d5ea9138ea0fac82b2f63f7f7c6877986a69", size = 100262, upload-time = "2025-10-14T04:40:32.108Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/ed/27/c6491ff4954e58a10f69ad90aca8a1b6fe9c5d3c6f380907af3c37435b59/charset_normalizer-3.4.4-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:6e1fcf0720908f200cd21aa4e6750a48ff6ce4afe7ff5a79a90d5ed8a08296f8", size = 206988, upload-time = "2025-10-14T04:40:33.79Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/94/59/2e87300fe67ab820b5428580a53cad894272dbb97f38a7a814a2a1ac1011/charset_normalizer-3.4.4-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5f819d5fe9234f9f82d75bdfa9aef3a3d72c4d24a6e57aeaebba32a704553aa0", size = 147324, upload-time = "2025-10-14T04:40:34.961Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/07/fb/0cf61dc84b2b088391830f6274cb57c82e4da8bbc2efeac8c025edb88772/charset_normalizer-3.4.4-cp311-cp311-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:a59cb51917aa591b1c4e6a43c132f0cdc3c76dbad6155df4e28ee626cc77a0a3", size = 142742, upload-time = "2025-10-14T04:40:36.105Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/62/8b/171935adf2312cd745d290ed93cf16cf0dfe320863ab7cbeeae1dcd6535f/charset_normalizer-3.4.4-cp311-cp311-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:8ef3c867360f88ac904fd3f5e1f902f13307af9052646963ee08ff4f131adafc", size = 160863, upload-time = "2025-10-14T04:40:37.188Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/09/73/ad875b192bda14f2173bfc1bc9a55e009808484a4b256748d931b6948442/charset_normalizer-3.4.4-cp311-cp311-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:d9e45d7faa48ee908174d8fe84854479ef838fc6a705c9315372eacbc2f02897", size = 157837, upload-time = "2025-10-14T04:40:38.435Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/6d/fc/de9cce525b2c5b94b47c70a4b4fb19f871b24995c728e957ee68ab1671ea/charset_normalizer-3.4.4-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:840c25fb618a231545cbab0564a799f101b63b9901f2569faecd6b222ac72381", size = 151550, upload-time = "2025-10-14T04:40:40.053Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/55/c2/43edd615fdfba8c6f2dfbd459b25a6b3b551f24ea21981e23fb768503ce1/charset_normalizer-3.4.4-cp311-cp311-manylinux_2_31_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:ca5862d5b3928c4940729dacc329aa9102900382fea192fc5e52eb69d6093815", size = 149162, upload-time = "2025-10-14T04:40:41.163Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/03/86/bde4ad8b4d0e9429a4e82c1e8f5c659993a9a863ad62c7df05cf7b678d75/charset_normalizer-3.4.4-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d9c7f57c3d666a53421049053eaacdd14bbd0a528e2186fcb2e672effd053bb0", size = 150019, upload-time = "2025-10-14T04:40:42.276Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/1f/86/a151eb2af293a7e7bac3a739b81072585ce36ccfb4493039f49f1d3cae8c/charset_normalizer-3.4.4-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:277e970e750505ed74c832b4bf75dac7476262ee2a013f5574dd49075879e161", size = 143310, upload-time = "2025-10-14T04:40:43.439Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/b5/fe/43dae6144a7e07b87478fdfc4dbe9efd5defb0e7ec29f5f58a55aeef7bf7/charset_normalizer-3.4.4-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:31fd66405eaf47bb62e8cd575dc621c56c668f27d46a61d975a249930dd5e2a4", size = 162022, upload-time = "2025-10-14T04:40:44.547Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/80/e6/7aab83774f5d2bca81f42ac58d04caf44f0cc2b65fc6db2b3b2e8a05f3b3/charset_normalizer-3.4.4-cp311-cp311-musllinux_1_2_riscv64.whl", hash = "sha256:0d3d8f15c07f86e9ff82319b3d9ef6f4bf907608f53fe9d92b28ea9ae3d1fd89", size = 149383, upload-time = "2025-10-14T04:40:46.018Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/4f/e8/b289173b4edae05c0dde07f69f8db476a0b511eac556dfe0d6bda3c43384/charset_normalizer-3.4.4-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:9f7fcd74d410a36883701fafa2482a6af2ff5ba96b9a620e9e0721e28ead5569", size = 159098, upload-time = "2025-10-14T04:40:47.081Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/d8/df/fe699727754cae3f8478493c7f45f777b17c3ef0600e28abfec8619eb49c/charset_normalizer-3.4.4-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:ebf3e58c7ec8a8bed6d66a75d7fb37b55e5015b03ceae72a8e7c74495551e224", size = 152991, upload-time = "2025-10-14T04:40:48.246Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/1a/86/584869fe4ddb6ffa3bd9f491b87a01568797fb9bd8933f557dba9771beaf/charset_normalizer-3.4.4-cp311-cp311-win32.whl", hash = "sha256:eecbc200c7fd5ddb9a7f16c7decb07b566c29fa2161a16cf67b8d068bd21690a", size = 99456, upload-time = "2025-10-14T04:40:49.376Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/65/f6/62fdd5feb60530f50f7e38b4f6a1d5203f4d16ff4f9f0952962c044e919a/charset_normalizer-3.4.4-cp311-cp311-win_amd64.whl", hash = "sha256:5ae497466c7901d54b639cf42d5b8c1b6a4fead55215500d2f486d34db48d016", size = 106978, upload-time = "2025-10-14T04:40:50.844Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/7a/9d/0710916e6c82948b3be62d9d398cb4fcf4e97b56d6a6aeccd66c4b2f2bd5/charset_normalizer-3.4.4-cp311-cp311-win_arm64.whl", hash = "sha256:65e2befcd84bc6f37095f5961e68a6f077bf44946771354a28ad434c2cce0ae1", size = 99969, upload-time = "2025-10-14T04:40:52.272Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/f3/85/1637cd4af66fa687396e757dec650f28025f2a2f5a5531a3208dc0ec43f2/charset_normalizer-3.4.4-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:0a98e6759f854bd25a58a73fa88833fba3b7c491169f86ce1180c948ab3fd394", size = 208425, upload-time = "2025-10-14T04:40:53.353Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/9d/6a/04130023fef2a0d9c62d0bae2649b69f7b7d8d24ea5536feef50551029df/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b5b290ccc2a263e8d185130284f8501e3e36c5e02750fc6b6bdeb2e9e96f1e25", size = 148162, upload-time = "2025-10-14T04:40:54.558Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/78/29/62328d79aa60da22c9e0b9a66539feae06ca0f5a4171ac4f7dc285b83688/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:74bb723680f9f7a6234dcf67aea57e708ec1fbdf5699fb91dfd6f511b0a320ef", size = 144558, upload-time = "2025-10-14T04:40:55.677Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/86/bb/b32194a4bf15b88403537c2e120b817c61cd4ecffa9b6876e941c3ee38fe/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:f1e34719c6ed0b92f418c7c780480b26b5d9c50349e9a9af7d76bf757530350d", size = 161497, upload-time = "2025-10-14T04:40:57.217Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/19/89/a54c82b253d5b9b111dc74aca196ba5ccfcca8242d0fb64146d4d3183ff1/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:2437418e20515acec67d86e12bf70056a33abdacb5cb1655042f6538d6b085a8", size = 159240, upload-time = "2025-10-14T04:40:58.358Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/c0/10/d20b513afe03acc89ec33948320a5544d31f21b05368436d580dec4e234d/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:11d694519d7f29d6cd09f6ac70028dba10f92f6cdd059096db198c283794ac86", size = 153471, upload-time = "2025-10-14T04:40:59.468Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/61/fa/fbf177b55bdd727010f9c0a3c49eefa1d10f960e5f09d1d887bf93c2e698/charset_normalizer-3.4.4-cp312-cp312-manylinux_2_31_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:ac1c4a689edcc530fc9d9aa11f5774b9e2f33f9a0c6a57864e90908f5208d30a", size = 150864, upload-time = "2025-10-14T04:41:00.623Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/05/12/9fbc6a4d39c0198adeebbde20b619790e9236557ca59fc40e0e3cebe6f40/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:21d142cc6c0ec30d2efee5068ca36c128a30b0f2c53c1c07bd78cb6bc1d3be5f", size = 150647, upload-time = "2025-10-14T04:41:01.754Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/ad/1f/6a9a593d52e3e8c5d2b167daf8c6b968808efb57ef4c210acb907c365bc4/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:5dbe56a36425d26d6cfb40ce79c314a2e4dd6211d51d6d2191c00bed34f354cc", size = 145110, upload-time = "2025-10-14T04:41:03.231Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/30/42/9a52c609e72471b0fc54386dc63c3781a387bb4fe61c20231a4ebcd58bdd/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:5bfbb1b9acf3334612667b61bd3002196fe2a1eb4dd74d247e0f2a4d50ec9bbf", size = 162839, upload-time = "2025-10-14T04:41:04.715Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/c4/5b/c0682bbf9f11597073052628ddd38344a3d673fda35a36773f7d19344b23/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_riscv64.whl", hash = "sha256:d055ec1e26e441f6187acf818b73564e6e6282709e9bcb5b63f5b23068356a15", size = 150667, upload-time = "2025-10-14T04:41:05.827Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/e4/24/a41afeab6f990cf2daf6cb8c67419b63b48cf518e4f56022230840c9bfb2/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:af2d8c67d8e573d6de5bc30cdb27e9b95e49115cd9baad5ddbd1a6207aaa82a9", size = 160535, upload-time = "2025-10-14T04:41:06.938Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/2a/e5/6a4ce77ed243c4a50a1fecca6aaaab419628c818a49434be428fe24c9957/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:780236ac706e66881f3b7f2f32dfe90507a09e67d1d454c762cf642e6e1586e0", size = 154816, upload-time = "2025-10-14T04:41:08.101Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/a8/ef/89297262b8092b312d29cdb2517cb1237e51db8ecef2e9af5edbe7b683b1/charset_normalizer-3.4.4-cp312-cp312-win32.whl", hash = "sha256:5833d2c39d8896e4e19b689ffc198f08ea58116bee26dea51e362ecc7cd3ed26", size = 99694, upload-time = "2025-10-14T04:41:09.23Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/3d/2d/1e5ed9dd3b3803994c155cd9aacb60c82c331bad84daf75bcb9c91b3295e/charset_normalizer-3.4.4-cp312-cp312-win_amd64.whl", hash = "sha256:a79cfe37875f822425b89a82333404539ae63dbdddf97f84dcbc3d339aae9525", size = 107131, upload-time = "2025-10-14T04:41:10.467Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/d0/d9/0ed4c7098a861482a7b6a95603edce4c0d9db2311af23da1fb2b75ec26fc/charset_normalizer-3.4.4-cp312-cp312-win_arm64.whl", hash = "sha256:376bec83a63b8021bb5c8ea75e21c4ccb86e7e45ca4eb81146091b56599b80c3", size = 100390, upload-time = "2025-10-14T04:41:11.915Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/0a/4c/925909008ed5a988ccbb72dcc897407e5d6d3bd72410d69e051fc0c14647/charset_normalizer-3.4.4-py3-none-any.whl", hash = "sha256:7a32c560861a02ff789ad905a2fe94e3f840803362c84fecf1851cb4cf3dc37f", size = 53402, upload-time = "2025-10-14T04:42:31.76Z" },
+]
+
+[[package]]
+name = "colorama"
+version = "0.4.6"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/d8/53/6f443c9a4a8358a93a6792e2acffb9d9d5cb0a5cfd8802644b7b1c9a02e4/colorama-0.4.6.tar.gz", hash = "sha256:08695f5cb7ed6e0531a20572697297273c47b8cae5a63ffc6d6ed5c201be6e44", size = 27697, upload-time = "2022-10-25T02:36:22.414Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/d1/d6/3965ed04c63042e047cb6a3e6ed1a63a35087b6a609aa3a15ed8ac56c221/colorama-0.4.6-py2.py3-none-any.whl", hash = "sha256:4f1d9991f5acc0ca119f9d443620b77f9d6b33703e51011c16baf57afb285fc6", size = 25335, upload-time = "2022-10-25T02:36:20.889Z" },
+]
+
+[[package]]
+name = "exceptiongroup"
+version = "1.3.1"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+dependencies = [
+    { name = "typing-extensions" },
+]
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/50/79/66800aadf48771f6b62f7eb014e352e5d06856655206165d775e675a02c9/exceptiongroup-1.3.1.tar.gz", hash = "sha256:8b412432c6055b0b7d14c310000ae93352ed6754f70fa8f7c34141f91c4e3219", size = 30371, upload-time = "2025-11-21T23:01:54.787Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/8a/0e/97c33bf5009bdbac74fd2beace167cab3f978feb69cc36f1ef79360d6c4e/exceptiongroup-1.3.1-py3-none-any.whl", hash = "sha256:a7a39a3bd276781e98394987d3a5701d0c4edffb633bb7a5144577f82c773598", size = 16740, upload-time = "2025-11-21T23:01:53.443Z" },
+]
+
+[[package]]
+name = "idna"
+version = "3.11"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/6f/6d/0703ccc57f3a7233505399edb88de3cbd678da106337b9fcde432b65ed60/idna-3.11.tar.gz", hash = "sha256:795dafcc9c04ed0c1fb032c2aa73654d8e8c5023a7df64a53f39190ada629902", size = 194582, upload-time = "2025-10-12T14:55:20.501Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/0e/61/66938bbb5fc52dbdf84594873d5b51fb1f7c7794e9c0f5bd885f30bc507b/idna-3.11-py3-none-any.whl", hash = "sha256:771a87f49d9defaf64091e6e6fe9c18d4833f140bd19464795bc32d966ca37ea", size = 71008, upload-time = "2025-10-12T14:55:18.883Z" },
+]
+
+[[package]]
+name = "iniconfig"
+version = "2.3.0"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/72/34/14ca021ce8e5dfedc35312d08ba8bf51fdd999c576889fc2c24cb97f4f10/iniconfig-2.3.0.tar.gz", hash = "sha256:c76315c77db068650d49c5b56314774a7804df16fee4402c1f19d6d15d8c4730", size = 20503, upload-time = "2025-10-18T21:55:43.219Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/cb/b1/3846dd7f199d53cb17f49cba7e651e9ce294d8497c8c150530ed11865bb8/iniconfig-2.3.0-py3-none-any.whl", hash = "sha256:f631c04d2c48c52b84d0d0549c99ff3859c98df65b3101406327ecc7d53fbf12", size = 7484, upload-time = "2025-10-18T21:55:41.639Z" },
+]
+
+[[package]]
+name = "lark"
+version = "1.3.1"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/da/34/28fff3ab31ccff1fd4f6c7c7b0ceb2b6968d8ea4950663eadcb5720591a0/lark-1.3.1.tar.gz", hash = "sha256:b426a7a6d6d53189d318f2b6236ab5d6429eaf09259f1ca33eb716eed10d2905", size = 382732, upload-time = "2025-10-27T18:25:56.653Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/82/3d/14ce75ef66813643812f3093ab17e46d3a206942ce7376d31ec2d36229e7/lark-1.3.1-py3-none-any.whl", hash = "sha256:c629b661023a014c37da873b4ff58a817398d12635d3bbb2c5a03be7fe5d1e12", size = 113151, upload-time = "2025-10-27T18:25:54.882Z" },
+]
+
+[[package]]
+name = "packaging"
+version = "25.0"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/a1/d4/1fc4078c65507b51b96ca8f8c3ba19e6a61c8253c72794544580a7b6c24d/packaging-25.0.tar.gz", hash = "sha256:d443872c98d677bf60f6a1f2f8c1cb748e8fe762d2bf9d3148b5599295b0fc4f", size = 165727, upload-time = "2025-04-19T11:48:59.673Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/20/12/38679034af332785aac8774540895e234f4d07f7545804097de4b666afd8/packaging-25.0-py3-none-any.whl", hash = "sha256:29572ef2b1f17581046b3a2227d5c611fb25ec70ca1ba8554b24b0e69331a484", size = 66469, upload-time = "2025-04-19T11:48:57.875Z" },
+]
+
+[[package]]
+name = "pluggy"
+version = "1.6.0"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/f9/e2/3e91f31a7d2b083fe6ef3fa267035b518369d9511ffab804f839851d2779/pluggy-1.6.0.tar.gz", hash = "sha256:7dcc130b76258d33b90f61b658791dede3486c3e6bfb003ee5c9bfb396dd22f3", size = 69412, upload-time = "2025-05-15T12:30:07.975Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/54/20/4d324d65cc6d9205fabedc306948156824eb9f0ee1633355a8f7ec5c66bf/pluggy-1.6.0-py3-none-any.whl", hash = "sha256:e920276dd6813095e9377c0bc5566d94c932c33b27a3e3945d8389c374dd4746", size = 20538, upload-time = "2025-05-15T12:30:06.134Z" },
+]
+
+[[package]]
+name = "pycryptodomex"
+version = "3.23.0"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/c9/85/e24bf90972a30b0fcd16c73009add1d7d7cd9140c2498a68252028899e41/pycryptodomex-3.23.0.tar.gz", hash = "sha256:71909758f010c82bc99b0abf4ea12012c98962fbf0583c2164f8b84533c2e4da", size = 4922157, upload-time = "2025-05-17T17:23:41.434Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/dd/9c/1a8f35daa39784ed8adf93a694e7e5dc15c23c741bbda06e1d45f8979e9e/pycryptodomex-3.23.0-cp37-abi3-macosx_10_9_universal2.whl", hash = "sha256:06698f957fe1ab229a99ba2defeeae1c09af185baa909a31a5d1f9d42b1aaed6", size = 2499240, upload-time = "2025-05-17T17:22:46.953Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/7a/62/f5221a191a97157d240cf6643747558759126c76ee92f29a3f4aee3197a5/pycryptodomex-3.23.0-cp37-abi3-macosx_10_9_x86_64.whl", hash = "sha256:b2c2537863eccef2d41061e82a881dcabb04944c5c06c5aa7110b577cc487545", size = 1644042, upload-time = "2025-05-17T17:22:49.098Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/8c/fd/5a054543c8988d4ed7b612721d7e78a4b9bf36bc3c5ad45ef45c22d0060e/pycryptodomex-3.23.0-cp37-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:43c446e2ba8df8889e0e16f02211c25b4934898384c1ec1ec04d7889c0333587", size = 2186227, upload-time = "2025-05-17T17:22:51.139Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/c8/a9/8862616a85cf450d2822dbd4fff1fcaba90877907a6ff5bc2672cafe42f8/pycryptodomex-3.23.0-cp37-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f489c4765093fb60e2edafdf223397bc716491b2b69fe74367b70d6999257a5c", size = 2272578, upload-time = "2025-05-17T17:22:53.676Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/46/9f/bda9c49a7c1842820de674ab36c79f4fbeeee03f8ff0e4f3546c3889076b/pycryptodomex-3.23.0-cp37-abi3-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:bdc69d0d3d989a1029df0eed67cc5e8e5d968f3724f4519bd03e0ec68df7543c", size = 2312166, upload-time = "2025-05-17T17:22:56.585Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/03/cc/870b9bf8ca92866ca0186534801cf8d20554ad2a76ca959538041b7a7cf4/pycryptodomex-3.23.0-cp37-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:6bbcb1dd0f646484939e142462d9e532482bc74475cecf9c4903d4e1cd21f003", size = 2185467, upload-time = "2025-05-17T17:22:59.237Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/96/e3/ce9348236d8e669fea5dd82a90e86be48b9c341210f44e25443162aba187/pycryptodomex-3.23.0-cp37-abi3-musllinux_1_2_i686.whl", hash = "sha256:8a4fcd42ccb04c31268d1efeecfccfd1249612b4de6374205376b8f280321744", size = 2346104, upload-time = "2025-05-17T17:23:02.112Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/a5/e9/e869bcee87beb89040263c416a8a50204f7f7a83ac11897646c9e71e0daf/pycryptodomex-3.23.0-cp37-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:55ccbe27f049743a4caf4f4221b166560d3438d0b1e5ab929e07ae1702a4d6fd", size = 2271038, upload-time = "2025-05-17T17:23:04.872Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/8d/67/09ee8500dd22614af5fbaa51a4aee6e342b5fa8aecf0a6cb9cbf52fa6d45/pycryptodomex-3.23.0-cp37-abi3-win32.whl", hash = "sha256:189afbc87f0b9f158386bf051f720e20fa6145975f1e76369303d0f31d1a8d7c", size = 1771969, upload-time = "2025-05-17T17:23:07.115Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/69/96/11f36f71a865dd6df03716d33bd07a67e9d20f6b8d39820470b766af323c/pycryptodomex-3.23.0-cp37-abi3-win_amd64.whl", hash = "sha256:52e5ca58c3a0b0bd5e100a9fbc8015059b05cffc6c66ce9d98b4b45e023443b9", size = 1803124, upload-time = "2025-05-17T17:23:09.267Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/f9/93/45c1cdcbeb182ccd2e144c693eaa097763b08b38cded279f0053ed53c553/pycryptodomex-3.23.0-cp37-abi3-win_arm64.whl", hash = "sha256:02d87b80778c171445d67e23d1caef279bf4b25c3597050ccd2e13970b57fd51", size = 1707161, upload-time = "2025-05-17T17:23:11.414Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/f3/b8/3e76d948c3c4ac71335bbe75dac53e154b40b0f8f1f022dfa295257a0c96/pycryptodomex-3.23.0-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:ebfff755c360d674306e5891c564a274a47953562b42fb74a5c25b8fc1fb1cb5", size = 1627695, upload-time = "2025-05-17T17:23:17.38Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/6a/cf/80f4297a4820dfdfd1c88cf6c4666a200f204b3488103d027b5edd9176ec/pycryptodomex-3.23.0-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:eca54f4bb349d45afc17e3011ed4264ef1cc9e266699874cdd1349c504e64798", size = 1675772, upload-time = "2025-05-17T17:23:19.202Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/d1/42/1e969ee0ad19fe3134b0e1b856c39bd0b70d47a4d0e81c2a8b05727394c9/pycryptodomex-3.23.0-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4f2596e643d4365e14d0879dc5aafe6355616c61c2176009270f3048f6d9a61f", size = 1668083, upload-time = "2025-05-17T17:23:21.867Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/6e/c3/1de4f7631fea8a992a44ba632aa40e0008764c0fb9bf2854b0acf78c2cf2/pycryptodomex-3.23.0-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:fdfac7cda115bca3a5abb2f9e43bc2fb66c2b65ab074913643803ca7083a79ea", size = 1706056, upload-time = "2025-05-17T17:23:24.031Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/f2/5f/af7da8e6f1e42b52f44a24d08b8e4c726207434e2593732d39e7af5e7256/pycryptodomex-3.23.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:14c37aaece158d0ace436f76a7bb19093db3b4deade9797abfc39ec6cd6cc2fe", size = 1806478, upload-time = "2025-05-17T17:23:26.066Z" },
+]
+
+[[package]]
+name = "pygments"
+version = "2.19.2"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/b0/77/a5b8c569bf593b0140bde72ea885a803b82086995367bf2037de0159d924/pygments-2.19.2.tar.gz", hash = "sha256:636cb2477cec7f8952536970bc533bc43743542f70392ae026374600add5b887", size = 4968631, upload-time = "2025-06-21T13:39:12.283Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/c7/21/705964c7812476f378728bdf590ca4b771ec72385c533964653c68e86bdc/pygments-2.19.2-py3-none-any.whl", hash = "sha256:86540386c03d588bb81d44bc3928634ff26449851e99741617ecb9037ee5ec0b", size = 1225217, upload-time = "2025-06-21T13:39:07.939Z" },
+]
+
+[[package]]
+name = "pytest"
+version = "9.0.1"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+dependencies = [
+    { name = "colorama", marker = "sys_platform == 'win32'" },
+    { name = "exceptiongroup", marker = "python_full_version < '3.11'" },
+    { name = "iniconfig" },
+    { name = "packaging" },
+    { name = "pluggy" },
+    { name = "pygments" },
+    { name = "tomli", marker = "python_full_version < '3.11'" },
+]
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/07/56/f013048ac4bc4c1d9be45afd4ab209ea62822fb1598f40687e6bf45dcea4/pytest-9.0.1.tar.gz", hash = "sha256:3e9c069ea73583e255c3b21cf46b8d3c56f6e3a1a8f6da94ccb0fcf57b9d73c8", size = 1564125, upload-time = "2025-11-12T13:05:09.333Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/0b/8b/6300fb80f858cda1c51ffa17075df5d846757081d11ab4aa35cef9e6258b/pytest-9.0.1-py3-none-any.whl", hash = "sha256:67be0030d194df2dfa7b556f2e56fb3c3315bd5c8822c6951162b92b32ce7dad", size = 373668, upload-time = "2025-11-12T13:05:07.379Z" },
+]
+
+[[package]]
+name = "ragflow-cli"
+version = "0.23.1"
+source = { virtual = "." }
+dependencies = [
+    { name = "beartype" },
+    { name = "lark" },
+    { name = "pycryptodomex" },
+    { name = "requests" },
+]
+
+[package.dev-dependencies]
+test = [
+    { name = "pytest" },
+    { name = "requests" },
+    { name = "requests-toolbelt" },
+]
+
+[package.metadata]
+requires-dist = [
+    { name = "beartype", specifier = ">=0.20.0,<1.0.0" },
+    { name = "lark", specifier = ">=1.1.0" },
+    { name = "pycryptodomex", specifier = ">=3.10.0" },
+    { name = "requests", specifier = ">=2.30.0,<3.0.0" },
+]
+
+[package.metadata.requires-dev]
+test = [
+    { name = "pytest", specifier = ">=8.3.5" },
+    { name = "requests", specifier = ">=2.32.3" },
+    { name = "requests-toolbelt", specifier = ">=1.0.0" },
+]
+
+[[package]]
+name = "requests"
+version = "2.32.5"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+dependencies = [
+    { name = "certifi" },
+    { name = "charset-normalizer" },
+    { name = "idna" },
+    { name = "urllib3" },
+]
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/c9/74/b3ff8e6c8446842c3f5c837e9c3dfcfe2018ea6ecef224c710c85ef728f4/requests-2.32.5.tar.gz", hash = "sha256:dbba0bac56e100853db0ea71b82b4dfd5fe2bf6d3754a8893c3af500cec7d7cf", size = 134517, upload-time = "2025-08-18T20:46:02.573Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/1e/db/4254e3eabe8020b458f1a747140d32277ec7a271daf1d235b70dc0b4e6e3/requests-2.32.5-py3-none-any.whl", hash = "sha256:2462f94637a34fd532264295e186976db0f5d453d1cdd31473c85a6a161affb6", size = 64738, upload-time = "2025-08-18T20:46:00.542Z" },
+]
+
+[[package]]
+name = "requests-toolbelt"
+version = "1.0.0"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+dependencies = [
+    { name = "requests" },
+]
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/f3/61/d7545dafb7ac2230c70d38d31cbfe4cc64f7144dc41f6e4e4b78ecd9f5bb/requests-toolbelt-1.0.0.tar.gz", hash = "sha256:7681a0a3d047012b5bdc0ee37d7f8f07ebe76ab08caeccfc3921ce23c88d5bc6", size = 206888, upload-time = "2023-05-01T04:11:33.229Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/3f/51/d4db610ef29373b879047326cbf6fa98b6c1969d6f6dc423279de2b1be2c/requests_toolbelt-1.0.0-py2.py3-none-any.whl", hash = "sha256:cccfdd665f0a24fcf4726e690f65639d272bb0637b9b92dfd91a5568ccf6bd06", size = 54481, upload-time = "2023-05-01T04:11:28.427Z" },
+]
+
+[[package]]
+name = "tomli"
+version = "2.3.0"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/52/ed/3f73f72945444548f33eba9a87fc7a6e969915e7b1acc8260b30e1f76a2f/tomli-2.3.0.tar.gz", hash = "sha256:64be704a875d2a59753d80ee8a533c3fe183e3f06807ff7dc2232938ccb01549", size = 17392, upload-time = "2025-10-08T22:01:47.119Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/b3/2e/299f62b401438d5fe1624119c723f5d877acc86a4c2492da405626665f12/tomli-2.3.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:88bd15eb972f3664f5ed4b57c1634a97153b4bac4479dcb6a495f41921eb7f45", size = 153236, upload-time = "2025-10-08T22:01:00.137Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/86/7f/d8fffe6a7aefdb61bced88fcb5e280cfd71e08939da5894161bd71bea022/tomli-2.3.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:883b1c0d6398a6a9d29b508c331fa56adbcdff647f6ace4dfca0f50e90dfd0ba", size = 148084, upload-time = "2025-10-08T22:01:01.63Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/47/5c/24935fb6a2ee63e86d80e4d3b58b222dafaf438c416752c8b58537c8b89a/tomli-2.3.0-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d1381caf13ab9f300e30dd8feadb3de072aeb86f1d34a8569453ff32a7dea4bf", size = 234832, upload-time = "2025-10-08T22:01:02.543Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/89/da/75dfd804fc11e6612846758a23f13271b76d577e299592b4371a4ca4cd09/tomli-2.3.0-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a0e285d2649b78c0d9027570d4da3425bdb49830a6156121360b3f8511ea3441", size = 242052, upload-time = "2025-10-08T22:01:03.836Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/70/8c/f48ac899f7b3ca7eb13af73bacbc93aec37f9c954df3c08ad96991c8c373/tomli-2.3.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:0a154a9ae14bfcf5d8917a59b51ffd5a3ac1fd149b71b47a3a104ca4edcfa845", size = 239555, upload-time = "2025-10-08T22:01:04.834Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/ba/28/72f8afd73f1d0e7829bfc093f4cb98ce0a40ffc0cc997009ee1ed94ba705/tomli-2.3.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:74bf8464ff93e413514fefd2be591c3b0b23231a77f901db1eb30d6f712fc42c", size = 245128, upload-time = "2025-10-08T22:01:05.84Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/b6/eb/a7679c8ac85208706d27436e8d421dfa39d4c914dcf5fa8083a9305f58d9/tomli-2.3.0-cp311-cp311-win32.whl", hash = "sha256:00b5f5d95bbfc7d12f91ad8c593a1659b6387b43f054104cda404be6bda62456", size = 96445, upload-time = "2025-10-08T22:01:06.896Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/0a/fe/3d3420c4cb1ad9cb462fb52967080575f15898da97e21cb6f1361d505383/tomli-2.3.0-cp311-cp311-win_amd64.whl", hash = "sha256:4dc4ce8483a5d429ab602f111a93a6ab1ed425eae3122032db7e9acf449451be", size = 107165, upload-time = "2025-10-08T22:01:08.107Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/ff/b7/40f36368fcabc518bb11c8f06379a0fd631985046c038aca08c6d6a43c6e/tomli-2.3.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:d7d86942e56ded512a594786a5ba0a5e521d02529b3826e7761a05138341a2ac", size = 154891, upload-time = "2025-10-08T22:01:09.082Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/f9/3f/d9dd692199e3b3aab2e4e4dd948abd0f790d9ded8cd10cbaae276a898434/tomli-2.3.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:73ee0b47d4dad1c5e996e3cd33b8a76a50167ae5f96a2607cbe8cc773506ab22", size = 148796, upload-time = "2025-10-08T22:01:10.266Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/60/83/59bff4996c2cf9f9387a0f5a3394629c7efa5ef16142076a23a90f1955fa/tomli-2.3.0-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:792262b94d5d0a466afb5bc63c7daa9d75520110971ee269152083270998316f", size = 242121, upload-time = "2025-10-08T22:01:11.332Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/45/e5/7c5119ff39de8693d6baab6c0b6dcb556d192c165596e9fc231ea1052041/tomli-2.3.0-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4f195fe57ecceac95a66a75ac24d9d5fbc98ef0962e09b2eddec5d39375aae52", size = 250070, upload-time = "2025-10-08T22:01:12.498Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/45/12/ad5126d3a278f27e6701abde51d342aa78d06e27ce2bb596a01f7709a5a2/tomli-2.3.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:e31d432427dcbf4d86958c184b9bfd1e96b5b71f8eb17e6d02531f434fd335b8", size = 245859, upload-time = "2025-10-08T22:01:13.551Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/fb/a1/4d6865da6a71c603cfe6ad0e6556c73c76548557a8d658f9e3b142df245f/tomli-2.3.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:7b0882799624980785240ab732537fcfc372601015c00f7fc367c55308c186f6", size = 250296, upload-time = "2025-10-08T22:01:14.614Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/a0/b7/a7a7042715d55c9ba6e8b196d65d2cb662578b4d8cd17d882d45322b0d78/tomli-2.3.0-cp312-cp312-win32.whl", hash = "sha256:ff72b71b5d10d22ecb084d345fc26f42b5143c5533db5e2eaba7d2d335358876", size = 97124, upload-time = "2025-10-08T22:01:15.629Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/06/1e/f22f100db15a68b520664eb3328fb0ae4e90530887928558112c8d1f4515/tomli-2.3.0-cp312-cp312-win_amd64.whl", hash = "sha256:1cb4ed918939151a03f33d4242ccd0aa5f11b3547d0cf30f7c74a408a5b99878", size = 107698, upload-time = "2025-10-08T22:01:16.51Z" },
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/77/b8/0135fadc89e73be292b473cb820b4f5a08197779206b33191e801feeae40/tomli-2.3.0-py3-none-any.whl", hash = "sha256:e95b1af3c5b07d9e643909b5abbec77cd9f1217e6d0bca72b0234736b9fb1f1b", size = 14408, upload-time = "2025-10-08T22:01:46.04Z" },
+]
+
+[[package]]
+name = "typing-extensions"
+version = "4.15.0"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/72/94/1a15dd82efb362ac84269196e94cf00f187f7ed21c242792a923cdb1c61f/typing_extensions-4.15.0.tar.gz", hash = "sha256:0cea48d173cc12fa28ecabc3b837ea3cf6f38c6d1136f85cbaaf598984861466", size = 109391, upload-time = "2025-08-25T13:49:26.313Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/18/67/36e9267722cc04a6b9f15c7f3441c2363321a3ea07da7ae0c0707beb2a9c/typing_extensions-4.15.0-py3-none-any.whl", hash = "sha256:f0fa19c6845758ab08074a0cfa8b7aecb71c999ca73d62883bc25cc018c4e548", size = 44614, upload-time = "2025-08-25T13:49:24.86Z" },
+]
+
+[[package]]
+name = "urllib3"
+version = "2.5.0"
+source = { registry = "https://pypi.tuna.tsinghua.edu.cn/simple" }
+sdist = { url = "https://pypi.tuna.tsinghua.edu.cn/packages/15/22/9ee70a2574a4f4599c47dd506532914ce044817c7752a79b6a51286319bc/urllib3-2.5.0.tar.gz", hash = "sha256:3fc47733c7e419d4bc3f6b3dc2b4f890bb743906a30d56ba4a5bfa4bbff92760", size = 393185, upload-time = "2025-06-18T14:07:41.644Z" }
+wheels = [
+    { url = "https://pypi.tuna.tsinghua.edu.cn/packages/a7/c2/fe1e52489ae3122415c51f387e221dd0773709bad6c6cdaa599e8a2c5185/urllib3-2.5.0-py3-none-any.whl", hash = "sha256:e6b01673c0fa6a13e374b50871808eb3bf7046c4b125b216f6bf1cc604cff0dc", size = 129795, upload-time = "2025-06-18T14:07:40.39Z" },
+]

admin/server/admin_server.py

@@ -14,12 +14,15 @@
 #  limitations under the License.
 #
 
+import time
+start_ts = time.time()
+
 import os
 import signal
 import logging
-import time
 import threading
 import traceback
+import faulthandler
 
 from flask import Flask
 from flask_login import LoginManager
@@ -37,6 +40,7 @@ from common.versions import get_ragflow_version
 stop_event = threading.Event()
 
 if __name__ == '__main__':
+    faulthandler.enable()
     init_root_logger("admin_service")
     logging.info(r"""
         ____  ___   ______________                 ___       __          _     
@@ -64,7 +68,7 @@ if __name__ == '__main__':
     SERVICE_CONFIGS.configs = load_configurations(SERVICE_CONF)
 
     try:
-        logging.info("RAGFlow Admin service start...")
+        logging.info(f"RAGFlow admin is ready after {time.time() - start_ts}s initialization.")
         run_simple(
             hostname="0.0.0.0",
             port=9381,

admin/server/auth.py

@@ -176,11 +176,11 @@ def login_verify(f):
                     "message": "Access denied",
                     "data": None
                 }), 200
-        except Exception as e:
-            error_msg = str(e)
+        except Exception:
+            logging.exception("An error occurred during admin login verification.")
             return jsonify({
                 "code": 500,
-                "message": error_msg
+                "message": "An internal server error occurred."
             }), 200
 
         return f(*args, **kwargs)

admin/server/routes.py

@@ -15,24 +15,33 @@
 #
 
 import secrets
+from typing import Any
 
-from flask import Blueprint, request
+from common.time_utils import current_timestamp, datetime_format
+from datetime import datetime
+from flask import Blueprint, Response, request
 from flask_login import current_user, login_required, logout_user
 
 from auth import login_verify, login_admin, check_admin_auth
 from responses import success_response, error_response
-from services import UserMgr, ServiceMgr, UserServiceMgr
+from services import UserMgr, ServiceMgr, UserServiceMgr, SettingsMgr, ConfigMgr, EnvironmentsMgr
 from roles import RoleMgr
 from api.common.exceptions import AdminException
 from common.versions import get_ragflow_version
+from api.utils.api_utils import generate_confirmation_token
 
-admin_bp = Blueprint('admin', __name__, url_prefix='/api/v1/admin')
+admin_bp = Blueprint("admin", __name__, url_prefix="/api/v1/admin")
 
 
-@admin_bp.route('/login', methods=['POST'])
+@admin_bp.route("/ping", methods=["GET"])
+def ping():
+    return success_response("PONG")
+
+
+@admin_bp.route("/login", methods=["POST"])
 def login():
     if not request.json:
-        return error_response('Authorize admin failed.' ,400)
+        return error_response("Authorize admin failed.", 400)
     try:
         email = request.json.get("email", "")
         password = request.json.get("password", "")
@@ -41,7 +50,7 @@ def login():
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/logout', methods=['GET'])
+@admin_bp.route("/logout", methods=["GET"])
 @login_required
 def logout():
     try:
@@ -53,7 +62,7 @@ def logout():
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/auth', methods=['GET'])
+@admin_bp.route("/auth", methods=["GET"])
 @login_verify
 def auth_admin():
     try:
@@ -62,7 +71,7 @@ def auth_admin():
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/users', methods=['GET'])
+@admin_bp.route("/users", methods=["GET"])
 @login_required
 @check_admin_auth
 def list_users():
@@ -73,18 +82,18 @@ def list_users():
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/users', methods=['POST'])
+@admin_bp.route("/users", methods=["POST"])
 @login_required
 @check_admin_auth
 def create_user():
     try:
         data = request.get_json()
-        if not data or 'username' not in data or 'password' not in data:
+        if not data or "username" not in data or "password" not in data:
             return error_response("Username and password are required", 400)
 
-        username = data['username']
-        password = data['password']
-        role = data.get('role', 'user')
+        username = data["username"]
+        password = data["password"]
+        role = data.get("role", "user")
 
         res = UserMgr.create_user(username, password, role)
         if res["success"]:
@@ -100,7 +109,7 @@ def create_user():
         return error_response(str(e))
 
 
-@admin_bp.route('/users/<username>', methods=['DELETE'])
+@admin_bp.route("/users/<username>", methods=["DELETE"])
 @login_required
 @check_admin_auth
 def delete_user(username):
@@ -117,16 +126,16 @@ def delete_user(username):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/users/<username>/password', methods=['PUT'])
+@admin_bp.route("/users/<username>/password", methods=["PUT"])
 @login_required
 @check_admin_auth
 def change_password(username):
     try:
         data = request.get_json()
-        if not data or 'new_password' not in data:
+        if not data or "new_password" not in data:
             return error_response("New password is required", 400)
 
-        new_password = data['new_password']
+        new_password = data["new_password"]
         msg = UserMgr.update_user_password(username, new_password)
         return success_response(None, msg)
 
@@ -136,15 +145,15 @@ def change_password(username):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/users/<username>/activate', methods=['PUT'])
+@admin_bp.route("/users/<username>/activate", methods=["PUT"])
 @login_required
 @check_admin_auth
 def alter_user_activate_status(username):
     try:
         data = request.get_json()
-        if not data or 'activate_status' not in data:
+        if not data or "activate_status" not in data:
             return error_response("Activation status is required", 400)
-        activate_status = data['activate_status']
+        activate_status = data["activate_status"]
         msg = UserMgr.update_user_activate_status(username, activate_status)
         return success_response(None, msg)
     except AdminException as e:
@@ -153,7 +162,39 @@ def alter_user_activate_status(username):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/users/<username>', methods=['GET'])
+@admin_bp.route("/users/<username>/admin", methods=["PUT"])
+@login_required
+@check_admin_auth
+def grant_admin(username):
+    try:
+        if current_user.email == username:
+            return error_response(f"can't grant current user: {username}", 409)
+        msg = UserMgr.grant_admin(username)
+        return success_response(None, msg)
+
+    except AdminException as e:
+        return error_response(e.message, e.code)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route("/users/<username>/admin", methods=["DELETE"])
+@login_required
+@check_admin_auth
+def revoke_admin(username):
+    try:
+        if current_user.email == username:
+            return error_response(f"can't grant current user: {username}", 409)
+        msg = UserMgr.revoke_admin(username)
+        return success_response(None, msg)
+
+    except AdminException as e:
+        return error_response(e.message, e.code)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route("/users/<username>", methods=["GET"])
 @login_required
 @check_admin_auth
 def get_user_details(username):
@@ -167,7 +208,7 @@ def get_user_details(username):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/users/<username>/datasets', methods=['GET'])
+@admin_bp.route("/users/<username>/datasets", methods=["GET"])
 @login_required
 @check_admin_auth
 def get_user_datasets(username):
@@ -181,7 +222,7 @@ def get_user_datasets(username):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/users/<username>/agents', methods=['GET'])
+@admin_bp.route("/users/<username>/agents", methods=["GET"])
 @login_required
 @check_admin_auth
 def get_user_agents(username):
@@ -195,7 +236,7 @@ def get_user_agents(username):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/services', methods=['GET'])
+@admin_bp.route("/services", methods=["GET"])
 @login_required
 @check_admin_auth
 def get_services():
@@ -206,7 +247,7 @@ def get_services():
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/service_types/<service_type>', methods=['GET'])
+@admin_bp.route("/service_types/<service_type>", methods=["GET"])
 @login_required
 @check_admin_auth
 def get_services_by_type(service_type_str):
@@ -217,7 +258,7 @@ def get_services_by_type(service_type_str):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/services/<service_id>', methods=['GET'])
+@admin_bp.route("/services/<service_id>", methods=["GET"])
 @login_required
 @check_admin_auth
 def get_service(service_id):
@@ -228,7 +269,7 @@ def get_service(service_id):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/services/<service_id>', methods=['DELETE'])
+@admin_bp.route("/services/<service_id>", methods=["DELETE"])
 @login_required
 @check_admin_auth
 def shutdown_service(service_id):
@@ -239,7 +280,7 @@ def shutdown_service(service_id):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/services/<service_id>', methods=['PUT'])
+@admin_bp.route("/services/<service_id>", methods=["PUT"])
 @login_required
 @check_admin_auth
 def restart_service(service_id):
@@ -250,38 +291,38 @@ def restart_service(service_id):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/roles', methods=['POST'])
+@admin_bp.route("/roles", methods=["POST"])
 @login_required
 @check_admin_auth
 def create_role():
     try:
         data = request.get_json()
-        if not data or 'role_name' not in data:
+        if not data or "role_name" not in data:
             return error_response("Role name is required", 400)
-        role_name: str = data['role_name']
-        description: str = data['description']
+        role_name: str = data["role_name"]
+        description: str = data["description"]
         res = RoleMgr.create_role(role_name, description)
         return success_response(res)
     except Exception as e:
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/roles/<role_name>', methods=['PUT'])
+@admin_bp.route("/roles/<role_name>", methods=["PUT"])
 @login_required
 @check_admin_auth
 def update_role(role_name: str):
     try:
         data = request.get_json()
-        if not data or 'description' not in data:
+        if not data or "description" not in data:
             return error_response("Role description is required", 400)
-        description: str = data['description']
+        description: str = data["description"]
         res = RoleMgr.update_role_description(role_name, description)
         return success_response(res)
     except Exception as e:
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/roles/<role_name>', methods=['DELETE'])
+@admin_bp.route("/roles/<role_name>", methods=["DELETE"])
 @login_required
 @check_admin_auth
 def delete_role(role_name: str):
@@ -292,7 +333,7 @@ def delete_role(role_name: str):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/roles', methods=['GET'])
+@admin_bp.route("/roles", methods=["GET"])
 @login_required
 @check_admin_auth
 def list_roles():
@@ -303,7 +344,7 @@ def list_roles():
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/roles/<role_name>/permission', methods=['GET'])
+@admin_bp.route("/roles/<role_name>/permission", methods=["GET"])
 @login_required
 @check_admin_auth
 def get_role_permission(role_name: str):
@@ -314,54 +355,54 @@ def get_role_permission(role_name: str):
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/roles/<role_name>/permission', methods=['POST'])
+@admin_bp.route("/roles/<role_name>/permission", methods=["POST"])
 @login_required
 @check_admin_auth
 def grant_role_permission(role_name: str):
     try:
         data = request.get_json()
-        if not data or 'actions' not in data or 'resource' not in data:
+        if not data or "actions" not in data or "resource" not in data:
             return error_response("Permission is required", 400)
-        actions: list = data['actions']
-        resource: str = data['resource']
+        actions: list = data["actions"]
+        resource: str = data["resource"]
         res = RoleMgr.grant_role_permission(role_name, actions, resource)
         return success_response(res)
     except Exception as e:
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/roles/<role_name>/permission', methods=['DELETE'])
+@admin_bp.route("/roles/<role_name>/permission", methods=["DELETE"])
 @login_required
 @check_admin_auth
 def revoke_role_permission(role_name: str):
     try:
         data = request.get_json()
-        if not data or 'actions' not in data or 'resource' not in data:
+        if not data or "actions" not in data or "resource" not in data:
             return error_response("Permission is required", 400)
-        actions: list = data['actions']
-        resource: str = data['resource']
+        actions: list = data["actions"]
+        resource: str = data["resource"]
         res = RoleMgr.revoke_role_permission(role_name, actions, resource)
         return success_response(res)
     except Exception as e:
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/users/<user_name>/role', methods=['PUT'])
+@admin_bp.route("/users/<user_name>/role", methods=["PUT"])
 @login_required
 @check_admin_auth
 def update_user_role(user_name: str):
     try:
         data = request.get_json()
-        if not data or 'role_name' not in data:
+        if not data or "role_name" not in data:
             return error_response("Role name is required", 400)
-        role_name: str = data['role_name']
+        role_name: str = data["role_name"]
         res = RoleMgr.update_user_role(user_name, role_name)
         return success_response(res)
     except Exception as e:
         return error_response(str(e), 500)
 
 
-@admin_bp.route('/users/<user_name>/permission', methods=['GET'])
+@admin_bp.route("/users/<user_name>/permission", methods=["GET"])
 @login_required
 @check_admin_auth
 def get_user_permission(user_name: str):
@@ -371,7 +412,140 @@ def get_user_permission(user_name: str):
     except Exception as e:
         return error_response(str(e), 500)
 
-@admin_bp.route('/version', methods=['GET'])
+
+@admin_bp.route("/variables", methods=["PUT"])
+@login_required
+@check_admin_auth
+def set_variable():
+    try:
+        data = request.get_json()
+        if not data and "var_name" not in data:
+            return error_response("Var name is required", 400)
+
+        if "var_value" not in data:
+            return error_response("Var value is required", 400)
+        var_name: str = data["var_name"]
+        var_value: str = data["var_value"]
+
+        SettingsMgr.update_by_name(var_name, var_value)
+        return success_response(None, "Set variable successfully")
+    except AdminException as e:
+        return error_response(str(e), 400)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route("/variables", methods=["GET"])
+@login_required
+@check_admin_auth
+def get_variable():
+    try:
+        if request.content_length is None or request.content_length == 0:
+            # list variables
+            res = list(SettingsMgr.get_all())
+            return success_response(res)
+
+        # get var
+        data = request.get_json()
+        if not data and "var_name" not in data:
+            return error_response("Var name is required", 400)
+        var_name: str = data["var_name"]
+        res = SettingsMgr.get_by_name(var_name)
+        return success_response(res)
+    except AdminException as e:
+        return error_response(str(e), 400)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route("/configs", methods=["GET"])
+@login_required
+@check_admin_auth
+def get_config():
+    try:
+        res = list(ConfigMgr.get_all())
+        return success_response(res)
+    except AdminException as e:
+        return error_response(str(e), 400)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route("/environments", methods=["GET"])
+@login_required
+@check_admin_auth
+def get_environments():
+    try:
+        res = list(EnvironmentsMgr.get_all())
+        return success_response(res)
+    except AdminException as e:
+        return error_response(str(e), 400)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route("/users/<username>/keys", methods=["POST"])
+@login_required
+@check_admin_auth
+def generate_user_api_key(username: str) -> tuple[Response, int]:
+    try:
+        user_details: list[dict[str, Any]] = UserMgr.get_user_details(username)
+        if not user_details:
+            return error_response("User not found!", 404)
+        tenants: list[dict[str, Any]] = UserServiceMgr.get_user_tenants(username)
+        if not tenants:
+            return error_response("Tenant not found!", 404)
+        tenant_id: str = tenants[0]["tenant_id"]
+        key: str = generate_confirmation_token()
+        obj: dict[str, Any] = {
+            "tenant_id": tenant_id,
+            "token": key,
+            "beta": generate_confirmation_token().replace("ragflow-", "")[:32],
+            "create_time": current_timestamp(),
+            "create_date": datetime_format(datetime.now()),
+            "update_time": None,
+            "update_date": None,
+        }
+
+        if not UserMgr.save_api_key(obj):
+            return error_response("Failed to generate API key!", 500)
+        return success_response(obj, "API key generated successfully")
+    except AdminException as e:
+        return error_response(e.message, e.code)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route("/users/<username>/keys", methods=["GET"])
+@login_required
+@check_admin_auth
+def get_user_api_keys(username: str) -> tuple[Response, int]:
+    try:
+        api_keys: list[dict[str, Any]] = UserMgr.get_user_api_key(username)
+        return success_response(api_keys, "Get user API keys")
+    except AdminException as e:
+        return error_response(e.message, e.code)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route("/users/<username>/keys/<key>", methods=["DELETE"])
+@login_required
+@check_admin_auth
+def delete_user_api_key(username: str, key: str) -> tuple[Response, int]:
+    try:
+        deleted = UserMgr.delete_api_key(username, key)
+        if deleted:
+            return success_response(None, "API key deleted successfully")
+        else:
+            return error_response("API key not found or could not be deleted", 404)
+    except AdminException as e:
+        return error_response(e.message, e.code)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route("/version", methods=["GET"])
 @login_required
 @check_admin_auth
 def show_version():

admin/server/services.py

@@ -13,15 +13,22 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+
+import os
 import logging
 import re
+from typing import Any
+
 from werkzeug.security import check_password_hash
 from common.constants import ActiveEnum
 from api.db.services import UserService
 from api.db.joint_services.user_account_service import create_new_user, delete_user_data
 from api.db.services.canvas_service import UserCanvasService
-from api.db.services.user_service import TenantService
+from api.db.services.user_service import TenantService, UserTenantService
 from api.db.services.knowledgebase_service import KnowledgebaseService
+from api.db.services.system_settings_service import SystemSettingsService
+from api.db.services.api_service import APITokenService
+from api.db.db_models import APIToken
 from api.utils.crypt import decrypt
 from api.utils import health_utils
 
@@ -35,13 +42,15 @@ class UserMgr:
         users = UserService.get_all_users()
         result = []
         for user in users:
-            result.append({
-                'email': user.email,
-                'nickname': user.nickname,
-                'create_date': user.create_date,
-                'is_active': user.is_active,
-                'is_superuser': user.is_superuser,
-            })
+            result.append(
+                {
+                    "email": user.email,
+                    "nickname": user.nickname,
+                    "create_date": user.create_date,
+                    "is_active": user.is_active,
+                    "is_superuser": user.is_superuser,
+                }
+            )
         return result
 
     @staticmethod
@@ -50,19 +59,21 @@ class UserMgr:
         users = UserService.query_user_by_email(username)
         result = []
         for user in users:
-            result.append({
-                'avatar': user.avatar,
-                'email': user.email,
-                'language': user.language,
-                'last_login_time': user.last_login_time,
-                'is_active': user.is_active,
-                'is_anonymous': user.is_anonymous,
-                'login_channel': user.login_channel,
-                'status': user.status,
-                'is_superuser': user.is_superuser,
-                'create_date': user.create_date,
-                'update_date': user.update_date
-            })
+            result.append(
+                {
+                    "avatar": user.avatar,
+                    "email": user.email,
+                    "language": user.language,
+                    "last_login_time": user.last_login_time,
+                    "is_active": user.is_active,
+                    "is_anonymous": user.is_anonymous,
+                    "login_channel": user.login_channel,
+                    "status": user.status,
+                    "is_superuser": user.is_superuser,
+                    "create_date": user.create_date,
+                    "update_date": user.update_date,
+                }
+            )
         return result
 
     @staticmethod
@@ -124,8 +135,8 @@ class UserMgr:
         # format activate_status before handle
         _activate_status = activate_status.lower()
         target_status = {
-            'on': ActiveEnum.ACTIVE.value,
-            'off': ActiveEnum.INACTIVE.value,
+            "on": ActiveEnum.ACTIVE.value,
+            "off": ActiveEnum.INACTIVE.value,
         }.get(_activate_status)
         if not target_status:
             raise AdminException(f"Invalid activate_status: {activate_status}")
@@ -135,9 +146,84 @@ class UserMgr:
         UserService.update_user(usr.id, {"is_active": target_status})
         return f"Turn {_activate_status} user activate status successfully!"
 
+    @staticmethod
+    def get_user_api_key(username: str) -> list[dict[str, Any]]:
+        # use email to find user. check exist and unique.
+        user_list: list[Any] = UserService.query_user_by_email(username)
+        if not user_list:
+            raise UserNotFoundError(username)
+        elif len(user_list) > 1:
+            raise AdminException(f"More than one user with username '{username}' found!")
+
+        usr: Any = user_list[0]
+        # tenant_id is typically the same as user_id for the owner tenant
+        tenant_id: str = usr.id
+
+        # Query all API keys for this tenant
+        api_keys: Any = APITokenService.query(tenant_id=tenant_id)
+
+        result: list[dict[str, Any]] = []
+        for key in api_keys:
+            result.append(key.to_dict())
+
+        return result
+
+    @staticmethod
+    def save_api_key(api_key: dict[str, Any]) -> bool:
+        return APITokenService.save(**api_key)
+
+    @staticmethod
+    def delete_api_key(username: str, key: str) -> bool:
+        # use email to find user. check exist and unique.
+        user_list: list[Any] = UserService.query_user_by_email(username)
+        if not user_list:
+            raise UserNotFoundError(username)
+        elif len(user_list) > 1:
+            raise AdminException(f"Exist more than 1 user: {username}!")
+
+        usr: Any = user_list[0]
+        # tenant_id is typically the same as user_id for the owner tenant
+        tenant_id: str = usr.id
+
+        # Delete the API key
+        deleted_count: int = APITokenService.filter_delete([APIToken.tenant_id == tenant_id, APIToken.token == key])
+        return deleted_count > 0
+
+    @staticmethod
+    def grant_admin(username: str):
+        # use email to find user. check exist and unique.
+        user_list = UserService.query_user_by_email(username)
+        if not user_list:
+            raise UserNotFoundError(username)
+        elif len(user_list) > 1:
+            raise AdminException(f"Exist more than 1 user: {username}!")
+
+        # check activate status different from new
+        usr = user_list[0]
+        if usr.is_superuser:
+            return f"{usr} is already superuser!"
+        # update is_active
+        UserService.update_user(usr.id, {"is_superuser": True})
+        return "Grant successfully!"
+
+    @staticmethod
+    def revoke_admin(username: str):
+        # use email to find user. check exist and unique.
+        user_list = UserService.query_user_by_email(username)
+        if not user_list:
+            raise UserNotFoundError(username)
+        elif len(user_list) > 1:
+            raise AdminException(f"Exist more than 1 user: {username}!")
+        # check activate status different from new
+        usr = user_list[0]
+        if not usr.is_superuser:
+            return f"{usr} isn't superuser, yet!"
+        # update is_active
+        UserService.update_user(usr.id, {"is_superuser": False})
+        return "Revoke successfully!"
+
 
 class UserServiceMgr:
-
     @staticmethod
     def get_user_datasets(username):
         # use email to find user.
@@ -167,35 +253,43 @@ class UserServiceMgr:
         tenant_ids = [m["tenant_id"] for m in tenants]
         # filter permitted agents and owned agents
         res = UserCanvasService.get_all_agents_by_tenant_ids(tenant_ids, usr.id)
-        return [{
-            'title': r['title'],
-            'permission': r['permission'],
-            'canvas_category': r['canvas_category'].split('_')[0],
-            'avatar': r['avatar']
-        } for r in res]
+        return [{"title": r["title"], "permission": r["permission"], "canvas_category": r["canvas_category"].split("_")[0], "avatar": r["avatar"]} for r in res]
+
+    @staticmethod
+    def get_user_tenants(email: str) -> list[dict[str, Any]]:
+        users: list[Any] = UserService.query_user_by_email(email)
+        if not users:
+            raise UserNotFoundError(email)
+        user: Any = users[0]
+
+        tenants: list[dict[str, Any]] = UserTenantService.get_tenants_by_user_id(user.id)
+        return tenants
 
 
 class ServiceMgr:
-
     @staticmethod
     def get_all_services():
+        doc_engine = os.getenv("DOC_ENGINE", "elasticsearch")
         result = []
         configs = SERVICE_CONFIGS.configs
         for service_id, config in enumerate(configs):
             config_dict = config.to_dict()
+            if config_dict["service_type"] == "retrieval":
+                if config_dict["extra"]["retrieval_type"] != doc_engine:
+                    continue
             try:
                 service_detail = ServiceMgr.get_service_details(service_id)
                 if "status" in service_detail:
-                    config_dict['status'] = service_detail['status']
+                    config_dict["status"] = service_detail["status"]
                 else:
-                    config_dict['status'] = 'timeout'
+                    config_dict["status"] = "timeout"
             except Exception as e:
                 logging.warning(f"Can't get service details, error: {e}")
-                config_dict['status'] = 'timeout'
-            if not config_dict['host']:
-                config_dict['host'] = '-'
-            if not config_dict['port']:
-                config_dict['port'] = '-'
+                config_dict["status"] = "timeout"
+            if not config_dict["host"]:
+                config_dict["host"] = "-"
+            if not config_dict["port"]:
+                config_dict["port"] = "-"
             result.append(config_dict)
         return result
 
@@ -211,11 +305,18 @@ class ServiceMgr:
             raise AdminException(f"invalid service_index: {service_idx}")
 
         service_config = configs[service_idx]
-        service_info = {'name': service_config.name, 'detail_func_name': service_config.detail_func_name}
 
-        detail_func = getattr(health_utils, service_info.get('detail_func_name'))
+        # exclude retrieval service if retrieval_type is not matched
+        doc_engine = os.getenv("DOC_ENGINE", "elasticsearch")
+        if service_config.service_type == "retrieval":
+            if service_config.retrieval_type != doc_engine:
+                raise AdminException(f"invalid service_index: {service_idx}")
+
+        service_info = {"name": service_config.name, "detail_func_name": service_config.detail_func_name}
+
+        detail_func = getattr(health_utils, service_info.get("detail_func_name"))
         res = detail_func()
-        res.update({'service_name': service_info.get('name')})
+        res.update({"service_name": service_info.get("name")})
         return res
 
     @staticmethod
@@ -225,3 +326,84 @@ class ServiceMgr:
     @staticmethod
     def restart_service(service_id: int):
         raise AdminException("restart_service: not implemented")
+
+
+class SettingsMgr:
+    @staticmethod
+    def get_all():
+        settings = SystemSettingsService.get_all()
+        result = []
+        for setting in settings:
+            result.append(
+                {
+                    "name": setting.name,
+                    "source": setting.source,
+                    "data_type": setting.data_type,
+                    "value": setting.value,
+                }
+            )
+        return result
+
+    @staticmethod
+    def get_by_name(name: str):
+        settings = SystemSettingsService.get_by_name(name)
+        if len(settings) == 0:
+            raise AdminException(f"Can't get setting: {name}")
+        result = []
+        for setting in settings:
+            result.append(
+                {
+                    "name": setting.name,
+                    "source": setting.source,
+                    "data_type": setting.data_type,
+                    "value": setting.value,
+                }
+            )
+        return result
+
+    @staticmethod
+    def update_by_name(name: str, value: str):
+        settings = SystemSettingsService.get_by_name(name)
+        if len(settings) == 1:
+            setting = settings[0]
+            setting.value = value
+            setting_dict = setting.to_dict()
+            SystemSettingsService.update_by_name(name, setting_dict)
+        elif len(settings) > 1:
+            raise AdminException(f"Can't update more than 1 setting: {name}")
+        else:
+            raise AdminException(f"No setting: {name}")
+
+
+class ConfigMgr:
+    @staticmethod
+    def get_all():
+        result = []
+        configs = SERVICE_CONFIGS.configs
+        for config in configs:
+            config_dict = config.to_dict()
+            result.append(config_dict)
+        return result
+
+
+class EnvironmentsMgr:
+    @staticmethod
+    def get_all():
+        result = []
+
+        env_kv = {"env": "DOC_ENGINE", "value": os.getenv("DOC_ENGINE")}
+        result.append(env_kv)
+
+        env_kv = {"env": "DEFAULT_SUPERUSER_EMAIL", "value": os.getenv("DEFAULT_SUPERUSER_EMAIL", "admin@ragflow.io")}
+        result.append(env_kv)
+
+        env_kv = {"env": "DB_TYPE", "value": os.getenv("DB_TYPE", "mysql")}
+        result.append(env_kv)
+
+        env_kv = {"env": "DEVICE", "value": os.getenv("DEVICE", "cpu")}
+        result.append(env_kv)
+
+        env_kv = {"env": "STORAGE_IMPL", "value": os.getenv("STORAGE_IMPL", "MINIO")}
+        result.append(env_kv)
+
+        return result

agent/__init__.py

@@ -13,6 +13,3 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-
-from beartype.claw import beartype_this_package
-beartype_this_package()

agent/canvas.py

@@ -13,7 +13,10 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import asyncio
 import base64
+import inspect
+import binascii
 import json
 import logging
 import re
@@ -25,7 +28,10 @@ from typing import Any, Union, Tuple
 
 from agent.component import component_class
 from agent.component.base import ComponentBase
+from api.db.services.file_service import FileService
+from api.db.services.llm_service import LLMBundle
 from api.db.services.task_service import has_canceled
+from common.constants import LLMType
 from common.misc_utils import get_uuid, hash_str2int
 from common.exceptions import TaskCanceledException
 from rag.prompts.generator import chunks_format
@@ -79,14 +85,12 @@ class Graph:
         self.dsl = json.loads(dsl)
         self._tenant_id = tenant_id
         self.task_id = task_id if task_id else get_uuid()
+        self._thread_pool = ThreadPoolExecutor(max_workers=5)
         self.load()
 
     def load(self):
         self.components = self.dsl["components"]
         cpn_nms = set([])
-        for k, cpn in self.components.items():
-            cpn_nms.add(cpn["obj"]["component_name"])
-
         for k, cpn in self.components.items():
             cpn_nms.add(cpn["obj"]["component_name"])
             param = component_class(cpn["obj"]["component_name"] + "Param")()
@@ -156,7 +160,7 @@ class Graph:
         return self._tenant_id
 
     def get_value_with_variable(self,value: str) -> Any:
-        pat = re.compile(r"\{* *\{([a-zA-Z:0-9]+@[A-Za-z0-9_.]+|sys\.[A-Za-z0-9_.]+|env\.[A-Za-z0-9_.]+)\} *\}*")
+        pat = re.compile(r"\{* *\{([a-zA-Z:0-9]+@[A-Za-z0-9_.-]+|sys\.[A-Za-z0-9_.]+|env\.[A-Za-z0-9_.]+)\} *\}*")
         out_parts = []
         last = 0
 
@@ -206,17 +210,28 @@ class Graph:
         for key in path.split('.'):
             if cur is None:
                 return None
+
             if isinstance(cur, str):
                 try:
                     cur = json.loads(cur)
                 except Exception:
                     return None
+
             if isinstance(cur, dict):
                 cur = cur.get(key)
-            else:
-                cur = getattr(cur, key, None)
+                continue
+
+            if isinstance(cur, (list, tuple)):
+                try:
+                    idx = int(key)
+                    cur = cur[idx]
+                except Exception:
+                    return None
+                continue
+
+            cur = getattr(cur, key, None)
         return cur
-    
+
     def set_variable_value(self, exp: str,value):
         exp = exp.strip("{").strip("}").strip(" ").strip("{").strip("}")
         if exp.find("@") < 0:
@@ -263,14 +278,16 @@ class Graph:
 
 class Canvas(Graph):
 
-    def __init__(self, dsl: str, tenant_id=None, task_id=None):
+    def __init__(self, dsl: str, tenant_id=None, task_id=None, canvas_id=None):
         self.globals = {
             "sys.query": "",
             "sys.user_id": tenant_id,
             "sys.conversation_turns": 0,
             "sys.files": []
         }
+        self.variables = {}
         super().__init__(dsl, tenant_id, task_id)
+        self._id = canvas_id
 
     def load(self):
         super().load()
@@ -284,6 +301,10 @@ class Canvas(Graph):
             "sys.conversation_turns": 0,
             "sys.files": []
         }
+        if "variables" in self.dsl:
+            self.variables = self.dsl["variables"]
+        else:
+            self.variables = {}
 
         self.retrieval = self.dsl["retrieval"]
         self.memory = self.dsl.get("memory", [])
@@ -300,8 +321,9 @@ class Canvas(Graph):
             self.history = []
             self.retrieval = []
             self.memory = []
+        print(self.variables)
         for k in self.globals.keys():
-            if k.startswith("sys.") or k.startswith("env."):
+            if k.startswith("sys."):
                 if isinstance(self.globals[k], str):
                     self.globals[k] = ""
                 elif isinstance(self.globals[k], int):
@@ -314,9 +336,31 @@ class Canvas(Graph):
                     self.globals[k] = {}
                 else:
                     self.globals[k] = None
+            if k.startswith("env."):
+                key = k[4:]
+                if key in self.variables:
+                    variable = self.variables[key]
+                    if variable["value"]:
+                        self.globals[k] = variable["value"]
+                    else:
+                        if variable["type"] == "string":
+                            self.globals[k] = ""
+                        elif variable["type"] == "number":
+                            self.globals[k] = 0
+                        elif variable["type"] == "boolean":
+                            self.globals[k] = False
+                        elif variable["type"] == "object":
+                            self.globals[k] = {}
+                        elif variable["type"].startswith("array"):
+                            self.globals[k] = []
+                        else:
+                            self.globals[k] = ""
+                else:
+                    self.globals[k] = ""
 
     async def run(self, **kwargs):
         st = time.perf_counter()
+        self._loop = asyncio.get_running_loop()
         self.message_id = get_uuid()
         created_at = int(time.time())
         self.add_user_input(kwargs.get("query"))
@@ -325,14 +369,19 @@ class Canvas(Graph):
 
         if kwargs.get("webhook_payload"):
             for k, cpn in self.components.items():
-                if self.components[k]["obj"].component_name.lower() == "webhook":
-                    for kk, vv in kwargs["webhook_payload"].items():
+                if self.components[k]["obj"].component_name.lower() == "begin"  and self.components[k]["obj"]._param.mode == "Webhook":
+                    payload = kwargs.get("webhook_payload", {})
+                    if "input" in payload:
+                        self.components[k]["obj"].set_input_value("request", payload["input"])
+                    for kk, vv in payload.items():
+                        if kk == "input":
+                            continue
                         self.components[k]["obj"].set_output(kk, vv)
 
         for k in kwargs.keys():
             if k in ["query", "user_id", "files"] and kwargs[k]:
                 if k == "files":
-                    self.globals[f"sys.{k}"] = self.get_files(kwargs[k])
+                    self.globals[f"sys.{k}"] = await self.get_files_async(kwargs[k])
                 else:
                     self.globals[f"sys.{k}"] = kwargs[k]
         if not self.globals["sys.conversation_turns"] :
@@ -362,31 +411,50 @@ class Canvas(Graph):
         yield decorate("workflow_started", {"inputs": kwargs.get("inputs")})
         self.retrieval.append({"chunks": {}, "doc_aggs": {}})
 
-        def _run_batch(f, t):
+        async def _run_batch(f, t):
             if self.is_canceled():
                 msg = f"Task {self.task_id} has been canceled during batch execution."
                 logging.info(msg)
                 raise TaskCanceledException(msg)
 
-            with ThreadPoolExecutor(max_workers=5) as executor:
-                thr = []
-                i = f
-                while i < t:
-                    cpn = self.get_component_obj(self.path[i])
-                    if cpn.component_name.lower() in ["begin", "userfillup"]:
-                        thr.append(executor.submit(cpn.invoke, inputs=kwargs.get("inputs", {})))
-                        i += 1
+            loop = asyncio.get_running_loop()
+            tasks = []
+
+            def _run_async_in_thread(coro_func, **call_kwargs):
+                return asyncio.run(coro_func(**call_kwargs))
+
+            i = f
+            while i < t:
+                cpn = self.get_component_obj(self.path[i])
+                task_fn = None
+                call_kwargs = None
+
+                if cpn.component_name.lower() in ["begin", "userfillup"]:
+                    call_kwargs = {"inputs": kwargs.get("inputs", {})}
+                    task_fn = cpn.invoke
+                    i += 1
+                else:
+                    for _, ele in cpn.get_input_elements().items():
+                        if isinstance(ele, dict) and ele.get("_cpn_id") and ele.get("_cpn_id") not in self.path[:i] and self.path[0].lower().find("userfillup") < 0:
+                            self.path.pop(i)
+                            t -= 1
+                            break
                     else:
-                        for _, ele in cpn.get_input_elements().items():
-                            if isinstance(ele, dict) and ele.get("_cpn_id") and ele.get("_cpn_id") not in self.path[:i] and self.path[0].lower().find("userfillup") < 0:
-                                self.path.pop(i)
-                                t -= 1
-                                break
-                        else:
-                            thr.append(executor.submit(cpn.invoke, **cpn.get_input()))
-                            i += 1
-                for t in thr:
-                    t.result()
+                        call_kwargs = cpn.get_input()
+                        task_fn = cpn.invoke
+                        i += 1
+
+                if task_fn is None:
+                    continue
+
+                invoke_async = getattr(cpn, "invoke_async", None)
+                if invoke_async and asyncio.iscoroutinefunction(invoke_async):
+                    tasks.append(loop.run_in_executor(self._thread_pool, partial(_run_async_in_thread, invoke_async, **(call_kwargs or {}))))
+                else:
+                    tasks.append(loop.run_in_executor(self._thread_pool, partial(task_fn, **(call_kwargs or {}))))
+
+            if tasks:
+                await asyncio.gather(*tasks)
 
         def _node_finished(cpn_obj):
             return decorate("node_finished",{
@@ -403,6 +471,7 @@ class Canvas(Graph):
         self.error = ""
         idx = len(self.path) - 1
         partials = []
+        tts_mdl = None
         while idx < len(self.path):
             to = len(self.path)
             for i in range(idx, to):
@@ -413,35 +482,72 @@ class Canvas(Graph):
                     "component_type": self.get_component_type(self.path[i]),
                     "thoughts": self.get_component_thoughts(self.path[i])
                 })
-            _run_batch(idx, to)
+            await _run_batch(idx, to)
             to = len(self.path)
-            # post processing of components invocation
+            # post-processing of components invocation
             for i in range(idx, to):
                 cpn = self.get_component(self.path[i])
                 cpn_obj = self.get_component_obj(self.path[i])
                 if cpn_obj.component_name.lower() == "message":
+                    if cpn_obj.get_param("auto_play"):
+                        tts_mdl = LLMBundle(self._tenant_id, LLMType.TTS)
                     if isinstance(cpn_obj.output("content"), partial):
                         _m = ""
-                        for m in cpn_obj.output("content")():
+                        buff_m = ""
+                        stream = cpn_obj.output("content")()
+                        async def _process_stream(m):
+                            nonlocal buff_m, _m, tts_mdl
                             if not m:
-                                continue
+                                return
                             if m == "<think>":
-                                yield decorate("message", {"content": "", "start_to_think": True})
+                                return decorate("message", {"content": "", "start_to_think": True})
+
                             elif m == "</think>":
-                                yield decorate("message", {"content": "", "end_to_think": True})
-                            else:
-                                yield decorate("message", {"content": m})
-                                _m += m
+                                return decorate("message", {"content": "", "end_to_think": True})
+
+                            buff_m += m
+                            _m += m
+
+                            if len(buff_m) > 16:
+                                ev = decorate(
+                                    "message",
+                                    {
+                                        "content": m,
+                                        "audio_binary": self.tts(tts_mdl, buff_m)
+                                    }
+                                )
+                                buff_m = ""
+                                return ev
+
+                            return decorate("message", {"content": m})
+
+                        if inspect.isasyncgen(stream):
+                            async for m in stream:
+                                ev= await _process_stream(m)
+                                if ev:
+                                    yield ev
+                        else:
+                            for m in stream:
+                                ev= await _process_stream(m)
+                                if ev:
+                                    yield ev
+                        if buff_m:
+                            yield decorate("message", {"content": "", "audio_binary": self.tts(tts_mdl, buff_m)})
+                            buff_m = ""
                         cpn_obj.set_output("content", _m)
                         cite = re.search(r"\[ID:[ 0-9]+\]", _m)
                     else:
                         yield decorate("message", {"content": cpn_obj.output("content")})
                         cite = re.search(r"\[ID:[ 0-9]+\]",  cpn_obj.output("content"))
 
-                    if isinstance(cpn_obj.output("attachment"), tuple):
-                        yield decorate("message", {"attachment": cpn_obj.output("attachment")})
-                        
-                    yield decorate("message_end", {"reference": self.get_reference() if cite else None})
+                    message_end = {}
+                    if cpn_obj.get_param("status"):
+                        message_end["status"] = cpn_obj.get_param("status")
+                    if isinstance(cpn_obj.output("attachment"), dict):
+                        message_end["attachment"] = cpn_obj.output("attachment")
+                    if cite:
+                        message_end["reference"] = self.get_reference()
+                    yield decorate("message_end", message_end)
 
                     while partials:
                         _cpn_obj = self.get_component_obj(partials[0])
@@ -462,7 +568,7 @@ class Canvas(Graph):
                     else:
                         self.error = cpn_obj.error()
 
-                if cpn_obj.component_name.lower() != "iteration":
+                if cpn_obj.component_name.lower() not in ("iteration","loop"):
                     if isinstance(cpn_obj.output("content"), partial):
                         if self.error:
                             cpn_obj.set_output("content", None)
@@ -487,14 +593,16 @@ class Canvas(Graph):
                     for cpn_id in cpn_ids:
                         _append_path(cpn_id)
 
-                if cpn_obj.component_name.lower() == "iterationitem" and cpn_obj.end():
+                if cpn_obj.component_name.lower() in ("iterationitem","loopitem") and cpn_obj.end():
                     iter = cpn_obj.get_parent()
                     yield _node_finished(iter)
                     _extend_path(self.get_component(cpn["parent_id"])["downstream"])
                 elif cpn_obj.component_name.lower() in ["categorize", "switch"]:
                     _extend_path(cpn_obj.output("_next"))
-                elif cpn_obj.component_name.lower() == "iteration":
+                elif cpn_obj.component_name.lower() in ("iteration", "loop"):
                     _append_path(cpn_obj.get_start())
+                elif cpn_obj.component_name.lower() == "exitloop" and cpn_obj.get_parent().component_name.lower() == "loop":
+                    _extend_path(self.get_component(cpn["parent_id"])["downstream"])
                 elif not cpn["downstream"] and cpn_obj.get_parent():
                     _append_path(cpn_obj.get_parent().get_start())
                 else:
@@ -550,6 +658,50 @@ class Canvas(Graph):
             return False
         return True
 
+
+    def tts(self,tts_mdl, text):
+        def clean_tts_text(text: str) -> str:
+            if not text:
+                return ""
+
+            text = text.encode("utf-8", "ignore").decode("utf-8", "ignore")
+
+            text = re.sub(r"[\x00-\x08\x0B-\x0C\x0E-\x1F\x7F]", "", text)
+
+            emoji_pattern = re.compile(
+                "[\U0001F600-\U0001F64F"
+                "\U0001F300-\U0001F5FF"
+                "\U0001F680-\U0001F6FF"
+                "\U0001F1E0-\U0001F1FF"
+                "\U00002700-\U000027BF"
+                "\U0001F900-\U0001F9FF"
+                "\U0001FA70-\U0001FAFF"
+                "\U0001FAD0-\U0001FAFF]+",
+                flags=re.UNICODE
+            )
+            text = emoji_pattern.sub("", text)
+
+            text = re.sub(r"\s+", " ", text).strip()
+
+            MAX_LEN = 500
+            if len(text) > MAX_LEN:
+                text = text[:MAX_LEN]
+
+            return text
+        if not tts_mdl or not text:
+            return None
+        text = clean_tts_text(text)
+        if not text:
+            return None
+        bin = b""
+        try:
+            for chunk in tts_mdl.tts(text):
+                bin += chunk
+        except Exception as e:
+            logging.error(f"TTS failed: {e}, text={text!r}")
+            return None
+        return binascii.hexlify(bin).decode("utf-8")
+
     def get_history(self, window_size):
         convs = []
         if window_size <= 0:
@@ -570,6 +722,9 @@ class Canvas(Graph):
     def get_mode(self):
         return self.components["begin"]["obj"]._param.mode
 
+    def get_sys_query(self):
+        return self.globals.get("sys.query", "")
+
     def set_global_param(self, **kwargs):
         self.globals.update(kwargs)
 
@@ -579,21 +734,30 @@ class Canvas(Graph):
     def get_component_input_elements(self, cpnnm):
         return self.components[cpnnm]["obj"].get_input_elements()
 
-    def get_files(self, files: Union[None, list[dict]]) -> list[str]:
-        from api.db.services.file_service import FileService
+    async def get_files_async(self, files: Union[None, list[dict]]) -> list[str]:
         if not files:
             return  []
         def image_to_base64(file):
             return "data:{};base64,{}".format(file["mime_type"],
                                         base64.b64encode(FileService.get_blob(file["created_by"], file["id"])).decode("utf-8"))
-        exe = ThreadPoolExecutor(max_workers=5)
-        threads = []
+        loop = asyncio.get_running_loop()
+        tasks = []
         for file in files:
             if file["mime_type"].find("image") >=0:
-                threads.append(exe.submit(image_to_base64, file))
+                tasks.append(loop.run_in_executor(self._thread_pool, image_to_base64, file))
                 continue
-            threads.append(exe.submit(FileService.parse, file["name"], FileService.get_blob(file["created_by"], file["id"]), True, file["created_by"]))
-        return [th.result() for th in threads]
+            tasks.append(loop.run_in_executor(self._thread_pool, FileService.parse, file["name"], FileService.get_blob(file["created_by"], file["id"]), True, file["created_by"]))
+        return await asyncio.gather(*tasks)
+
+    def get_files(self, files: Union[None, list[dict]]) -> list[str]:
+        """
+        Synchronous wrapper for get_files_async, used by sync component invoke paths.
+        """
+        loop = getattr(self, "_loop", None)
+        if loop and loop.is_running():
+            return asyncio.run_coroutine_threadsafe(self.get_files_async(files), loop).result()
+
+        return asyncio.run(self.get_files_async(files))
 
     def tool_use_callback(self, agent_id: str, func_name: str, params: dict, result: Any, elapsed_time=None):
         agent_ids = agent_id.split("-->")
@@ -647,4 +811,3 @@ class Canvas(Graph):
 
     def get_component_thoughts(self, cpn_id) -> str:
         return self.components.get(cpn_id)["obj"].thoughts()
-

agent/component/agent_with_tools.py

@@ -13,10 +13,11 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import asyncio
+import json
 import logging
 import os
 import re
-from concurrent.futures import ThreadPoolExecutor
 from copy import deepcopy
 from functools import partial
 from typing import Any
@@ -28,8 +29,8 @@ from api.db.services.llm_service import LLMBundle
 from api.db.services.tenant_llm_service import TenantLLMService
 from api.db.services.mcp_server_service import MCPServerService
 from common.connection_utils import timeout
-from rag.prompts.generator import next_step, COMPLETE_TASK, analyze_task, \
-    citation_prompt, reflect, rank_memories, kb_prompt, citation_plus, full_question, message_fit_in
+from rag.prompts.generator import next_step_async, COMPLETE_TASK, \
+    citation_prompt, kb_prompt, citation_plus, full_question, message_fit_in, structured_output_prompt
 from common.mcp_tool_call_conn import MCPToolCallSession, mcp_tool_metadata_to_openai_tool
 from agent.component.llm import LLMParam, LLM
 
@@ -83,9 +84,11 @@ class Agent(LLM, ToolBase):
     def __init__(self, canvas, id, param: LLMParam):
         LLM.__init__(self, canvas, id, param)
         self.tools = {}
-        for cpn in self._param.tools:
+        for idx, cpn in enumerate(self._param.tools):
             cpn = self._load_tool_obj(cpn)
-            self.tools[cpn.get_meta()["function"]["name"]] = cpn
+            original_name = cpn.get_meta()["function"]["name"]
+            indexed_name = f"{original_name}_{idx}"
+            self.tools[indexed_name] = cpn
 
         self.chat_mdl = LLMBundle(self._canvas.get_tenant_id(), TenantLLMService.llm_id2llm_type(self._param.llm_id), self._param.llm_id,
                                   max_retries=self._param.max_retries,
@@ -93,7 +96,12 @@ class Agent(LLM, ToolBase):
                                   max_rounds=self._param.max_rounds,
                                   verbose_tool_use=True
                                   )
-        self.tool_meta = [v.get_meta() for _,v in self.tools.items()]
+        self.tool_meta = []
+        for indexed_name, tool_obj in self.tools.items():
+            original_meta = tool_obj.get_meta()
+            indexed_meta = deepcopy(original_meta)
+            indexed_meta["function"]["name"] = indexed_name
+            self.tool_meta.append(indexed_meta)
 
         for mcp in self._param.mcp:
             _, mcp_server = MCPServerService.get_by_id(mcp["mcp_id"])
@@ -107,7 +115,8 @@ class Agent(LLM, ToolBase):
 
     def _load_tool_obj(self, cpn: dict) -> object:
         from agent.component import component_class
-        param = component_class(cpn["component_name"] + "Param")()
+        tool_name = cpn["component_name"]
+        param = component_class(tool_name + "Param")()
         param.update(cpn["params"])
         try:
             param.check()
@@ -137,8 +146,34 @@ class Agent(LLM, ToolBase):
             res.update(cpn.get_input_form())
         return res
 
-    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 20*60)))
+    def _get_output_schema(self):
+        try:
+            cand = self._param.outputs.get("structured")
+        except Exception:
+            return None
+
+        if isinstance(cand, dict):
+            if isinstance(cand.get("properties"), dict) and len(cand["properties"]) > 0:
+                return cand
+            for k in ("schema", "structured"):
+                if isinstance(cand.get(k), dict) and isinstance(cand[k].get("properties"), dict) and len(cand[k]["properties"]) > 0:
+                    return cand[k]
+
+        return None
+
+    async def _force_format_to_schema_async(self, text: str, schema_prompt: str) -> str:
+        fmt_msgs = [
+            {"role": "system", "content": schema_prompt + "\nIMPORTANT: Output ONLY valid JSON. No markdown, no extra text."},
+            {"role": "user", "content": text},
+        ]
+        _, fmt_msgs = message_fit_in(fmt_msgs, int(self.chat_mdl.max_length * 0.97))
+        return await self._generate_async(fmt_msgs)
+
     def _invoke(self, **kwargs):
+        return asyncio.run(self._invoke_async(**kwargs))
+
+    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 20*60)))
+    async def _invoke_async(self, **kwargs):
         if self.check_if_canceled("Agent processing"):
             return
 
@@ -157,20 +192,25 @@ class Agent(LLM, ToolBase):
         if not self.tools:
             if self.check_if_canceled("Agent processing"):
                 return
-            return LLM._invoke(self, **kwargs)
+            return await LLM._invoke_async(self, **kwargs)
 
         prompt, msg, user_defined_prompt = self._prepare_prompt_variables()
+        output_schema = self._get_output_schema()
+        schema_prompt = ""
+        if output_schema:
+            schema = json.dumps(output_schema, ensure_ascii=False, indent=2)
+            schema_prompt = structured_output_prompt(schema)
 
         downstreams = self._canvas.get_component(self._id)["downstream"] if self._canvas.get_component(self._id) else []
         ex = self.exception_handler()
-        if any([self._canvas.get_component_obj(cid).component_name.lower()=="message" for cid in downstreams]) and not (ex and ex["goto"]):
-            self.set_output("content", partial(self.stream_output_with_tools, prompt, msg, user_defined_prompt))
+        if any([self._canvas.get_component_obj(cid).component_name.lower()=="message" for cid in downstreams]) and not (ex and ex["goto"]) and not output_schema:
+            self.set_output("content", partial(self.stream_output_with_tools_async, prompt, deepcopy(msg), user_defined_prompt))
             return
 
         _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
         use_tools = []
         ans = ""
-        for delta_ans, tk in self._react_with_tools_streamly(prompt, msg, use_tools, user_defined_prompt):
+        async for delta_ans, _tk in self._react_with_tools_streamly_async_simple(prompt, msg, use_tools, user_defined_prompt,schema_prompt=schema_prompt):
             if self.check_if_canceled("Agent processing"):
                 return
             ans += delta_ans
@@ -183,16 +223,38 @@ class Agent(LLM, ToolBase):
                 self.set_output("_ERROR", ans)
             return
 
+        if output_schema:
+            error = ""
+            for _ in range(self._param.max_retries + 1):
+                try:
+                    def clean_formated_answer(ans: str) -> str:
+                        ans = re.sub(r"^.*</think>", "", ans, flags=re.DOTALL)
+                        ans = re.sub(r"^.*```json", "", ans, flags=re.DOTALL)
+                        return re.sub(r"```\n*$", "", ans, flags=re.DOTALL)
+                    obj = json_repair.loads(clean_formated_answer(ans))
+                    self.set_output("structured", obj)
+                    if use_tools:
+                        self.set_output("use_tools", use_tools)
+                    return obj
+                except Exception:
+                    error = "The answer cannot be parsed as JSON"
+                    ans = await self._force_format_to_schema_async(ans, schema_prompt)
+                    if ans.find("**ERROR**") >= 0:
+                        continue
+
+            self.set_output("_ERROR", error)
+            return
+
         self.set_output("content", ans)
         if use_tools:
             self.set_output("use_tools", use_tools)
         return ans
 
-    def stream_output_with_tools(self, prompt, msg, user_defined_prompt={}):
+    async def stream_output_with_tools_async(self, prompt, msg, user_defined_prompt={}):
         _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
         answer_without_toolcall = ""
         use_tools = []
-        for delta_ans,_ in self._react_with_tools_streamly(prompt, msg, use_tools, user_defined_prompt):
+        async for delta_ans, _ in self._react_with_tools_streamly_async_simple(prompt, msg, use_tools, user_defined_prompt):
             if self.check_if_canceled("Agent streaming"):
                 return
 
@@ -210,55 +272,58 @@ class Agent(LLM, ToolBase):
         if use_tools:
             self.set_output("use_tools", use_tools)
 
-    def _gen_citations(self, text):
-        retrievals = self._canvas.get_reference()
-        retrievals = {"chunks": list(retrievals["chunks"].values()), "doc_aggs": list(retrievals["doc_aggs"].values())}
-        formated_refer = kb_prompt(retrievals, self.chat_mdl.max_length, True)
-        for delta_ans in self._generate_streamly([{"role": "system", "content": citation_plus("\n\n".join(formated_refer))},
-                                                  {"role": "user", "content": text}
-                                                  ]):
-            yield delta_ans
-
-    def _react_with_tools_streamly(self, prompt, history: list[dict], use_tools, user_defined_prompt={}):
+    async def _react_with_tools_streamly_async_simple(self, prompt, history: list[dict], use_tools, user_defined_prompt={}, schema_prompt: str = ""):
         token_count = 0
         tool_metas = self.tool_meta
         hist = deepcopy(history)
         last_calling = ""
         if len(hist) > 3:
             st = timer()
-            user_request = full_question(messages=history, chat_mdl=self.chat_mdl)
+            user_request = await full_question(messages=history, chat_mdl=self.chat_mdl)
             self.callback("Multi-turn conversation optimization", {}, user_request, elapsed_time=timer()-st)
         else:
             user_request = history[-1]["content"]
 
-        def use_tool(name, args):
-            nonlocal hist, use_tools, token_count,last_calling,user_request
+        def build_task_desc(prompt: str, user_request: str, user_defined_prompt: dict | None = None) -> str:
+            """Build a minimal task_desc by concatenating prompt, query, and tool schemas."""
+            user_defined_prompt = user_defined_prompt or {}
+
+            task_desc = (
+                "### Agent Prompt\n"
+                f"{prompt}\n\n"
+                "### User Request\n"
+                f"{user_request}\n\n"
+            )
+
+            if user_defined_prompt:
+                udp_json = json.dumps(user_defined_prompt, ensure_ascii=False, indent=2)
+                task_desc += "\n### User Defined Prompts\n" + udp_json + "\n"
+
+            return task_desc
+
+
+        async def use_tool_async(name, args):
+            nonlocal hist, use_tools, last_calling
             logging.info(f"{last_calling=} == {name=}")
-            # Summarize of function calling
-            #if all([
-            #    isinstance(self.toolcall_session.get_tool_obj(name), Agent),
-            #    last_calling,
-            #    last_calling != name
-            #]):
-            #    self.toolcall_session.get_tool_obj(name).add2system_prompt(f"The chat history with other agents are as following: \n" + self.get_useful_memory(user_request, str(args["user_prompt"]),user_defined_prompt))
             last_calling = name
-            tool_response = self.toolcall_session.tool_call(name, args)
+            tool_response = await self.toolcall_session.tool_call_async(name, args)
             use_tools.append({
                 "name": name,
                 "arguments": args,
                 "results": tool_response
             })
-            # self.callback("add_memory", {}, "...")
-            #self.add_memory(hist[-2]["content"], hist[-1]["content"], name, args, str(tool_response), user_defined_prompt)
-
             return name, tool_response
 
-        def complete():
+        async def complete():
             nonlocal hist
             need2cite = self._param.cite and self._canvas.get_reference()["chunks"] and self._id.find("-->") < 0
+            if schema_prompt:
+                need2cite = False
             cited = False
-            if hist[0]["role"] == "system" and need2cite:
-                if len(hist) < 7:
+            if hist and hist[0]["role"] == "system":
+                if schema_prompt:
+                    hist[0]["content"] += "\n" + schema_prompt
+                if need2cite and len(hist) < 7:
                     hist[0]["content"] += citation_prompt()
                     cited = True
             yield "", token_count
@@ -267,7 +332,7 @@ class Agent(LLM, ToolBase):
             if len(hist) > 12:
                 _hist = [hist[0], hist[1], *hist[-10:]]
             entire_txt = ""
-            for delta_ans in self._generate_streamly(_hist):
+            async for delta_ans in self._generate_streamly(_hist):
                 if not need2cite or cited:
                     yield delta_ans, 0
                 entire_txt += delta_ans
@@ -276,7 +341,7 @@ class Agent(LLM, ToolBase):
 
             st = timer()
             txt = ""
-            for delta_ans in self._gen_citations(entire_txt):
+            async for delta_ans in self._gen_citations_async(entire_txt):
                 if self.check_if_canceled("Agent streaming"):
                     return
                 yield delta_ans, 0
@@ -284,6 +349,21 @@ class Agent(LLM, ToolBase):
 
             self.callback("gen_citations", {}, txt, elapsed_time=timer()-st)
 
+        def build_observation(tool_call_res: list[tuple]) -> str:
+            """
+            Build a Observation from tool call results.
+            No LLM involved.
+            """
+            if not tool_call_res:
+                return ""
+
+            lines = ["Observation:"]
+            for name, result in tool_call_res:
+                lines.append(f"[{name} result]")
+                lines.append(str(result))
+
+            return "\n".join(lines)
+
         def append_user_content(hist, content):
             if hist[-1]["role"] == "user":
                 hist[-1]["content"] += content
@@ -291,14 +371,14 @@ class Agent(LLM, ToolBase):
                 hist.append({"role": "user", "content": content})
 
         st = timer()
-        task_desc = analyze_task(self.chat_mdl, prompt, user_request, tool_metas, user_defined_prompt)
+        task_desc = build_task_desc(prompt, user_request, user_defined_prompt)
         self.callback("analyze_task", {}, task_desc, elapsed_time=timer()-st)
         for _ in range(self._param.max_rounds + 1):
             if self.check_if_canceled("Agent streaming"):
                 return
-            response, tk = next_step(self.chat_mdl, hist, tool_metas, task_desc, user_defined_prompt)
+            response, tk = await next_step_async(self.chat_mdl, hist, tool_metas, task_desc, user_defined_prompt)
             # self.callback("next_step", {}, str(response)[:256]+"...")
-            token_count += tk
+            token_count += tk or 0
             hist.append({"role": "assistant", "content": response})
             try:
                 functions = json_repair.loads(re.sub(r"```.*", "", response))
@@ -307,23 +387,24 @@ class Agent(LLM, ToolBase):
                 for f in functions:
                     if not isinstance(f, dict):
                         raise TypeError(f"An object type should be returned, but `{f}`")
-                with ThreadPoolExecutor(max_workers=5) as executor:
-                    thr = []
-                    for func in functions:
-                        name = func["name"]
-                        args = func["arguments"]
-                        if name == COMPLETE_TASK:
-                            append_user_content(hist, f"Respond with a formal answer. FORGET(DO NOT mention) about `{COMPLETE_TASK}`. The language for the response MUST be as the same as the first user request.\n")
-                            for txt, tkcnt in complete():
-                                yield txt, tkcnt
-                            return
 
-                        thr.append(executor.submit(use_tool, name, args))
+                tool_tasks = []
+                for func in functions:
+                    name = func["name"]
+                    args = func["arguments"]
+                    if name == COMPLETE_TASK:
+                        append_user_content(hist, f"Respond with a formal answer. FORGET(DO NOT mention) about `{COMPLETE_TASK}`. The language for the response MUST be as the same as the first user request.\n")
+                        async for txt, tkcnt in complete():
+                            yield txt, tkcnt
+                        return
 
-                    st = timer()
-                    reflection = reflect(self.chat_mdl, hist, [th.result() for th in thr], user_defined_prompt)
-                    append_user_content(hist, reflection)
-                    self.callback("reflection", {}, str(reflection), elapsed_time=timer()-st)
+                    tool_tasks.append(asyncio.create_task(use_tool_async(name, args)))
+
+                results = await asyncio.gather(*tool_tasks) if tool_tasks else []
+                st = timer()
+                reflection = build_observation(results)
+                append_user_content(hist, reflection)
+                self.callback("reflection", {}, str(reflection), elapsed_time=timer()-st)
 
             except Exception as e:
                 logging.exception(msg=f"Wrong JSON argument format in LLM ReAct response: {e}")
@@ -347,21 +428,146 @@ Respond immediately with your final comprehensive answer.
             return
         append_user_content(hist, final_instruction)
 
-        for txt, tkcnt in complete():
+        async for txt, tkcnt in complete():
             yield txt, tkcnt
 
-    def get_useful_memory(self, goal: str, sub_goal:str, topn=3, user_defined_prompt:dict={}) -> str:
-        # self.callback("get_useful_memory", {"topn": 3}, "...")
-        mems = self._canvas.get_memory()
-        rank = rank_memories(self.chat_mdl, goal, sub_goal, [summ for (user, assist, summ) in mems], user_defined_prompt)
-        try:
-            rank = json_repair.loads(re.sub(r"```.*", "", rank))[:topn]
-            mems = [mems[r] for r in rank]
-            return "\n\n".join([f"User: {u}\nAgent: {a}" for u, a,_ in mems])
-        except Exception as e:
-            logging.exception(e)
+#     async def _react_with_tools_streamly_async(self, prompt, history: list[dict], use_tools, user_defined_prompt={}, schema_prompt: str = ""):
+#         token_count = 0
+#         tool_metas = self.tool_meta
+#         hist = deepcopy(history)
+#         last_calling = ""
+#         if len(hist) > 3:
+#             st = timer()
+#             user_request = await full_question(messages=history, chat_mdl=self.chat_mdl)
+#             self.callback("Multi-turn conversation optimization", {}, user_request, elapsed_time=timer()-st)
+#         else:
+#             user_request = history[-1]["content"]
 
-        return "Error occurred."
+#         async def use_tool_async(name, args):
+#             nonlocal hist, use_tools, last_calling
+#             logging.info(f"{last_calling=} == {name=}")
+#             last_calling = name
+#             tool_response = await self.toolcall_session.tool_call_async(name, args)
+#             use_tools.append({
+#                 "name": name,
+#                 "arguments": args,
+#                 "results": tool_response
+#             })
+#             # self.callback("add_memory", {}, "...")
+#             #self.add_memory(hist[-2]["content"], hist[-1]["content"], name, args, str(tool_response), user_defined_prompt)
+
+#             return name, tool_response
+
+#         async def complete():
+#             nonlocal hist
+#             need2cite = self._param.cite and self._canvas.get_reference()["chunks"] and self._id.find("-->") < 0
+#             if schema_prompt:
+#                 need2cite = False
+#             cited = False
+#             if hist and hist[0]["role"] == "system":
+#                 if schema_prompt:
+#                     hist[0]["content"] += "\n" + schema_prompt
+#                 if need2cite and len(hist) < 7:
+#                     hist[0]["content"] += citation_prompt()
+#                     cited = True
+#             yield "", token_count
+
+#             _hist = hist
+#             if len(hist) > 12:
+#                 _hist = [hist[0], hist[1], *hist[-10:]]
+#             entire_txt = ""
+#             async for delta_ans in self._generate_streamly(_hist):
+#                 if not need2cite or cited:
+#                     yield delta_ans, 0
+#                 entire_txt += delta_ans
+#             if not need2cite or cited:
+#                 return
+
+#             st = timer()
+#             txt = ""
+#             async for delta_ans in self._gen_citations_async(entire_txt):
+#                 if self.check_if_canceled("Agent streaming"):
+#                     return
+#                 yield delta_ans, 0
+#                 txt += delta_ans
+
+#             self.callback("gen_citations", {}, txt, elapsed_time=timer()-st)
+
+#         def append_user_content(hist, content):
+#             if hist[-1]["role"] == "user":
+#                 hist[-1]["content"] += content
+#             else:
+#                 hist.append({"role": "user", "content": content})
+
+#         st = timer()
+#         task_desc = await analyze_task_async(self.chat_mdl, prompt, user_request, tool_metas, user_defined_prompt)
+#         self.callback("analyze_task", {}, task_desc, elapsed_time=timer()-st)
+#         for _ in range(self._param.max_rounds + 1):
+#             if self.check_if_canceled("Agent streaming"):
+#                 return
+#             response, tk = await next_step_async(self.chat_mdl, hist, tool_metas, task_desc, user_defined_prompt)
+#             # self.callback("next_step", {}, str(response)[:256]+"...")
+#             token_count += tk or 0
+#             hist.append({"role": "assistant", "content": response})
+#             try:
+#                 functions = json_repair.loads(re.sub(r"```.*", "", response))
+#                 if not isinstance(functions, list):
+#                     raise TypeError(f"List should be returned, but `{functions}`")
+#                 for f in functions:
+#                     if not isinstance(f, dict):
+#                         raise TypeError(f"An object type should be returned, but `{f}`")
+
+#                 tool_tasks = []
+#                 for func in functions:
+#                     name = func["name"]
+#                     args = func["arguments"]
+#                     if name == COMPLETE_TASK:
+#                         append_user_content(hist, f"Respond with a formal answer. FORGET(DO NOT mention) about `{COMPLETE_TASK}`. The language for the response MUST be as the same as the first user request.\n")
+#                         async for txt, tkcnt in complete():
+#                             yield txt, tkcnt
+#                         return
+
+#                     tool_tasks.append(asyncio.create_task(use_tool_async(name, args)))
+
+#                 results = await asyncio.gather(*tool_tasks) if tool_tasks else []
+#                 st = timer()
+#                 reflection = await reflect_async(self.chat_mdl, hist, results, user_defined_prompt)
+#                 append_user_content(hist, reflection)
+#                 self.callback("reflection", {}, str(reflection), elapsed_time=timer()-st)
+
+#             except Exception as e:
+#                 logging.exception(msg=f"Wrong JSON argument format in LLM ReAct response: {e}")
+#                 e = f"\nTool call error, please correct the input parameter of response format and call it again.\n *** Exception ***\n{e}"
+#                 append_user_content(hist, str(e))
+
+#         logging.warning( f"Exceed max rounds: {self._param.max_rounds}")
+#         final_instruction = f"""
+# {user_request}
+# IMPORTANT: You have reached the conversation limit. Based on ALL the information and research you have gathered so far, please provide a DIRECT and COMPREHENSIVE final answer to the original request.
+# Instructions:
+# 1. SYNTHESIZE all information collected during this conversation
+# 2. Provide a COMPLETE response using existing data - do not suggest additional research
+# 3. Structure your response as a FINAL DELIVERABLE, not a plan
+# 4. If information is incomplete, state what you found and provide the best analysis possible with available data
+# 5. DO NOT mention conversation limits or suggest further steps
+# 6. Focus on delivering VALUE with the information already gathered
+# Respond immediately with your final comprehensive answer.
+#         """
+#         if self.check_if_canceled("Agent final instruction"):
+#             return
+#         append_user_content(hist, final_instruction)
+
+#         async for txt, tkcnt in complete():
+#             yield txt, tkcnt
+
+    async def _gen_citations_async(self, text):
+        retrievals = self._canvas.get_reference()
+        retrievals = {"chunks": list(retrievals["chunks"].values()), "doc_aggs": list(retrievals["doc_aggs"].values())}
+        formated_refer = kb_prompt(retrievals, self.chat_mdl.max_length, True)
+        async for delta_ans in self._generate_streamly([{"role": "system", "content": citation_plus("\n\n".join(formated_refer))},
+                                                  {"role": "user", "content": text}
+                                                  ]):
+            yield delta_ans
 
     def reset(self, only_output=False):
         """
@@ -369,7 +575,7 @@ Respond immediately with your final comprehensive answer.
         """
         for k in self._param.outputs.keys():
             self._param.outputs[k]["value"] = None
-            
+
         for k, cpn in self.tools.items():
             if hasattr(cpn, "reset") and callable(cpn.reset):
                 cpn.reset()
@@ -378,4 +584,3 @@ Respond immediately with your final comprehensive answer.
         for k in self._param.inputs.keys():
             self._param.inputs[k]["value"] = None
         self._param.debug_inputs = {}
-

agent/component/base.py

@@ -14,6 +14,7 @@
 #  limitations under the License.
 #
 
+import asyncio
 import re
 import time
 from abc import ABC
@@ -23,11 +24,13 @@ import os
 import logging
 from typing import Any, List, Union
 import pandas as pd
-import trio
 from agent import settings
 from common.connection_utils import timeout
 
 
+
+from common.misc_utils import thread_pool_exec
+
 _FEEDED_DEPRECATED_PARAMS = "_feeded_deprecated_params"
 _DEPRECATED_PARAMS = "_deprecated_params"
 _USER_FEEDED_PARAMS = "_user_feeded_params"
@@ -97,7 +100,7 @@ class ComponentParamBase(ABC):
         def _recursive_convert_obj_to_dict(obj):
             ret_dict = {}
             if isinstance(obj, dict):
-                for k,v in obj.items():
+                for k, v in obj.items():
                     if isinstance(v, dict) or (v and type(v).__name__ not in dir(builtins)):
                         ret_dict[k] = _recursive_convert_obj_to_dict(v)
                     else:
@@ -253,96 +256,65 @@ class ComponentParamBase(ABC):
                 self._validate_param(attr, validation_json)
 
     @staticmethod
-    def check_string(param, descr):
+    def check_string(param, description):
         if type(param).__name__ not in ["str"]:
-            raise ValueError(
-                descr + " {} not supported, should be string type".format(param)
-            )
+            raise ValueError(description + " {} not supported, should be string type".format(param))
 
     @staticmethod
-    def check_empty(param, descr):
+    def check_empty(param, description):
         if not param:
-            raise ValueError(
-                descr + " does not support empty value."
-            )
+            raise ValueError(description + " does not support empty value.")
 
     @staticmethod
-    def check_positive_integer(param, descr):
+    def check_positive_integer(param, description):
         if type(param).__name__ not in ["int", "long"] or param <= 0:
-            raise ValueError(
-                descr + " {} not supported, should be positive integer".format(param)
-            )
+            raise ValueError(description + " {} not supported, should be positive integer".format(param))
 
     @staticmethod
-    def check_positive_number(param, descr):
+    def check_positive_number(param, description):
         if type(param).__name__ not in ["float", "int", "long"] or param <= 0:
-            raise ValueError(
-                descr + " {} not supported, should be positive numeric".format(param)
-            )
+            raise ValueError(description + " {} not supported, should be positive numeric".format(param))
 
     @staticmethod
-    def check_nonnegative_number(param, descr):
+    def check_nonnegative_number(param, description):
         if type(param).__name__ not in ["float", "int", "long"] or param < 0:
-            raise ValueError(
-                descr
-                + " {} not supported, should be non-negative numeric".format(param)
-            )
+            raise ValueError(description + " {} not supported, should be non-negative numeric".format(param))
 
     @staticmethod
-    def check_decimal_float(param, descr):
+    def check_decimal_float(param, description):
         if type(param).__name__ not in ["float", "int"] or param < 0 or param > 1:
-            raise ValueError(
-                descr
-                + " {} not supported, should be a float number in range [0, 1]".format(
-                    param
-                )
-            )
+            raise ValueError(description + " {} not supported, should be a float number in range [0, 1]".format(param))
 
     @staticmethod
-    def check_boolean(param, descr):
+    def check_boolean(param, description):
         if type(param).__name__ != "bool":
-            raise ValueError(
-                descr + " {} not supported, should be bool type".format(param)
-            )
+            raise ValueError(description + " {} not supported, should be bool type".format(param))
 
     @staticmethod
-    def check_open_unit_interval(param, descr):
+    def check_open_unit_interval(param, description):
         if type(param).__name__ not in ["float"] or param <= 0 or param >= 1:
-            raise ValueError(
-                descr + " should be a numeric number between 0 and 1 exclusively"
-            )
+            raise ValueError(description + " should be a numeric number between 0 and 1 exclusively")
 
     @staticmethod
-    def check_valid_value(param, descr, valid_values):
+    def check_valid_value(param, description, valid_values):
         if param not in valid_values:
-            raise ValueError(
-                descr
-                + " {} is not supported, it should be in {}".format(param, valid_values)
-            )
+            raise ValueError(description + " {} is not supported, it should be in {}".format(param, valid_values))
 
     @staticmethod
-    def check_defined_type(param, descr, types):
+    def check_defined_type(param, description, types):
         if type(param).__name__ not in types:
-            raise ValueError(
-                descr + " {} not supported, should be one of {}".format(param, types)
-            )
+            raise ValueError(description + " {} not supported, should be one of {}".format(param, types))
 
     @staticmethod
-    def check_and_change_lower(param, valid_list, descr=""):
+    def check_and_change_lower(param, valid_list, description=""):
         if type(param).__name__ != "str":
-            raise ValueError(
-                descr
-                + " {} not supported, should be one of {}".format(param, valid_list)
-            )
+            raise ValueError(description + " {} not supported, should be one of {}".format(param, valid_list))
 
         lower_param = param.lower()
         if lower_param in valid_list:
             return lower_param
         else:
-            raise ValueError(
-                descr
-                + " {} not supported, should be one of {}".format(param, valid_list)
-            )
+            raise ValueError(description + " {} not supported, should be one of {}".format(param, valid_list))
 
     @staticmethod
     def _greater_equal_than(value, limit):
@@ -374,16 +346,16 @@ class ComponentParamBase(ABC):
     def _not_in(value, wrong_value_list):
         return value not in wrong_value_list
 
-    def _warn_deprecated_param(self, param_name, descr):
+    def _warn_deprecated_param(self, param_name, description):
         if self._deprecated_params_set.get(param_name):
             logging.warning(
-                f"{descr} {param_name} is deprecated and ignored in this version."
+                f"{description} {param_name} is deprecated and ignored in this version."
             )
 
-    def _warn_to_deprecate_param(self, param_name, descr, new_param):
+    def _warn_to_deprecate_param(self, param_name, description, new_param):
         if self._deprecated_params_set.get(param_name):
             logging.warning(
-                f"{descr} {param_name} will be deprecated in future release; "
+                f"{description} {param_name} will be deprecated in future release; "
                 f"please use {new_param} instead."
             )
             return True
@@ -392,8 +364,8 @@ class ComponentParamBase(ABC):
 
 class ComponentBase(ABC):
     component_name: str
-    thread_limiter = trio.CapacityLimiter(int(os.environ.get('MAX_CONCURRENT_CHATS', 10)))
-    variable_ref_patt = r"\{* *\{([a-zA-Z:0-9]+@[A-Za-z0-9_.]+|sys\.[A-Za-z0-9_.]+|env\.[A-Za-z0-9_.]+)\} *\}*"
+    thread_limiter = asyncio.Semaphore(int(os.environ.get("MAX_CONCURRENT_CHATS", 10)))
+    variable_ref_patt = r"\{* *\{([a-zA-Z:0-9]+@[A-Za-z0-9_.-]+|sys\.[A-Za-z0-9_.]+|env\.[A-Za-z0-9_.]+)\} *\}*"
 
     def __str__(self):
         """
@@ -407,10 +379,11 @@ class ComponentBase(ABC):
             "params": {}
         }}""".format(self.component_name,
                      self._param
-        )
+                     )
 
     def __init__(self, canvas, id, param: ComponentParamBase):
         from agent.canvas import Graph  # Local import to avoid cyclic dependency
+
         assert isinstance(canvas, Graph), "canvas must be an instance of Canvas"
         self._canvas = canvas
         self._id = id
@@ -445,14 +418,42 @@ class ComponentBase(ABC):
         self.set_output("_elapsed_time", time.perf_counter() - self.output("_created_time"))
         return self.output()
 
-    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60)))
+    async def invoke_async(self, **kwargs) -> dict[str, Any]:
+        """
+        Async wrapper for component invocation.
+        Prefers coroutine `_invoke_async` if present; otherwise falls back to `_invoke`.
+        Handles timing and error recording consistently with `invoke`.
+        """
+        self.set_output("_created_time", time.perf_counter())
+        try:
+            if self.check_if_canceled("Component processing"):
+                return
+
+            fn_async = getattr(self, "_invoke_async", None)
+            if fn_async and asyncio.iscoroutinefunction(fn_async):
+                await fn_async(**kwargs)
+            elif asyncio.iscoroutinefunction(self._invoke):
+                await self._invoke(**kwargs)
+            else:
+                await thread_pool_exec(self._invoke, **kwargs)
+        except Exception as e:
+            if self.get_exception_default_value():
+                self.set_exception_default_value()
+            else:
+                self.set_output("_ERROR", str(e))
+            logging.exception(e)
+        self._param.debug_inputs = {}
+        self.set_output("_elapsed_time", time.perf_counter() - self.output("_created_time"))
+        return self.output()
+
+    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10 * 60)))
     def _invoke(self, **kwargs):
         raise NotImplementedError()
 
-    def output(self, var_nm: str=None) -> Union[dict[str, Any], Any]:
+    def output(self, var_nm: str = None) -> Union[dict[str, Any], Any]:
         if var_nm:
             return self._param.outputs.get(var_nm, {}).get("value", "")
-        return {k: o.get("value") for k,o in self._param.outputs.items()}
+        return {k: o.get("value") for k, o in self._param.outputs.items()}
 
     def set_output(self, key: str, value: Any):
         if key not in self._param.outputs:
@@ -463,18 +464,18 @@ class ComponentBase(ABC):
         return self._param.outputs.get("_ERROR", {}).get("value")
 
     def reset(self, only_output=False):
-        outputs: dict = self._param.outputs # for better performance
+        outputs: dict = self._param.outputs  # for better performance
         for k in outputs.keys():
             outputs[k]["value"] = None
         if only_output:
             return
 
-        inputs: dict = self._param.inputs # for better performance
+        inputs: dict = self._param.inputs  # for better performance
         for k in inputs.keys():
             inputs[k]["value"] = None
         self._param.debug_inputs = {}
 
-    def get_input(self, key: str=None) -> Union[Any, dict[str, Any]]:
+    def get_input(self, key: str = None) -> Union[Any, dict[str, Any]]:
         if key:
             return self._param.inputs.get(key, {}).get("value")
 
@@ -498,13 +499,13 @@ class ComponentBase(ABC):
 
     def get_input_elements_from_text(self, txt: str) -> dict[str, dict[str, str]]:
         res = {}
-        for r in re.finditer(self.variable_ref_patt, txt, flags=re.IGNORECASE|re.DOTALL):
+        for r in re.finditer(self.variable_ref_patt, txt, flags=re.IGNORECASE | re.DOTALL):
             exp = r.group(1)
-            cpn_id, var_nm = exp.split("@") if exp.find("@")>0 else ("", exp)
+            cpn_id, var_nm = exp.split("@") if exp.find("@") > 0 else ("", exp)
             res[exp] = {
-                "name": (self._canvas.get_component_name(cpn_id) +f"@{var_nm}") if cpn_id else exp,
+                "name": (self._canvas.get_component_name(cpn_id) + f"@{var_nm}") if cpn_id else exp,
                 "value": self._canvas.get_variable_value(exp),
-                "_retrival": self._canvas.get_variable_value(f"{cpn_id}@_references") if cpn_id else None,
+                "_retrieval": self._canvas.get_variable_value(f"{cpn_id}@_references") if cpn_id else None,
                 "_cpn_id": cpn_id
             }
         return res
@@ -555,6 +556,7 @@ class ComponentBase(ABC):
         for n, v in kv.items():
             def repl(_match, val=v):
                 return str(val) if val is not None else ""
+
             content = re.sub(
                 r"\{%s\}" % re.escape(n),
                 repl,

agent/component/begin.py

@@ -14,6 +14,7 @@
 #  limitations under the License.
 #
 from agent.component.fillup import UserFillUpParam, UserFillUp
+from api.db.services.file_service import FileService
 
 
 class BeginParam(UserFillUpParam):
@@ -27,7 +28,7 @@ class BeginParam(UserFillUpParam):
         self.prologue = "Hi! I'm your smart assistant. What can I do for you?"
 
     def check(self):
-        self.check_valid_value(self.mode, "The 'mode' should be either `conversational` or `task`", ["conversational", "task"])
+        self.check_valid_value(self.mode, "The 'mode' should be either `conversational` or `task`", ["conversational", "task","Webhook"])
 
     def get_input_form(self) -> dict[str, dict]:
         return getattr(self, "inputs")
@@ -48,7 +49,7 @@ class Begin(UserFillUp):
                 if v.get("optional") and v.get("value", None) is None:
                     v = None
                 else:
-                    v = self._canvas.get_files([v["value"]])
+                    v = FileService.get_files([v["value"]])
             else:
                 v = v.get("value")
             self.set_output(k, v)

agent/component/categorize.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import asyncio
 import logging
 import os
 import re
@@ -96,20 +97,30 @@ Here's description of each category:
 class Categorize(LLM, ABC):
     component_name = "Categorize"
 
+    def get_input_elements(self) -> dict[str, dict]:
+        query_key = self._param.query or "sys.query"
+        elements = self.get_input_elements_from_text(f"{{{query_key}}}")
+        if not elements:
+            logging.warning(f"[Categorize] input element not detected for query key: {query_key}")
+        return elements
+
     @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60)))
-    def _invoke(self, **kwargs):
+    async def _invoke_async(self, **kwargs):
         if self.check_if_canceled("Categorize processing"):
             return
 
         msg = self._canvas.get_history(self._param.message_history_window_size)
         if not msg:
             msg = [{"role": "user", "content": ""}]
-        if kwargs.get("sys.query"):
-            msg[-1]["content"] = kwargs["sys.query"]
-            self.set_input_value("sys.query", kwargs["sys.query"])
+        query_key = self._param.query or "sys.query"
+        if query_key in kwargs:
+            query_value = kwargs[query_key]
         else:
-            msg[-1]["content"] = self._canvas.get_variable_value(self._param.query)
-            self.set_input_value(self._param.query, msg[-1]["content"])
+            query_value = self._canvas.get_variable_value(query_key)
+        if query_value is None:
+            query_value = ""
+        msg[-1]["content"] = query_value
+        self.set_input_value(query_key, msg[-1]["content"])
         self._param.update_prompt()
         chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
 
@@ -121,7 +132,7 @@ class Categorize(LLM, ABC):
         if self.check_if_canceled("Categorize processing"):
             return
 
-        ans = chat_mdl.chat(self._param.sys_prompt, [{"role": "user", "content": user_prompt}], self._param.gen_conf())
+        ans = await chat_mdl.async_chat(self._param.sys_prompt, [{"role": "user", "content": user_prompt}], self._param.gen_conf())
         logging.info(f"input: {user_prompt}, answer: {str(ans)}")
         if ERROR_PREFIX in ans:
             raise Exception(ans)
@@ -144,5 +155,9 @@ class Categorize(LLM, ABC):
         self.set_output("category_name", max_category)
         self.set_output("_next", cpn_ids)
 
+    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60)))
+    def _invoke(self, **kwargs):
+        return asyncio.run(self._invoke_async(**kwargs))
+
     def thoughts(self) -> str:
         return "Which should it falls into {}? ...".format(",".join([f"`{c}`" for c, _ in self._param.category_description.items()]))

agent/component/excel_processor.py

@@ -0,0 +1,401 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+
+"""
+ExcelProcessor Component
+
+A component for reading, processing, and generating Excel files in RAGFlow agents.
+Supports multiple Excel file inputs, data transformation, and Excel output generation.
+"""
+
+import logging
+import os
+from abc import ABC
+from io import BytesIO
+
+import pandas as pd
+
+from agent.component.base import ComponentBase, ComponentParamBase
+from api.db.services.file_service import FileService
+from api.utils.api_utils import timeout
+from common import settings
+from common.misc_utils import get_uuid
+
+
+class ExcelProcessorParam(ComponentParamBase):
+    """
+    Define the ExcelProcessor component parameters.
+    """
+    def __init__(self):
+        super().__init__()
+        # Input configuration
+        self.input_files = []  # Variable references to uploaded files
+        self.operation = "read"  # read, merge, transform, output
+        
+        # Processing options
+        self.sheet_selection = "all"  # all, first, or comma-separated sheet names
+        self.merge_strategy = "concat"  # concat, join
+        self.join_on = ""  # Column name for join operations
+        
+        # Transform options (for LLM-guided transformations)
+        self.transform_instructions = ""
+        self.transform_data = ""  # Variable reference to transformation data
+        
+        # Output options
+        self.output_format = "xlsx"  # xlsx, csv
+        self.output_filename = "output"
+        
+        # Component outputs
+        self.outputs = {
+            "data": {
+                "type": "object",
+                "value": {}
+            },
+            "summary": {
+                "type": "str",
+                "value": ""
+            },
+            "markdown": {
+                "type": "str",
+                "value": ""
+            }
+        }
+    
+    def check(self):
+        self.check_valid_value(
+            self.operation, 
+            "[ExcelProcessor] Operation", 
+            ["read", "merge", "transform", "output"]
+        )
+        self.check_valid_value(
+            self.output_format,
+            "[ExcelProcessor] Output format",
+            ["xlsx", "csv"]
+        )
+        return True
+
+
+class ExcelProcessor(ComponentBase, ABC):
+    """
+    Excel processing component for RAGFlow agents.
+    
+    Operations:
+    - read: Parse Excel files into structured data
+    - merge: Combine multiple Excel files
+    - transform: Apply data transformations based on instructions
+    - output: Generate Excel file output
+    """
+    component_name = "ExcelProcessor"
+
+    def get_input_form(self) -> dict[str, dict]:
+        """Define input form for the component."""
+        res = {}
+        for ref in (self._param.input_files or []):
+            for k, o in self.get_input_elements_from_text(ref).items():
+                res[k] = {"name": o.get("name", ""), "type": "file"}
+        if self._param.transform_data:
+            for k, o in self.get_input_elements_from_text(self._param.transform_data).items():
+                res[k] = {"name": o.get("name", ""), "type": "object"}
+        return res
+
+    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60)))
+    def _invoke(self, **kwargs):
+        if self.check_if_canceled("ExcelProcessor processing"):
+            return
+
+        operation = self._param.operation.lower()
+        
+        if operation == "read":
+            self._read_excels()
+        elif operation == "merge":
+            self._merge_excels()
+        elif operation == "transform":
+            self._transform_data()
+        elif operation == "output":
+            self._output_excel()
+        else:
+            self.set_output("summary", f"Unknown operation: {operation}")
+
+    def _get_file_content(self, file_ref: str) -> tuple[bytes, str]:
+        """
+        Get file content from a variable reference.
+        Returns (content_bytes, filename).
+        """
+        value = self._canvas.get_variable_value(file_ref)
+        if value is None:
+            return None, None
+            
+        # Handle different value formats
+        if isinstance(value, dict):
+            # File reference from Begin/UserFillUp component
+            file_id = value.get("id") or value.get("file_id")
+            created_by = value.get("created_by") or self._canvas.get_tenant_id()
+            filename = value.get("name") or value.get("filename", "unknown.xlsx")
+            if file_id:
+                content = FileService.get_blob(created_by, file_id)
+                return content, filename
+        elif isinstance(value, list) and len(value) > 0:
+            # List of file references - return first
+            return self._get_file_content_from_list(value[0])
+        elif isinstance(value, str):
+            # Could be base64 encoded or a path
+            if value.startswith("data:"):
+                import base64
+                # Extract base64 content
+                _, encoded = value.split(",", 1)
+                return base64.b64decode(encoded), "uploaded.xlsx"
+                
+        return None, None
+    
+    def _get_file_content_from_list(self, item) -> tuple[bytes, str]:
+        """Extract file content from a list item."""
+        if isinstance(item, dict):
+            return self._get_file_content(item)
+        return None, None
+
+    def _parse_excel_to_dataframes(self, content: bytes, filename: str) -> dict[str, pd.DataFrame]:
+        """Parse Excel content into a dictionary of DataFrames (one per sheet)."""
+        try:
+            excel_file = BytesIO(content)
+            
+            if filename.lower().endswith(".csv"):
+                df = pd.read_csv(excel_file)
+                return {"Sheet1": df}
+            else:
+                # Read all sheets
+                xlsx = pd.ExcelFile(excel_file, engine='openpyxl')
+                sheet_selection = self._param.sheet_selection
+                
+                if sheet_selection == "all":
+                    sheets_to_read = xlsx.sheet_names
+                elif sheet_selection == "first":
+                    sheets_to_read = [xlsx.sheet_names[0]] if xlsx.sheet_names else []
+                else:
+                    # Comma-separated sheet names
+                    requested = [s.strip() for s in sheet_selection.split(",")]
+                    sheets_to_read = [s for s in requested if s in xlsx.sheet_names]
+                
+                dfs = {}
+                for sheet in sheets_to_read:
+                    dfs[sheet] = pd.read_excel(xlsx, sheet_name=sheet)
+                return dfs
+                
+        except Exception as e:
+            logging.error(f"Error parsing Excel file {filename}: {e}")
+            return {}
+
+    def _read_excels(self):
+        """Read and parse Excel files into structured data."""
+        all_data = {}
+        summaries = []
+        markdown_parts = []
+        
+        for file_ref in (self._param.input_files or []):
+            if self.check_if_canceled("ExcelProcessor reading"):
+                return
+                
+            # Get variable value
+            value = self._canvas.get_variable_value(file_ref)
+            self.set_input_value(file_ref, str(value)[:200] if value else "")
+            
+            if value is None:
+                continue
+            
+            # Handle file content
+            content, filename = self._get_file_content(file_ref)
+            if content is None:
+                continue
+                
+            # Parse Excel
+            dfs = self._parse_excel_to_dataframes(content, filename)
+            
+            for sheet_name, df in dfs.items():
+                key = f"{filename}_{sheet_name}" if len(dfs) > 1 else filename
+                all_data[key] = df.to_dict(orient="records")
+                
+                # Build summary
+                summaries.append(f"**{key}**: {len(df)} rows, {len(df.columns)} columns ({', '.join(df.columns.tolist()[:5])}{'...' if len(df.columns) > 5 else ''})")
+                
+                # Build markdown table
+                markdown_parts.append(f"### {key}\n\n{df.head(10).to_markdown(index=False)}\n")
+        
+        # Set outputs
+        self.set_output("data", all_data)
+        self.set_output("summary", "\n".join(summaries) if summaries else "No Excel files found")
+        self.set_output("markdown", "\n\n".join(markdown_parts) if markdown_parts else "No data")
+
+    def _merge_excels(self):
+        """Merge multiple Excel files/sheets into one."""
+        all_dfs = []
+        
+        for file_ref in (self._param.input_files or []):
+            if self.check_if_canceled("ExcelProcessor merging"):
+                return
+                
+            value = self._canvas.get_variable_value(file_ref)
+            self.set_input_value(file_ref, str(value)[:200] if value else "")
+            
+            if value is None:
+                continue
+                
+            content, filename = self._get_file_content(file_ref)
+            if content is None:
+                continue
+                
+            dfs = self._parse_excel_to_dataframes(content, filename)
+            all_dfs.extend(dfs.values())
+        
+        if not all_dfs:
+            self.set_output("data", {})
+            self.set_output("summary", "No data to merge")
+            return
+        
+        # Merge strategy
+        if self._param.merge_strategy == "concat":
+            merged_df = pd.concat(all_dfs, ignore_index=True)
+        elif self._param.merge_strategy == "join" and self._param.join_on:
+            # Join on specified column
+            merged_df = all_dfs[0]
+            for df in all_dfs[1:]:
+                merged_df = merged_df.merge(df, on=self._param.join_on, how="outer")
+        else:
+            merged_df = pd.concat(all_dfs, ignore_index=True)
+        
+        self.set_output("data", {"merged": merged_df.to_dict(orient="records")})
+        self.set_output("summary", f"Merged {len(all_dfs)} sources into {len(merged_df)} rows, {len(merged_df.columns)} columns")
+        self.set_output("markdown", merged_df.head(20).to_markdown(index=False))
+
+    def _transform_data(self):
+        """Apply transformations to data based on instructions or input data."""
+        # Get the data to transform
+        transform_ref = self._param.transform_data
+        if not transform_ref:
+            self.set_output("summary", "No transform data reference provided")
+            return
+            
+        data = self._canvas.get_variable_value(transform_ref)
+        self.set_input_value(transform_ref, str(data)[:300] if data else "")
+        
+        if data is None:
+            self.set_output("summary", "Transform data is empty")
+            return
+        
+        # Convert to DataFrame
+        if isinstance(data, dict):
+            # Could be {"sheet": [rows]} format
+            if all(isinstance(v, list) for v in data.values()):
+                # Multiple sheets
+                all_markdown = []
+                for sheet_name, rows in data.items():
+                    df = pd.DataFrame(rows)
+                    all_markdown.append(f"### {sheet_name}\n\n{df.to_markdown(index=False)}")
+                self.set_output("data", data)
+                self.set_output("markdown", "\n\n".join(all_markdown))
+            else:
+                df = pd.DataFrame([data])
+                self.set_output("data", df.to_dict(orient="records"))
+                self.set_output("markdown", df.to_markdown(index=False))
+        elif isinstance(data, list):
+            df = pd.DataFrame(data)
+            self.set_output("data", df.to_dict(orient="records"))
+            self.set_output("markdown", df.to_markdown(index=False))
+        else:
+            self.set_output("data", {"raw": str(data)})
+            self.set_output("markdown", str(data))
+        
+        self.set_output("summary", "Transformed data ready for processing")
+
+    def _output_excel(self):
+        """Generate Excel file output from data."""
+        # Get data from transform_data reference
+        transform_ref = self._param.transform_data
+        if not transform_ref:
+            self.set_output("summary", "No data reference for output")
+            return
+            
+        data = self._canvas.get_variable_value(transform_ref)
+        self.set_input_value(transform_ref, str(data)[:300] if data else "")
+        
+        if data is None:
+            self.set_output("summary", "No data to output")
+            return
+        
+        try:
+            # Prepare DataFrames
+            if isinstance(data, dict):
+                if all(isinstance(v, list) for v in data.values()):
+                    # Multi-sheet format
+                    dfs = {k: pd.DataFrame(v) for k, v in data.items()}
+                else:
+                    dfs = {"Sheet1": pd.DataFrame([data])}
+            elif isinstance(data, list):
+                dfs = {"Sheet1": pd.DataFrame(data)}
+            else:
+                self.set_output("summary", "Invalid data format for Excel output")
+                return
+            
+            # Generate output
+            doc_id = get_uuid()
+            
+            if self._param.output_format == "csv":
+                # For CSV, only output first sheet
+                first_df = list(dfs.values())[0]
+                binary_content = first_df.to_csv(index=False).encode("utf-8")
+                filename = f"{self._param.output_filename}.csv"
+            else:
+                # Excel output
+                excel_io = BytesIO()
+                with pd.ExcelWriter(excel_io, engine='openpyxl') as writer:
+                    for sheet_name, df in dfs.items():
+                        # Sanitize sheet name (max 31 chars, no special chars)
+                        safe_name = sheet_name[:31].replace("/", "_").replace("\\", "_")
+                        df.to_excel(writer, sheet_name=safe_name, index=False)
+                excel_io.seek(0)
+                binary_content = excel_io.read()
+                filename = f"{self._param.output_filename}.xlsx"
+            
+            # Store file
+            settings.STORAGE_IMPL.put(self._canvas._tenant_id, doc_id, binary_content)
+            
+            # Set attachment output
+            self.set_output("attachment", {
+                "doc_id": doc_id,
+                "format": self._param.output_format,
+                "file_name": filename
+            })
+            
+            total_rows = sum(len(df) for df in dfs.values())
+            self.set_output("summary", f"Generated {filename} with {len(dfs)} sheet(s), {total_rows} total rows")
+            self.set_output("data", {k: v.to_dict(orient="records") for k, v in dfs.items()})
+            
+            logging.info(f"ExcelProcessor: Generated {filename} as {doc_id}")
+            
+        except Exception as e:
+            logging.error(f"ExcelProcessor output error: {e}")
+            self.set_output("summary", f"Error generating output: {str(e)}")
+
+    def thoughts(self) -> str:
+        """Return component thoughts for UI display."""
+        op = self._param.operation
+        if op == "read":
+            return "Reading Excel files..."
+        elif op == "merge":
+            return "Merging Excel data..."
+        elif op == "transform":
+            return "Transforming data..."
+        elif op == "output":
+            return "Generating Excel output..."
+        return "Processing Excel..."

agent/component/exit_loop.py

@@ -13,26 +13,20 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-from agent.component.base import ComponentParamBase, ComponentBase
+from abc import ABC
+from agent.component.base import ComponentBase, ComponentParamBase
 
 
-class WebhookParam(ComponentParamBase):
-
-    """
-    Define the Begin component parameters.
-    """
-    def __init__(self):
-        super().__init__()
-
-    def get_input_form(self) -> dict[str, dict]:
-        return getattr(self, "inputs")
+class ExitLoopParam(ComponentParamBase, ABC):
+    def check(self):
+        return True
 
 
-class Webhook(ComponentBase):
-    component_name = "Webhook"
+class ExitLoop(ComponentBase, ABC):
+    component_name = "ExitLoop"
 
     def _invoke(self, **kwargs):
         pass
 
     def thoughts(self) -> str:
-        return ""
+        return ""

agent/component/fillup.py

@@ -18,6 +18,7 @@ import re
 from functools import partial
 
 from agent.component.base import ComponentParamBase, ComponentBase
+from api.db.services.file_service import FileService
 
 
 class UserFillUpParam(ComponentParamBase):
@@ -63,6 +64,13 @@ class UserFillUp(ComponentBase):
         for k, v in kwargs.get("inputs", {}).items():
             if self.check_if_canceled("UserFillUp processing"):
                 return
+            if isinstance(v, dict) and v.get("type", "").lower().find("file") >=0:
+                if v.get("optional") and v.get("value", None) is None:
+                    v = None
+                else:
+                    v = FileService.get_files([v["value"]])
+            else:
+                v = v.get("value")
             self.set_output(k, v)
 
     def thoughts(self) -> str:

agent/component/llm.py

@@ -13,12 +13,13 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import asyncio
 import json
 import logging
 import os
 import re
 from copy import deepcopy
-from typing import Any, Generator
+from typing import Any, AsyncGenerator
 import json_repair
 from functools import partial
 from common.constants import LLMType
@@ -55,7 +56,6 @@ class LLMParam(ComponentParamBase):
         self.check_nonnegative_number(int(self.max_tokens), "[Agent] Max tokens")
         self.check_decimal_float(float(self.top_p), "[Agent] Top P")
         self.check_empty(self.llm_id, "[Agent] LLM")
-        self.check_empty(self.sys_prompt, "[Agent] System prompt")
         self.check_empty(self.prompts, "[Agent] User prompt")
 
     def gen_conf(self):
@@ -166,25 +166,67 @@ class LLM(ComponentBase):
             sys_prompt = re.sub(rf"<{tag}>(.*?)</{tag}>", "", sys_prompt, flags=re.DOTALL|re.IGNORECASE)
         return pts, sys_prompt
 
-    def _generate(self, msg:list[dict], **kwargs) -> str:
+    async def _generate_async(self, msg: list[dict], **kwargs) -> str:
         if not self.imgs:
-            return self.chat_mdl.chat(msg[0]["content"], msg[1:], self._param.gen_conf(), **kwargs)
-        return self.chat_mdl.chat(msg[0]["content"], msg[1:], self._param.gen_conf(), images=self.imgs, **kwargs)
+            return await self.chat_mdl.async_chat(msg[0]["content"], msg[1:], self._param.gen_conf(), **kwargs)
+        return await self.chat_mdl.async_chat(msg[0]["content"], msg[1:], self._param.gen_conf(), images=self.imgs, **kwargs)
 
-    def _generate_streamly(self, msg:list[dict], **kwargs) -> Generator[str, None, None]:
-        ans = ""
+    async def _generate_streamly(self, msg: list[dict], **kwargs) -> AsyncGenerator[str, None]:
+        async def delta_wrapper(txt_iter):
+            ans = ""
+            last_idx = 0
+            endswith_think = False
+
+            def delta(txt):
+                nonlocal ans, last_idx, endswith_think
+                delta_ans = txt[last_idx:]
+                ans = txt
+
+                if delta_ans.find("<think>") == 0:
+                    last_idx += len("<think>")
+                    return "<think>"
+                elif delta_ans.find("<think>") > 0:
+                    delta_ans = txt[last_idx:last_idx + delta_ans.find("<think>")]
+                    last_idx += delta_ans.find("<think>")
+                    return delta_ans
+                elif delta_ans.endswith("</think>"):
+                    endswith_think = True
+                elif endswith_think:
+                    endswith_think = False
+                    return "</think>"
+
+                last_idx = len(ans)
+                if ans.endswith("</think>"):
+                    last_idx -= len("</think>")
+                return re.sub(r"(<think>|</think>)", "", delta_ans)
+
+            async for t in txt_iter:
+                yield delta(t)
+
+        if not self.imgs:
+            async for t in delta_wrapper(self.chat_mdl.async_chat_streamly(msg[0]["content"], msg[1:], self._param.gen_conf(), **kwargs)):
+                yield t
+            return
+
+        async for t in delta_wrapper(self.chat_mdl.async_chat_streamly(msg[0]["content"], msg[1:], self._param.gen_conf(), images=self.imgs, **kwargs)):
+            yield t
+
+    async def _stream_output_async(self, prompt, msg):
+        _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
+        answer = ""
         last_idx = 0
         endswith_think = False
+
         def delta(txt):
-            nonlocal ans, last_idx, endswith_think
+            nonlocal answer, last_idx, endswith_think
             delta_ans = txt[last_idx:]
-            ans = txt
+            answer = txt
 
             if delta_ans.find("<think>") == 0:
                 last_idx += len("<think>")
                 return "<think>"
             elif delta_ans.find("<think>") > 0:
-                delta_ans = txt[last_idx:last_idx+delta_ans.find("<think>")]
+                delta_ans = txt[last_idx:last_idx + delta_ans.find("<think>")]
                 last_idx += delta_ans.find("<think>")
                 return delta_ans
             elif delta_ans.endswith("</think>"):
@@ -193,20 +235,33 @@ class LLM(ComponentBase):
                 endswith_think = False
                 return "</think>"
 
-            last_idx = len(ans)
-            if ans.endswith("</think>"):
+            last_idx = len(answer)
+            if answer.endswith("</think>"):
                 last_idx -= len("</think>")
             return re.sub(r"(<think>|</think>)", "", delta_ans)
 
-        if not self.imgs:
-            for txt in self.chat_mdl.chat_streamly(msg[0]["content"], msg[1:], self._param.gen_conf(), **kwargs):
-                yield delta(txt)
-        else:
-            for txt in self.chat_mdl.chat_streamly(msg[0]["content"], msg[1:], self._param.gen_conf(), images=self.imgs, **kwargs):
-                yield delta(txt)
+        stream_kwargs = {"images": self.imgs} if self.imgs else {}
+        async for ans in self.chat_mdl.async_chat_streamly(msg[0]["content"], msg[1:], self._param.gen_conf(), **stream_kwargs):
+            if self.check_if_canceled("LLM streaming"):
+                return
+
+            if isinstance(ans, int):
+                continue
+
+            if ans.find("**ERROR**") >= 0:
+                if self.get_exception_default_value():
+                    self.set_output("content", self.get_exception_default_value())
+                    yield self.get_exception_default_value()
+                else:
+                    self.set_output("_ERROR", ans)
+                return
+
+            yield delta(ans)
+
+        self.set_output("content", answer)
 
     @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60)))
-    def _invoke(self, **kwargs):
+    async def _invoke_async(self, **kwargs):
         if self.check_if_canceled("LLM processing"):
             return
 
@@ -217,22 +272,25 @@ class LLM(ComponentBase):
 
         prompt, msg, _ = self._prepare_prompt_variables()
         error: str = ""
-        output_structure=None
+        output_structure = None
         try:
-            output_structure = self._param.outputs['structured']
+            output_structure = self._param.outputs["structured"]
         except Exception:
             pass
-        if output_structure and isinstance(output_structure, dict) and output_structure.get("properties"):
-            schema=json.dumps(output_structure, ensure_ascii=False, indent=2)
-            prompt += structured_output_prompt(schema)
-            for _ in range(self._param.max_retries+1):
+        if output_structure and isinstance(output_structure, dict) and output_structure.get("properties") and len(output_structure["properties"]) > 0:
+            schema = json.dumps(output_structure, ensure_ascii=False, indent=2)
+            prompt_with_schema = prompt + structured_output_prompt(schema)
+            for _ in range(self._param.max_retries + 1):
                 if self.check_if_canceled("LLM processing"):
                     return
 
-                _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
+                _, msg_fit = message_fit_in(
+                    [{"role": "system", "content": prompt_with_schema}, *deepcopy(msg)],
+                    int(self.chat_mdl.max_length * 0.97),
+                )
                 error = ""
-                ans = self._generate(msg)
-                msg.pop(0)
+                ans = await self._generate_async(msg_fit)
+                msg_fit.pop(0)
                 if ans.find("**ERROR**") >= 0:
                     logging.error(f"LLM response error: {ans}")
                     error = ans
@@ -241,7 +299,7 @@ class LLM(ComponentBase):
                     self.set_output("structured", json_repair.loads(clean_formated_answer(ans)))
                     return
                 except Exception:
-                    msg.append({"role": "user", "content": "The answer can't not be parsed as JSON"})
+                    msg_fit.append({"role": "user", "content": "The answer can't not be parsed as JSON"})
                     error = "The answer can't not be parsed as JSON"
             if error:
                 self.set_output("_ERROR", error)
@@ -249,18 +307,23 @@ class LLM(ComponentBase):
 
         downstreams = self._canvas.get_component(self._id)["downstream"] if self._canvas.get_component(self._id) else []
         ex = self.exception_handler()
-        if any([self._canvas.get_component_obj(cid).component_name.lower()=="message" for cid in downstreams]) and not (ex and ex["goto"]):
-            self.set_output("content", partial(self._stream_output, prompt, msg))
+        if any([self._canvas.get_component_obj(cid).component_name.lower() == "message" for cid in downstreams]) and not (
+            ex and ex["goto"]
+        ):
+            self.set_output("content", partial(self._stream_output_async, prompt, deepcopy(msg)))
             return
 
-        for _ in range(self._param.max_retries+1):
+        error = ""
+        for _ in range(self._param.max_retries + 1):
             if self.check_if_canceled("LLM processing"):
                 return
 
-            _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
+            _, msg_fit = message_fit_in(
+                [{"role": "system", "content": prompt}, *deepcopy(msg)], int(self.chat_mdl.max_length * 0.97)
+            )
             error = ""
-            ans = self._generate(msg)
-            msg.pop(0)
+            ans = await self._generate_async(msg_fit)
+            msg_fit.pop(0)
             if ans.find("**ERROR**") >= 0:
                 logging.error(f"LLM response error: {ans}")
                 error = ans
@@ -274,26 +337,12 @@ class LLM(ComponentBase):
             else:
                 self.set_output("_ERROR", error)
 
-    def _stream_output(self, prompt, msg):
-        _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
-        answer = ""
-        for ans in self._generate_streamly(msg):
-            if self.check_if_canceled("LLM streaming"):
-                return
+    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60)))
+    def _invoke(self, **kwargs):
+        return asyncio.run(self._invoke_async(**kwargs))
 
-            if ans.find("**ERROR**") >= 0:
-                if self.get_exception_default_value():
-                    self.set_output("content", self.get_exception_default_value())
-                    yield self.get_exception_default_value()
-                else:
-                    self.set_output("_ERROR", ans)
-                return
-            yield ans
-            answer += ans
-        self.set_output("content", answer)
-
-    def add_memory(self, user:str, assist:str, func_name: str, params: dict, results: str, user_defined_prompt:dict={}):
-        summ = tool_call_summary(self.chat_mdl, func_name, params, results, user_defined_prompt)
+    async def add_memory(self, user:str, assist:str, func_name: str, params: dict, results: str, user_defined_prompt:dict={}):
+        summ = await tool_call_summary(self.chat_mdl, func_name, params, results, user_defined_prompt)
         logging.info(f"[MEMORY]: {summ}")
         self._canvas.add_memory(user, assist, summ)
 

agent/component/loop.py

@@ -0,0 +1,80 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+from abc import ABC
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class LoopParam(ComponentParamBase):
+    """
+    Define the Loop component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.loop_variables = []
+        self.loop_termination_condition=[]
+        self.maximum_loop_count = 0
+
+    def get_input_form(self) -> dict[str, dict]:
+        return {
+            "items": {
+                "type": "json",
+                "name": "Items"
+            }
+        }
+
+    def check(self):
+        return True
+
+
+class Loop(ComponentBase, ABC):
+    component_name = "Loop"
+
+    def get_start(self):
+        for cid in self._canvas.components.keys():
+            if self._canvas.get_component(cid)["obj"].component_name.lower() != "loopitem":
+                continue
+            if self._canvas.get_component(cid)["parent_id"] == self._id:
+                return cid
+
+    def _invoke(self, **kwargs):
+        if self.check_if_canceled("Loop processing"):
+            return
+
+        for item in self._param.loop_variables:
+            if any([not item.get("variable"), not item.get("input_mode"), not item.get("value"),not item.get("type")]):
+                assert "Loop Variable is not complete."
+            if item["input_mode"]=="variable":
+                self.set_output(item["variable"],self._canvas.get_variable_value(item["value"]))
+            elif item["input_mode"]=="constant":
+                self.set_output(item["variable"],item["value"])
+            else:
+                if item["type"] == "number":
+                    self.set_output(item["variable"], 0)
+                elif item["type"] == "string":
+                    self.set_output(item["variable"], "")
+                elif item["type"] == "boolean":
+                    self.set_output(item["variable"], False)
+                elif item["type"].startswith("object"):
+                    self.set_output(item["variable"], {})
+                elif item["type"].startswith("array"):
+                    self.set_output(item["variable"], [])
+                else:
+                    self.set_output(item["variable"], "")
+
+
+    def thoughts(self) -> str:
+        return "Loop from canvas."

agent/component/loopitem.py

@@ -0,0 +1,167 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+from abc import ABC
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class LoopItemParam(ComponentParamBase):
+    """
+    Define the LoopItem component parameters.
+    """
+    def check(self):
+        return True
+
+class LoopItem(ComponentBase, ABC):
+    component_name = "LoopItem"
+
+    def __init__(self, canvas, id, param: ComponentParamBase):
+        super().__init__(canvas, id, param)
+        self._idx = 0
+
+
+    def _invoke(self, **kwargs):
+        if self.check_if_canceled("LoopItem processing"):
+            return
+        parent = self.get_parent()
+        maximum_loop_count = parent._param.maximum_loop_count
+        if self._idx >= maximum_loop_count:
+            self._idx = -1
+            return
+        if self._idx > 0:
+            if self.check_if_canceled("LoopItem processing"):
+                return
+        self._idx += 1
+
+    def evaluate_condition(self,var, operator, value):
+        if isinstance(var, str):
+            if operator == "contains":
+                return value in var
+            elif operator == "not contains":
+                return value not in var
+            elif operator == "start with":
+                return var.startswith(value)
+            elif operator == "end with":
+                return var.endswith(value)
+            elif operator == "is":
+                return var == value
+            elif operator == "is not":
+                return var != value
+            elif operator == "empty":
+                return var == ""
+            elif operator == "not empty":
+                return var != ""
+
+        elif isinstance(var, (int, float)):
+            if operator == "=":
+                return var == value
+            elif operator == "≠":
+                return var != value
+            elif operator == ">":
+                return var > value
+            elif operator == "<":
+                return var < value
+            elif operator == "≥":
+                return var >= value
+            elif operator == "≤":
+                return var <= value
+            elif operator == "empty":
+                return var is None
+            elif operator == "not empty":
+                return var is not None
+
+        elif isinstance(var, bool):
+            if operator == "is":
+                return var is value
+            elif operator == "is not":
+                return var is not value
+            elif operator == "empty":
+                return var is None
+            elif operator == "not empty":
+                return var is not None
+
+        elif isinstance(var, dict):
+            if operator == "empty":
+                return len(var) == 0
+            elif operator == "not empty":
+                return len(var) > 0
+
+        elif isinstance(var, list):
+            if operator == "contains":
+                return value in var
+            elif operator == "not contains":
+                return value not in var
+
+            elif operator == "is":
+                return var == value
+            elif operator == "is not":
+                return var != value
+
+            elif operator == "empty":
+                return len(var) == 0
+            elif operator == "not empty":
+                return len(var) > 0
+        elif var is None:
+            if operator == "empty":
+                return True
+            return False
+
+        raise Exception(f"Invalid operator: {operator}")
+
+    def end(self):
+        if self._idx == -1:
+            return True
+        parent = self.get_parent()
+        logical_operator = parent._param.logical_operator if hasattr(parent._param, "logical_operator") else "and"
+        conditions = []
+        for item in parent._param.loop_termination_condition:
+            if not item.get("variable") or not item.get("operator"):
+                raise ValueError("Loop condition is incomplete.")
+            var = self._canvas.get_variable_value(item["variable"])
+            operator = item["operator"]
+            input_mode = item.get("input_mode", "constant")
+
+            if input_mode == "variable":
+                value = self._canvas.get_variable_value(item.get("value", ""))
+            elif input_mode == "constant":
+                value = item.get("value", "")
+            else:
+                raise ValueError("Invalid input mode.")
+            conditions.append(self.evaluate_condition(var, operator, value))
+        should_end = (
+            all(conditions) if logical_operator == "and"
+            else any(conditions) if logical_operator == "or"
+            else None
+        )
+        if should_end is None:
+            raise ValueError("Invalid logical operator,should be 'and' or 'or'.")
+
+        if should_end:
+            self._idx = -1
+            return True
+
+        return False
+
+    def next(self):
+        if self._idx == -1:
+            self._idx = 0
+        else:
+            self._idx += 1
+            if self._idx >= len(self._items):
+                self._idx = -1
+        return False
+
+    def thoughts(self) -> str:
+        return "Next turn..."

agent/component/message.py

@@ -13,6 +13,10 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import asyncio
+import nest_asyncio
+nest_asyncio.apply()
+import inspect
 import json
 import os
 import random
@@ -29,6 +33,8 @@ from common.connection_utils import timeout
 from common.misc_utils import get_uuid
 from common import settings
 
+from api.db.joint_services.memory_message_service import queue_save_to_memory_task
+
 
 class MessageParam(ComponentParamBase):
     """
@@ -39,6 +45,7 @@ class MessageParam(ComponentParamBase):
         self.content = []
         self.stream = True
         self.output_format = None  # default output format
+        self.auto_play = False
         self.outputs = {
             "content": {
                 "type": "str"
@@ -66,8 +73,12 @@ class Message(ComponentBase):
                 v = ""
             ans = ""
             if isinstance(v, partial):
-                for t in v():
-                    ans += t
+                iter_obj = v()
+                if inspect.isasyncgen(iter_obj):
+                    ans = asyncio.run(self._consume_async_gen(iter_obj))
+                else:
+                    for t in iter_obj:
+                        ans += t
             elif isinstance(v, list) and delimiter:
                 ans = delimiter.join([str(vv) for vv in v])
             elif not isinstance(v, str):
@@ -89,7 +100,13 @@ class Message(ComponentBase):
             _kwargs[_n] = v
         return script, _kwargs
 
-    def _stream(self, rand_cnt:str):
+    async def _consume_async_gen(self, agen):
+        buf = ""
+        async for t in agen:
+            buf += t
+        return buf
+
+    async def _stream(self, rand_cnt:str):
         s = 0
         all_content = ""
         cache = {}
@@ -111,15 +128,27 @@ class Message(ComponentBase):
                 v = ""
             if isinstance(v, partial):
                 cnt = ""
-                for t in v():
-                    if self.check_if_canceled("Message streaming"):
-                        return
+                iter_obj = v()
+                if inspect.isasyncgen(iter_obj):
+                    async for t in iter_obj:
+                        if self.check_if_canceled("Message streaming"):
+                            return
 
-                    all_content += t
-                    cnt += t
-                    yield t
+                        all_content += t
+                        cnt += t
+                        yield t
+                else:
+                    for t in iter_obj:
+                        if self.check_if_canceled("Message streaming"):
+                            return
+
+                        all_content += t
+                        cnt += t
+                        yield t
                 self.set_input_value(exp, cnt)
                 continue
+            elif inspect.isawaitable(v):
+                v = await v
             elif not isinstance(v, str):
                 try:
                     v = json.dumps(v, ensure_ascii=False)
@@ -139,6 +168,7 @@ class Message(ComponentBase):
 
         self.set_output("content", all_content)
         self._convert_content(all_content)
+        await self._save_to_memory(all_content)
 
     def _is_jinjia2(self, content:str) -> bool:
         patt = [
@@ -171,18 +201,61 @@ class Message(ComponentBase):
 
         self.set_output("content", content)
         self._convert_content(content)
+        self._save_to_memory(content)
 
     def thoughts(self) -> str:
         return ""
 
+    def _parse_markdown_table_lines(self, table_lines: list):
+        """
+        Parse a list of Markdown table lines into a pandas DataFrame.
+        
+        Args:
+            table_lines: List of strings, each representing a row in the Markdown table
+                        (excluding separator lines like |---|---|)
+        
+        Returns:
+            pandas DataFrame with the table data, or None if parsing fails
+        """
+        import pandas as pd
+        
+        if not table_lines:
+            return None
+        
+        rows = []
+        headers = None
+        
+        for line in table_lines:
+            # Split by | and clean up
+            cells = [cell.strip() for cell in line.split('|')]
+            # Remove empty first and last elements from split (caused by leading/trailing |)
+            cells = [c for c in cells if c]
+            
+            if headers is None:
+                headers = cells
+            else:
+                rows.append(cells)
+        
+        if headers and rows:
+            # Ensure all rows have same number of columns as headers
+            normalized_rows = []
+            for row in rows:
+                while len(row) < len(headers):
+                    row.append('')
+                normalized_rows.append(row[:len(headers)])
+            
+            return pd.DataFrame(normalized_rows, columns=headers)
+        
+        return None
+
     def _convert_content(self, content):
         if not self._param.output_format:
             return
 
         import pypandoc
         doc_id = get_uuid()
-        
-        if self._param.output_format.lower() not in {"markdown", "html", "pdf", "docx"}:
+
+        if self._param.output_format.lower() not in {"markdown", "html", "pdf", "docx", "xlsx"}:
             self._param.output_format = "markdown"
 
         try:
@@ -202,6 +275,119 @@ class Message(ComponentBase):
 
                 binary_content = converted.encode("utf-8")
 
+            elif self._param.output_format == "xlsx":
+                import pandas as pd
+                from io import BytesIO
+
+                # Debug: log the content being parsed
+                logging.info(f"XLSX Parser: Content length={len(content) if content else 0}, first 500 chars: {content[:500] if content else 'None'}")
+                
+                # Try to parse ALL Markdown tables from the content
+                # Each table will be written to a separate sheet
+                tables = []  # List of (sheet_name, dataframe)
+                
+                if isinstance(content, str):
+                    lines = content.strip().split('\n')
+                    logging.info(f"XLSX Parser: Total lines={len(lines)}, lines starting with '|': {sum(1 for line in lines if line.strip().startswith('|'))}")
+                    current_table_lines = []
+                    current_table_title = None
+                    pending_title = None
+                    in_table = False
+                    table_count = 0
+                    
+                    for i, line in enumerate(lines):
+                        stripped = line.strip()
+                        
+                        # Check for potential table title (lines before a table)
+                        # Look for patterns like "Table 1:", "## Table", or markdown headers
+                        if not in_table and stripped and not stripped.startswith('|'):
+                            # Check if this could be a table title
+                            lower_stripped = stripped.lower()
+                            if (lower_stripped.startswith('table') or 
+                                stripped.startswith('#') or
+                                ':' in stripped):
+                                pending_title = stripped.lstrip('#').strip()
+                        
+                        if stripped.startswith('|') and '|' in stripped[1:]:
+                            # Check if this is a separator line (|---|---|)
+                            cleaned = stripped.replace(' ', '').replace('|', '').replace('-', '').replace(':', '')
+                            if cleaned == '':
+                                continue  # Skip separator line
+                            
+                            if not in_table:
+                                # Starting a new table
+                                in_table = True
+                                current_table_lines = []
+                                current_table_title = pending_title
+                                pending_title = None
+                            
+                            current_table_lines.append(stripped)
+                        
+                        elif in_table and not stripped.startswith('|'):
+                            # End of current table - save it
+                            if current_table_lines:
+                                df = self._parse_markdown_table_lines(current_table_lines)
+                                if df is not None and not df.empty:
+                                    table_count += 1
+                                    # Generate sheet name
+                                    if current_table_title:
+                                        # Clean and truncate title for sheet name
+                                        sheet_name = current_table_title[:31]
+                                        sheet_name = sheet_name.replace('/', '_').replace('\\', '_').replace('*', '').replace('?', '').replace('[', '').replace(']', '').replace(':', '')
+                                    else:
+                                        sheet_name = f"Table_{table_count}"
+                                    tables.append((sheet_name, df))
+                            
+                            # Reset for next table
+                            in_table = False
+                            current_table_lines = []
+                            current_table_title = None
+                            
+                            # Check if this line could be a title for the next table
+                            if stripped:
+                                lower_stripped = stripped.lower()
+                                if (lower_stripped.startswith('table') or 
+                                    stripped.startswith('#') or
+                                    ':' in stripped):
+                                    pending_title = stripped.lstrip('#').strip()
+                    
+                    # Don't forget the last table if content ends with a table
+                    if in_table and current_table_lines:
+                        df = self._parse_markdown_table_lines(current_table_lines)
+                        if df is not None and not df.empty:
+                            table_count += 1
+                            if current_table_title:
+                                sheet_name = current_table_title[:31]
+                                sheet_name = sheet_name.replace('/', '_').replace('\\', '_').replace('*', '').replace('?', '').replace('[', '').replace(']', '').replace(':', '')
+                            else:
+                                sheet_name = f"Table_{table_count}"
+                            tables.append((sheet_name, df))
+                
+                # Fallback: if no tables found, create single sheet with content
+                if not tables:
+                    df = pd.DataFrame({"Content": [content if content else ""]})
+                    tables = [("Data", df)]
+
+                # Write all tables to Excel, each in a separate sheet
+                excel_io = BytesIO()
+                with pd.ExcelWriter(excel_io, engine='openpyxl') as writer:
+                    used_names = set()
+                    for sheet_name, df in tables:
+                        # Ensure unique sheet names
+                        original_name = sheet_name
+                        counter = 1
+                        while sheet_name in used_names:
+                            suffix = f"_{counter}"
+                            sheet_name = original_name[:31-len(suffix)] + suffix
+                            counter += 1
+                        used_names.add(sheet_name)
+                        df.to_excel(writer, sheet_name=sheet_name, index=False)
+                
+                excel_io.seek(0)
+                binary_content = excel_io.read()
+                
+                logging.info(f"Generated Excel with {len(tables)} sheet(s): {[t[0] for t in tables]}")
+
             else:  # pdf, docx
                 with tempfile.NamedTemporaryFile(suffix=f".{self._param.output_format}", delete=False) as tmp:
                     tmp_name = tmp.name
@@ -231,11 +417,24 @@ class Message(ComponentBase):
 
             settings.STORAGE_IMPL.put(self._canvas._tenant_id, doc_id, binary_content)
             self.set_output("attachment", {
-                "doc_id":doc_id, 
-                "format":self._param.output_format, 
+                "doc_id":doc_id,
+                "format":self._param.output_format,
                 "file_name":f"{doc_id[:8]}.{self._param.output_format}"})
 
             logging.info(f"Converted content uploaded as {doc_id} (format={self._param.output_format})")
 
         except Exception as e:
-            logging.error(f"Error converting content to {self._param.output_format}: {e}")
+            logging.error(f"Error converting content to {self._param.output_format}: {e}")
+
+    async def _save_to_memory(self, content):
+        if not hasattr(self._param, "memory_ids") or not self._param.memory_ids:
+            return True, "No memory selected."
+
+        message_dict = {
+            "user_id": self._canvas._tenant_id,
+            "agent_id": self._canvas._id,
+            "session_id": self._canvas.task_id,
+            "user_input": self._canvas.get_sys_query(),
+            "agent_response": content
+        }
+        return await queue_save_to_memory_task(self._param.memory_ids, message_dict)

agent/templates/advanced_ingestion_pipeline.json

@@ -193,7 +193,7 @@
                         "presence_penalty": 0.4,
                         "prompts": [
                             {
-                                "content": "Text Content:\n{Splitter:KindDingosJam@chunks}\n",
+                                "content": "Text Content:\n{Splitter:NineTiesSin@chunks}\n",
                                 "role": "user"
                             }
                         ],
@@ -226,7 +226,7 @@
                         "presence_penalty": 0.4,
                         "prompts": [
                             {
-                                "content": "Text Content:\n\n{Splitter:KindDingosJam@chunks}\n",
+                                "content": "Text Content:\n\n{Splitter:TastyPointsLay@chunks}\n",
                                 "role": "user"
                             }
                         ],
@@ -259,7 +259,7 @@
                         "presence_penalty": 0.4,
                         "prompts": [
                             {
-                                "content": "Content: \n\n{Splitter:KindDingosJam@chunks}",
+                                "content": "Content: \n\n{Splitter:CuteBusesBet@chunks}",
                                 "role": "user"
                             }
                         ],
@@ -485,7 +485,7 @@
                             "outputs": {},
                             "presencePenaltyEnabled": false,
                             "presence_penalty": 0.4,
-                            "prompts": "Text Content:\n{Splitter:KindDingosJam@chunks}\n",
+                            "prompts": "Text Content:\n{Splitter:NineTiesSin@chunks}\n",
                             "sys_prompt": "Role\nYou are a text analyzer.\n\nTask\nExtract the most important keywords/phrases of a given piece of text content.\n\nRequirements\n- Summarize the text content, and give the top 5 important keywords/phrases.\n- The keywords MUST be in the same language as the given piece of text content.\n- The keywords are delimited by ENGLISH COMMA.\n- Output keywords ONLY.",
                             "temperature": 0.1,
                             "temperatureEnabled": false,
@@ -522,7 +522,7 @@
                             "outputs": {},
                             "presencePenaltyEnabled": false,
                             "presence_penalty": 0.4,
-                            "prompts": "Text Content:\n\n{Splitter:KindDingosJam@chunks}\n",
+                            "prompts": "Text Content:\n\n{Splitter:TastyPointsLay@chunks}\n",
                             "sys_prompt": "Role\nYou are a text analyzer.\n\nTask\nPropose 3 questions about a given piece of text content.\n\nRequirements\n- Understand and summarize the text content, and propose the top 3 important questions.\n- The questions SHOULD NOT have overlapping meanings.\n- The questions SHOULD cover the main content of the text as much as possible.\n- The questions MUST be in the same language as the given piece of text content.\n- One question per line.\n- Output questions ONLY.",
                             "temperature": 0.1,
                             "temperatureEnabled": false,
@@ -559,7 +559,7 @@
                             "outputs": {},
                             "presencePenaltyEnabled": false,
                             "presence_penalty": 0.4,
-                            "prompts": "Content: \n\n{Splitter:KindDingosJam@chunks}",
+                            "prompts": "Content: \n\n{Splitter:BlueResultsWink@chunks}",
                             "sys_prompt": "Extract important structured information from the given content. Output ONLY a valid JSON string with no additional text. If no important structured information is found, output an empty JSON object: {}.\n\nImportant structured information may include: names, dates, locations, events, key facts, numerical data, or other extractable entities.",
                             "temperature": 0.1,
                             "temperatureEnabled": false,

agent/templates/sql_assistant.json

@@ -578,7 +578,7 @@
                     {
                         "data": {
                             "form": {
-                                "text": "Searches for relevant database creation statements.\n\nIt should label with a knowledgebase to which the schema is dumped in. You could use \" General \" as parsing method, \" 2 \" as chunk size and \" ; \"  as delimiter."
+                                "text": "Searches for relevant database creation statements.\n\nIt should label with a dataset to which the schema is dumped in. You could use \" General \" as parsing method, \" 2 \" as chunk size and \" ; \"  as delimiter."
                             },
                             "label": "Note",
                             "name": "Note Schema"

agent/test/dsl_examples/categorize_and_agent_with_tavily.json

@@ -75,7 +75,7 @@
   },
   "history": [],
   "path": [],
-  "retrival": {"chunks": [], "doc_aggs": []},
+  "retrieval": {"chunks": [], "doc_aggs": []},
   "globals": {
     "sys.query": "",
     "sys.user_id": "",

agent/test/dsl_examples/iteration.json

@@ -82,7 +82,7 @@
   },
   "history": [],
   "path": [],
-  "retrival": {"chunks": [], "doc_aggs": []},
+  "retrieval": {"chunks": [], "doc_aggs": []},
   "globals": {
     "sys.query": "",
     "sys.user_id": "",

agent/test/dsl_examples/retrieval_and_generate.json

@@ -31,7 +31,7 @@
                     "component_name": "LLM",
                     "params": {
                       "llm_id": "deepseek-chat",
-                      "sys_prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {retrieval:0@formalized_content}\n      The above is the knowledge base.",
+                      "sys_prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {retrieval:0@formalized_content}\n      Above is the knowledge base.",
                       "temperature": 0.2
                     }
                 },
@@ -51,7 +51,7 @@
   },
   "history": [],
   "path": [],
-  "retrival": {"chunks": [], "doc_aggs": []},
+  "retrieval": {"chunks": [], "doc_aggs": []},
   "globals": {
     "sys.query": "",
     "sys.user_id": "",

agent/test/dsl_examples/retrieval_categorize_and_generate.json

@@ -65,7 +65,7 @@
                     "component_name": "Agent",
                     "params": {
                       "llm_id": "deepseek-chat",
-                      "sys_prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {retrieval:0@formalized_content}\n      The above is the knowledge base.",
+                      "sys_prompt": "You are an intelligent assistant. Please summarize the content of the dataset to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {retrieval:0@formalized_content}\n      The above is the knowledge base.",
                       "temperature": 0.2
                     }
                 },
@@ -85,7 +85,7 @@
   },
   "history": [],
   "path": [],
-  "retrival": {"chunks": [], "doc_aggs": []},
+  "retrieval": {"chunks": [], "doc_aggs": []},
   "globals": {
     "sys.query": "",
     "sys.user_id": "",

agent/test/dsl_examples/tavily_and_generate.json

@@ -25,7 +25,7 @@
                     "component_name": "LLM",
                     "params": {
                       "llm_id": "deepseek-chat",
-                      "sys_prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {tavily:0@formalized_content}\n      The above is the knowledge base.",
+                      "sys_prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {tavily:0@formalized_content}\n      Above is the knowledge base.",
                       "temperature": 0.2
                     }
                 },
@@ -45,7 +45,7 @@
   },
   "history": [],
   "path": [],
-  "retrival": {"chunks": [], "doc_aggs": []},
+  "retrieval": {"chunks": [], "doc_aggs": []},
   "globals": {
     "sys.query": "",
     "sys.user_id": "",

agent/tools/base.py

@@ -17,6 +17,7 @@ import logging
 import re
 import time
 from copy import deepcopy
+import asyncio
 from functools import partial
 from typing import TypedDict, List, Any
 from agent.component.base import ComponentParamBase, ComponentBase
@@ -26,6 +27,10 @@ from common.mcp_tool_call_conn import MCPToolCallSession, ToolCallSession
 from timeit import default_timer as timer
 
 
+
+
+from common.misc_utils import thread_pool_exec
+
 class ToolParameter(TypedDict):
     type: str
     description: str
@@ -48,12 +53,19 @@ class LLMToolPluginCallSession(ToolCallSession):
         self.callback = callback
 
     def tool_call(self, name: str, arguments: dict[str, Any]) -> Any:
+        return asyncio.run(self.tool_call_async(name, arguments))
+
+    async def tool_call_async(self, name: str, arguments: dict[str, Any]) -> Any:
         assert name in self.tools_map, f"LLM tool {name} does not exist"
         st = timer()
-        if isinstance(self.tools_map[name], MCPToolCallSession):
-            resp = self.tools_map[name].tool_call(name, arguments, 60)
+        tool_obj = self.tools_map[name]
+        if isinstance(tool_obj, MCPToolCallSession):
+            resp = await thread_pool_exec(tool_obj.tool_call, name, arguments, 60)
         else:
-            resp = self.tools_map[name].invoke(**arguments)
+            if hasattr(tool_obj, "invoke_async") and asyncio.iscoroutinefunction(tool_obj.invoke_async):
+                resp = await tool_obj.invoke_async(**arguments)
+            else:
+                resp = await thread_pool_exec(tool_obj.invoke, **arguments)
 
         self.callback(name, arguments, resp, elapsed_time=timer()-st)
         return resp
@@ -114,6 +126,7 @@ class ToolParamBase(ComponentParamBase):
 class ToolBase(ComponentBase):
     def __init__(self, canvas, id, param: ComponentParamBase):
         from agent.canvas import Canvas  # Local import to avoid cyclic dependency
+
         assert isinstance(canvas, Canvas), "canvas must be an instance of Canvas"
         self._canvas = canvas
         self._id = id
@@ -139,6 +152,33 @@ class ToolBase(ComponentBase):
         self.set_output("_elapsed_time", time.perf_counter() - self.output("_created_time"))
         return res
 
+    async def invoke_async(self, **kwargs):
+        """
+        Async wrapper for tool invocation.
+        If `_invoke` is a coroutine, await it directly; otherwise run in a thread to avoid blocking.
+        Mirrors the exception handling of `invoke`.
+        """
+        if self.check_if_canceled("Tool processing"):
+            return
+
+        self.set_output("_created_time", time.perf_counter())
+        try:
+            fn_async = getattr(self, "_invoke_async", None)
+            if fn_async and asyncio.iscoroutinefunction(fn_async):
+                res = await fn_async(**kwargs)
+            elif asyncio.iscoroutinefunction(self._invoke):
+                res = await self._invoke(**kwargs)
+            else:
+                res = await thread_pool_exec(self._invoke, **kwargs)
+        except Exception as e:
+            self._param.outputs["_ERROR"] = {"value": str(e)}
+            logging.exception(e)
+            res = str(e)
+        self._param.debug_inputs = []
+
+        self.set_output("_elapsed_time", time.perf_counter() - self.output("_created_time"))
+        return res
+
     def _retrieve_chunks(self, res_list: list, get_title, get_url, get_content, get_score=None):
         chunks = []
         aggs = []

agent/tools/code_exec.py

@@ -13,16 +13,20 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import ast
 import base64
+import json
 import logging
 import os
 from abc import ABC
-from strenum import StrEnum
 from typing import Optional
+
 from pydantic import BaseModel, Field, field_validator
-from agent.tools.base import ToolParamBase, ToolBase, ToolMeta
-from common.connection_utils import timeout
+from strenum import StrEnum
+
+from agent.tools.base import ToolBase, ToolMeta, ToolParamBase
 from common import settings
+from common.connection_utils import timeout
 
 
 class Language(StrEnum):
@@ -62,10 +66,10 @@ class CodeExecParam(ToolParamBase):
     """
 
     def __init__(self):
-        self.meta:ToolMeta = {
+        self.meta: ToolMeta = {
             "name": "execute_code",
             "description": """
-This tool has a sandbox that can execute code written in 'Python'/'Javascript'. It recieves a piece of code and return a Json string.
+This tool has a sandbox that can execute code written in 'Python'/'Javascript'. It receives a piece of code and return a Json string.
 Here's a code example for Python(`main` function MUST be included):
 def main() -> dict:
     \"\"\"
@@ -99,16 +103,12 @@ module.exports = { main };
                     "enum": ["python", "javascript"],
                     "required": True,
                 },
-                "script": {
-                    "type": "string",
-                    "description": "A piece of code in right format. There MUST be main function.",
-                    "required": True
-                }
-            }
+                "script": {"type": "string", "description": "A piece of code in right format. There MUST be main function.", "required": True},
+            },
         }
         super().__init__()
         self.lang = Language.PYTHON.value
-        self.script = "def main(arg1: str, arg2: str) -> dict: return {\"result\": arg1 + arg2}"
+        self.script = 'def main(arg1: str, arg2: str) -> dict: return {"result": arg1 + arg2}'
         self.arguments = {}
         self.outputs = {"result": {"value": "", "type": "string"}}
 
@@ -119,17 +119,14 @@ module.exports = { main };
     def get_input_form(self) -> dict[str, dict]:
         res = {}
         for k, v in self.arguments.items():
-            res[k] = {
-                "type": "line",
-                "name": k
-            }
+            res[k] = {"type": "line", "name": k}
         return res
 
 
 class CodeExec(ToolBase, ABC):
     component_name = "CodeExec"
 
-    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60)))
+    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10 * 60)))
     def _invoke(self, **kwargs):
         if self.check_if_canceled("CodeExec processing"):
             return
@@ -138,17 +135,12 @@ class CodeExec(ToolBase, ABC):
         script = kwargs.get("script", self._param.script)
         arguments = {}
         for k, v in self._param.arguments.items():
-
             if kwargs.get(k):
                 arguments[k] = kwargs[k]
                 continue
             arguments[k] = self._canvas.get_variable_value(v) if v else None
 
-        self._execute_code(
-            language=lang,
-            code=script,
-            arguments=arguments
-        )
+        self._execute_code(language=lang, code=script, arguments=arguments)
 
     def _execute_code(self, language: str, code: str, arguments: dict):
         import requests
@@ -169,7 +161,7 @@ class CodeExec(ToolBase, ABC):
             if self.check_if_canceled("CodeExec execution"):
                 return "Task has been canceled"
 
-            resp = requests.post(url=f"http://{settings.SANDBOX_HOST}:9385/run", json=code_req, timeout=int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60)))
+            resp = requests.post(url=f"http://{settings.SANDBOX_HOST}:9385/run", json=code_req, timeout=int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10 * 60)))
             logging.info(f"http://{settings.SANDBOX_HOST}:9385/run,  code_req: {code_req}, resp.status_code {resp.status_code}:")
 
             if self.check_if_canceled("CodeExec execution"):
@@ -183,35 +175,10 @@ class CodeExec(ToolBase, ABC):
                 if stderr:
                     self.set_output("_ERROR", stderr)
                     return
-                try:
-                    rt = eval(body.get("stdout", ""))
-                except Exception:
-                    rt = body.get("stdout", "")
-                logging.info(f"http://{settings.SANDBOX_HOST}:9385/run -> {rt}")
-                if isinstance(rt, tuple):
-                    for i, (k, o) in enumerate(self._param.outputs.items()):
-                        if self.check_if_canceled("CodeExec execution"):
-                            return
-
-                        if k.find("_") == 0:
-                            continue
-                        o["value"] = rt[i]
-                elif isinstance(rt, dict):
-                    for i, (k, o) in enumerate(self._param.outputs.items()):
-                        if self.check_if_canceled("CodeExec execution"):
-                            return
-
-                        if k not in rt or k.find("_") == 0:
-                            continue
-                        o["value"] = rt[k]
-                else:
-                    for i, (k, o) in enumerate(self._param.outputs.items()):
-                        if self.check_if_canceled("CodeExec execution"):
-                            return
-
-                        if k.find("_") == 0:
-                            continue
-                        o["value"] = rt
+                raw_stdout = body.get("stdout", "")
+                parsed_stdout = self._deserialize_stdout(raw_stdout)
+                logging.info(f"[CodeExec]: http://{settings.SANDBOX_HOST}:9385/run -> {parsed_stdout}")
+                self._populate_outputs(parsed_stdout, raw_stdout)
             else:
                 self.set_output("_ERROR", "There is no response from sandbox")
 
@@ -228,3 +195,149 @@ class CodeExec(ToolBase, ABC):
 
     def thoughts(self) -> str:
         return "Running a short script to process data."
+
+    def _deserialize_stdout(self, stdout: str):
+        text = str(stdout).strip()
+        if not text:
+            return ""
+        for loader in (json.loads, ast.literal_eval):
+            try:
+                return loader(text)
+            except Exception:
+                continue
+        return text
+
+    def _coerce_output_value(self, value, expected_type: Optional[str]):
+        if expected_type is None:
+            return value
+
+        etype = expected_type.strip().lower()
+        inner_type = None
+        if etype.startswith("array<") and etype.endswith(">"):
+            inner_type = etype[6:-1].strip()
+            etype = "array"
+
+        try:
+            if etype == "string":
+                return "" if value is None else str(value)
+
+            if etype == "number":
+                if value is None or value == "":
+                    return None
+                if isinstance(value, (int, float)):
+                    return value
+                if isinstance(value, str):
+                    try:
+                        return float(value)
+                    except Exception:
+                        return value
+                return float(value)
+
+            if etype == "boolean":
+                if isinstance(value, bool):
+                    return value
+                if isinstance(value, str):
+                    lv = value.lower()
+                    if lv in ("true", "1", "yes", "y", "on"):
+                        return True
+                    if lv in ("false", "0", "no", "n", "off"):
+                        return False
+                return bool(value)
+
+            if etype == "array":
+                candidate = value
+                if isinstance(candidate, str):
+                    parsed = self._deserialize_stdout(candidate)
+                    candidate = parsed
+                if isinstance(candidate, tuple):
+                    candidate = list(candidate)
+                if not isinstance(candidate, list):
+                    candidate = [] if candidate is None else [candidate]
+
+                if inner_type == "string":
+                    return ["" if v is None else str(v) for v in candidate]
+                if inner_type == "number":
+                    coerced = []
+                    for v in candidate:
+                        try:
+                            if v is None or v == "":
+                                coerced.append(None)
+                            elif isinstance(v, (int, float)):
+                                coerced.append(v)
+                            else:
+                                coerced.append(float(v))
+                        except Exception:
+                            coerced.append(v)
+                    return coerced
+                return candidate
+
+            if etype == "object":
+                if isinstance(value, dict):
+                    return value
+                if isinstance(value, str):
+                    parsed = self._deserialize_stdout(value)
+                    if isinstance(parsed, dict):
+                        return parsed
+                return value
+        except Exception:
+            return value
+
+        return value
+
+    def _populate_outputs(self, parsed_stdout, raw_stdout: str):
+        outputs_items = list(self._param.outputs.items())
+        logging.info(f"[CodeExec]: outputs schema keys: {[k for k, _ in outputs_items]}")
+        if not outputs_items:
+            return
+
+        if isinstance(parsed_stdout, dict):
+            for key, meta in outputs_items:
+                if key.startswith("_"):
+                    continue
+                val = self._get_by_path(parsed_stdout, key)
+                coerced = self._coerce_output_value(val, meta.get("type"))
+                logging.info(f"[CodeExec]: populate dict key='{key}' raw='{val}' coerced='{coerced}'")
+                self.set_output(key, coerced)
+            return
+
+        if isinstance(parsed_stdout, (list, tuple)):
+            for idx, (key, meta) in enumerate(outputs_items):
+                if key.startswith("_"):
+                    continue
+                val = parsed_stdout[idx] if idx < len(parsed_stdout) else None
+                coerced = self._coerce_output_value(val, meta.get("type"))
+                logging.info(f"[CodeExec]: populate list key='{key}' raw='{val}' coerced='{coerced}'")
+                self.set_output(key, coerced)
+            return
+
+        default_val = parsed_stdout if parsed_stdout is not None else raw_stdout
+        for idx, (key, meta) in enumerate(outputs_items):
+            if key.startswith("_"):
+                continue
+            val = default_val if idx == 0 else None
+            coerced = self._coerce_output_value(val, meta.get("type"))
+            logging.info(f"[CodeExec]: populate scalar key='{key}' raw='{val}' coerced='{coerced}'")
+            self.set_output(key, coerced)
+
+    def _get_by_path(self, data, path: str):
+        if not path:
+            return None
+        cur = data
+        for part in path.split("."):
+            part = part.strip()
+            if not part:
+                return None
+            if isinstance(cur, dict):
+                cur = cur.get(part)
+            elif isinstance(cur, list):
+                try:
+                    idx = int(part)
+                    cur = cur[idx]
+                except Exception:
+                    return None
+            else:
+                return None
+            if cur is None:
+                return None
+        logging.info(f"[CodeExec]: resolve path '{path}' -> {cur}")
+        return cur

agent/tools/exesql.py

@@ -86,6 +86,12 @@ class ExeSQL(ToolBase, ABC):
 
         def convert_decimals(obj):
             from decimal import Decimal
+            import math
+            if isinstance(obj, float):
+                # Handle NaN and Infinity which are not valid JSON values
+                if math.isnan(obj) or math.isinf(obj):
+                    return None
+                return obj
             if isinstance(obj, Decimal):
                 return float(obj)  # 或 str(obj)
             elif isinstance(obj, dict):

agent/tools/retrieval.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import asyncio
 from functools import partial
 import json
 import os
@@ -21,13 +22,15 @@ from abc import ABC
 from agent.tools.base import ToolParamBase, ToolBase, ToolMeta
 from common.constants import LLMType
 from api.db.services.document_service import DocumentService
-from api.db.services.dialog_service import meta_filter
+from common.metadata_utils import apply_meta_data_filter
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMBundle
+from api.db.services.memory_service import MemoryService
+from api.db.joint_services import memory_message_service
 from common import settings
 from common.connection_utils import timeout
 from rag.app.tag import label_question
-from rag.prompts.generator import cross_languages, kb_prompt, gen_meta_filter
+from rag.prompts.generator import cross_languages, kb_prompt, memory_prompt
 
 
 class RetrievalParam(ToolParamBase):
@@ -56,6 +59,7 @@ class RetrievalParam(ToolParamBase):
         self.top_n = 8
         self.top_k = 1024
         self.kb_ids = []
+        self.memory_ids = []
         self.kb_vars = []
         self.rerank_id = ""
         self.empty_response = ""
@@ -80,15 +84,7 @@ class RetrievalParam(ToolParamBase):
 class Retrieval(ToolBase, ABC):
     component_name = "Retrieval"
 
-    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 12)))
-    def _invoke(self, **kwargs):
-        if self.check_if_canceled("Retrieval processing"):
-            return
-
-        if not kwargs.get("query"):
-            self.set_output("formalized_content", self._param.empty_response)
-            return
-
+    async def _retrieve_kb(self, query_text: str):
         kb_ids: list[str] = []
         for id in self._param.kb_ids:
             if id.find("@") < 0:
@@ -123,58 +119,62 @@ class Retrieval(ToolBase, ABC):
         if self._param.rerank_id:
             rerank_mdl = LLMBundle(kbs[0].tenant_id, LLMType.RERANK, self._param.rerank_id)
 
-        vars = self.get_input_elements_from_text(kwargs["query"])
-        vars = {k:o["value"] for k,o in vars.items()}
-        query = self.string_format(kwargs["query"], vars)
+        vars = self.get_input_elements_from_text(query_text)
+        vars = {k: o["value"] for k, o in vars.items()}
+        query = self.string_format(query_text, vars)
 
-        doc_ids=[]
-        if self._param.meta_data_filter!={}:
+        doc_ids = []
+        if self._param.meta_data_filter != {}:
             metas = DocumentService.get_meta_by_kbs(kb_ids)
-            if self._param.meta_data_filter.get("method") == "auto":
+
+            def _resolve_manual_filter(flt: dict) -> dict:
+                pat = re.compile(self.variable_ref_patt)
+                s = flt.get("value", "")
+                out_parts = []
+                last = 0
+
+                for m in pat.finditer(s):
+                    out_parts.append(s[last:m.start()])
+                    key = m.group(1)
+                    v = self._canvas.get_variable_value(key)
+                    if v is None:
+                        rep = ""
+                    elif isinstance(v, partial):
+                        buf = []
+                        for chunk in v():
+                            buf.append(chunk)
+                        rep = "".join(buf)
+                    elif isinstance(v, str):
+                        rep = v
+                    else:
+                        rep = json.dumps(v, ensure_ascii=False)
+
+                    out_parts.append(rep)
+                    last = m.end()
+
+                out_parts.append(s[last:])
+                flt["value"] = "".join(out_parts)
+                return flt
+
+            chat_mdl = None
+            if self._param.meta_data_filter.get("method") in ["auto", "semi_auto"]:
                 chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT)
-                filters: dict = gen_meta_filter(chat_mdl, metas, query)
-                doc_ids.extend(meta_filter(metas, filters["conditions"], filters.get("logic", "and")))
-                if not doc_ids:
-                    doc_ids = None
-            elif self._param.meta_data_filter.get("method") == "manual":
-                filters = self._param.meta_data_filter["manual"]
-                for flt in filters:
-                    pat = re.compile(self.variable_ref_patt)
-                    s = flt["value"]
-                    out_parts = []
-                    last = 0
 
-                    for m in pat.finditer(s):
-                        out_parts.append(s[last:m.start()])
-                        key = m.group(1)
-                        v = self._canvas.get_variable_value(key)
-                        if v is None:
-                            rep = ""
-                        elif isinstance(v, partial):
-                            buf = []
-                            for chunk in v():
-                                buf.append(chunk)
-                            rep = "".join(buf)
-                        elif isinstance(v, str):
-                            rep = v
-                        else:
-                            rep = json.dumps(v, ensure_ascii=False)
-
-                        out_parts.append(rep)
-                        last = m.end()
-
-                    out_parts.append(s[last:])
-                    flt["value"] = "".join(out_parts)
-                doc_ids.extend(meta_filter(metas, filters, self._param.meta_data_filter.get("logic", "and")))
-                if filters and not doc_ids:
-                    doc_ids = ["-999"]
+            doc_ids = await apply_meta_data_filter(
+                self._param.meta_data_filter,
+                metas,
+                query,
+                chat_mdl,
+                doc_ids,
+                _resolve_manual_filter if self._param.meta_data_filter.get("method") == "manual" else None,
+            )
 
         if self._param.cross_languages:
-            query = cross_languages(kbs[0].tenant_id, None, query, self._param.cross_languages)
+            query = await cross_languages(kbs[0].tenant_id, None, query, self._param.cross_languages)
 
         if kbs:
             query = re.sub(r"^user[:：\s]*", "", query, flags=re.IGNORECASE)
-            kbinfos = settings.retriever.retrieval(
+            kbinfos = await settings.retriever.retrieval(
                 query,
                 embd_mdl,
                 [kb.tenant_id for kb in kbs],
@@ -193,17 +193,20 @@ class Retrieval(ToolBase, ABC):
 
             if self._param.toc_enhance:
                 chat_mdl = LLMBundle(self._canvas._tenant_id, LLMType.CHAT)
-                cks = settings.retriever.retrieval_by_toc(query, kbinfos["chunks"], [kb.tenant_id for kb in kbs], chat_mdl, self._param.top_n)
+                cks = await settings.retriever.retrieval_by_toc(query, kbinfos["chunks"], [kb.tenant_id for kb in kbs],
+                                                          chat_mdl, self._param.top_n)
                 if self.check_if_canceled("Retrieval processing"):
                     return
                 if cks:
                     kbinfos["chunks"] = cks
+            kbinfos["chunks"] = settings.retriever.retrieval_by_children(kbinfos["chunks"],
+                                                                         [kb.tenant_id for kb in kbs])
             if self._param.use_kg:
-                ck = settings.kg_retriever.retrieval(query,
-                                                       [kb.tenant_id for kb in kbs],
-                                                       kb_ids,
-                                                       embd_mdl,
-                                                       LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT))
+                ck = await settings.kg_retriever.retrieval(query,
+                                                     [kb.tenant_id for kb in kbs],
+                                                     kb_ids,
+                                                     embd_mdl,
+                                                     LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT))
                 if self.check_if_canceled("Retrieval processing"):
                     return
                 if ck["content_with_weight"]:
@@ -212,7 +215,8 @@ class Retrieval(ToolBase, ABC):
             kbinfos = {"chunks": [], "doc_aggs": []}
 
         if self._param.use_kg and kbs:
-            ck = settings.kg_retriever.retrieval(query, [kb.tenant_id for kb in kbs], filtered_kb_ids, embd_mdl, LLMBundle(kbs[0].tenant_id, LLMType.CHAT))
+            ck = await settings.kg_retriever.retrieval(query, [kb.tenant_id for kb in kbs], filtered_kb_ids, embd_mdl,
+                                                 LLMBundle(kbs[0].tenant_id, LLMType.CHAT))
             if self.check_if_canceled("Retrieval processing"):
                 return
             if ck["content_with_weight"]:
@@ -242,6 +246,58 @@ class Retrieval(ToolBase, ABC):
 
         return form_cnt
 
+    async def _retrieve_memory(self, query_text: str):
+        memory_ids: list[str] = [memory_id for memory_id in self._param.memory_ids]
+        memory_list = MemoryService.get_by_ids(memory_ids)
+        if not memory_list:
+            raise Exception("No memory is selected.")
+
+        embd_names = list({memory.embd_id for memory in memory_list})
+        assert len(embd_names) == 1, "Memory use different embedding models."
+
+        vars = self.get_input_elements_from_text(query_text)
+        vars = {k: o["value"] for k, o in vars.items()}
+        query = self.string_format(query_text, vars)
+        # query message
+        message_list = memory_message_service.query_message({"memory_id": memory_ids}, {
+            "query": query,
+            "similarity_threshold": self._param.similarity_threshold,
+            "keywords_similarity_weight": self._param.keywords_similarity_weight,
+            "top_n": self._param.top_n
+        })
+        if not message_list:
+            self.set_output("formalized_content", self._param.empty_response)
+            return ""
+        formated_content = "\n".join(memory_prompt(message_list, 200000))
+        # set formalized_content output
+        self.set_output("formalized_content", formated_content)
+
+        return formated_content
+
+    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 12)))
+    async def _invoke_async(self, **kwargs):
+        if self.check_if_canceled("Retrieval processing"):
+            return
+        if not kwargs.get("query"):
+            self.set_output("formalized_content", self._param.empty_response)
+            return
+
+        if hasattr(self._param, "retrieval_from") and self._param.retrieval_from == "dataset":
+            return await self._retrieve_kb(kwargs["query"])
+        elif hasattr(self._param, "retrieval_from") and self._param.retrieval_from == "memory":
+            return await self._retrieve_memory(kwargs["query"])
+        elif self._param.kb_ids:
+            return await self._retrieve_kb(kwargs["query"])
+        elif hasattr(self._param, "memory_ids") and self._param.memory_ids:
+            return await self._retrieve_memory(kwargs["query"])
+        else:
+            self.set_output("formalized_content", self._param.empty_response)
+            return
+
+    @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 12)))
+    def _invoke(self, **kwargs):
+        return asyncio.run(self._invoke_async(**kwargs))
+
     def thoughts(self) -> str:
         return """
 Keywords: {}

agent/tools/yahoofinance.py

@@ -75,7 +75,7 @@ class YahooFinance(ToolBase, ABC):
     @timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 60)))
     def _invoke(self, **kwargs):
         if self.check_if_canceled("YahooFinance processing"):
-            return
+            return None
 
         if not kwargs.get("stock_code"):
             self.set_output("report", "")
@@ -84,33 +84,33 @@ class YahooFinance(ToolBase, ABC):
         last_e = ""
         for _ in range(self._param.max_retries+1):
             if self.check_if_canceled("YahooFinance processing"):
-                return
+                return None
 
-            yohoo_res = []
+            yahoo_res = []
             try:
                 msft = yf.Ticker(kwargs["stock_code"])
                 if self.check_if_canceled("YahooFinance processing"):
-                    return
+                    return None
 
                 if self._param.info:
-                    yohoo_res.append("# Information:\n" + pd.Series(msft.info).to_markdown() + "\n")
+                    yahoo_res.append("# Information:\n" + pd.Series(msft.info).to_markdown() + "\n")
                 if self._param.history:
-                    yohoo_res.append("# History:\n" + msft.history().to_markdown() + "\n")
+                    yahoo_res.append("# History:\n" + msft.history().to_markdown() + "\n")
                 if self._param.financials:
-                    yohoo_res.append("# Calendar:\n" + pd.DataFrame(msft.calendar).to_markdown() + "\n")
+                    yahoo_res.append("# Calendar:\n" + pd.DataFrame(msft.calendar).to_markdown() + "\n")
                 if self._param.balance_sheet:
-                    yohoo_res.append("# Balance sheet:\n" + msft.balance_sheet.to_markdown() + "\n")
-                    yohoo_res.append("# Quarterly balance sheet:\n" + msft.quarterly_balance_sheet.to_markdown() + "\n")
+                    yahoo_res.append("# Balance sheet:\n" + msft.balance_sheet.to_markdown() + "\n")
+                    yahoo_res.append("# Quarterly balance sheet:\n" + msft.quarterly_balance_sheet.to_markdown() + "\n")
                 if self._param.cash_flow_statement:
-                    yohoo_res.append("# Cash flow statement:\n" + msft.cashflow.to_markdown() + "\n")
-                    yohoo_res.append("# Quarterly cash flow statement:\n" + msft.quarterly_cashflow.to_markdown() + "\n")
+                    yahoo_res.append("# Cash flow statement:\n" + msft.cashflow.to_markdown() + "\n")
+                    yahoo_res.append("# Quarterly cash flow statement:\n" + msft.quarterly_cashflow.to_markdown() + "\n")
                 if self._param.news:
-                    yohoo_res.append("# News:\n" + pd.DataFrame(msft.news).to_markdown() + "\n")
-                self.set_output("report", "\n\n".join(yohoo_res))
+                    yahoo_res.append("# News:\n" + pd.DataFrame(msft.news).to_markdown() + "\n")
+                self.set_output("report", "\n\n".join(yahoo_res))
                 return self.output("report")
             except Exception as e:
                 if self.check_if_canceled("YahooFinance processing"):
-                    return
+                    return None
 
                 last_e = e
                 logging.exception(f"YahooFinance error: {e}")

agentic_reasoning/__init__.py

@@ -1 +0,0 @@
-from .deep_research import DeepResearcher as DeepResearcher

agentic_reasoning/deep_research.py

@@ -1,236 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-import re
-from functools import partial
-from agentic_reasoning.prompts import BEGIN_SEARCH_QUERY, BEGIN_SEARCH_RESULT, END_SEARCH_RESULT, MAX_SEARCH_LIMIT, \
-    END_SEARCH_QUERY, REASON_PROMPT, RELEVANT_EXTRACTION_PROMPT
-from api.db.services.llm_service import LLMBundle
-from rag.nlp import extract_between
-from rag.prompts import kb_prompt
-from rag.utils.tavily_conn import Tavily
-
-
-class DeepResearcher:
-    def __init__(self,
-                 chat_mdl: LLMBundle,
-                 prompt_config: dict,
-                 kb_retrieve: partial = None,
-                 kg_retrieve: partial = None
-                 ):
-        self.chat_mdl = chat_mdl
-        self.prompt_config = prompt_config
-        self._kb_retrieve = kb_retrieve
-        self._kg_retrieve = kg_retrieve
-
-    def _remove_tags(text: str, start_tag: str, end_tag: str) -> str:
-        """General Tag Removal Method"""
-        pattern = re.escape(start_tag) + r"(.*?)" + re.escape(end_tag)
-        return re.sub(pattern, "", text)
-
-    @staticmethod
-    def _remove_query_tags(text: str) -> str:
-        """Remove Query Tags"""
-        return DeepResearcher._remove_tags(text, BEGIN_SEARCH_QUERY, END_SEARCH_QUERY)
-
-    @staticmethod
-    def _remove_result_tags(text: str) -> str:
-        """Remove Result Tags"""
-        return DeepResearcher._remove_tags(text, BEGIN_SEARCH_RESULT, END_SEARCH_RESULT)
-
-    def _generate_reasoning(self, msg_history):
-        """Generate reasoning steps"""
-        query_think = ""
-        if msg_history[-1]["role"] != "user":
-            msg_history.append({"role": "user", "content": "Continues reasoning with the new information.\n"})
-        else:
-            msg_history[-1]["content"] += "\n\nContinues reasoning with the new information.\n"
-            
-        for ans in self.chat_mdl.chat_streamly(REASON_PROMPT, msg_history, {"temperature": 0.7}):
-            ans = re.sub(r"^.*</think>", "", ans, flags=re.DOTALL)
-            if not ans:
-                continue
-            query_think = ans
-            yield query_think
-        return query_think
-
-    def _extract_search_queries(self, query_think, question, step_index):
-        """Extract search queries from thinking"""
-        queries = extract_between(query_think, BEGIN_SEARCH_QUERY, END_SEARCH_QUERY)
-        if not queries and step_index == 0:
-            # If this is the first step and no queries are found, use the original question as the query
-            queries = [question]
-        return queries
-
-    def _truncate_previous_reasoning(self, all_reasoning_steps):
-        """Truncate previous reasoning steps to maintain a reasonable length"""
-        truncated_prev_reasoning = ""
-        for i, step in enumerate(all_reasoning_steps):
-            truncated_prev_reasoning += f"Step {i + 1}: {step}\n\n"
-
-        prev_steps = truncated_prev_reasoning.split('\n\n')
-        if len(prev_steps) <= 5:
-            truncated_prev_reasoning = '\n\n'.join(prev_steps)
-        else:
-            truncated_prev_reasoning = ''
-            for i, step in enumerate(prev_steps):
-                if i == 0 or i >= len(prev_steps) - 4 or BEGIN_SEARCH_QUERY in step or BEGIN_SEARCH_RESULT in step:
-                    truncated_prev_reasoning += step + '\n\n'
-                else:
-                    if truncated_prev_reasoning[-len('\n\n...\n\n'):] != '\n\n...\n\n':
-                        truncated_prev_reasoning += '...\n\n'
-        
-        return truncated_prev_reasoning.strip('\n')
-
-    def _retrieve_information(self, search_query):
-        """Retrieve information from different sources"""
-        # 1. Knowledge base retrieval
-        kbinfos = []
-        try:
-            kbinfos = self._kb_retrieve(question=search_query) if self._kb_retrieve else {"chunks": [], "doc_aggs": []}
-        except Exception as e:
-            logging.error(f"Knowledge base retrieval error: {e}")
-
-        # 2. Web retrieval (if Tavily API is configured)
-        try:
-            if self.prompt_config.get("tavily_api_key"):
-                tav = Tavily(self.prompt_config["tavily_api_key"])
-                tav_res = tav.retrieve_chunks(search_query)
-                kbinfos["chunks"].extend(tav_res["chunks"])
-                kbinfos["doc_aggs"].extend(tav_res["doc_aggs"])
-        except Exception as e:
-            logging.error(f"Web retrieval error: {e}")
-
-        # 3. Knowledge graph retrieval (if configured)
-        try:
-            if self.prompt_config.get("use_kg") and self._kg_retrieve:
-                ck = self._kg_retrieve(question=search_query)
-                if ck["content_with_weight"]:
-                    kbinfos["chunks"].insert(0, ck)
-        except Exception as e:
-            logging.error(f"Knowledge graph retrieval error: {e}")
-
-        return kbinfos
-
-    def _update_chunk_info(self, chunk_info, kbinfos):
-        """Update chunk information for citations"""
-        if not chunk_info["chunks"]:
-            # If this is the first retrieval, use the retrieval results directly
-            for k in chunk_info.keys():
-                chunk_info[k] = kbinfos[k]
-        else:
-            # Merge newly retrieved information, avoiding duplicates
-            cids = [c["chunk_id"] for c in chunk_info["chunks"]]
-            for c in kbinfos["chunks"]:
-                if c["chunk_id"] not in cids:
-                    chunk_info["chunks"].append(c)
-                    
-            dids = [d["doc_id"] for d in chunk_info["doc_aggs"]]
-            for d in kbinfos["doc_aggs"]:
-                if d["doc_id"] not in dids:
-                    chunk_info["doc_aggs"].append(d)
-
-    def _extract_relevant_info(self, truncated_prev_reasoning, search_query, kbinfos):
-        """Extract and summarize relevant information"""
-        summary_think = ""
-        for ans in self.chat_mdl.chat_streamly(
-                RELEVANT_EXTRACTION_PROMPT.format(
-                    prev_reasoning=truncated_prev_reasoning,
-                    search_query=search_query,
-                    document="\n".join(kb_prompt(kbinfos, 4096))
-                ),
-                [{"role": "user",
-                  "content": f'Now you should analyze each web page and find helpful information based on the current search query "{search_query}" and previous reasoning steps.'}],
-                {"temperature": 0.7}):
-            ans = re.sub(r"^.*</think>", "", ans, flags=re.DOTALL)
-            if not ans:
-                continue
-            summary_think = ans
-            yield summary_think
-        
-        return summary_think
-
-    def thinking(self, chunk_info: dict, question: str):
-        executed_search_queries = []
-        msg_history = [{"role": "user", "content": f'Question:\"{question}\"\n'}]
-        all_reasoning_steps = []
-        think = "<think>"
-        
-        for step_index in range(MAX_SEARCH_LIMIT + 1):
-            # Check if the maximum search limit has been reached
-            if step_index == MAX_SEARCH_LIMIT - 1:
-                summary_think = f"\n{BEGIN_SEARCH_RESULT}\nThe maximum search limit is exceeded. You are not allowed to search.\n{END_SEARCH_RESULT}\n"
-                yield {"answer": think + summary_think + "</think>", "reference": {}, "audio_binary": None}
-                all_reasoning_steps.append(summary_think)
-                msg_history.append({"role": "assistant", "content": summary_think})
-                break
-
-            # Step 1: Generate reasoning
-            query_think = ""
-            for ans in self._generate_reasoning(msg_history):
-                query_think = ans
-                yield {"answer": think + self._remove_query_tags(query_think) + "</think>", "reference": {}, "audio_binary": None}
-
-            think += self._remove_query_tags(query_think)
-            all_reasoning_steps.append(query_think)
-            
-            # Step 2: Extract search queries
-            queries = self._extract_search_queries(query_think, question, step_index)
-            if not queries and step_index > 0:
-                # If not the first step and no queries, end the search process
-                break
-
-            # Process each search query
-            for search_query in queries:
-                logging.info(f"[THINK]Query: {step_index}. {search_query}")
-                msg_history.append({"role": "assistant", "content": search_query})
-                think += f"\n\n> {step_index + 1}. {search_query}\n\n"
-                yield {"answer": think + "</think>", "reference": {}, "audio_binary": None}
-
-                # Check if the query has already been executed
-                if search_query in executed_search_queries:
-                    summary_think = f"\n{BEGIN_SEARCH_RESULT}\nYou have searched this query. Please refer to previous results.\n{END_SEARCH_RESULT}\n"
-                    yield {"answer": think + summary_think + "</think>", "reference": {}, "audio_binary": None}
-                    all_reasoning_steps.append(summary_think)
-                    msg_history.append({"role": "user", "content": summary_think})
-                    think += summary_think
-                    continue
-                
-                executed_search_queries.append(search_query)
-                
-                # Step 3: Truncate previous reasoning steps
-                truncated_prev_reasoning = self._truncate_previous_reasoning(all_reasoning_steps)
-                
-                # Step 4: Retrieve information
-                kbinfos = self._retrieve_information(search_query)
-                
-                # Step 5: Update chunk information
-                self._update_chunk_info(chunk_info, kbinfos)
-                
-                # Step 6: Extract relevant information
-                think += "\n\n"
-                summary_think = ""
-                for ans in self._extract_relevant_info(truncated_prev_reasoning, search_query, kbinfos):
-                    summary_think = ans
-                    yield {"answer": think + self._remove_result_tags(summary_think) + "</think>", "reference": {}, "audio_binary": None}
-
-                all_reasoning_steps.append(summary_think)
-                msg_history.append(
-                    {"role": "user", "content": f"\n\n{BEGIN_SEARCH_RESULT}{summary_think}{END_SEARCH_RESULT}\n\n"})
-                think += self._remove_result_tags(summary_think)
-                logging.info(f"[THINK]Summary: {step_index}. {summary_think}")
-
-        yield think + "</think>"

agentic_reasoning/prompts.py

@@ -1,147 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-
-BEGIN_SEARCH_QUERY = "<|begin_search_query|>"
-END_SEARCH_QUERY = "<|end_search_query|>"
-BEGIN_SEARCH_RESULT = "<|begin_search_result|>"
-END_SEARCH_RESULT = "<|end_search_result|>"
-MAX_SEARCH_LIMIT = 6
-
-REASON_PROMPT = f"""You are an advanced reasoning agent. Your goal is to answer the user's question by breaking it down into a series of verifiable steps.
-
-You have access to a powerful search tool to find information.
-
-**Your Task:**
-1.  Analyze the user's question.
-2.  If you need information, issue a search query to find a specific fact.
-3.  Review the search results.
-4.  Repeat the search process until you have all the facts needed to answer the question.
-5.  Once you have gathered sufficient information, synthesize the facts and provide the final answer directly.
-
-**Tool Usage:**
-- To search, you MUST write your query between the special tokens: {BEGIN_SEARCH_QUERY}your query{END_SEARCH_QUERY}.
-- The system will provide results between {BEGIN_SEARCH_RESULT}search results{END_SEARCH_RESULT}.
-- You have a maximum of {MAX_SEARCH_LIMIT} search attempts.
-
----
-**Example 1: Multi-hop Question**
-
-**Question:** "Are both the directors of Jaws and Casino Royale from the same country?"
-
-**Your Thought Process & Actions:**
-First, I need to identify the director of Jaws.
-{BEGIN_SEARCH_QUERY}who is the director of Jaws?{END_SEARCH_QUERY}
-[System returns search results]
-{BEGIN_SEARCH_RESULT}
-Jaws is a 1975 American thriller film directed by Steven Spielberg.
-{END_SEARCH_RESULT}
-Okay, the director of Jaws is Steven Spielberg. Now I need to find out his nationality.
-{BEGIN_SEARCH_QUERY}where is Steven Spielberg from?{END_SEARCH_QUERY}
-[System returns search results]
-{BEGIN_SEARCH_RESULT}
-Steven Allan Spielberg is an American filmmaker. Born in Cincinnati, Ohio...
-{END_SEARCH_RESULT}
-So, Steven Spielberg is from the USA. Next, I need to find the director of Casino Royale.
-{BEGIN_SEARCH_QUERY}who is the director of Casino Royale 2006?{END_SEARCH_QUERY}
-[System returns search results]
-{BEGIN_SEARCH_RESULT}
-Casino Royale is a 2006 spy film directed by Martin Campbell.
-{END_SEARCH_RESULT}
-The director of Casino Royale is Martin Campbell. Now I need his nationality.
-{BEGIN_SEARCH_QUERY}where is Martin Campbell from?{END_SEARCH_QUERY}
-[System returns search results]
-{BEGIN_SEARCH_RESULT}
-Martin Campbell (born 24 October 1943) is a New Zealand film and television director.
-{END_SEARCH_RESULT}
-I have all the information. Steven Spielberg is from the USA, and Martin Campbell is from New Zealand. They are not from the same country.
-
-Final Answer: No, the directors of Jaws and Casino Royale are not from the same country. Steven Spielberg is from the USA, and Martin Campbell is from New Zealand.
-
----
-**Example 2: Simple Fact Retrieval**
-
-**Question:** "When was the founder of craigslist born?"
-
-**Your Thought Process & Actions:**
-First, I need to know who founded craigslist.
-{BEGIN_SEARCH_QUERY}who founded craigslist?{END_SEARCH_QUERY}
-[System returns search results]
-{BEGIN_SEARCH_RESULT}
-Craigslist was founded in 1995 by Craig Newmark.
-{END_SEARCH_RESULT}
-The founder is Craig Newmark. Now I need his birth date.
-{BEGIN_SEARCH_QUERY}when was Craig Newmark born?{END_SEARCH_QUERY}
-[System returns search results]
-{BEGIN_SEARCH_RESULT}
-Craig Newmark was born on December 6, 1952.
-{END_SEARCH_RESULT}
-I have found the answer.
-
-Final Answer: The founder of craigslist, Craig Newmark, was born on December 6, 1952.
-
----
-**Important Rules:**
-- **One Fact at a Time:** Decompose the problem and issue one search query at a time to find a single, specific piece of information.
-- **Be Precise:** Formulate clear and precise search queries. If a search fails, rephrase it.
-- **Synthesize at the End:** Do not provide the final answer until you have completed all necessary searches.
-- **Language Consistency:** Your search queries should be in the same language as the user's question.
-
-Now, begin your work. Please answer the following question by thinking step-by-step.
-"""
-
-RELEVANT_EXTRACTION_PROMPT = """You are a highly efficient information extraction module. Your sole purpose is to extract the single most relevant piece of information from the provided `Searched Web Pages` that directly answers the `Current Search Query`.
-
-**Your Task:**
-1.  Read the `Current Search Query` to understand what specific information is needed.
-2.  Scan the `Searched Web Pages` to find the answer to that query.
-3.  Extract only the essential, factual information that answers the query. Be concise.
-
-**Context (For Your Information Only):**
-The `Previous Reasoning Steps` are provided to give you context on the overall goal, but your primary focus MUST be on answering the `Current Search Query`. Do not use information from the previous steps in your output.
-
-**Output Format:**
-Your response must follow one of two formats precisely.
-
-1.  **If a direct and relevant answer is found:**
-    - Start your response immediately with `Final Information`.
-    - Provide only the extracted fact(s). Do not add any extra conversational text.
-
-    *Example:*
-    `Current Search Query`: Where is Martin Campbell from?
-    `Searched Web Pages`: [Long article snippet about Martin Campbell's career, which includes the sentence "Martin Campbell (born 24 October 1943) is a New Zealand film and television director..."]
-    
-    *Your Output:*
-    Final Information
-    Martin Campbell is a New Zealand film and television director.
-
-2.  **If no relevant answer that directly addresses the query is found in the web pages:**
-    - Start your response immediately with `Final Information`.
-    - Write the exact phrase: `No helpful information found.`
-
----
-**BEGIN TASK**
-
-**Inputs:**
-
-- **Previous Reasoning Steps:**
-{prev_reasoning}
-
-- **Current Search Query:**
-{search_query}
-
-- **Searched Web Pages:**
-{document}
-"""

api/__init__.py

@@ -14,5 +14,5 @@
 #  limitations under the License.
 #
 
-from beartype.claw import beartype_this_package
-beartype_this_package()
+# from beartype.claw import beartype_this_package
+# beartype_this_package()

api/apps/__init__.py

@@ -13,26 +13,26 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 import os
 import sys
-import logging
+import time
 from importlib.util import module_from_spec, spec_from_file_location
 from pathlib import Path
-from quart import Blueprint, Quart, request, g, current_app, session
-from werkzeug.wrappers.request import Request
-from flasgger import Swagger
+from quart import Blueprint, Quart, request, g, current_app, session, jsonify
 from itsdangerous.url_safe import URLSafeTimedSerializer as Serializer
 from quart_cors import cors
-from common.constants import StatusEnum
+from common.constants import StatusEnum, RetCode
 from api.db.db_models import close_connection, APIToken
 from api.db.services import UserService
 from api.utils.json_encode import CustomJSONEncoder
 from api.utils import commands
 
-from flask_mail import Mail
-from quart_auth import Unauthorized
+from quart_auth import Unauthorized as QuartAuthUnauthorized
+from werkzeug.exceptions import Unauthorized as WerkzeugUnauthorized
+from quart_schema import QuartSchema
 from common import settings
-from api.utils.api_utils import server_error_response
+from api.utils.api_utils import server_error_response, get_json_result
 from api.constants import API_VERSION
 from common.misc_utils import get_uuid
 
@@ -40,48 +40,37 @@ settings.init_settings()
 
 __all__ = ["app"]
 
-Request.json = property(lambda self: self.get_json(force=True, silent=True))
+UNAUTHORIZED_MESSAGE = "<Unauthorized '401: Unauthorized'>"
+
+
+def _unauthorized_message(error):
+    if error is None:
+        return UNAUTHORIZED_MESSAGE
+    try:
+        msg = repr(error)
+    except Exception:
+        return UNAUTHORIZED_MESSAGE
+    if msg == UNAUTHORIZED_MESSAGE:
+        return msg
+    if "Unauthorized" in msg and "401" in msg:
+        return msg
+    return UNAUTHORIZED_MESSAGE
 
 app = Quart(__name__)
 app = cors(app, allow_origin="*")
-smtp_mail_server = Mail()
 
-# Add this at the beginning of your file to configure Swagger UI
-swagger_config = {
-    "headers": [],
-    "specs": [
-        {
-            "endpoint": "apispec",
-            "route": "/apispec.json",
-            "rule_filter": lambda rule: True,  # Include all endpoints
-            "model_filter": lambda tag: True,  # Include all models
-        }
-    ],
-    "static_url_path": "/flasgger_static",
-    "swagger_ui": True,
-    "specs_route": "/apidocs/",
-}
-
-swagger = Swagger(
-    app,
-    config=swagger_config,
-    template={
-        "swagger": "2.0",
-        "info": {
-            "title": "RAGFlow API",
-            "description": "",
-            "version": "1.0.0",
-        },
-        "securityDefinitions": {
-            "ApiKeyAuth": {"type": "apiKey", "name": "Authorization", "in": "header"}
-        },
-    },
-)
+# openapi supported
+QuartSchema(app)
 
 app.url_map.strict_slashes = False
 app.json_encoder = CustomJSONEncoder
 app.errorhandler(Exception)(server_error_response)
 
+# Configure Quart timeouts for slow LLM responses (e.g., local Ollama on CPU)
+# Default Quart timeouts are 60 seconds which is too short for many LLM backends
+app.config["RESPONSE_TIMEOUT"] = int(os.environ.get("QUART_RESPONSE_TIMEOUT", 600))
+app.config["BODY_TIMEOUT"] = int(os.environ.get("QUART_BODY_TIMEOUT", 600))
+
 ## convince for dev and debug
 # app.config["LOGIN_DISABLED"] = True
 app.config["SESSION_PERMANENT"] = False
@@ -102,12 +91,13 @@ from werkzeug.local import LocalProxy
 T = TypeVar("T")
 P = ParamSpec("P")
 
+
 def _load_user():
     jwt = Serializer(secret_key=settings.SECRET_KEY)
     authorization = request.headers.get("Authorization")
     g.user = None
     if not authorization:
-        return
+        return None
 
     try:
         access_token = str(jwt.loads(authorization))
@@ -124,18 +114,28 @@ def _load_user():
         user = UserService.query(
             access_token=access_token, status=StatusEnum.VALID.value
         )
-        if not user and len(authorization.split()) == 2:
-            objs = APIToken.query(token=authorization.split()[1])
-            if objs:
-                user = UserService.query(id=objs[0].tenant_id, status=StatusEnum.VALID.value)
         if user:
             if not user[0].access_token or not user[0].access_token.strip():
                 logging.warning(f"User {user[0].email} has empty access_token in database")
                 return None
             g.user = user[0]
             return user[0]
-    except Exception as e:
-        logging.warning(f"load_user got exception {e}")
+    except Exception as e_auth:
+        logging.warning(f"load_user got exception {e_auth}")
+        try:
+            authorization = request.headers.get("Authorization")
+            if len(authorization.split()) == 2:
+                objs = APIToken.query(token=authorization.split()[1])
+                if objs:
+                    user = UserService.query(id=objs[0].tenant_id, status=StatusEnum.VALID.value)
+                    if user:
+                        if not user[0].access_token or not user[0].access_token.strip():
+                            logging.warning(f"User {user[0].email} has empty access_token in database")
+                            return None
+                        g.user = user[0]
+                        return user[0]
+        except Exception as e_api_token:
+            logging.warning(f"load_user got exception {e_api_token}")
 
 
 current_user = LocalProxy(_load_user)
@@ -163,10 +163,18 @@ def login_required(func: Callable[P, Awaitable[T]]) -> Callable[P, Awaitable[T]]
 
     @wraps(func)
     async def wrapper(*args: P.args, **kwargs: P.kwargs) -> T:
-        if not current_user:# or not session.get("_user_id"):
-            raise Unauthorized()
-        else:
-            return await current_app.ensure_async(func)(*args, **kwargs)
+        timing_enabled = os.getenv("RAGFLOW_API_TIMING")
+        t_start = time.perf_counter() if timing_enabled else None
+        user = current_user
+        if timing_enabled:
+            logging.info(
+                "api_timing login_required auth_ms=%.2f path=%s",
+                (time.perf_counter() - t_start) * 1000,
+                request.path,
+            )
+        if not user:  # or not session.get("_user_id"):
+            raise QuartAuthUnauthorized()
+        return await current_app.ensure_async(func)(*args, **kwargs)
 
     return wrapper
 
@@ -177,7 +185,7 @@ def login_user(user, remember=False, duration=None, force=False, fresh=True):
     user's `is_active` property is ``False``, they will not be logged in
     unless `force` is ``True``.
 
-    This will return ``True`` if the log in attempt succeeds, and ``False`` if
+    This will return ``True`` if the login attempt succeeds, and ``False`` if
     it fails (i.e. because the user is inactive).
 
     :param user: The user object to log in.
@@ -227,6 +235,7 @@ def logout_user():
 
     return True
 
+
 def search_pages_path(page_path):
     app_path_list = [
         path for path in page_path.glob("*_app.py") if not path.name.startswith(".")
@@ -273,6 +282,36 @@ client_urls_prefix = [
 ]
 
 
+@app.errorhandler(404)
+async def not_found(error):
+    logging.error(f"The requested URL {request.path} was not found")
+    message = f"Not Found: {request.path}"
+    response = {
+        "code": RetCode.NOT_FOUND,
+        "message": message,
+        "data": None,
+        "error": "Not Found",
+    }
+    return jsonify(response), RetCode.NOT_FOUND
+
+
+@app.errorhandler(401)
+async def unauthorized(error):
+    logging.warning("Unauthorized request")
+    return get_json_result(code=RetCode.UNAUTHORIZED, message=_unauthorized_message(error)), RetCode.UNAUTHORIZED
+
+
+@app.errorhandler(QuartAuthUnauthorized)
+async def unauthorized_quart_auth(error):
+    logging.warning("Unauthorized request (quart_auth)")
+    return get_json_result(code=RetCode.UNAUTHORIZED, message=repr(error)), RetCode.UNAUTHORIZED
+
+
+@app.errorhandler(WerkzeugUnauthorized)
+async def unauthorized_werkzeug(error):
+    logging.warning("Unauthorized request (werkzeug)")
+    return get_json_result(code=RetCode.UNAUTHORIZED, message=_unauthorized_message(error)), RetCode.UNAUTHORIZED
+
 @app.teardown_request
 def _db_close(exception):
     if exception:

api/apps/api_app.py

@@ -18,8 +18,7 @@ from quart import request
 from api.db.db_models import APIToken
 from api.db.services.api_service import APITokenService, API4ConversationService
 from api.db.services.user_service import UserTenantService
-from api.utils.api_utils import server_error_response, get_data_error_result, get_json_result, validate_request, \
-    generate_confirmation_token
+from api.utils.api_utils import generate_confirmation_token, get_data_error_result, get_json_result, get_request_json, server_error_response, validate_request
 from common.time_utils import current_timestamp, datetime_format
 from api.apps import login_required, current_user
 
@@ -27,7 +26,7 @@ from api.apps import login_required, current_user
 @manager.route('/new_token', methods=['POST'])  # noqa: F821
 @login_required
 async def new_token():
-    req = await request.json
+    req = await get_request_json()
     try:
         tenants = UserTenantService.query(user_id=current_user.id)
         if not tenants:
@@ -73,7 +72,7 @@ def token_list():
 @validate_request("tokens", "tenant_id")
 @login_required
 async def rm():
-    req = await request.json
+    req = await get_request_json()
     try:
         for token in req["tokens"]:
             APITokenService.filter_delete(
@@ -116,4 +115,3 @@ def stats():
         return get_json_result(data=res)
     except Exception as e:
         return server_error_response(e)
-

api/apps/auth/github.py

@@ -14,7 +14,7 @@
 #  limitations under the License.
 #
 
-import requests
+from common.http_client import async_request, sync_request
 from .oauth import OAuthClient, UserInfo
 
 
@@ -34,24 +34,49 @@ class GithubOAuthClient(OAuthClient):
 
     def fetch_user_info(self, access_token, **kwargs):
         """
-        Fetch GitHub user info.
+        Fetch GitHub user info (synchronous).
         """
         user_info = {}
         try:
             headers = {"Authorization": f"Bearer {access_token}"}
-            # user info
-            response = requests.get(self.userinfo_url, headers=headers, timeout=self.http_request_timeout)
+            response = sync_request("GET", self.userinfo_url, headers=headers, timeout=self.http_request_timeout)
             response.raise_for_status()
             user_info.update(response.json())
-            # email info
-            response = requests.get(self.userinfo_url+"/emails", headers=headers, timeout=self.http_request_timeout)
-            response.raise_for_status()
-            email_info = response.json()
-            user_info["email"] = next(
-                (email for email in email_info if email["primary"]), None
-            )["email"]
+            email_response = sync_request(
+                "GET", self.userinfo_url + "/emails", headers=headers, timeout=self.http_request_timeout
+            )
+            email_response.raise_for_status()
+            email_info = email_response.json()
+            user_info["email"] = next((email for email in email_info if email["primary"]), None)["email"]
             return self.normalize_user_info(user_info)
-        except requests.exceptions.RequestException as e:
+        except Exception as e:
+            raise ValueError(f"Failed to fetch github user info: {e}")
+
+    async def async_fetch_user_info(self, access_token, **kwargs):
+        """Async variant of fetch_user_info using httpx."""
+        user_info = {}
+        headers = {"Authorization": f"Bearer {access_token}"}
+        try:
+            response = await async_request(
+                "GET",
+                self.userinfo_url,
+                headers=headers,
+                timeout=self.http_request_timeout,
+            )
+            response.raise_for_status()
+            user_info.update(response.json())
+
+            email_response = await async_request(
+                "GET",
+                self.userinfo_url + "/emails",
+                headers=headers,
+                timeout=self.http_request_timeout,
+            )
+            email_response.raise_for_status()
+            email_info = email_response.json()
+            user_info["email"] = next((email for email in email_info if email["primary"]), None)["email"]
+            return self.normalize_user_info(user_info)
+        except Exception as e:
             raise ValueError(f"Failed to fetch github user info: {e}")
 
 

api/apps/auth/oauth.py

@@ -14,8 +14,8 @@
 #  limitations under the License.
 #
 
-import requests
 import urllib.parse
+from common.http_client import async_request, sync_request
 
 
 class UserInfo:
@@ -74,15 +74,40 @@ class OAuthClient:
                 "redirect_uri": self.redirect_uri,
                 "grant_type": "authorization_code"
             }
-            response = requests.post(
+            response = sync_request(
+                "POST",
                 self.token_url,
                 data=payload,
                 headers={"Accept": "application/json"},
-                timeout=self.http_request_timeout
+                timeout=self.http_request_timeout,
             )
             response.raise_for_status()
             return response.json()
-        except requests.exceptions.RequestException as e:
+        except Exception as e:
+            raise ValueError(f"Failed to exchange authorization code for token: {e}")
+
+    async def async_exchange_code_for_token(self, code):
+        """
+        Async variant of exchange_code_for_token using httpx.
+        """
+        payload = {
+            "client_id": self.client_id,
+            "client_secret": self.client_secret,
+            "code": code,
+            "redirect_uri": self.redirect_uri,
+            "grant_type": "authorization_code",
+        }
+        try:
+            response = await async_request(
+                "POST",
+                self.token_url,
+                data=payload,
+                headers={"Accept": "application/json"},
+                timeout=self.http_request_timeout,
+            )
+            response.raise_for_status()
+            return response.json()
+        except Exception as e:
             raise ValueError(f"Failed to exchange authorization code for token: {e}")
 
 
@@ -92,11 +117,27 @@ class OAuthClient:
         """
         try:
             headers = {"Authorization": f"Bearer {access_token}"}
-            response = requests.get(self.userinfo_url, headers=headers, timeout=self.http_request_timeout)
+            response = sync_request("GET", self.userinfo_url, headers=headers, timeout=self.http_request_timeout)
             response.raise_for_status()
             user_info = response.json()
             return self.normalize_user_info(user_info)
-        except requests.exceptions.RequestException as e:
+        except Exception as e:
+            raise ValueError(f"Failed to fetch user info: {e}")
+
+    async def async_fetch_user_info(self, access_token, **kwargs):
+        """Async variant of fetch_user_info using httpx."""
+        headers = {"Authorization": f"Bearer {access_token}"}
+        try:
+            response = await async_request(
+                "GET",
+                self.userinfo_url,
+                headers=headers,
+                timeout=self.http_request_timeout,
+            )
+            response.raise_for_status()
+            user_info = response.json()
+            return self.normalize_user_info(user_info)
+        except Exception as e:
             raise ValueError(f"Failed to fetch user info: {e}")
 
 

api/apps/auth/oidc.py

@@ -15,7 +15,7 @@
 #
 
 import jwt
-import requests
+from common.http_client import sync_request
 from .oauth import OAuthClient
 
 
@@ -50,10 +50,10 @@ class OIDCClient(OAuthClient):
         """
         try:
             metadata_url = f"{issuer}/.well-known/openid-configuration"
-            response = requests.get(metadata_url, timeout=7)
+            response = sync_request("GET", metadata_url, timeout=7)
             response.raise_for_status()
             return response.json()
-        except requests.exceptions.RequestException as e:
+        except Exception as e:
             raise ValueError(f"Failed to fetch OIDC metadata: {e}")
 
 
@@ -95,6 +95,13 @@ class OIDCClient(OAuthClient):
         user_info.update(super().fetch_user_info(access_token).to_dict())
         return self.normalize_user_info(user_info)
 
+    async def async_fetch_user_info(self, access_token, id_token=None, **kwargs):
+        user_info = {}
+        if id_token:
+            user_info = self.parse_id_token(id_token)
+        user_info.update((await super().async_fetch_user_info(access_token)).to_dict())
+        return self.normalize_user_info(user_info)
+
 
     def normalize_user_info(self, user_info):
         return super().normalize_user_info(user_info)

api/apps/canvas_app.py

@@ -13,15 +13,13 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import inspect
 import json
 import logging
-import re
-import sys
 from functools import partial
-import trio
 from quart import request, Response, make_response
 from agent.component import LLM
-from api.db import CanvasCategory, FileType
+from api.db import CanvasCategory
 from api.db.services.canvas_service import CanvasTemplateService, UserCanvasService, API4ConversationService
 from api.db.services.document_service import DocumentService
 from api.db.services.file_service import FileService
@@ -30,15 +28,19 @@ from api.db.services.task_service import queue_dataflow, CANVAS_DEBUG_DOC_ID, Ta
 from api.db.services.user_service import TenantService
 from api.db.services.user_canvas_version import UserCanvasVersionService
 from common.constants import RetCode
-from common.misc_utils import get_uuid
-from api.utils.api_utils import get_json_result, server_error_response, validate_request, get_data_error_result, \
-    request_json
+from common.misc_utils import get_uuid, thread_pool_exec
+from api.utils.api_utils import (
+    get_json_result,
+    server_error_response,
+    validate_request,
+    get_data_error_result,
+    get_request_json,
+)
 from agent.canvas import Canvas
 from peewee import MySQLDatabase, PostgresqlDatabase
 from api.db.db_models import APIToken, Task
 import time
 
-from api.utils.file_utils import filename_type, read_potential_broken_pdf
 from rag.flow.pipeline import Pipeline
 from rag.nlp import search
 from rag.utils.redis_conn import REDIS_CONN
@@ -56,7 +58,7 @@ def templates():
 @validate_request("canvas_ids")
 @login_required
 async def rm():
-    req = await request_json()
+    req = await get_request_json()
     for i in req["canvas_ids"]:
         if not UserCanvasService.accessible(i, current_user.id):
             return get_json_result(
@@ -70,7 +72,7 @@ async def rm():
 @validate_request("dsl", "title")
 @login_required
 async def save():
-    req = await request_json()
+    req = await get_request_json()
     if not isinstance(req["dsl"], str):
         req["dsl"] = json.dumps(req["dsl"], ensure_ascii=False)
     req["dsl"] = json.loads(req["dsl"])
@@ -129,17 +131,17 @@ def getsse(canvas_id):
 @validate_request("id")
 @login_required
 async def run():
-    req = await request_json()
+    req = await get_request_json()
     query = req.get("query", "")
     files = req.get("files", [])
     inputs = req.get("inputs", {})
     user_id = req.get("user_id", current_user.id)
-    if not UserCanvasService.accessible(req["id"], current_user.id):
+    if not await thread_pool_exec(UserCanvasService.accessible, req["id"], current_user.id):
         return get_json_result(
             data=False, message='Only owner of canvas authorized for this operation.',
             code=RetCode.OPERATING_ERROR)
 
-    e, cvs = UserCanvasService.get_by_id(req["id"])
+    e, cvs = await thread_pool_exec(UserCanvasService.get_by_id, req["id"])
     if not e:
         return get_data_error_result(message="canvas not found.")
 
@@ -149,13 +151,13 @@ async def run():
     if cvs.canvas_category == CanvasCategory.DataFlow:
         task_id = get_uuid()
         Pipeline(cvs.dsl, tenant_id=current_user.id, doc_id=CANVAS_DEBUG_DOC_ID, task_id=task_id, flow_id=req["id"])
-        ok, error_message = queue_dataflow(tenant_id=user_id, flow_id=req["id"], task_id=task_id, file=files[0], priority=0)
+        ok, error_message = await thread_pool_exec(queue_dataflow, user_id, req["id"], task_id, CANVAS_DEBUG_DOC_ID, files[0], 0)
         if not ok:
             return get_data_error_result(message=error_message)
         return get_json_result(data={"message_id": task_id})
 
     try:
-        canvas = Canvas(cvs.dsl, current_user.id)
+        canvas = Canvas(cvs.dsl, current_user.id, canvas_id=cvs.id)
     except Exception as e:
         return server_error_response(e)
 
@@ -186,7 +188,7 @@ async def run():
 @validate_request("id", "dsl", "component_id")
 @login_required
 async def rerun():
-    req = await request_json()
+    req = await get_request_json()
     doc = PipelineOperationLogService.get_documents_info(req["id"])
     if not doc:
         return get_data_error_result(message="Document not found.")
@@ -194,7 +196,7 @@ async def rerun():
     if 0 < doc["progress"] < 1:
         return get_data_error_result(message=f"`{doc['name']}` is processing...")
 
-    if settings.docStoreConn.indexExist(search.index_name(current_user.id), doc["kb_id"]):
+    if settings.docStoreConn.index_exist(search.index_name(current_user.id), doc["kb_id"]):
         settings.docStoreConn.delete({"doc_id": doc["id"]}, search.index_name(current_user.id), doc["kb_id"])
     doc["progress_msg"] = ""
     doc["chunk_num"] = 0
@@ -224,7 +226,7 @@ def cancel(task_id):
 @validate_request("id")
 @login_required
 async def reset():
-    req = await request_json()
+    req = await get_request_json()
     if not UserCanvasService.accessible(req["id"], current_user.id):
         return get_json_result(
             data=False, message='Only owner of canvas authorized for this operation.',
@@ -234,7 +236,7 @@ async def reset():
         if not e:
             return get_data_error_result(message="canvas not found.")
 
-        canvas = Canvas(json.dumps(user_canvas.dsl), current_user.id)
+        canvas = Canvas(json.dumps(user_canvas.dsl), current_user.id, canvas_id=user_canvas.id)
         canvas.reset()
         req["dsl"] = json.loads(str(canvas))
         UserCanvasService.update_by_id(req["id"], {"dsl": req["dsl"]})
@@ -250,71 +252,10 @@ async def upload(canvas_id):
         return get_data_error_result(message="canvas not found.")
 
     user_id = cvs["user_id"]
-    def structured(filename, filetype, blob, content_type):
-        nonlocal user_id
-        if filetype == FileType.PDF.value:
-            blob = read_potential_broken_pdf(blob)
-
-        location = get_uuid()
-        FileService.put_blob(user_id, location, blob)
-
-        return {
-            "id": location,
-            "name": filename,
-            "size": sys.getsizeof(blob),
-            "extension": filename.split(".")[-1].lower(),
-            "mime_type": content_type,
-            "created_by": user_id,
-            "created_at": time.time(),
-            "preview_url": None
-        }
-
-    if request.args.get("url"):
-        from crawl4ai import (
-            AsyncWebCrawler,
-            BrowserConfig,
-            CrawlerRunConfig,
-            DefaultMarkdownGenerator,
-            PruningContentFilter,
-            CrawlResult
-        )
-        try:
-            url = request.args.get("url")
-            filename = re.sub(r"\?.*", "", url.split("/")[-1])
-            async def adownload():
-                browser_config = BrowserConfig(
-                    headless=True,
-                    verbose=False,
-                )
-                async with AsyncWebCrawler(config=browser_config) as crawler:
-                    crawler_config = CrawlerRunConfig(
-                        markdown_generator=DefaultMarkdownGenerator(
-                            content_filter=PruningContentFilter()
-                        ),
-                        pdf=True,
-                        screenshot=False
-                    )
-                    result: CrawlResult = await crawler.arun(
-                        url=url,
-                        config=crawler_config
-                    )
-                    return result
-            page = trio.run(adownload())
-            if page.pdf:
-                if filename.split(".")[-1].lower() != "pdf":
-                    filename += ".pdf"
-                return get_json_result(data=structured(filename, "pdf", page.pdf, page.response_headers["content-type"]))
-
-            return get_json_result(data=structured(filename, "html", str(page.markdown).encode("utf-8"), page.response_headers["content-type"], user_id))
-
-        except Exception as e:
-            return  server_error_response(e)
-
     files = await request.files
-    file = files['file']
+    file = files['file'] if files and files.get("file") else None
     try:
-        DocumentService.check_doc_health(user_id, file.filename)
-        return get_json_result(data=structured(file.filename, filename_type(file.filename), file.read(), file.content_type))
+        return get_json_result(data=FileService.upload_info(user_id, file, request.args.get("url")))
     except Exception as e:
         return  server_error_response(e)
 
@@ -333,7 +274,7 @@ def input_form():
                 data=False, message='Only owner of canvas authorized for this operation.',
                 code=RetCode.OPERATING_ERROR)
 
-        canvas = Canvas(json.dumps(user_canvas.dsl), current_user.id)
+        canvas = Canvas(json.dumps(user_canvas.dsl), current_user.id, canvas_id=user_canvas.id)
         return get_json_result(data=canvas.get_component_input_form(cpn_id))
     except Exception as e:
         return server_error_response(e)
@@ -343,14 +284,14 @@ def input_form():
 @validate_request("id", "component_id", "params")
 @login_required
 async def debug():
-    req = await request_json()
+    req = await get_request_json()
     if not UserCanvasService.accessible(req["id"], current_user.id):
         return get_json_result(
             data=False, message='Only owner of canvas authorized for this operation.',
             code=RetCode.OPERATING_ERROR)
     try:
         e, user_canvas = UserCanvasService.get_by_id(req["id"])
-        canvas = Canvas(json.dumps(user_canvas.dsl), current_user.id)
+        canvas = Canvas(json.dumps(user_canvas.dsl), current_user.id, canvas_id=user_canvas.id)
         canvas.reset()
         canvas.message_id = get_uuid()
         component = canvas.get_component(req["component_id"])["obj"]
@@ -363,8 +304,13 @@ async def debug():
         for k in outputs.keys():
             if isinstance(outputs[k], partial):
                 txt = ""
-                for c in outputs[k]():
-                    txt += c
+                iter_obj = outputs[k]()
+                if inspect.isasyncgen(iter_obj):
+                    async for c in iter_obj:
+                        txt += c
+                else:
+                    for c in iter_obj:
+                        txt += c
                 outputs[k] = txt
         return get_json_result(data=outputs)
     except Exception as e:
@@ -375,7 +321,7 @@ async def debug():
 @validate_request("db_type", "database", "username", "host", "port", "password")
 @login_required
 async def test_db_connect():
-    req = await request_json()
+    req = await get_request_json()
     try:
         if req["db_type"] in ["mysql", "mariadb"]:
             db = MySQLDatabase(req["database"], user=req["username"], host=req["host"], port=req["port"],
@@ -406,7 +352,15 @@ async def test_db_connect():
                 f"UID={req['username']};"
                 f"PWD={req['password']};"
             )
-            logging.info(conn_str)
+            redacted_conn_str = (
+                f"DATABASE={req['database']};"
+                f"HOSTNAME={req['host']};"
+                f"PORT={req['port']};"
+                f"PROTOCOL=TCPIP;"
+                f"UID={req['username']};"
+                f"PWD=****;"
+            )
+            logging.info(redacted_conn_str)
             conn = ibm_db.connect(conn_str, "", "")
             stmt = ibm_db.exec_immediate(conn, "SELECT 1 FROM sysibm.sysdummy1")
             ibm_db.fetch_assoc(stmt)
@@ -520,7 +474,7 @@ def list_canvas():
 @validate_request("id", "title", "permission")
 @login_required
 async def setting():
-    req = await request_json()
+    req = await get_request_json()
     req["user_id"] = current_user.id
 
     if not UserCanvasService.accessible(req["id"], current_user.id):
@@ -590,6 +544,7 @@ def sessions(canvas_id):
 @login_required
 def prompts():
     from rag.prompts.generator import ANALYZE_TASK_SYSTEM, ANALYZE_TASK_USER, NEXT_STEP, REFLECT, CITATION_PROMPT_TEMPLATE
+
     return get_json_result(data={
         "task_analysis": ANALYZE_TASK_SYSTEM +"\n\n"+ ANALYZE_TASK_USER,
         "plan_generation": NEXT_STEP,

api/apps/chunk_app.py

@@ -13,36 +13,42 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import base64
 import datetime
 import json
+import logging
 import re
-
 import xxhash
 from quart import request
 
-from api.db.services.dialog_service import meta_filter
 from api.db.services.document_service import DocumentService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMBundle
+from common.metadata_utils import apply_meta_data_filter
 from api.db.services.search_service import SearchService
 from api.db.services.user_service import UserTenantService
-from api.utils.api_utils import get_data_error_result, get_json_result, server_error_response, validate_request, \
-    request_json
+from api.utils.api_utils import (
+    get_data_error_result,
+    get_json_result,
+    server_error_response,
+    validate_request,
+    get_request_json,
+)
+from common.misc_utils import thread_pool_exec
 from rag.app.qa import beAdoc, rmPrefix
 from rag.app.tag import label_question
 from rag.nlp import rag_tokenizer, search
-from rag.prompts.generator import gen_meta_filter, cross_languages, keyword_extraction
+from rag.prompts.generator import cross_languages, keyword_extraction
 from common.string_utils import remove_redundant_spaces
 from common.constants import RetCode, LLMType, ParserType, PAGERANK_FLD
 from common import settings
 from api.apps import login_required, current_user
 
-
 @manager.route('/list', methods=['POST'])  # noqa: F821
 @login_required
 @validate_request("doc_id")
 async def list_chunk():
-    req = await request_json()
+    req = await get_request_json()
     doc_id = req["doc_id"]
     page = int(req.get("page", 1))
     size = int(req.get("size", 30))
@@ -60,7 +66,7 @@ async def list_chunk():
         }
         if "available_int" in req:
             query["available_int"] = int(req["available_int"])
-        sres = settings.retriever.search(query, search.index_name(tenant_id), kb_ids, highlight=["content_ltks"])
+        sres = await settings.retriever.search(query, search.index_name(tenant_id), kb_ids, highlight=["content_ltks"])
         res = {"total": sres.total, "chunks": [], "doc": doc.to_dict()}
         for id in sres.ids:
             d = {
@@ -75,6 +81,7 @@ async def list_chunk():
                 "image_id": sres.field[id].get("img_id", ""),
                 "available_int": int(sres.field[id].get("available_int", 1)),
                 "positions": sres.field[id].get("position_int", []),
+                "doc_type_kwd": sres.field[id].get("doc_type_kwd")
             }
             assert isinstance(d["positions"], list)
             assert len(d["positions"]) == 0 or (isinstance(d["positions"][0], list) and len(d["positions"][0]) == 5)
@@ -123,11 +130,16 @@ def get():
 @login_required
 @validate_request("doc_id", "chunk_id", "content_with_weight")
 async def set():
-    req = await request_json()
+    req = await get_request_json()
+    content_with_weight = req["content_with_weight"]
+    if not isinstance(content_with_weight, (str, bytes)):
+        raise TypeError("expected string or bytes-like object")
+    if isinstance(content_with_weight, bytes):
+        content_with_weight = content_with_weight.decode("utf-8", errors="ignore")
     d = {
         "id": req["chunk_id"],
-        "content_with_weight": req["content_with_weight"]}
-    d["content_ltks"] = rag_tokenizer.tokenize(req["content_with_weight"])
+        "content_with_weight": content_with_weight}
+    d["content_ltks"] = rag_tokenizer.tokenize(content_with_weight)
     d["content_sm_ltks"] = rag_tokenizer.fine_grained_tokenize(d["content_ltks"])
     if "important_kwd" in req:
         if not isinstance(req["important_kwd"], list):
@@ -147,31 +159,43 @@ async def set():
         d["available_int"] = req["available_int"]
 
     try:
-        tenant_id = DocumentService.get_tenant_id(req["doc_id"])
-        if not tenant_id:
-            return get_data_error_result(message="Tenant not found!")
+        def _set_sync():
+            tenant_id = DocumentService.get_tenant_id(req["doc_id"])
+            if not tenant_id:
+                return get_data_error_result(message="Tenant not found!")
 
-        embd_id = DocumentService.get_embd_id(req["doc_id"])
-        embd_mdl = LLMBundle(tenant_id, LLMType.EMBEDDING, embd_id)
+            embd_id = DocumentService.get_embd_id(req["doc_id"])
+            embd_mdl = LLMBundle(tenant_id, LLMType.EMBEDDING, embd_id)
 
-        e, doc = DocumentService.get_by_id(req["doc_id"])
-        if not e:
-            return get_data_error_result(message="Document not found!")
+            e, doc = DocumentService.get_by_id(req["doc_id"])
+            if not e:
+                return get_data_error_result(message="Document not found!")
 
-        if doc.parser_id == ParserType.QA:
-            arr = [
-                t for t in re.split(
-                    r"[\n\t]",
-                    req["content_with_weight"]) if len(t) > 1]
-            q, a = rmPrefix(arr[0]), rmPrefix("\n".join(arr[1:]))
-            d = beAdoc(d, q, a, not any(
-                [rag_tokenizer.is_chinese(t) for t in q + a]))
+            _d = d
+            if doc.parser_id == ParserType.QA:
+                arr = [
+                    t for t in re.split(
+                        r"[\n\t]",
+                        req["content_with_weight"]) if len(t) > 1]
+                q, a = rmPrefix(arr[0]), rmPrefix("\n".join(arr[1:]))
+                _d = beAdoc(d, q, a, not any(
+                    [rag_tokenizer.is_chinese(t) for t in q + a]))
 
-        v, c = embd_mdl.encode([doc.name, req["content_with_weight"] if not d.get("question_kwd") else "\n".join(d["question_kwd"])])
-        v = 0.1 * v[0] + 0.9 * v[1] if doc.parser_id != ParserType.QA else v[1]
-        d["q_%d_vec" % len(v)] = v.tolist()
-        settings.docStoreConn.update({"id": req["chunk_id"]}, d, search.index_name(tenant_id), doc.kb_id)
-        return get_json_result(data=True)
+            v, c = embd_mdl.encode([doc.name, content_with_weight if not _d.get("question_kwd") else "\n".join(_d["question_kwd"])])
+            v = 0.1 * v[0] + 0.9 * v[1] if doc.parser_id != ParserType.QA else v[1]
+            _d["q_%d_vec" % len(v)] = v.tolist()
+            settings.docStoreConn.update({"id": req["chunk_id"]}, _d, search.index_name(tenant_id), doc.kb_id)
+
+            # update image
+            image_base64 = req.get("image_base64", None)
+            img_id = req.get("img_id", "")
+            if image_base64 and img_id and "-" in img_id:
+                bkt, name = img_id.split("-", 1)
+                image_binary = base64.b64decode(image_base64)
+                settings.STORAGE_IMPL.put(bkt, name, image_binary)
+            return get_json_result(data=True)
+
+        return await thread_pool_exec(_set_sync)
     except Exception as e:
         return server_error_response(e)
 
@@ -180,18 +204,21 @@ async def set():
 @login_required
 @validate_request("chunk_ids", "available_int", "doc_id")
 async def switch():
-    req = await request_json()
+    req = await get_request_json()
     try:
-        e, doc = DocumentService.get_by_id(req["doc_id"])
-        if not e:
-            return get_data_error_result(message="Document not found!")
-        for cid in req["chunk_ids"]:
-            if not settings.docStoreConn.update({"id": cid},
-                                                {"available_int": int(req["available_int"])},
-                                                search.index_name(DocumentService.get_tenant_id(req["doc_id"])),
-                                                doc.kb_id):
-                return get_data_error_result(message="Index updating failure")
-        return get_json_result(data=True)
+        def _switch_sync():
+            e, doc = DocumentService.get_by_id(req["doc_id"])
+            if not e:
+                return get_data_error_result(message="Document not found!")
+            for cid in req["chunk_ids"]:
+                if not settings.docStoreConn.update({"id": cid},
+                                                    {"available_int": int(req["available_int"])},
+                                                    search.index_name(DocumentService.get_tenant_id(req["doc_id"])),
+                                                    doc.kb_id):
+                    return get_data_error_result(message="Index updating failure")
+            return get_json_result(data=True)
+
+        return await thread_pool_exec(_switch_sync)
     except Exception as e:
         return server_error_response(e)
 
@@ -200,22 +227,40 @@ async def switch():
 @login_required
 @validate_request("chunk_ids", "doc_id")
 async def rm():
-    req = await request_json()
+    req = await get_request_json()
     try:
-        e, doc = DocumentService.get_by_id(req["doc_id"])
-        if not e:
-            return get_data_error_result(message="Document not found!")
-        if not settings.docStoreConn.delete({"id": req["chunk_ids"]},
-                                            search.index_name(DocumentService.get_tenant_id(req["doc_id"])),
-                                            doc.kb_id):
-            return get_data_error_result(message="Chunk deleting failure")
-        deleted_chunk_ids = req["chunk_ids"]
-        chunk_number = len(deleted_chunk_ids)
-        DocumentService.decrement_chunk_num(doc.id, doc.kb_id, 1, chunk_number, 0)
-        for cid in deleted_chunk_ids:
-            if settings.STORAGE_IMPL.obj_exist(doc.kb_id, cid):
-                settings.STORAGE_IMPL.rm(doc.kb_id, cid)
-        return get_json_result(data=True)
+        def _rm_sync():
+            e, doc = DocumentService.get_by_id(req["doc_id"])
+            if not e:
+                return get_data_error_result(message="Document not found!")
+            condition = {"id": req["chunk_ids"], "doc_id": req["doc_id"]}
+            try:
+                deleted_count = settings.docStoreConn.delete(condition,
+                                                             search.index_name(DocumentService.get_tenant_id(req["doc_id"])),
+                                                             doc.kb_id)
+            except Exception:
+                return get_data_error_result(message="Chunk deleting failure")
+            deleted_chunk_ids = req["chunk_ids"]
+            if isinstance(deleted_chunk_ids, list):
+                unique_chunk_ids = list(dict.fromkeys(deleted_chunk_ids))
+                has_ids = len(unique_chunk_ids) > 0
+            else:
+                unique_chunk_ids = [deleted_chunk_ids]
+                has_ids = deleted_chunk_ids not in (None, "")
+            if has_ids and deleted_count == 0:
+                return get_data_error_result(message="Index updating failure")
+            if deleted_count > 0 and deleted_count < len(unique_chunk_ids):
+                deleted_count += settings.docStoreConn.delete({"doc_id": req["doc_id"]},
+                                                              search.index_name(DocumentService.get_tenant_id(req["doc_id"])),
+                                                              doc.kb_id)
+            chunk_number = deleted_count
+            DocumentService.decrement_chunk_num(doc.id, doc.kb_id, 1, chunk_number, 0)
+            for cid in deleted_chunk_ids:
+                if settings.STORAGE_IMPL.obj_exist(doc.kb_id, cid):
+                    settings.STORAGE_IMPL.rm(doc.kb_id, cid)
+            return get_json_result(data=True)
+
+        return await thread_pool_exec(_rm_sync)
     except Exception as e:
         return server_error_response(e)
 
@@ -224,7 +269,8 @@ async def rm():
 @login_required
 @validate_request("doc_id", "content_with_weight")
 async def create():
-    req = await request_json()
+    req = await get_request_json()
+    req_id = request.headers.get("X-Request-ID")
     chunck_id = xxhash.xxh64((req["content_with_weight"] + req["doc_id"]).encode("utf-8")).hexdigest()
     d = {"id": chunck_id, "content_ltks": rag_tokenizer.tokenize(req["content_with_weight"]),
          "content_with_weight": req["content_with_weight"]}
@@ -241,40 +287,59 @@ async def create():
     d["create_timestamp_flt"] = datetime.datetime.now().timestamp()
     if "tag_feas" in req:
         d["tag_feas"] = req["tag_feas"]
-    if "tag_feas" in req:
-        d["tag_feas"] = req["tag_feas"]
 
     try:
-        e, doc = DocumentService.get_by_id(req["doc_id"])
-        if not e:
-            return get_data_error_result(message="Document not found!")
-        d["kb_id"] = [doc.kb_id]
-        d["docnm_kwd"] = doc.name
-        d["title_tks"] = rag_tokenizer.tokenize(doc.name)
-        d["doc_id"] = doc.id
+        def _log_response(resp, code, message):
+            logging.info(
+                "chunk_create response req_id=%s status=%s code=%s message=%s",
+                req_id,
+                getattr(resp, "status_code", None),
+                code,
+                message,
+            )
 
-        tenant_id = DocumentService.get_tenant_id(req["doc_id"])
-        if not tenant_id:
-            return get_data_error_result(message="Tenant not found!")
+        def _create_sync():
+            e, doc = DocumentService.get_by_id(req["doc_id"])
+            if not e:
+                resp = get_data_error_result(message="Document not found!")
+                _log_response(resp, RetCode.DATA_ERROR, "Document not found!")
+                return resp
+            d["kb_id"] = [doc.kb_id]
+            d["docnm_kwd"] = doc.name
+            d["title_tks"] = rag_tokenizer.tokenize(doc.name)
+            d["doc_id"] = doc.id
 
-        e, kb = KnowledgebaseService.get_by_id(doc.kb_id)
-        if not e:
-            return get_data_error_result(message="Knowledgebase not found!")
-        if kb.pagerank:
-            d[PAGERANK_FLD] = kb.pagerank
+            tenant_id = DocumentService.get_tenant_id(req["doc_id"])
+            if not tenant_id:
+                resp = get_data_error_result(message="Tenant not found!")
+                _log_response(resp, RetCode.DATA_ERROR, "Tenant not found!")
+                return resp
 
-        embd_id = DocumentService.get_embd_id(req["doc_id"])
-        embd_mdl = LLMBundle(tenant_id, LLMType.EMBEDDING.value, embd_id)
+            e, kb = KnowledgebaseService.get_by_id(doc.kb_id)
+            if not e:
+                resp = get_data_error_result(message="Knowledgebase not found!")
+                _log_response(resp, RetCode.DATA_ERROR, "Knowledgebase not found!")
+                return resp
+            if kb.pagerank:
+                d[PAGERANK_FLD] = kb.pagerank
 
-        v, c = embd_mdl.encode([doc.name, req["content_with_weight"] if not d["question_kwd"] else "\n".join(d["question_kwd"])])
-        v = 0.1 * v[0] + 0.9 * v[1]
-        d["q_%d_vec" % len(v)] = v.tolist()
-        settings.docStoreConn.insert([d], search.index_name(tenant_id), doc.kb_id)
+            embd_id = DocumentService.get_embd_id(req["doc_id"])
+            embd_mdl = LLMBundle(tenant_id, LLMType.EMBEDDING.value, embd_id)
 
-        DocumentService.increment_chunk_num(
-            doc.id, doc.kb_id, c, 1, 0)
-        return get_json_result(data={"chunk_id": chunck_id})
+            v, c = embd_mdl.encode([doc.name, req["content_with_weight"] if not d["question_kwd"] else "\n".join(d["question_kwd"])])
+            v = 0.1 * v[0] + 0.9 * v[1]
+            d["q_%d_vec" % len(v)] = v.tolist()
+            settings.docStoreConn.insert([d], search.index_name(tenant_id), doc.kb_id)
+
+            DocumentService.increment_chunk_num(
+                doc.id, doc.kb_id, c, 1, 0)
+            resp = get_json_result(data={"chunk_id": chunck_id})
+            _log_response(resp, RetCode.SUCCESS, "success")
+            return resp
+
+        return await thread_pool_exec(_create_sync)
     except Exception as e:
+        logging.info("chunk_create exception req_id=%s error=%r", req_id, e)
         return server_error_response(e)
 
 
@@ -282,7 +347,7 @@ async def create():
 @login_required
 @validate_request("kb_id", "question")
 async def retrieval_test():
-    req = await request_json()
+    req = await get_request_json()
     page = int(req.get("page", 1))
     size = int(req.get("size", 30))
     question = req["question"]
@@ -297,25 +362,29 @@ async def retrieval_test():
     use_kg = req.get("use_kg", False)
     top = int(req.get("top_k", 1024))
     langs = req.get("cross_languages", [])
-    tenant_ids = []
+    user_id = current_user.id
 
-    if req.get("search_id", ""):
-        search_config = SearchService.get_detail(req.get("search_id", "")).get("search_config", {})
-        meta_data_filter = search_config.get("meta_data_filter", {})
-        metas = DocumentService.get_meta_by_kbs(kb_ids)
-        if meta_data_filter.get("method") == "auto":
-            chat_mdl = LLMBundle(current_user.id, LLMType.CHAT, llm_name=search_config.get("chat_id", ""))
-            filters: dict = gen_meta_filter(chat_mdl, metas, question)
-            doc_ids.extend(meta_filter(metas, filters["conditions"], filters.get("logic", "and")))
-            if not doc_ids:
-                doc_ids = None
-        elif meta_data_filter.get("method") == "manual":
-            doc_ids.extend(meta_filter(metas, meta_data_filter["manual"], meta_data_filter.get("logic", "and")))
-            if meta_data_filter["manual"] and not doc_ids:
-                doc_ids = ["-999"]
+    async def _retrieval():
+        local_doc_ids = list(doc_ids) if doc_ids else []
+        tenant_ids = []
 
-    try:
-        tenants = UserTenantService.query(user_id=current_user.id)
+        meta_data_filter = {}
+        chat_mdl = None
+        if req.get("search_id", ""):
+            search_config = SearchService.get_detail(req.get("search_id", "")).get("search_config", {})
+            meta_data_filter = search_config.get("meta_data_filter", {})
+            if meta_data_filter.get("method") in ["auto", "semi_auto"]:
+                chat_mdl = LLMBundle(user_id, LLMType.CHAT, llm_name=search_config.get("chat_id", ""))
+        else:
+            meta_data_filter = req.get("meta_data_filter") or {}
+            if meta_data_filter.get("method") in ["auto", "semi_auto"]:
+                chat_mdl = LLMBundle(user_id, LLMType.CHAT)
+
+        if meta_data_filter:
+            metas = DocumentService.get_meta_by_kbs(kb_ids)
+            local_doc_ids = await apply_meta_data_filter(meta_data_filter, metas, question, chat_mdl, local_doc_ids)
+
+        tenants = UserTenantService.query(user_id=user_id)
         for kb_id in kb_ids:
             for tenant in tenants:
                 if KnowledgebaseService.query(
@@ -324,15 +393,16 @@ async def retrieval_test():
                     break
             else:
                 return get_json_result(
-                    data=False, message='Only owner of knowledgebase authorized for this operation.',
+                    data=False, message='Only owner of dataset authorized for this operation.',
                     code=RetCode.OPERATING_ERROR)
 
         e, kb = KnowledgebaseService.get_by_id(kb_ids[0])
         if not e:
             return get_data_error_result(message="Knowledgebase not found!")
 
+        _question = question
         if langs:
-            question = cross_languages(kb.tenant_id, None, question, langs)
+            _question = await cross_languages(kb.tenant_id, None, _question, langs)
 
         embd_mdl = LLMBundle(kb.tenant_id, LLMType.EMBEDDING.value, llm_name=kb.embd_id)
 
@@ -342,31 +412,42 @@ async def retrieval_test():
 
         if req.get("keyword", False):
             chat_mdl = LLMBundle(kb.tenant_id, LLMType.CHAT)
-            question += keyword_extraction(chat_mdl, question)
+            _question += await keyword_extraction(chat_mdl, _question)
+
+        labels = label_question(_question, [kb])
+        ranks = await settings.retriever.retrieval(
+                        _question,
+                        embd_mdl,
+                        tenant_ids,
+                        kb_ids,
+                        page,
+                        size,
+                        float(req.get("similarity_threshold", 0.0)),
+                        float(req.get("vector_similarity_weight", 0.3)),
+                        doc_ids=local_doc_ids,
+                        top=top,
+                        rerank_mdl=rerank_mdl,
+                        rank_feature=labels
+                    )
 
-        labels = label_question(question, [kb])
-        ranks = settings.retriever.retrieval(question, embd_mdl, tenant_ids, kb_ids, page, size,
-                               float(req.get("similarity_threshold", 0.0)),
-                               float(req.get("vector_similarity_weight", 0.3)),
-                               top,
-                               doc_ids, rerank_mdl=rerank_mdl,
-                                             highlight=req.get("highlight", False),
-                               rank_feature=labels
-                               )
         if use_kg:
-            ck = settings.kg_retriever.retrieval(question,
+            ck = await settings.kg_retriever.retrieval(_question,
                                                    tenant_ids,
                                                    kb_ids,
                                                    embd_mdl,
                                                    LLMBundle(kb.tenant_id, LLMType.CHAT))
             if ck["content_with_weight"]:
                 ranks["chunks"].insert(0, ck)
+        ranks["chunks"] = settings.retriever.retrieval_by_children(ranks["chunks"], tenant_ids)
 
         for c in ranks["chunks"]:
             c.pop("vector", None)
         ranks["labels"] = labels
 
         return get_json_result(data=ranks)
+
+    try:
+        return await _retrieval()
     except Exception as e:
         if str(e).find("not_found") > 0:
             return get_json_result(data=False, message='No chunk found! Check the chunk status please!',
@@ -376,7 +457,7 @@ async def retrieval_test():
 
 @manager.route('/knowledge_graph', methods=['GET'])  # noqa: F821
 @login_required
-def knowledge_graph():
+async def knowledge_graph():
     doc_id = request.args["doc_id"]
     tenant_id = DocumentService.get_tenant_id(doc_id)
     kb_ids = KnowledgebaseService.get_kb_ids(tenant_id)
@@ -384,7 +465,7 @@ def knowledge_graph():
         "doc_ids": [doc_id],
         "knowledge_graph_kwd": ["graph", "mind_map"]
     }
-    sres = settings.retriever.search(req, search.index_name(tenant_id), kb_ids)
+    sres = await settings.retriever.search(req, search.index_name(tenant_id), kb_ids)
     obj = {"graph": {}, "mind_map": {}}
     for id in sres.ids[:2]:
         ty = sres.field[id]["knowledge_graph_kwd"]

api/apps/connector_app.py

@@ -26,19 +26,20 @@ from google_auth_oauthlib.flow import Flow
 
 from api.db import InputType
 from api.db.services.connector_service import ConnectorService, SyncLogsService
-from api.utils.api_utils import get_data_error_result, get_json_result, validate_request
+from api.utils.api_utils import get_data_error_result, get_json_result, get_request_json, validate_request
 from common.constants import RetCode, TaskStatus
-from common.data_source.config import GOOGLE_DRIVE_WEB_OAUTH_REDIRECT_URI, DocumentSource
-from common.data_source.google_util.constant import GOOGLE_DRIVE_WEB_OAUTH_POPUP_TEMPLATE, GOOGLE_SCOPES
+from common.data_source.config import GOOGLE_DRIVE_WEB_OAUTH_REDIRECT_URI, GMAIL_WEB_OAUTH_REDIRECT_URI, BOX_WEB_OAUTH_REDIRECT_URI, DocumentSource
+from common.data_source.google_util.constant import WEB_OAUTH_POPUP_TEMPLATE, GOOGLE_SCOPES
 from common.misc_utils import get_uuid
 from rag.utils.redis_conn import REDIS_CONN
 from api.apps import login_required, current_user
+from box_sdk_gen import BoxOAuth, OAuthConfig, GetAuthorizeUrlOptions
 
 
 @manager.route("/set", methods=["POST"])  # noqa: F821
 @login_required
 async def set_connector():
-    req = await request.json
+    req = await get_request_json()
     if req.get("id"):
         conn = {fld: req[fld] for fld in ["prune_freq", "refresh_freq", "config", "timeout_secs"] if fld in req}
         ConnectorService.update_by_id(req["id"], conn)
@@ -90,7 +91,7 @@ def list_logs(connector_id):
 @manager.route("/<connector_id>/resume", methods=["PUT"])  # noqa: F821
 @login_required
 async def resume(connector_id):
-    req = await request.json
+    req = await get_request_json()
     if req.get("resume"):
         ConnectorService.resume(connector_id, TaskStatus.SCHEDULE)
     else:
@@ -102,7 +103,7 @@ async def resume(connector_id):
 @login_required
 @validate_request("kb_id")
 async def rebuild(connector_id):
-    req = await request.json
+    req = await get_request_json()
     err = ConnectorService.rebuild(req["kb_id"], connector_id, current_user.id)
     if err:
         return get_json_result(data=False, message=err, code=RetCode.SERVER_ERROR)
@@ -117,17 +118,27 @@ def rm_connector(connector_id):
     return get_json_result(data=True)
 
 
-GOOGLE_WEB_FLOW_STATE_PREFIX = "google_drive_web_flow_state"
-GOOGLE_WEB_FLOW_RESULT_PREFIX = "google_drive_web_flow_result"
 WEB_FLOW_TTL_SECS = 15 * 60
 
 
-def _web_state_cache_key(flow_id: str) -> str:
-    return f"{GOOGLE_WEB_FLOW_STATE_PREFIX}:{flow_id}"
+def _web_state_cache_key(flow_id: str, source_type: str | None = None) -> str:
+    """Return Redis key for web OAuth state.
+
+    The default prefix keeps backward compatibility for Google Drive.
+    When source_type == "gmail", a different prefix is used so that
+    Drive/Gmail flows don't clash in Redis.
+    """
+    prefix = f"{source_type}_web_flow_state"
+    return f"{prefix}:{flow_id}"
 
 
-def _web_result_cache_key(flow_id: str) -> str:
-    return f"{GOOGLE_WEB_FLOW_RESULT_PREFIX}:{flow_id}"
+def _web_result_cache_key(flow_id: str, source_type: str | None = None) -> str:
+    """Return Redis key for web OAuth result.
+
+    Mirrors _web_state_cache_key logic for result storage.
+    """
+    prefix = f"{source_type}_web_flow_result"
+    return f"{prefix}:{flow_id}"
 
 
 def _load_credentials(payload: str | dict[str, Any]) -> dict[str, Any]:
@@ -146,19 +157,24 @@ def _get_web_client_config(credentials: dict[str, Any]) -> dict[str, Any]:
     return {"web": web_section}
 
 
-async def _render_web_oauth_popup(flow_id: str, success: bool, message: str):
+async def _render_web_oauth_popup(flow_id: str, success: bool, message: str, source="drive"):
     status = "success" if success else "error"
     auto_close = "window.close();" if success else ""
     escaped_message = escape(message)
+    #   Drive: ragflow-google-drive-oauth
+    #   Gmail: ragflow-gmail-oauth
+    payload_type = f"ragflow-{source}-oauth"
     payload_json = json.dumps(
         {
-            "type": "ragflow-google-drive-oauth",
+            "type": payload_type,
             "status": status,
             "flowId": flow_id or "",
             "message": message,
         }
     )
-    html = GOOGLE_DRIVE_WEB_OAUTH_POPUP_TEMPLATE.format(
+    # TODO(google-oauth): title/heading/message may need to reflect drive/gmail based on cached type
+    html = WEB_OAUTH_POPUP_TEMPLATE.format(
+        title=f"Google {source.capitalize()} Authorization",
         heading="Authorization complete" if success else "Authorization failed",
         message=escaped_message,
         payload_json=payload_json,
@@ -169,20 +185,33 @@ async def _render_web_oauth_popup(flow_id: str, success: bool, message: str):
     return response
 
 
-@manager.route("/google-drive/oauth/web/start", methods=["POST"])  # noqa: F821
+@manager.route("/google/oauth/web/start", methods=["POST"])  # noqa: F821
 @login_required
 @validate_request("credentials")
-async def start_google_drive_web_oauth():
-    if not GOOGLE_DRIVE_WEB_OAUTH_REDIRECT_URI:
+async def start_google_web_oauth():
+    source = request.args.get("type", "google-drive")
+    if source not in ("google-drive", "gmail"):
+        return get_json_result(code=RetCode.ARGUMENT_ERROR, message="Invalid Google OAuth type.")
+
+    if source == "gmail":
+        redirect_uri = GMAIL_WEB_OAUTH_REDIRECT_URI
+        scopes = GOOGLE_SCOPES[DocumentSource.GMAIL]
+    else:
+        redirect_uri = GOOGLE_DRIVE_WEB_OAUTH_REDIRECT_URI
+        scopes = GOOGLE_SCOPES[DocumentSource.GOOGLE_DRIVE]
+
+    if not redirect_uri:
         return get_json_result(
             code=RetCode.SERVER_ERROR,
-            message="Google Drive OAuth redirect URI is not configured on the server.",
+            message="Google OAuth redirect URI is not configured on the server.",
         )
 
-    req = await request.json or {}
+    req = await get_request_json()
     raw_credentials = req.get("credentials", "")
+
     try:
         credentials = _load_credentials(raw_credentials)
+        print(credentials)
     except ValueError as exc:
         return get_json_result(code=RetCode.ARGUMENT_ERROR, message=str(exc))
 
@@ -199,8 +228,8 @@ async def start_google_drive_web_oauth():
 
     flow_id = str(uuid.uuid4())
     try:
-        flow = Flow.from_client_config(client_config, scopes=GOOGLE_SCOPES[DocumentSource.GOOGLE_DRIVE])
-        flow.redirect_uri = GOOGLE_DRIVE_WEB_OAUTH_REDIRECT_URI
+        flow = Flow.from_client_config(client_config, scopes=scopes)
+        flow.redirect_uri = redirect_uri
         authorization_url, _ = flow.authorization_url(
             access_type="offline",
             include_granted_scopes="true",
@@ -219,7 +248,7 @@ async def start_google_drive_web_oauth():
         "client_config": client_config,
         "created_at": int(time.time()),
     }
-    REDIS_CONN.set_obj(_web_state_cache_key(flow_id), cache_payload, WEB_FLOW_TTL_SECS)
+    REDIS_CONN.set_obj(_web_state_cache_key(flow_id, source), cache_payload, WEB_FLOW_TTL_SECS)
 
     return get_json_result(
         data={
@@ -230,60 +259,115 @@ async def start_google_drive_web_oauth():
     )
 
 
-@manager.route("/google-drive/oauth/web/callback", methods=["GET"])  # noqa: F821
-async def google_drive_web_oauth_callback():
+@manager.route("/gmail/oauth/web/callback", methods=["GET"])  # noqa: F821
+async def google_gmail_web_oauth_callback():
     state_id = request.args.get("state")
     error = request.args.get("error")
+    source = "gmail"
+
     error_description = request.args.get("error_description") or error
 
     if not state_id:
-        return await _render_web_oauth_popup("", False, "Missing OAuth state parameter.")
+        return await _render_web_oauth_popup("", False, "Missing OAuth state parameter.", source)
 
-    state_cache = REDIS_CONN.get(_web_state_cache_key(state_id))
+    state_cache = REDIS_CONN.get(_web_state_cache_key(state_id, source))
     if not state_cache:
-        return await _render_web_oauth_popup(state_id, False, "Authorization session expired. Please restart from the main window.")
+        return await _render_web_oauth_popup(state_id, False, "Authorization session expired. Please restart from the main window.", source)
 
     state_obj = json.loads(state_cache)
     client_config = state_obj.get("client_config")
     if not client_config:
-        REDIS_CONN.delete(_web_state_cache_key(state_id))
-        return await _render_web_oauth_popup(state_id, False, "Authorization session was invalid. Please retry.")
+        REDIS_CONN.delete(_web_state_cache_key(state_id, source))
+        return await _render_web_oauth_popup(state_id, False, "Authorization session was invalid. Please retry.", source)
 
     if error:
-        REDIS_CONN.delete(_web_state_cache_key(state_id))
-        return await _render_web_oauth_popup(state_id, False, error_description or "Authorization was cancelled.")
+        REDIS_CONN.delete(_web_state_cache_key(state_id, source))
+        return await _render_web_oauth_popup(state_id, False, error_description or "Authorization was cancelled.", source)
 
     code = request.args.get("code")
     if not code:
-        return await _render_web_oauth_popup(state_id, False, "Missing authorization code from Google.")
+        return await _render_web_oauth_popup(state_id, False, "Missing authorization code from Google.", source)
 
     try:
-        flow = Flow.from_client_config(client_config, scopes=GOOGLE_SCOPES[DocumentSource.GOOGLE_DRIVE])
-        flow.redirect_uri = GOOGLE_DRIVE_WEB_OAUTH_REDIRECT_URI
+        # TODO(google-oauth): branch scopes/redirect_uri based on source_type (drive vs gmail)
+        flow = Flow.from_client_config(client_config, scopes=GOOGLE_SCOPES[DocumentSource.GMAIL])
+        flow.redirect_uri = GMAIL_WEB_OAUTH_REDIRECT_URI
         flow.fetch_token(code=code)
     except Exception as exc:  # pragma: no cover - defensive
         logging.exception("Failed to exchange Google OAuth code: %s", exc)
-        REDIS_CONN.delete(_web_state_cache_key(state_id))
-        return await _render_web_oauth_popup(state_id, False, "Failed to exchange tokens with Google. Please retry.")
+        REDIS_CONN.delete(_web_state_cache_key(state_id, source))
+        return await _render_web_oauth_popup(state_id, False, "Failed to exchange tokens with Google. Please retry.", source)
 
     creds_json = flow.credentials.to_json()
     result_payload = {
         "user_id": state_obj.get("user_id"),
         "credentials": creds_json,
     }
-    REDIS_CONN.set_obj(_web_result_cache_key(state_id), result_payload, WEB_FLOW_TTL_SECS)
-    REDIS_CONN.delete(_web_state_cache_key(state_id))
+    REDIS_CONN.set_obj(_web_result_cache_key(state_id, source), result_payload, WEB_FLOW_TTL_SECS)
+    REDIS_CONN.delete(_web_state_cache_key(state_id, source))
 
-    return await _render_web_oauth_popup(state_id, True, "Authorization completed successfully.")
+    return await _render_web_oauth_popup(state_id, True, "Authorization completed successfully.", source)
 
 
-@manager.route("/google-drive/oauth/web/result", methods=["POST"])  # noqa: F821
+@manager.route("/google-drive/oauth/web/callback", methods=["GET"])  # noqa: F821
+async def google_drive_web_oauth_callback():
+    state_id = request.args.get("state")
+    error = request.args.get("error")
+    source = "google-drive"
+
+    error_description = request.args.get("error_description") or error
+
+    if not state_id:
+        return await _render_web_oauth_popup("", False, "Missing OAuth state parameter.", source)
+
+    state_cache = REDIS_CONN.get(_web_state_cache_key(state_id, source))
+    if not state_cache:
+        return await _render_web_oauth_popup(state_id, False, "Authorization session expired. Please restart from the main window.", source)
+
+    state_obj = json.loads(state_cache)
+    client_config = state_obj.get("client_config")
+    if not client_config:
+        REDIS_CONN.delete(_web_state_cache_key(state_id, source))
+        return await _render_web_oauth_popup(state_id, False, "Authorization session was invalid. Please retry.", source)
+
+    if error:
+        REDIS_CONN.delete(_web_state_cache_key(state_id, source))
+        return await _render_web_oauth_popup(state_id, False, error_description or "Authorization was cancelled.", source)
+
+    code = request.args.get("code")
+    if not code:
+        return await _render_web_oauth_popup(state_id, False, "Missing authorization code from Google.", source)
+
+    try:
+        # TODO(google-oauth): branch scopes/redirect_uri based on source_type (drive vs gmail)
+        flow = Flow.from_client_config(client_config, scopes=GOOGLE_SCOPES[DocumentSource.GOOGLE_DRIVE])
+        flow.redirect_uri = GOOGLE_DRIVE_WEB_OAUTH_REDIRECT_URI
+        flow.fetch_token(code=code)
+    except Exception as exc:  # pragma: no cover - defensive
+        logging.exception("Failed to exchange Google OAuth code: %s", exc)
+        REDIS_CONN.delete(_web_state_cache_key(state_id, source))
+        return await _render_web_oauth_popup(state_id, False, "Failed to exchange tokens with Google. Please retry.", source)
+
+    creds_json = flow.credentials.to_json()
+    result_payload = {
+        "user_id": state_obj.get("user_id"),
+        "credentials": creds_json,
+    }
+    REDIS_CONN.set_obj(_web_result_cache_key(state_id, source), result_payload, WEB_FLOW_TTL_SECS)
+    REDIS_CONN.delete(_web_state_cache_key(state_id, source))
+
+    return await _render_web_oauth_popup(state_id, True, "Authorization completed successfully.", source)
+
+@manager.route("/google/oauth/web/result", methods=["POST"])  # noqa: F821
 @login_required
 @validate_request("flow_id")
-async def poll_google_drive_web_result():
+async def poll_google_web_result():
     req = await request.json or {}
+    source = request.args.get("type")
+    if source not in ("google-drive", "gmail"):
+        return get_json_result(code=RetCode.ARGUMENT_ERROR, message="Invalid Google OAuth type.")
     flow_id = req.get("flow_id")
-    cache_raw = REDIS_CONN.get(_web_result_cache_key(flow_id))
+    cache_raw = REDIS_CONN.get(_web_result_cache_key(flow_id, source))
     if not cache_raw:
         return get_json_result(code=RetCode.RUNNING, message="Authorization is still pending.")
 
@@ -291,5 +375,109 @@ async def poll_google_drive_web_result():
     if result.get("user_id") != current_user.id:
         return get_json_result(code=RetCode.PERMISSION_ERROR, message="You are not allowed to access this authorization result.")
 
-    REDIS_CONN.delete(_web_result_cache_key(flow_id))
+    REDIS_CONN.delete(_web_result_cache_key(flow_id, source))
     return get_json_result(data={"credentials": result.get("credentials")})
+
+@manager.route("/box/oauth/web/start", methods=["POST"])  # noqa: F821
+@login_required
+async def start_box_web_oauth():
+    req = await get_request_json()
+
+    client_id = req.get("client_id")
+    client_secret = req.get("client_secret")    
+    redirect_uri = req.get("redirect_uri", BOX_WEB_OAUTH_REDIRECT_URI)
+
+    if not client_id or not client_secret:
+        return get_json_result(code=RetCode.ARGUMENT_ERROR, message="Box client_id and client_secret are required.")
+
+    flow_id = str(uuid.uuid4())
+
+    box_auth = BoxOAuth(
+        OAuthConfig(
+            client_id=client_id,
+            client_secret=client_secret,
+        )
+    )
+
+    auth_url = box_auth.get_authorize_url(
+        options=GetAuthorizeUrlOptions(
+            redirect_uri=redirect_uri,
+            state=flow_id,
+        )
+    )
+
+    cache_payload = {
+        "user_id": current_user.id,
+        "auth_url": auth_url,
+        "client_id": client_id,
+        "client_secret": client_secret,
+        "created_at": int(time.time()),
+    }
+    REDIS_CONN.set_obj(_web_state_cache_key(flow_id, "box"), cache_payload, WEB_FLOW_TTL_SECS)
+    return get_json_result(
+        data = {
+            "flow_id": flow_id,
+            "authorization_url": auth_url,
+            "expires_in": WEB_FLOW_TTL_SECS,}
+    )
+
+@manager.route("/box/oauth/web/callback", methods=["GET"])  # noqa: F821
+async def box_web_oauth_callback():
+    flow_id = request.args.get("state")
+    if not flow_id:
+        return await _render_web_oauth_popup("", False, "Missing OAuth parameters.", "box")
+    
+    code = request.args.get("code")
+    if not code:
+        return await _render_web_oauth_popup(flow_id, False, "Missing authorization code from Box.", "box")
+
+    cache_payload = json.loads(REDIS_CONN.get(_web_state_cache_key(flow_id, "box")))
+    if not cache_payload:
+        return get_json_result(code=RetCode.ARGUMENT_ERROR, message="Box OAuth session expired or invalid.")
+
+    error = request.args.get("error")
+    error_description = request.args.get("error_description") or error
+    if error:
+        REDIS_CONN.delete(_web_state_cache_key(flow_id, "box"))
+        return await _render_web_oauth_popup(flow_id, False, error_description or "Authorization failed.", "box")
+    
+    auth = BoxOAuth(
+        OAuthConfig(
+            client_id=cache_payload.get("client_id"),
+            client_secret=cache_payload.get("client_secret"),
+        )
+    )
+
+    auth.get_tokens_authorization_code_grant(code)
+    token = auth.retrieve_token()
+    result_payload = {
+        "user_id": cache_payload.get("user_id"),
+        "client_id": cache_payload.get("client_id"),
+        "client_secret": cache_payload.get("client_secret"),
+        "access_token": token.access_token,
+        "refresh_token": token.refresh_token,
+    }
+
+    REDIS_CONN.set_obj(_web_result_cache_key(flow_id, "box"), result_payload, WEB_FLOW_TTL_SECS)
+    REDIS_CONN.delete(_web_state_cache_key(flow_id, "box"))
+
+    return await _render_web_oauth_popup(flow_id, True, "Authorization completed successfully.", "box")
+
+@manager.route("/box/oauth/web/result", methods=["POST"])  # noqa: F821
+@login_required
+@validate_request("flow_id")
+async def poll_box_web_result():
+    req = await get_request_json()
+    flow_id = req.get("flow_id")
+
+    cache_blob = REDIS_CONN.get(_web_result_cache_key(flow_id, "box"))
+    if not cache_blob:
+        return get_json_result(code=RetCode.RUNNING, message="Authorization is still pending.")
+
+    cache_raw = json.loads(cache_blob)
+    if cache_raw.get("user_id") != current_user.id:
+        return get_json_result(code=RetCode.PERMISSION_ERROR, message="You are not allowed to access this authorization result.")
+    
+    REDIS_CONN.delete(_web_result_cache_key(flow_id, "box"))
+
+    return get_json_result(data={"credentials": cache_raw})

api/apps/conversation_app.py

@@ -14,19 +14,21 @@
 #  limitations under the License.
 #
 import json
+import os
 import re
 import logging
 from copy import deepcopy
+import tempfile
 from quart import Response, request
 from api.apps import current_user, login_required
 from api.db.db_models import APIToken
 from api.db.services.conversation_service import ConversationService, structure_answer
-from api.db.services.dialog_service import DialogService, ask, chat, gen_mindmap
+from api.db.services.dialog_service import DialogService, async_ask, async_chat, gen_mindmap
 from api.db.services.llm_service import LLMBundle
 from api.db.services.search_service import SearchService
 from api.db.services.tenant_llm_service import TenantLLMService
 from api.db.services.user_service import TenantService, UserTenantService
-from api.utils.api_utils import get_data_error_result, get_json_result, server_error_response, validate_request
+from api.utils.api_utils import get_data_error_result, get_json_result, get_request_json, server_error_response, validate_request
 from rag.prompts.template import load_prompt
 from rag.prompts.generator import chunks_format
 from common.constants import RetCode, LLMType
@@ -35,7 +37,7 @@ from common.constants import RetCode, LLMType
 @manager.route("/set", methods=["POST"])  # noqa: F821
 @login_required
 async def set_conversation():
-    req = await request.json
+    req = await get_request_json()
     conv_id = req.get("conversation_id")
     is_new = req.get("is_new")
     name = req.get("name", "New conversation")
@@ -78,7 +80,7 @@ async def set_conversation():
 
 @manager.route("/get", methods=["GET"])  # noqa: F821
 @login_required
-def get():
+async def get():
     conv_id = request.args["conversation_id"]
     try:
         e, conv = ConversationService.get_by_id(conv_id)
@@ -129,7 +131,7 @@ def getsse(dialog_id):
 @manager.route("/rm", methods=["POST"])  # noqa: F821
 @login_required
 async def rm():
-    req = await request.json
+    req = await get_request_json()
     conv_ids = req["conversation_ids"]
     try:
         for cid in conv_ids:
@@ -150,7 +152,7 @@ async def rm():
 
 @manager.route("/list", methods=["GET"])  # noqa: F821
 @login_required
-def list_conversation():
+async def list_conversation():
     dialog_id = request.args["dialog_id"]
     try:
         if not DialogService.query(tenant_id=current_user.id, id=dialog_id):
@@ -167,7 +169,7 @@ def list_conversation():
 @login_required
 @validate_request("conversation_id", "messages")
 async def completion():
-    req = await request.json
+    req = await get_request_json()
     msg = []
     for m in req["messages"]:
         if m["role"] == "system":
@@ -216,10 +218,10 @@ async def completion():
             dia.llm_setting = chat_model_config
 
         is_embedded = bool(chat_model_id)
-        def stream():
+        async def stream():
             nonlocal dia, msg, req, conv
             try:
-                for ans in chat(dia, msg, True, **req):
+                async for ans in async_chat(dia, msg, True, **req):
                     ans = structure_answer(conv, ans, message_id, conv.id)
                     yield "data:" + json.dumps({"code": 0, "message": "", "data": ans}, ensure_ascii=False) + "\n\n"
                 if not is_embedded:
@@ -239,7 +241,7 @@ async def completion():
 
         else:
             answer = None
-            for ans in chat(dia, msg, **req):
+            async for ans in async_chat(dia, msg, **req):
                 answer = structure_answer(conv, ans, message_id, conv.id)
                 if not is_embedded:
                     ConversationService.update_by_id(conv.id, conv.to_dict())
@@ -248,11 +250,69 @@ async def completion():
     except Exception as e:
         return server_error_response(e)
 
+@manager.route("/sequence2txt", methods=["POST"])  # noqa: F821
+@login_required
+async def sequence2txt():
+    req = await request.form
+    stream_mode = req.get("stream", "false").lower() == "true"
+    files = await request.files
+    if "file" not in files:
+        return get_data_error_result(message="Missing 'file' in multipart form-data")
+
+    uploaded = files["file"]
+
+    ALLOWED_EXTS = {
+        ".wav", ".mp3", ".m4a", ".aac",
+        ".flac", ".ogg", ".webm",
+        ".opus", ".wma"
+    }
+
+    filename = uploaded.filename or ""
+    suffix = os.path.splitext(filename)[-1].lower()
+    if suffix not in ALLOWED_EXTS:
+        return get_data_error_result(message=
+            f"Unsupported audio format: {suffix}. "
+            f"Allowed: {', '.join(sorted(ALLOWED_EXTS))}"
+        )
+    fd, temp_audio_path = tempfile.mkstemp(suffix=suffix)
+    os.close(fd)
+    await uploaded.save(temp_audio_path)
+
+    tenants = TenantService.get_info_by(current_user.id)
+    if not tenants:
+        return get_data_error_result(message="Tenant not found!")
+
+    asr_id = tenants[0]["asr_id"]
+    if not asr_id:
+        return get_data_error_result(message="No default ASR model is set")
+
+    asr_mdl=LLMBundle(tenants[0]["tenant_id"], LLMType.SPEECH2TEXT, asr_id)
+    if not stream_mode:
+        text = asr_mdl.transcription(temp_audio_path)
+        try:
+            os.remove(temp_audio_path)
+        except Exception as e:
+            logging.error(f"Failed to remove temp audio file: {str(e)}")
+        return get_json_result(data={"text": text})
+    async def event_stream():
+        try:
+            for evt in asr_mdl.stream_transcription(temp_audio_path):
+                yield f"data: {json.dumps(evt, ensure_ascii=False)}\n\n"
+        except Exception as e:
+            err = {"event": "error", "text": str(e)}
+            yield f"data: {json.dumps(err, ensure_ascii=False)}\n\n"
+        finally:
+            try:
+                os.remove(temp_audio_path)
+            except Exception as e:
+                logging.error(f"Failed to remove temp audio file: {str(e)}")
+
+    return Response(event_stream(), content_type="text/event-stream")
 
 @manager.route("/tts", methods=["POST"])  # noqa: F821
 @login_required
 async def tts():
-    req = await request.json
+    req = await get_request_json()
     text = req["text"]
 
     tenants = TenantService.get_info_by(current_user.id)
@@ -285,7 +345,7 @@ async def tts():
 @login_required
 @validate_request("conversation_id", "message_id")
 async def delete_msg():
-    req = await request.json
+    req = await get_request_json()
     e, conv = ConversationService.get_by_id(req["conversation_id"])
     if not e:
         return get_data_error_result(message="Conversation not found!")
@@ -308,7 +368,7 @@ async def delete_msg():
 @login_required
 @validate_request("conversation_id", "message_id")
 async def thumbup():
-    req = await request.json
+    req = await get_request_json()
     e, conv = ConversationService.get_by_id(req["conversation_id"])
     if not e:
         return get_data_error_result(message="Conversation not found!")
@@ -335,7 +395,7 @@ async def thumbup():
 @login_required
 @validate_request("question", "kb_ids")
 async def ask_about():
-    req = await request.json
+    req = await get_request_json()
     uid = current_user.id
 
     search_id = req.get("search_id", "")
@@ -346,10 +406,10 @@ async def ask_about():
     if search_app:
         search_config = search_app.get("search_config", {})
 
-    def stream():
+    async def stream():
         nonlocal req, uid
         try:
-            for ans in ask(req["question"], req["kb_ids"], uid, search_config=search_config):
+            async for ans in async_ask(req["question"], req["kb_ids"], uid, search_config=search_config):
                 yield "data:" + json.dumps({"code": 0, "message": "", "data": ans}, ensure_ascii=False) + "\n\n"
         except Exception as e:
             yield "data:" + json.dumps({"code": 500, "message": str(e), "data": {"answer": "**ERROR**: " + str(e), "reference": []}}, ensure_ascii=False) + "\n\n"
@@ -367,7 +427,7 @@ async def ask_about():
 @login_required
 @validate_request("question", "kb_ids")
 async def mindmap():
-    req = await request.json
+    req = await get_request_json()
     search_id = req.get("search_id", "")
     search_app = SearchService.get_detail(search_id) if search_id else {}
     search_config = search_app.get("search_config", {}) if search_app else {}
@@ -375,7 +435,7 @@ async def mindmap():
     kb_ids.extend(req["kb_ids"])
     kb_ids = list(set(kb_ids))
 
-    mind_map = gen_mindmap(req["question"], kb_ids, search_app.get("tenant_id", current_user.id), search_config)
+    mind_map = await gen_mindmap(req["question"], kb_ids, search_app.get("tenant_id", current_user.id), search_config)
     if "error" in mind_map:
         return server_error_response(Exception(mind_map["error"]))
     return get_json_result(data=mind_map)
@@ -385,7 +445,7 @@ async def mindmap():
 @login_required
 @validate_request("question")
 async def related_questions():
-    req = await request.json
+    req = await get_request_json()
 
     search_id = req.get("search_id", "")
     search_config = {}
@@ -402,7 +462,7 @@ async def related_questions():
     if "parameter" in gen_conf:
         del gen_conf["parameter"]
     prompt = load_prompt("related_question")
-    ans = chat_mdl.chat(
+    ans = await chat_mdl.async_chat(
         prompt,
         [
             {

api/apps/dialog_app.py

@@ -21,18 +21,18 @@ from common.constants import StatusEnum
 from api.db.services.tenant_llm_service import TenantLLMService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.user_service import TenantService, UserTenantService
-from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
+from api.utils.api_utils import get_data_error_result, get_json_result, get_request_json, server_error_response, validate_request
 from common.misc_utils import get_uuid
 from common.constants import RetCode
-from api.utils.api_utils import get_json_result
 from api.apps import login_required, current_user
+import logging
 
 
 @manager.route('/set', methods=['POST'])  # noqa: F821
 @validate_request("prompt_config")
 @login_required
 async def set_dialog():
-    req = await request.json
+    req = await get_request_json()
     dialog_id = req.get("dialog_id", "")
     is_create = not dialog_id
     name = req.get("name", "New Dialog")
@@ -43,13 +43,19 @@ async def set_dialog():
     if len(name.encode("utf-8")) > 255:
         return get_data_error_result(message=f"Dialog name length is {len(name)} which is larger than 255")
 
-    if is_create and DialogService.query(tenant_id=current_user.id, name=name.strip()):
-        name = name.strip()
-        name = duplicate_name(
-            DialogService.query,
-            name=name,
-            tenant_id=current_user.id,
-            status=StatusEnum.VALID.value)
+    name = name.strip()
+    if is_create:
+        # only for chat creating
+        existing_names = {
+            d.name.casefold()
+            for d in DialogService.query(tenant_id=current_user.id, status=StatusEnum.VALID.value)
+            if d.name
+        }
+        if name.casefold() in existing_names:
+            def _name_exists(name: str, **_kwargs) -> bool:
+                return name.casefold() in existing_names
+
+            name = duplicate_name(_name_exists, name=name)
 
     description = req.get("description", "A helpful dialog")
     icon = req.get("icon", "")
@@ -64,16 +70,30 @@ async def set_dialog():
     meta_data_filter = req.get("meta_data_filter", {})
     prompt_config = req["prompt_config"]
 
-    if not is_create:
-        if not req.get("kb_ids", []) and not prompt_config.get("tavily_api_key") and "{knowledge}" in prompt_config['system']:
-            return get_data_error_result(message="Please remove `{knowledge}` in system prompt since no knowledge base / Tavily used here.")
+    # Set default parameters for datasets with knowledge retrieval
+    # All datasets with {knowledge} in system prompt need "knowledge" parameter to enable retrieval
+    kb_ids = req.get("kb_ids", [])
+    parameters = prompt_config.get("parameters")
+    logging.debug(f"set_dialog: kb_ids={kb_ids}, parameters={parameters}, is_create={not is_create}")
+    # Check if parameters is missing, None, or empty list
+    if kb_ids and not parameters:
+        # Check if system prompt uses {knowledge} placeholder
+        if "{knowledge}" in prompt_config.get("system", ""):
+            # Set default parameters for any dataset with knowledge placeholder
+            prompt_config["parameters"] = [{"key": "knowledge", "optional": False}]
+            logging.debug(f"Set default parameters for datasets with knowledge placeholder: {kb_ids}")
 
-        for p in prompt_config["parameters"]:
-            if p["optional"]:
-                continue
-            if prompt_config["system"].find("{%s}" % p["key"]) < 0:
-                return get_data_error_result(
-                    message="Parameter '{}' is not used".format(p["key"]))
+    if not is_create:
+        # only for chat updating
+        if not req.get("kb_ids", []) and not prompt_config.get("tavily_api_key") and "{knowledge}" in prompt_config.get("system", ""):
+            return get_data_error_result(message="Please remove `{knowledge}` in system prompt since no dataset / Tavily used here.")
+
+    for p in prompt_config.get("parameters", []):
+        if p["optional"]:
+            continue
+        if prompt_config.get("system", "").find("{%s}" % p["key"]) < 0:
+            return get_data_error_result(
+                message="Parameter '{}' is not used".format(p["key"]))
 
     try:
         e, tenant = TenantService.get_by_id(current_user.id)
@@ -181,7 +201,7 @@ async def list_dialogs_next():
     else:
         desc = True
 
-    req = await request.get_json()
+    req = await get_request_json()
     owner_ids = req.get("owner_ids", [])
     try:
         if not owner_ids:
@@ -209,7 +229,7 @@ async def list_dialogs_next():
 @login_required
 @validate_request("dialog_ids")
 async def rm():
-    req = await request.json
+    req = await get_request_json()
     dialog_list=[]
     tenants = UserTenantService.query(user_id=current_user.id)
     try:

api/apps/document_app.py

@@ -26,17 +26,19 @@ from api.db import VALID_FILE_TYPES, FileType
 from api.db.db_models import Task
 from api.db.services import duplicate_name
 from api.db.services.document_service import DocumentService, doc_upload_and_parse
+from common.metadata_utils import meta_filter, convert_conditions, turn2jsonschema
 from api.db.services.file2document_service import File2DocumentService
 from api.db.services.file_service import FileService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.task_service import TaskService, cancel_all_task_of
 from api.db.services.user_service import UserTenantService
-from common.misc_utils import get_uuid
+from common.misc_utils import get_uuid, thread_pool_exec
 from api.utils.api_utils import (
     get_data_error_result,
     get_json_result,
     server_error_response,
-    validate_request, request_json,
+    validate_request,
+    get_request_json,
 )
 from api.utils.file_utils import filename_type, thumbnail
 from common.file_utils import get_project_base_directory
@@ -60,20 +62,32 @@ async def upload():
         return get_json_result(data=False, message="No file part!", code=RetCode.ARGUMENT_ERROR)
 
     file_objs = files.getlist("file")
+    def _close_file_objs(objs):
+        for obj in objs:
+            try:
+                obj.close()
+            except Exception:
+                try:
+                    obj.stream.close()
+                except Exception:
+                    pass
     for file_obj in file_objs:
         if file_obj.filename == "":
+            _close_file_objs(file_objs)
             return get_json_result(data=False, message="No file selected!", code=RetCode.ARGUMENT_ERROR)
         if len(file_obj.filename.encode("utf-8")) > FILE_NAME_LEN_LIMIT:
+            _close_file_objs(file_objs)
             return get_json_result(data=False, message=f"File name must be {FILE_NAME_LEN_LIMIT} bytes or less.", code=RetCode.ARGUMENT_ERROR)
 
     e, kb = KnowledgebaseService.get_by_id(kb_id)
     if not e:
-        raise LookupError("Can't find this knowledgebase!")
+        raise LookupError("Can't find this dataset!")
     if not check_kb_team_permission(kb, current_user.id):
         return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
 
-    err, files = FileService.upload_document(kb, file_objs, current_user.id)
+    err, files = await thread_pool_exec(FileService.upload_document, kb, file_objs, current_user.id)
     if err:
+        files = [f[0] for f in files] if files else []
         return get_json_result(data=files, message="\n".join(err), code=RetCode.SERVER_ERROR)
 
     if not files:
@@ -97,7 +111,7 @@ async def web_crawl():
         return get_json_result(data=False, message="The URL format is invalid", code=RetCode.ARGUMENT_ERROR)
     e, kb = KnowledgebaseService.get_by_id(kb_id)
     if not e:
-        raise LookupError("Can't find this knowledgebase!")
+        raise LookupError("Can't find this dataset!")
     if check_kb_team_permission(kb, current_user.id):
         return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
 
@@ -153,7 +167,7 @@ async def web_crawl():
 @login_required
 @validate_request("name", "kb_id")
 async def create():
-    req = await request_json()
+    req = await get_request_json()
     kb_id = req["kb_id"]
     if not kb_id:
         return get_json_result(data=False, message='Lack of "KB ID"', code=RetCode.ARGUMENT_ERROR)
@@ -167,10 +181,10 @@ async def create():
     try:
         e, kb = KnowledgebaseService.get_by_id(kb_id)
         if not e:
-            return get_data_error_result(message="Can't find this knowledgebase!")
+            return get_data_error_result(message="Can't find this dataset!")
 
         if DocumentService.query(name=req["name"], kb_id=kb_id):
-            return get_data_error_result(message="Duplicated document name in the same knowledgebase.")
+            return get_data_error_result(message="Duplicated document name in the same dataset.")
 
         kb_root_folder = FileService.get_kb_folder(kb.tenant_id)
         if not kb_root_folder:
@@ -212,12 +226,13 @@ async def list_docs():
     kb_id = request.args.get("kb_id")
     if not kb_id:
         return get_json_result(data=False, message='Lack of "KB ID"', code=RetCode.ARGUMENT_ERROR)
+        
     tenants = UserTenantService.query(user_id=current_user.id)
     for tenant in tenants:
         if KnowledgebaseService.query(tenant_id=tenant.tenant_id, id=kb_id):
             break
     else:
-        return get_json_result(data=False, message="Only owner of knowledgebase authorized for this operation.", code=RetCode.OPERATING_ERROR)
+        return get_json_result(data=False, message="Only owner of dataset authorized for this operation.", code=RetCode.OPERATING_ERROR)
     keywords = request.args.get("keywords", "")
 
     page_number = int(request.args.get("page", 0))
@@ -230,7 +245,11 @@ async def list_docs():
     create_time_from = int(request.args.get("create_time_from", 0))
     create_time_to = int(request.args.get("create_time_to", 0))
 
-    req = await request.get_json()
+    req = await get_request_json()
+
+    return_empty_metadata = req.get("return_empty_metadata", False)
+    if isinstance(return_empty_metadata, str):
+        return_empty_metadata = return_empty_metadata.lower() == "true"
 
     run_status = req.get("run_status", [])
     if run_status:
@@ -245,9 +264,74 @@ async def list_docs():
             return get_data_error_result(message=f"Invalid filter conditions: {', '.join(invalid_types)} type{'s' if len(invalid_types) > 1 else ''}")
 
     suffix = req.get("suffix", [])
+    metadata_condition = req.get("metadata_condition", {}) or {}
+    metadata = req.get("metadata", {}) or {}
+    if isinstance(metadata, dict) and metadata.get("empty_metadata"):
+        return_empty_metadata = True
+        metadata = {k: v for k, v in metadata.items() if k != "empty_metadata"}
+    if return_empty_metadata:
+        metadata_condition = {}
+        metadata = {}
+    else:
+        if metadata_condition and not isinstance(metadata_condition, dict):
+            return get_data_error_result(message="metadata_condition must be an object.")
+        if metadata and not isinstance(metadata, dict):
+            return get_data_error_result(message="metadata must be an object.")
+
+    doc_ids_filter = None
+    metas = None
+    if metadata_condition or metadata:
+        metas = DocumentService.get_flatted_meta_by_kbs([kb_id])
+
+    if metadata_condition:
+        doc_ids_filter = set(meta_filter(metas, convert_conditions(metadata_condition), metadata_condition.get("logic", "and")))
+        if metadata_condition.get("conditions") and not doc_ids_filter:
+            return get_json_result(data={"total": 0, "docs": []})
+
+    if metadata:
+        metadata_doc_ids = None
+        for key, values in metadata.items():
+            if not values:
+                continue
+            if not isinstance(values, list):
+                values = [values]
+            values = [str(v) for v in values if v is not None and str(v).strip()]
+            if not values:
+                continue
+            key_doc_ids = set()
+            for value in values:
+                key_doc_ids.update(metas.get(key, {}).get(value, []))
+            if metadata_doc_ids is None:
+                metadata_doc_ids = key_doc_ids
+            else:
+                metadata_doc_ids &= key_doc_ids
+            if not metadata_doc_ids:
+                return get_json_result(data={"total": 0, "docs": []})
+        if metadata_doc_ids is not None:
+            if doc_ids_filter is None:
+                doc_ids_filter = metadata_doc_ids
+            else:
+                doc_ids_filter &= metadata_doc_ids
+            if not doc_ids_filter:
+                return get_json_result(data={"total": 0, "docs": []})
+
+    if doc_ids_filter is not None:
+        doc_ids_filter = list(doc_ids_filter)
 
     try:
-        docs, tol = DocumentService.get_by_kb_id(kb_id, page_number, items_per_page, orderby, desc, keywords, run_status, types, suffix)
+        docs, tol = DocumentService.get_by_kb_id(
+            kb_id,
+            page_number,
+            items_per_page,
+            orderby,
+            desc,
+            keywords,
+            run_status,
+            types,
+            suffix,
+            doc_ids_filter,
+            return_empty_metadata=return_empty_metadata,
+        )
 
         if create_time_from or create_time_to:
             filtered_docs = []
@@ -262,6 +346,8 @@ async def list_docs():
                 doc_item["thumbnail"] = f"/v1/document/image/{kb_id}-{doc_item['thumbnail']}"
             if doc_item.get("source_type"):
                 doc_item["source_type"] = doc_item["source_type"].split("/")[0]
+            if doc_item["parser_config"].get("metadata"):
+                doc_item["parser_config"]["metadata"] = turn2jsonschema(doc_item["parser_config"]["metadata"])
 
         return get_json_result(data={"total": tol, "docs": docs})
     except Exception as e:
@@ -271,7 +357,7 @@ async def list_docs():
 @manager.route("/filter", methods=["POST"])  # noqa: F821
 @login_required
 async def get_filter():
-    req = await request.get_json()
+    req = await get_request_json()
 
     kb_id = req.get("kb_id")
     if not kb_id:
@@ -281,7 +367,7 @@ async def get_filter():
         if KnowledgebaseService.query(tenant_id=tenant.tenant_id, id=kb_id):
             break
     else:
-        return get_json_result(data=False, message="Only owner of knowledgebase authorized for this operation.", code=RetCode.OPERATING_ERROR)
+        return get_json_result(data=False, message="Only owner of dataset authorized for this operation.", code=RetCode.OPERATING_ERROR)
 
     keywords = req.get("keywords", "")
 
@@ -309,7 +395,7 @@ async def get_filter():
 @manager.route("/infos", methods=["POST"])  # noqa: F821
 @login_required
 async def doc_infos():
-    req = await request_json()
+    req = await get_request_json()
     doc_ids = req["doc_ids"]
     for doc_id in doc_ids:
         if not DocumentService.accessible(doc_id, current_user.id):
@@ -318,6 +404,72 @@ async def doc_infos():
     return get_json_result(data=list(docs.dicts()))
 
 
+@manager.route("/metadata/summary", methods=["POST"])  # noqa: F821
+@login_required
+async def metadata_summary():
+    req = await get_request_json()
+    kb_id = req.get("kb_id")
+    doc_ids = req.get("doc_ids")
+    if not kb_id:
+        return get_json_result(data=False, message='Lack of "KB ID"', code=RetCode.ARGUMENT_ERROR)
+
+    tenants = UserTenantService.query(user_id=current_user.id)
+    for tenant in tenants:
+        if KnowledgebaseService.query(tenant_id=tenant.tenant_id, id=kb_id):
+            break
+    else:
+        return get_json_result(data=False, message="Only owner of dataset authorized for this operation.", code=RetCode.OPERATING_ERROR)
+
+    try:
+        summary = DocumentService.get_metadata_summary(kb_id, doc_ids)
+        return get_json_result(data={"summary": summary})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route("/metadata/update", methods=["POST"])  # noqa: F821
+@login_required
+@validate_request("doc_ids")
+async def metadata_update():
+    req = await get_request_json()
+    document_ids = req.get("doc_ids")
+    updates = req.get("updates", []) or []
+    deletes = req.get("deletes", []) or []
+
+    if not isinstance(updates, list) or not isinstance(deletes, list):
+        return get_json_result(data=False, message="updates and deletes must be lists.", code=RetCode.ARGUMENT_ERROR)
+
+    for upd in updates:
+        if not isinstance(upd, dict) or not upd.get("key") or "value" not in upd:
+            return get_json_result(data=False, message="Each update requires key and value.", code=RetCode.ARGUMENT_ERROR)
+    for d in deletes:
+        if not isinstance(d, dict) or not d.get("key"):
+            return get_json_result(data=False, message="Each delete requires key.", code=RetCode.ARGUMENT_ERROR)
+
+    updated = DocumentService.batch_update_metadata(None, document_ids, updates, deletes)
+    return get_json_result(data={"updated": updated})
+
+
+@manager.route("/update_metadata_setting", methods=["POST"])  # noqa: F821
+@login_required
+@validate_request("doc_id", "metadata")
+async def update_metadata_setting():
+    req = await get_request_json()
+    if not DocumentService.accessible(req["doc_id"], current_user.id):
+        return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
+
+    e, doc = DocumentService.get_by_id(req["doc_id"])
+    if not e:
+        return get_data_error_result(message="Document not found!")
+
+    DocumentService.update_parser_config(doc.id, {"metadata": req["metadata"]})
+    e, doc = DocumentService.get_by_id(doc.id)
+    if not e:
+        return get_data_error_result(message="Document not found!")
+
+    return get_json_result(data=doc.to_dict())
+
+
 @manager.route("/thumbnails", methods=["GET"])  # noqa: F821
 # @login_required
 def thumbnails():
@@ -341,7 +493,7 @@ def thumbnails():
 @login_required
 @validate_request("doc_ids", "status")
 async def change_status():
-    req = await request.get_json()
+    req = await get_request_json()
     doc_ids = req.get("doc_ids", [])
     status = str(req.get("status", ""))
 
@@ -349,31 +501,61 @@ async def change_status():
         return get_json_result(data=False, message='"Status" must be either 0 or 1!', code=RetCode.ARGUMENT_ERROR)
 
     result = {}
+    has_error = False
     for doc_id in doc_ids:
         if not DocumentService.accessible(doc_id, current_user.id):
             result[doc_id] = {"error": "No authorization."}
+            has_error = True
             continue
 
         try:
             e, doc = DocumentService.get_by_id(doc_id)
             if not e:
                 result[doc_id] = {"error": "No authorization."}
+                has_error = True
                 continue
             e, kb = KnowledgebaseService.get_by_id(doc.kb_id)
             if not e:
-                result[doc_id] = {"error": "Can't find this knowledgebase!"}
+                result[doc_id] = {"error": "Can't find this dataset!"}
+                has_error = True
+                continue
+            current_status = str(doc.status)
+            if current_status == status:
+                result[doc_id] = {"status": status}
                 continue
             if not DocumentService.update_by_id(doc_id, {"status": str(status)}):
                 result[doc_id] = {"error": "Database error (Document update)!"}
+                has_error = True
                 continue
 
             status_int = int(status)
-            if not settings.docStoreConn.update({"doc_id": doc_id}, {"available_int": status_int}, search.index_name(kb.tenant_id), doc.kb_id):
-                result[doc_id] = {"error": "Database error (docStore update)!"}
+            if getattr(doc, "chunk_num", 0) > 0:
+                try:
+                    ok = settings.docStoreConn.update(
+                        {"doc_id": doc_id},
+                        {"available_int": status_int},
+                        search.index_name(kb.tenant_id),
+                        doc.kb_id,
+                    )
+                except Exception as exc:
+                    msg = str(exc)
+                    if "3022" in msg:
+                        result[doc_id] = {"error": "Document store table missing."}
+                    else:
+                        result[doc_id] = {"error": f"Document store update failed: {msg}"}
+                    has_error = True
+                    continue
+                if not ok:
+                    result[doc_id] = {"error": "Database error (docStore update)!"}
+                    has_error = True
+                    continue
             result[doc_id] = {"status": status}
         except Exception as e:
             result[doc_id] = {"error": f"Internal server error: {str(e)}"}
+            has_error = True
 
+    if has_error:
+        return get_json_result(data=result, message="Partial failure", code=RetCode.SERVER_ERROR)
     return get_json_result(data=result)
 
 
@@ -381,7 +563,7 @@ async def change_status():
 @login_required
 @validate_request("doc_id")
 async def rm():
-    req = await request_json()
+    req = await get_request_json()
     doc_ids = req["doc_id"]
     if isinstance(doc_ids, str):
         doc_ids = [doc_ids]
@@ -390,7 +572,7 @@ async def rm():
         if not DocumentService.accessible4deletion(doc_id, current_user.id):
             return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
 
-    errors = FileService.delete_docs(doc_ids, current_user.id)
+    errors = await thread_pool_exec(FileService.delete_docs, doc_ids, current_user.id)
 
     if errors:
         return get_json_result(data=False, message=errors, code=RetCode.SERVER_ERROR)
@@ -402,45 +584,58 @@ async def rm():
 @login_required
 @validate_request("doc_ids", "run")
 async def run():
-    req = await request_json()
-    for doc_id in req["doc_ids"]:
-        if not DocumentService.accessible(doc_id, current_user.id):
-            return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
+    req = await get_request_json()
+    uid = current_user.id
     try:
-        kb_table_num_map = {}
-        for id in req["doc_ids"]:
-            info = {"run": str(req["run"]), "progress": 0}
-            if str(req["run"]) == TaskStatus.RUNNING.value and req.get("delete", False):
-                info["progress_msg"] = ""
-                info["chunk_num"] = 0
-                info["token_num"] = 0
+        def _run_sync():
+            for doc_id in req["doc_ids"]:
+                if not DocumentService.accessible(doc_id, uid):
+                    return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
 
-            tenant_id = DocumentService.get_tenant_id(id)
-            if not tenant_id:
-                return get_data_error_result(message="Tenant not found!")
-            e, doc = DocumentService.get_by_id(id)
-            if not e:
-                return get_data_error_result(message="Document not found!")
+            kb_table_num_map = {}
+            for id in req["doc_ids"]:
+                info = {"run": str(req["run"]), "progress": 0}
+                if str(req["run"]) == TaskStatus.RUNNING.value and req.get("delete", False):
+                    info["progress_msg"] = ""
+                    info["chunk_num"] = 0
+                    info["token_num"] = 0
 
-            if str(req["run"]) == TaskStatus.CANCEL.value:
-                if str(doc.run) == TaskStatus.RUNNING.value:
-                    cancel_all_task_of(id)
-                else:
-                    return get_data_error_result(message="Cannot cancel a task that is not in RUNNING status")
-            if all([("delete" not in req or req["delete"]), str(req["run"]) == TaskStatus.RUNNING.value, str(doc.run) == TaskStatus.DONE.value]):
-                DocumentService.clear_chunk_num_when_rerun(doc.id)
+                tenant_id = DocumentService.get_tenant_id(id)
+                if not tenant_id:
+                    return get_data_error_result(message="Tenant not found!")
+                e, doc = DocumentService.get_by_id(id)
+                if not e:
+                    return get_data_error_result(message="Document not found!")
 
-            DocumentService.update_by_id(id, info)
-            if req.get("delete", False):
-                TaskService.filter_delete([Task.doc_id == id])
-                if settings.docStoreConn.indexExist(search.index_name(tenant_id), doc.kb_id):
-                    settings.docStoreConn.delete({"doc_id": id}, search.index_name(tenant_id), doc.kb_id)
+                if str(req["run"]) == TaskStatus.CANCEL.value:
+                    if str(doc.run) == TaskStatus.RUNNING.value:
+                        cancel_all_task_of(id)
+                    else:
+                        return get_data_error_result(message="Cannot cancel a task that is not in RUNNING status")
+                if all([("delete" not in req or req["delete"]), str(req["run"]) == TaskStatus.RUNNING.value, str(doc.run) == TaskStatus.DONE.value]):
+                    DocumentService.clear_chunk_num_when_rerun(doc.id)
 
-            if str(req["run"]) == TaskStatus.RUNNING.value:
-                doc = doc.to_dict()
-                DocumentService.run(tenant_id, doc, kb_table_num_map)
+                DocumentService.update_by_id(id, info)
+                if req.get("delete", False):
+                    TaskService.filter_delete([Task.doc_id == id])
+                    if settings.docStoreConn.index_exist(search.index_name(tenant_id), doc.kb_id):
+                        settings.docStoreConn.delete({"doc_id": id}, search.index_name(tenant_id), doc.kb_id)
 
-        return get_json_result(data=True)
+                if str(req["run"]) == TaskStatus.RUNNING.value:
+                    if req.get("apply_kb"):
+                        e, kb = KnowledgebaseService.get_by_id(doc.kb_id)
+                        if not e:
+                            raise LookupError("Can't find this dataset!")
+                        doc.parser_config["llm_id"] = kb.parser_config.get("llm_id")
+                        doc.parser_config["enable_metadata"] = kb.parser_config.get("enable_metadata", False)
+                        doc.parser_config["metadata"] = kb.parser_config.get("metadata", {})
+                        DocumentService.update_parser_config(doc.id, doc.parser_config)
+                    doc_dict = doc.to_dict()
+                    DocumentService.run(tenant_id, doc_dict, kb_table_num_map)
+
+            return get_json_result(data=True)
+
+        return await thread_pool_exec(_run_sync)
     except Exception as e:
         return server_error_response(e)
 
@@ -449,46 +644,51 @@ async def run():
 @login_required
 @validate_request("doc_id", "name")
 async def rename():
-    req = await request_json()
-    if not DocumentService.accessible(req["doc_id"], current_user.id):
-        return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
+    req = await get_request_json()
+    uid = current_user.id
     try:
-        e, doc = DocumentService.get_by_id(req["doc_id"])
-        if not e:
-            return get_data_error_result(message="Document not found!")
-        if pathlib.Path(req["name"].lower()).suffix != pathlib.Path(doc.name.lower()).suffix:
-            return get_json_result(data=False, message="The extension of file can't be changed", code=RetCode.ARGUMENT_ERROR)
-        if len(req["name"].encode("utf-8")) > FILE_NAME_LEN_LIMIT:
-            return get_json_result(data=False, message=f"File name must be {FILE_NAME_LEN_LIMIT} bytes or less.", code=RetCode.ARGUMENT_ERROR)
+        def _rename_sync():
+            if not DocumentService.accessible(req["doc_id"], uid):
+                return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
 
-        for d in DocumentService.query(name=req["name"], kb_id=doc.kb_id):
-            if d.name == req["name"]:
-                return get_data_error_result(message="Duplicated document name in the same knowledgebase.")
+            e, doc = DocumentService.get_by_id(req["doc_id"])
+            if not e:
+                return get_data_error_result(message="Document not found!")
+            if pathlib.Path(req["name"].lower()).suffix != pathlib.Path(doc.name.lower()).suffix:
+                return get_json_result(data=False, message="The extension of file can't be changed", code=RetCode.ARGUMENT_ERROR)
+            if len(req["name"].encode("utf-8")) > FILE_NAME_LEN_LIMIT:
+                return get_json_result(data=False, message=f"File name must be {FILE_NAME_LEN_LIMIT} bytes or less.", code=RetCode.ARGUMENT_ERROR)
 
-        if not DocumentService.update_by_id(req["doc_id"], {"name": req["name"]}):
-            return get_data_error_result(message="Database error (Document rename)!")
+            for d in DocumentService.query(name=req["name"], kb_id=doc.kb_id):
+                if d.name == req["name"]:
+                    return get_data_error_result(message="Duplicated document name in the same dataset.")
 
-        informs = File2DocumentService.get_by_document_id(req["doc_id"])
-        if informs:
-            e, file = FileService.get_by_id(informs[0].file_id)
-            FileService.update_by_id(file.id, {"name": req["name"]})
+            if not DocumentService.update_by_id(req["doc_id"], {"name": req["name"]}):
+                return get_data_error_result(message="Database error (Document rename)!")
 
-        tenant_id = DocumentService.get_tenant_id(req["doc_id"])
-        title_tks = rag_tokenizer.tokenize(req["name"])
-        es_body = {
-            "docnm_kwd": req["name"],
-            "title_tks": title_tks,
-            "title_sm_tks": rag_tokenizer.fine_grained_tokenize(title_tks),
-        }
-        if settings.docStoreConn.indexExist(search.index_name(tenant_id), doc.kb_id):
-            settings.docStoreConn.update(
-                {"doc_id": req["doc_id"]},
-                es_body,
-                search.index_name(tenant_id),
-                doc.kb_id,
-            )
+            informs = File2DocumentService.get_by_document_id(req["doc_id"])
+            if informs:
+                e, file = FileService.get_by_id(informs[0].file_id)
+                FileService.update_by_id(file.id, {"name": req["name"]})
+
+            tenant_id = DocumentService.get_tenant_id(req["doc_id"])
+            title_tks = rag_tokenizer.tokenize(req["name"])
+            es_body = {
+                "docnm_kwd": req["name"],
+                "title_tks": title_tks,
+                "title_sm_tks": rag_tokenizer.fine_grained_tokenize(title_tks),
+            }
+            if settings.docStoreConn.index_exist(search.index_name(tenant_id), doc.kb_id):
+                settings.docStoreConn.update(
+                    {"doc_id": req["doc_id"]},
+                    es_body,
+                    search.index_name(tenant_id),
+                    doc.kb_id,
+                )
+            return get_json_result(data=True)
+
+        return await thread_pool_exec(_rename_sync)
 
-        return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)
 
@@ -502,7 +702,8 @@ async def get(doc_id):
             return get_data_error_result(message="Document not found!")
 
         b, n = File2DocumentService.get_storage_address(doc_id=doc_id)
-        response = await make_response(settings.STORAGE_IMPL.get(b, n))
+        data = await thread_pool_exec(settings.STORAGE_IMPL.get, b, n)
+        response = await make_response(data)
 
         ext = re.search(r"\.([^.]+)$", doc.name.lower())
         ext = ext.group(1) if ext else None
@@ -523,8 +724,7 @@ async def get(doc_id):
 async def download_attachment(attachment_id):
     try:
         ext = request.args.get("ext", "markdown")
-        data = settings.STORAGE_IMPL.get(current_user.id, attachment_id)
-        # data = settings.STORAGE_IMPL.get("eb500d50bb0411f0907561d2782adda5", attachment_id)
+        data = await thread_pool_exec(settings.STORAGE_IMPL.get, current_user.id, attachment_id)
         response = await make_response(data)
         response.headers.set("Content-Type", CONTENT_TYPE_MAP.get(ext, f"application/{ext}"))
 
@@ -539,7 +739,7 @@ async def download_attachment(attachment_id):
 @validate_request("doc_id")
 async def change_parser():
 
-    req = await request_json()
+    req = await get_request_json()
     if not DocumentService.accessible(req["doc_id"], current_user.id):
         return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
 
@@ -559,7 +759,8 @@ async def change_parser():
             tenant_id = DocumentService.get_tenant_id(req["doc_id"])
             if not tenant_id:
                 return get_data_error_result(message="Tenant not found!")
-            if settings.docStoreConn.indexExist(search.index_name(tenant_id), doc.kb_id):
+            DocumentService.delete_chunk_images(doc, tenant_id)
+            if settings.docStoreConn.index_exist(search.index_name(tenant_id), doc.kb_id):
                 settings.docStoreConn.delete({"doc_id": doc.id}, search.index_name(tenant_id), doc.kb_id)
         return None
 
@@ -596,7 +797,8 @@ async def get_image(image_id):
         if len(arr) != 2:
             return get_data_error_result(message="Image not found.")
         bkt, nm = image_id.split("-")
-        response = await make_response(settings.STORAGE_IMPL.get(bkt, nm))
+        data = await thread_pool_exec(settings.STORAGE_IMPL.get, bkt, nm)
+        response = await make_response(data)
         response.headers.set("Content-Type", "image/JPEG")
         return response
     except Exception as e:
@@ -607,7 +809,7 @@ async def get_image(image_id):
 @login_required
 @validate_request("conversation_id")
 async def upload_and_parse():
-    files = await request.file
+    files = await request.files
     if "file" not in files:
         return get_json_result(data=False, message="No file part!", code=RetCode.ARGUMENT_ERROR)
 
@@ -624,7 +826,8 @@ async def upload_and_parse():
 @manager.route("/parse", methods=["POST"])  # noqa: F821
 @login_required
 async def parse():
-    url = await request.json.get("url") if await request.json else ""
+    req = await get_request_json()
+    url = req.get("url", "")
     if url:
         if not is_valid_url(url):
             return get_json_result(data=False, message="The URL format is invalid", code=RetCode.ARGUMENT_ERROR)
@@ -679,7 +882,7 @@ async def parse():
 @login_required
 @validate_request("doc_id", "meta")
 async def set_meta():
-    req = await request_json()
+    req = await get_request_json()
     if not DocumentService.accessible(req["doc_id"], current_user.id):
         return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
     try:
@@ -687,7 +890,10 @@ async def set_meta():
         if not isinstance(meta, dict):
             return get_json_result(data=False, message="Only dictionary type supported.", code=RetCode.ARGUMENT_ERROR)
         for k, v in meta.items():
-            if not isinstance(v, str) and not isinstance(v, int) and not isinstance(v, float):
+            if isinstance(v, list):
+                if not all(isinstance(i, (str, int, float)) for i in v):
+                    return get_json_result(data=False, message=f"The type is not supported in list: {v}", code=RetCode.ARGUMENT_ERROR)
+            elif not isinstance(v, (str, int, float)):
                 return get_json_result(data=False, message=f"The type is not supported: {v}", code=RetCode.ARGUMENT_ERROR)
     except Exception as e:
         return get_json_result(data=False, message=f"Json syntax error: {e}", code=RetCode.ARGUMENT_ERROR)
@@ -705,3 +911,13 @@ async def set_meta():
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)
+
+
+@manager.route("/upload_info", methods=["POST"])  # noqa: F821
+async def upload_info():
+    files = await request.files
+    file = files['file'] if files and files.get("file") else None
+    try:
+        return get_json_result(data=FileService.upload_info(current_user.id, file, request.args.get("url")))
+    except Exception as e:
+        return  server_error_response(e)

api/apps/evaluation_app.py

@@ -0,0 +1,479 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+
+"""
+RAG Evaluation API Endpoints
+
+Provides REST API for RAG evaluation functionality including:
+- Dataset management
+- Test case management
+- Evaluation execution
+- Results retrieval
+- Configuration recommendations
+"""
+
+from quart import request
+from api.apps import login_required, current_user
+from api.db.services.evaluation_service import EvaluationService
+from api.utils.api_utils import (
+    get_data_error_result,
+    get_json_result,
+    get_request_json,
+    server_error_response,
+    validate_request
+)
+from common.constants import RetCode
+
+
+# ==================== Dataset Management ====================
+
+@manager.route('/dataset/create', methods=['POST'])  # noqa: F821
+@login_required
+@validate_request("name", "kb_ids")
+async def create_dataset():
+    """
+    Create a new evaluation dataset.
+    
+    Request body:
+    {
+        "name": "Dataset name",
+        "description": "Optional description",
+        "kb_ids": ["kb_id1", "kb_id2"]
+    }
+    """
+    try:
+        req = await get_request_json()
+        name = req.get("name", "").strip()
+        description = req.get("description", "")
+        kb_ids = req.get("kb_ids", [])
+        
+        if not name:
+            return get_data_error_result(message="Dataset name cannot be empty")
+        
+        if not kb_ids or not isinstance(kb_ids, list):
+            return get_data_error_result(message="kb_ids must be a non-empty list")
+        
+        success, result = EvaluationService.create_dataset(
+            name=name,
+            description=description,
+            kb_ids=kb_ids,
+            tenant_id=current_user.id,
+            user_id=current_user.id
+        )
+        
+        if not success:
+            return get_data_error_result(message=result)
+        
+        return get_json_result(data={"dataset_id": result})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/dataset/list', methods=['GET'])  # noqa: F821
+@login_required
+async def list_datasets():
+    """
+    List evaluation datasets for current tenant.
+    
+    Query params:
+    - page: Page number (default: 1)
+    - page_size: Items per page (default: 20)
+    """
+    try:
+        page = int(request.args.get("page", 1))
+        page_size = int(request.args.get("page_size", 20))
+        
+        result = EvaluationService.list_datasets(
+            tenant_id=current_user.id,
+            user_id=current_user.id,
+            page=page,
+            page_size=page_size
+        )
+        
+        return get_json_result(data=result)
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/dataset/<dataset_id>', methods=['GET'])  # noqa: F821
+@login_required
+async def get_dataset(dataset_id):
+    """Get dataset details by ID"""
+    try:
+        dataset = EvaluationService.get_dataset(dataset_id)
+        if not dataset:
+            return get_data_error_result(
+                message="Dataset not found",
+                code=RetCode.DATA_ERROR
+            )
+        
+        return get_json_result(data=dataset)
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/dataset/<dataset_id>', methods=['PUT'])  # noqa: F821
+@login_required
+async def update_dataset(dataset_id):
+    """
+    Update dataset.
+    
+    Request body:
+    {
+        "name": "New name",
+        "description": "New description",
+        "kb_ids": ["kb_id1", "kb_id2"]
+    }
+    """
+    try:
+        req = await get_request_json()
+        
+        # Remove fields that shouldn't be updated
+        req.pop("id", None)
+        req.pop("tenant_id", None)
+        req.pop("created_by", None)
+        req.pop("create_time", None)
+        
+        success = EvaluationService.update_dataset(dataset_id, **req)
+        
+        if not success:
+            return get_data_error_result(message="Failed to update dataset")
+        
+        return get_json_result(data={"dataset_id": dataset_id})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/dataset/<dataset_id>', methods=['DELETE'])  # noqa: F821
+@login_required
+async def delete_dataset(dataset_id):
+    """Delete dataset (soft delete)"""
+    try:
+        success = EvaluationService.delete_dataset(dataset_id)
+        
+        if not success:
+            return get_data_error_result(message="Failed to delete dataset")
+        
+        return get_json_result(data={"dataset_id": dataset_id})
+    except Exception as e:
+        return server_error_response(e)
+
+
+# ==================== Test Case Management ====================
+
+@manager.route('/dataset/<dataset_id>/case/add', methods=['POST'])  # noqa: F821
+@login_required
+@validate_request("question")
+async def add_test_case(dataset_id):
+    """
+    Add a test case to a dataset.
+    
+    Request body:
+    {
+        "question": "Test question",
+        "reference_answer": "Optional ground truth answer",
+        "relevant_doc_ids": ["doc_id1", "doc_id2"],
+        "relevant_chunk_ids": ["chunk_id1", "chunk_id2"],
+        "metadata": {"key": "value"}
+    }
+    """
+    try:
+        req = await get_request_json()
+        question = req.get("question", "").strip()
+        
+        if not question:
+            return get_data_error_result(message="Question cannot be empty")
+        
+        success, result = EvaluationService.add_test_case(
+            dataset_id=dataset_id,
+            question=question,
+            reference_answer=req.get("reference_answer"),
+            relevant_doc_ids=req.get("relevant_doc_ids"),
+            relevant_chunk_ids=req.get("relevant_chunk_ids"),
+            metadata=req.get("metadata")
+        )
+        
+        if not success:
+            return get_data_error_result(message=result)
+        
+        return get_json_result(data={"case_id": result})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/dataset/<dataset_id>/case/import', methods=['POST'])  # noqa: F821
+@login_required
+@validate_request("cases")
+async def import_test_cases(dataset_id):
+    """
+    Bulk import test cases.
+    
+    Request body:
+    {
+        "cases": [
+            {
+                "question": "Question 1",
+                "reference_answer": "Answer 1",
+                ...
+            },
+            {
+                "question": "Question 2",
+                ...
+            }
+        ]
+    }
+    """
+    try:
+        req = await get_request_json()
+        cases = req.get("cases", [])
+        
+        if not cases or not isinstance(cases, list):
+            return get_data_error_result(message="cases must be a non-empty list")
+        
+        success_count, failure_count = EvaluationService.import_test_cases(
+            dataset_id=dataset_id,
+            cases=cases
+        )
+        
+        return get_json_result(data={
+            "success_count": success_count,
+            "failure_count": failure_count,
+            "total": len(cases)
+        })
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/dataset/<dataset_id>/cases', methods=['GET'])  # noqa: F821
+@login_required
+async def get_test_cases(dataset_id):
+    """Get all test cases for a dataset"""
+    try:
+        cases = EvaluationService.get_test_cases(dataset_id)
+        return get_json_result(data={"cases": cases, "total": len(cases)})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/case/<case_id>', methods=['DELETE'])  # noqa: F821
+@login_required
+async def delete_test_case(case_id):
+    """Delete a test case"""
+    try:
+        success = EvaluationService.delete_test_case(case_id)
+        
+        if not success:
+            return get_data_error_result(message="Failed to delete test case")
+        
+        return get_json_result(data={"case_id": case_id})
+    except Exception as e:
+        return server_error_response(e)
+
+
+# ==================== Evaluation Execution ====================
+
+@manager.route('/run/start', methods=['POST'])  # noqa: F821
+@login_required
+@validate_request("dataset_id", "dialog_id")
+async def start_evaluation():
+    """
+    Start an evaluation run.
+    
+    Request body:
+    {
+        "dataset_id": "dataset_id",
+        "dialog_id": "dialog_id",
+        "name": "Optional run name"
+    }
+    """
+    try:
+        req = await get_request_json()
+        dataset_id = req.get("dataset_id")
+        dialog_id = req.get("dialog_id")
+        name = req.get("name")
+        
+        success, result = EvaluationService.start_evaluation(
+            dataset_id=dataset_id,
+            dialog_id=dialog_id,
+            user_id=current_user.id,
+            name=name
+        )
+        
+        if not success:
+            return get_data_error_result(message=result)
+        
+        return get_json_result(data={"run_id": result})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/run/<run_id>', methods=['GET'])  # noqa: F821
+@login_required
+async def get_evaluation_run(run_id):
+    """Get evaluation run details"""
+    try:
+        result = EvaluationService.get_run_results(run_id)
+        
+        if not result:
+            return get_data_error_result(
+                message="Evaluation run not found",
+                code=RetCode.DATA_ERROR
+            )
+        
+        return get_json_result(data=result)
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/run/<run_id>/results', methods=['GET'])  # noqa: F821
+@login_required
+async def get_run_results(run_id):
+    """Get detailed results for an evaluation run"""
+    try:
+        result = EvaluationService.get_run_results(run_id)
+        
+        if not result:
+            return get_data_error_result(
+                message="Evaluation run not found",
+                code=RetCode.DATA_ERROR
+            )
+        
+        return get_json_result(data=result)
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/run/list', methods=['GET'])  # noqa: F821
+@login_required
+async def list_evaluation_runs():
+    """
+    List evaluation runs.
+    
+    Query params:
+    - dataset_id: Filter by dataset (optional)
+    - dialog_id: Filter by dialog (optional)
+    - page: Page number (default: 1)
+    - page_size: Items per page (default: 20)
+    """
+    try:
+        # TODO: Implement list_runs in EvaluationService
+        return get_json_result(data={"runs": [], "total": 0})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/run/<run_id>', methods=['DELETE'])  # noqa: F821
+@login_required
+async def delete_evaluation_run(run_id):
+    """Delete an evaluation run"""
+    try:
+        # TODO: Implement delete_run in EvaluationService
+        return get_json_result(data={"run_id": run_id})
+    except Exception as e:
+        return server_error_response(e)
+
+
+# ==================== Analysis & Recommendations ====================
+
+@manager.route('/run/<run_id>/recommendations', methods=['GET'])  # noqa: F821
+@login_required
+async def get_recommendations(run_id):
+    """Get configuration recommendations based on evaluation results"""
+    try:
+        recommendations = EvaluationService.get_recommendations(run_id)
+        return get_json_result(data={"recommendations": recommendations})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/compare', methods=['POST'])  # noqa: F821
+@login_required
+@validate_request("run_ids")
+async def compare_runs():
+    """
+    Compare multiple evaluation runs.
+    
+    Request body:
+    {
+        "run_ids": ["run_id1", "run_id2", "run_id3"]
+    }
+    """
+    try:
+        req = await get_request_json()
+        run_ids = req.get("run_ids", [])
+        
+        if not run_ids or not isinstance(run_ids, list) or len(run_ids) < 2:
+            return get_data_error_result(
+                message="run_ids must be a list with at least 2 run IDs"
+            )
+        
+        # TODO: Implement compare_runs in EvaluationService
+        return get_json_result(data={"comparison": {}})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/run/<run_id>/export', methods=['GET'])  # noqa: F821
+@login_required
+async def export_results(run_id):
+    """Export evaluation results as JSON/CSV"""
+    try:
+        # format_type = request.args.get("format", "json")  # TODO: Use for CSV export
+        
+        result = EvaluationService.get_run_results(run_id)
+        
+        if not result:
+            return get_data_error_result(
+                message="Evaluation run not found",
+                code=RetCode.DATA_ERROR
+            )
+        
+        # TODO: Implement CSV export
+        return get_json_result(data=result)
+    except Exception as e:
+        return server_error_response(e)
+
+
+# ==================== Real-time Evaluation ====================
+
+@manager.route('/evaluate_single', methods=['POST'])  # noqa: F821
+@login_required
+@validate_request("question", "dialog_id")
+async def evaluate_single():
+    """
+    Evaluate a single question-answer pair in real-time.
+    
+    Request body:
+    {
+        "question": "Test question",
+        "dialog_id": "dialog_id",
+        "reference_answer": "Optional ground truth",
+        "relevant_chunk_ids": ["chunk_id1", "chunk_id2"]
+    }
+    """
+    try:
+        # req = await get_request_json()  # TODO: Use for single evaluation implementation
+        
+        # TODO: Implement single evaluation
+        # This would execute the RAG pipeline and return metrics immediately
+        
+        return get_json_result(data={
+            "answer": "",
+            "metrics": {},
+            "retrieved_chunks": []
+        })
+    except Exception as e:
+        return server_error_response(e)

api/apps/file2document_app.py

@@ -19,22 +19,20 @@ from pathlib import Path
 from api.db.services.file2document_service import File2DocumentService
 from api.db.services.file_service import FileService
 
-from quart import request
 from api.apps import login_required, current_user
 from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
+from api.utils.api_utils import get_data_error_result, get_json_result, get_request_json, server_error_response, validate_request
 from common.misc_utils import get_uuid
 from common.constants import RetCode
 from api.db import FileType
 from api.db.services.document_service import DocumentService
-from api.utils.api_utils import get_json_result
 
 
 @manager.route('/convert', methods=['POST'])  # noqa: F821
 @login_required
 @validate_request("file_ids", "kb_ids")
 async def convert():
-    req = await request.json
+    req = await get_request_json()
     kb_ids = req["kb_ids"]
     file_ids = req["file_ids"]
     file2documents = []
@@ -70,7 +68,7 @@ async def convert():
                     e, kb = KnowledgebaseService.get_by_id(kb_id)
                     if not e:
                         return get_data_error_result(
-                            message="Can't find this knowledgebase!")
+                            message="Can't find this dataset!")
                     e, file = FileService.get_by_id(id)
                     if not e:
                         return get_data_error_result(
@@ -79,7 +77,8 @@ async def convert():
                     doc = DocumentService.insert({
                         "id": get_uuid(),
                         "kb_id": kb.id,
-                        "parser_id": FileService.get_parser(file.type, file.name, kb.parser_id),
+                        "parser_id": kb.parser_id,
+                        "pipeline_id": kb.pipeline_id,
                         "parser_config": kb.parser_config,
                         "created_by": current_user.id,
                         "type": file.type,
@@ -104,7 +103,7 @@ async def convert():
 @login_required
 @validate_request("file_ids")
 async def rm():
-    req = await request.json
+    req = await get_request_json()
     file_ids = req["file_ids"]
     if not file_ids:
         return get_json_result(

api/apps/file_app.py

@@ -24,17 +24,16 @@ from api.common.check_team_permission import check_file_team_permission
 from api.db.services.document_service import DocumentService
 from api.db.services.file2document_service import File2DocumentService
 from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
-from common.misc_utils import get_uuid
+from common.misc_utils import get_uuid, thread_pool_exec
 from common.constants import RetCode, FileSource
 from api.db import FileType
 from api.db.services import duplicate_name
 from api.db.services.file_service import FileService
-from api.utils.api_utils import get_json_result
+from api.utils.api_utils import get_json_result, get_request_json
 from api.utils.file_utils import filename_type
 from api.utils.web_utils import CONTENT_TYPE_MAP
 from common import settings
 
-
 @manager.route('/upload', methods=['POST'])  # noqa: F821
 @login_required
 # @validate_request("parent_id")
@@ -61,9 +60,10 @@ async def upload():
         e, pf_folder = FileService.get_by_id(pf_id)
         if not e:
             return get_data_error_result( message="Can't find this folder!")
-        for file_obj in file_objs:
+
+        async def _handle_single_file(file_obj):
             MAX_FILE_NUM_PER_USER: int = int(os.environ.get('MAX_FILE_NUM_PER_USER', 0))
-            if 0 < MAX_FILE_NUM_PER_USER <= DocumentService.get_doc_count(current_user.id):
+            if 0 < MAX_FILE_NUM_PER_USER <= await thread_pool_exec(DocumentService.get_doc_count, current_user.id):
                 return get_data_error_result( message="Exceed the maximum file number of a free user!")
 
             # split file name path
@@ -75,35 +75,36 @@ async def upload():
             file_len = len(file_obj_names)
 
             # get folder
-            file_id_list = FileService.get_id_list_by_id(pf_id, file_obj_names, 1, [pf_id])
+            file_id_list = await thread_pool_exec(FileService.get_id_list_by_id, pf_id, file_obj_names, 1, [pf_id])
             len_id_list = len(file_id_list)
 
             # create folder
             if file_len != len_id_list:
-                e, file = FileService.get_by_id(file_id_list[len_id_list - 1])
+                e, file = await thread_pool_exec(FileService.get_by_id, file_id_list[len_id_list - 1])
                 if not e:
                     return get_data_error_result(message="Folder not found!")
-                last_folder = FileService.create_folder(file, file_id_list[len_id_list - 1], file_obj_names,
+                last_folder = await thread_pool_exec(FileService.create_folder, file, file_id_list[len_id_list - 1], file_obj_names,
                                                         len_id_list)
             else:
-                e, file = FileService.get_by_id(file_id_list[len_id_list - 2])
+                e, file = await thread_pool_exec(FileService.get_by_id, file_id_list[len_id_list - 2])
                 if not e:
                     return get_data_error_result(message="Folder not found!")
-                last_folder = FileService.create_folder(file, file_id_list[len_id_list - 2], file_obj_names,
+                last_folder = await thread_pool_exec(FileService.create_folder, file, file_id_list[len_id_list - 2], file_obj_names,
                                                         len_id_list)
 
             # file type
             filetype = filename_type(file_obj_names[file_len - 1])
             location = file_obj_names[file_len - 1]
-            while settings.STORAGE_IMPL.obj_exist(last_folder.id, location):
+            while await thread_pool_exec(settings.STORAGE_IMPL.obj_exist, last_folder.id, location):
                 location += "_"
-            blob = file_obj.read()
-            filename = duplicate_name(
+            blob = await thread_pool_exec(file_obj.read)
+            filename = await thread_pool_exec(
+                duplicate_name,
                 FileService.query,
                 name=file_obj_names[file_len - 1],
                 parent_id=last_folder.id)
-            settings.STORAGE_IMPL.put(last_folder.id, location, blob)
-            file = {
+            await thread_pool_exec(settings.STORAGE_IMPL.put, last_folder.id, location, blob)
+            file_data = {
                 "id": get_uuid(),
                 "parent_id": last_folder.id,
                 "tenant_id": current_user.id,
@@ -113,8 +114,13 @@ async def upload():
                 "location": location,
                 "size": len(blob),
             }
-            file = FileService.insert(file)
-            file_res.append(file.to_json())
+            inserted = await thread_pool_exec(FileService.insert, file_data)
+            return inserted.to_json()
+
+        for file_obj in file_objs:
+            res = await _handle_single_file(file_obj)
+            file_res.append(res)
+
         return get_json_result(data=file_res)
     except Exception as e:
         return server_error_response(e)
@@ -124,7 +130,7 @@ async def upload():
 @login_required
 @validate_request("name")
 async def create():
-    req = await request.json
+    req = await get_request_json()
     pf_id = req.get("parent_id")
     input_file_type = req.get("type")
     if not pf_id:
@@ -239,58 +245,62 @@ def get_all_parent_folders():
 @login_required
 @validate_request("file_ids")
 async def rm():
-    req = await request.json
+    req = await get_request_json()
     file_ids = req["file_ids"]
-
-    def _delete_single_file(file):
-        try:
-            if file.location:
-                settings.STORAGE_IMPL.rm(file.parent_id, file.location)
-        except Exception as e:
-            logging.exception(f"Fail to remove object: {file.parent_id}/{file.location}, error: {e}")
-
-        informs = File2DocumentService.get_by_file_id(file.id)
-        for inform in informs:
-            doc_id = inform.document_id
-            e, doc = DocumentService.get_by_id(doc_id)
-            if e and doc:
-                tenant_id = DocumentService.get_tenant_id(doc_id)
-                if tenant_id:
-                    DocumentService.remove_document(doc, tenant_id)
-            File2DocumentService.delete_by_file_id(file.id)
-
-        FileService.delete(file)
-
-    def _delete_folder_recursive(folder, tenant_id):
-        sub_files = FileService.list_all_files_by_parent_id(folder.id)
-        for sub_file in sub_files:
-            if sub_file.type == FileType.FOLDER.value:
-                _delete_folder_recursive(sub_file, tenant_id)
-            else:
-                _delete_single_file(sub_file)
-
-        FileService.delete(folder)
+    uid = current_user.id
 
     try:
-        for file_id in file_ids:
-            e, file = FileService.get_by_id(file_id)
-            if not e or not file:
-                return get_data_error_result(message="File or Folder not found!")
-            if not file.tenant_id:
-                return get_data_error_result(message="Tenant not found!")
-            if not check_file_team_permission(file, current_user.id):
-                return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
+        def _delete_single_file(file):
+            try:
+                if file.location:
+                    settings.STORAGE_IMPL.rm(file.parent_id, file.location)
+            except Exception as e:
+                logging.exception(f"Fail to remove object: {file.parent_id}/{file.location}, error: {e}")
 
-            if file.source_type == FileSource.KNOWLEDGEBASE:
-                continue
+            informs = File2DocumentService.get_by_file_id(file.id)
+            for inform in informs:
+                doc_id = inform.document_id
+                e, doc = DocumentService.get_by_id(doc_id)
+                if e and doc:
+                    tenant_id = DocumentService.get_tenant_id(doc_id)
+                    if tenant_id:
+                        DocumentService.remove_document(doc, tenant_id)
+                File2DocumentService.delete_by_file_id(file.id)
 
-            if file.type == FileType.FOLDER.value:
-                _delete_folder_recursive(file, current_user.id)
-                continue
+            FileService.delete(file)
 
-            _delete_single_file(file)
+        def _delete_folder_recursive(folder, tenant_id):
+            sub_files = FileService.list_all_files_by_parent_id(folder.id)
+            for sub_file in sub_files:
+                if sub_file.type == FileType.FOLDER.value:
+                    _delete_folder_recursive(sub_file, tenant_id)
+                else:
+                    _delete_single_file(sub_file)
 
-        return get_json_result(data=True)
+            FileService.delete(folder)
+
+        def _rm_sync():
+            for file_id in file_ids:
+                e, file = FileService.get_by_id(file_id)
+                if not e or not file:
+                    return get_data_error_result(message="File or Folder not found!")
+                if not file.tenant_id:
+                    return get_data_error_result(message="Tenant not found!")
+                if not check_file_team_permission(file, uid):
+                    return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)
+
+                if file.source_type == FileSource.KNOWLEDGEBASE:
+                    continue
+
+                if file.type == FileType.FOLDER.value:
+                    _delete_folder_recursive(file, uid)
+                    continue
+
+                _delete_single_file(file)
+
+            return get_json_result(data=True)
+
+        return await thread_pool_exec(_rm_sync)
 
     except Exception as e:
         return server_error_response(e)
@@ -300,7 +310,7 @@ async def rm():
 @login_required
 @validate_request("file_id", "name")
 async def rename():
-    req = await request.json
+    req = await get_request_json()
     try:
         e, file = FileService.get_by_id(req["file_id"])
         if not e:
@@ -346,10 +356,10 @@ async def get(file_id):
         if not check_file_team_permission(file, current_user.id):
             return get_json_result(data=False, message='No authorization.', code=RetCode.AUTHENTICATION_ERROR)
 
-        blob = settings.STORAGE_IMPL.get(file.parent_id, file.location)
+        blob = await thread_pool_exec(settings.STORAGE_IMPL.get, file.parent_id, file.location)
         if not blob:
             b, n = File2DocumentService.get_storage_address(file_id=file_id)
-            blob = settings.STORAGE_IMPL.get(b, n)
+            blob = await thread_pool_exec(settings.STORAGE_IMPL.get, b, n)
 
         response = await make_response(blob)
         ext = re.search(r"\.([^.]+)$", file.name.lower())
@@ -369,7 +379,7 @@ async def get(file_id):
 @login_required
 @validate_request("src_file_ids", "dest_file_id")
 async def move():
-    req = await request.json
+    req = await get_request_json()
     try:
         file_ids = req["src_file_ids"]
         dest_parent_id = req["dest_file_id"]
@@ -444,10 +454,12 @@ async def move():
                 },
             )
 
-        for file in files:
-            _move_entry_recursive(file, dest_folder)
+        def _move_sync():
+            for file in files:
+                _move_entry_recursive(file, dest_folder)
+            return get_json_result(data=True)
 
-        return get_json_result(data=True)
+        return await thread_pool_exec(_move_sync)
 
     except Exception as e:
         return server_error_response(e)

api/apps/kb_app.py

@@ -18,6 +18,7 @@ import logging
 import random
 import re
 
+from common.metadata_utils import turn2jsonschema
 from quart import request
 import numpy as np
 
@@ -29,8 +30,15 @@ from api.db.services.file_service import FileService
 from api.db.services.pipeline_operation_log_service import PipelineOperationLogService
 from api.db.services.task_service import TaskService, GRAPH_RAPTOR_FAKE_DOC_ID
 from api.db.services.user_service import TenantService, UserTenantService
-from api.utils.api_utils import get_error_data_result, server_error_response, get_data_error_result, validate_request, not_allowed_parameters, \
-    request_json
+from api.utils.api_utils import (
+    get_error_data_result,
+    server_error_response,
+    get_data_error_result,
+    validate_request,
+    not_allowed_parameters,
+    get_request_json,
+)
+from common.misc_utils import thread_pool_exec
 from api.db import VALID_FILE_TYPES
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.db_models import File
@@ -38,17 +46,16 @@ from api.utils.api_utils import get_json_result
 from rag.nlp import search
 from api.constants import DATASET_NAME_LIMIT
 from rag.utils.redis_conn import REDIS_CONN
-from rag.utils.doc_store_conn import OrderByExpr
 from common.constants import RetCode, PipelineTaskType, StatusEnum, VALID_TASK_STATUS, FileSource, LLMType, PAGERANK_FLD
 from common import settings
+from common.doc_store.doc_store_base import OrderByExpr
 from api.apps import login_required, current_user
 
-
 @manager.route('/create', methods=['post'])  # noqa: F821
 @login_required
 @validate_request("name")
 async def create():
-    req = await request_json()
+    req = await get_request_json()
     e, res = KnowledgebaseService.create_with_name(
         name = req.pop("name", None),
         tenant_id = current_user.id,
@@ -72,7 +79,7 @@ async def create():
 @validate_request("kb_id", "name", "description", "parser_id")
 @not_allowed_parameters("id", "tenant_id", "created_by", "create_time", "update_time", "create_date", "update_date", "created_by")
 async def update():
-    req = await request_json()
+    req = await get_request_json()
     if not isinstance(req["name"], str):
         return get_data_error_result(message="Dataset name must be string.")
     if req["name"].strip() == "":
@@ -81,6 +88,20 @@ async def update():
         return get_data_error_result(
             message=f"Dataset name length is {len(req['name'])} which is large than {DATASET_NAME_LIMIT}")
     req["name"] = req["name"].strip()
+    if settings.DOC_ENGINE_INFINITY:
+        parser_id = req.get("parser_id")
+        if isinstance(parser_id, str) and parser_id.lower() == "tag":
+            return get_json_result(
+                code=RetCode.OPERATING_ERROR,
+                message="The chunking method Tag has not been supported by Infinity yet.",
+                data=False,
+            )
+        if "pagerank" in req and req["pagerank"] > 0:
+            return get_json_result(
+                code=RetCode.DATA_ERROR,
+                message="'pagerank' can only be set when doc_engine is elasticsearch",
+                data=False,
+            )
 
     if not KnowledgebaseService.accessible4deletion(req["kb_id"], current_user.id):
         return get_json_result(
@@ -92,19 +113,32 @@ async def update():
         if not KnowledgebaseService.query(
                 created_by=current_user.id, id=req["kb_id"]):
             return get_json_result(
-                data=False, message='Only owner of knowledgebase authorized for this operation.',
+                data=False, message='Only owner of dataset authorized for this operation.',
                 code=RetCode.OPERATING_ERROR)
 
         e, kb = KnowledgebaseService.get_by_id(req["kb_id"])
+
+        # Rename folder in FileService
+        if e and req["name"].lower() != kb.name.lower():
+            FileService.filter_update(
+                [
+                    File.tenant_id == kb.tenant_id,
+                    File.source_type == FileSource.KNOWLEDGEBASE,
+                    File.type == "folder",
+                    File.name == kb.name,
+                ],
+                {"name": req["name"]},
+            )
+
         if not e:
             return get_data_error_result(
-                message="Can't find this knowledgebase!")
+                message="Can't find this dataset!")
 
         if req["name"].lower() != kb.name.lower() \
                 and len(
             KnowledgebaseService.query(name=req["name"], tenant_id=current_user.id, status=StatusEnum.VALID.value)) >= 1:
             return get_data_error_result(
-                message="Duplicated knowledgebase name.")
+                message="Duplicated dataset name.")
 
         del req["kb_id"]
         connectors = []
@@ -116,12 +150,22 @@ async def update():
 
         if kb.pagerank != req.get("pagerank", 0):
             if req.get("pagerank", 0) > 0:
-                settings.docStoreConn.update({"kb_id": kb.id}, {PAGERANK_FLD: req["pagerank"]},
-                                         search.index_name(kb.tenant_id), kb.id)
+                await thread_pool_exec(
+                    settings.docStoreConn.update,
+                    {"kb_id": kb.id},
+                    {PAGERANK_FLD: req["pagerank"]},
+                    search.index_name(kb.tenant_id),
+                    kb.id,
+                )
             else:
                 # Elasticsearch requires PAGERANK_FLD be non-zero!
-                settings.docStoreConn.update({"exists": PAGERANK_FLD}, {"remove": PAGERANK_FLD},
-                                         search.index_name(kb.tenant_id), kb.id)
+                await thread_pool_exec(
+                    settings.docStoreConn.update,
+                    {"exists": PAGERANK_FLD},
+                    {"remove": PAGERANK_FLD},
+                    search.index_name(kb.tenant_id),
+                    kb.id,
+                )
 
         e, kb = KnowledgebaseService.get_by_id(kb.id)
         if not e:
@@ -139,6 +183,22 @@ async def update():
         return server_error_response(e)
 
 
+@manager.route('/update_metadata_setting', methods=['post'])  # noqa: F821
+@login_required
+@validate_request("kb_id", "metadata")
+async def update_metadata_setting():
+    req = await get_request_json()
+    e, kb = KnowledgebaseService.get_by_id(req["kb_id"])
+    if not e:
+        return get_data_error_result(
+            message="Database error (Knowledgebase rename)!")
+    kb = kb.to_dict()
+    kb["parser_config"]["metadata"] = req["metadata"]
+    kb["parser_config"]["enable_metadata"] = req.get("enable_metadata", True)
+    KnowledgebaseService.update_by_id(kb["id"], kb)
+    return get_json_result(data=kb)
+
+
 @manager.route('/detail', methods=['GET'])  # noqa: F821
 @login_required
 def detail():
@@ -151,14 +211,16 @@ def detail():
                 break
         else:
             return get_json_result(
-                data=False, message='Only owner of knowledgebase authorized for this operation.',
+                data=False, message='Only owner of dataset authorized for this operation.',
                 code=RetCode.OPERATING_ERROR)
         kb = KnowledgebaseService.get_detail(kb_id)
         if not kb:
             return get_data_error_result(
-                message="Can't find this knowledgebase!")
+                message="Can't find this dataset!")
         kb["size"] = DocumentService.get_total_size_by_kb_id(kb_id=kb["id"],keywords="", run_status=[], types=[])
         kb["connectors"] = Connector2KbService.list_connectors(kb_id)
+        if kb["parser_config"].get("metadata"):
+            kb["parser_config"]["metadata"] = turn2jsonschema(kb["parser_config"]["metadata"])
 
         for key in ["graphrag_task_finish_at", "raptor_task_finish_at", "mindmap_task_finish_at"]:
             if finish_at := kb.get(key):
@@ -182,7 +244,7 @@ async def list_kbs():
     else:
         desc = True
 
-    req = await request_json()
+    req = await get_request_json()
     owner_ids = req.get("owner_ids", [])
     try:
         if not owner_ids:
@@ -209,8 +271,9 @@ async def list_kbs():
 @login_required
 @validate_request("kb_id")
 async def rm():
-    req = await request_json()
-    if not KnowledgebaseService.accessible4deletion(req["kb_id"], current_user.id):
+    req = await get_request_json()
+    uid = current_user.id
+    if not KnowledgebaseService.accessible4deletion(req["kb_id"], uid):
         return get_json_result(
             data=False,
             message='No authorization.',
@@ -218,31 +281,47 @@ async def rm():
         )
     try:
         kbs = KnowledgebaseService.query(
-            created_by=current_user.id, id=req["kb_id"])
+            created_by=uid, id=req["kb_id"])
         if not kbs:
             return get_json_result(
-                data=False, message='Only owner of knowledgebase authorized for this operation.',
+                data=False, message='Only owner of dataset authorized for this operation.',
                 code=RetCode.OPERATING_ERROR)
 
-        for doc in DocumentService.query(kb_id=req["kb_id"]):
-            if not DocumentService.remove_document(doc, kbs[0].tenant_id):
+        def _rm_sync():
+            for doc in DocumentService.query(kb_id=req["kb_id"]):
+                if not DocumentService.remove_document(doc, kbs[0].tenant_id):
+                    return get_data_error_result(
+                        message="Database error (Document removal)!")
+                f2d = File2DocumentService.get_by_document_id(doc.id)
+                if f2d:
+                    FileService.filter_delete([File.source_type == FileSource.KNOWLEDGEBASE, File.id == f2d[0].file_id])
+                File2DocumentService.delete_by_document_id(doc.id)
+            FileService.filter_delete(
+                [
+                    File.tenant_id == kbs[0].tenant_id,
+                    File.source_type == FileSource.KNOWLEDGEBASE,
+                    File.type == "folder",
+                    File.name == kbs[0].name,
+                ]
+            )
+            # Delete the table BEFORE deleting the database record
+            for kb in kbs:
+                try:
+                    settings.docStoreConn.delete({"kb_id": kb.id}, search.index_name(kb.tenant_id), kb.id)
+                    settings.docStoreConn.delete_idx(search.index_name(kb.tenant_id), kb.id)
+                    logging.info(f"Dropped index for dataset {kb.id}")
+                except Exception as e:
+                    logging.error(f"Failed to drop index for dataset {kb.id}: {e}")
+
+            if not KnowledgebaseService.delete_by_id(req["kb_id"]):
                 return get_data_error_result(
-                    message="Database error (Document removal)!")
-            f2d = File2DocumentService.get_by_document_id(doc.id)
-            if f2d:
-                FileService.filter_delete([File.source_type == FileSource.KNOWLEDGEBASE, File.id == f2d[0].file_id])
-            File2DocumentService.delete_by_document_id(doc.id)
-        FileService.filter_delete(
-            [File.source_type == FileSource.KNOWLEDGEBASE, File.type == "folder", File.name == kbs[0].name])
-        if not KnowledgebaseService.delete_by_id(req["kb_id"]):
-            return get_data_error_result(
-                message="Database error (Knowledgebase removal)!")
-        for kb in kbs:
-            settings.docStoreConn.delete({"kb_id": kb.id}, search.index_name(kb.tenant_id), kb.id)
-            settings.docStoreConn.deleteIdx(search.index_name(kb.tenant_id), kb.id)
-            if hasattr(settings.STORAGE_IMPL, 'remove_bucket'):
-                settings.STORAGE_IMPL.remove_bucket(kb.id)
-        return get_json_result(data=True)
+                    message="Database error (Knowledgebase removal)!")
+            for kb in kbs:
+                if hasattr(settings.STORAGE_IMPL, 'remove_bucket'):
+                    settings.STORAGE_IMPL.remove_bucket(kb.id)
+            return get_json_result(data=True)
+
+        return await thread_pool_exec(_rm_sync)
     except Exception as e:
         return server_error_response(e)
 
@@ -286,7 +365,7 @@ def list_tags_from_kbs():
 @manager.route('/<kb_id>/rm_tags', methods=['POST'])  # noqa: F821
 @login_required
 async def rm_tags(kb_id):
-    req = await request_json()
+    req = await get_request_json()
     if not KnowledgebaseService.accessible(kb_id, current_user.id):
         return get_json_result(
             data=False,
@@ -306,7 +385,7 @@ async def rm_tags(kb_id):
 @manager.route('/<kb_id>/rename_tag', methods=['POST'])  # noqa: F821
 @login_required
 async def rename_tags(kb_id):
-    req = await request_json()
+    req = await get_request_json()
     if not KnowledgebaseService.accessible(kb_id, current_user.id):
         return get_json_result(
             data=False,
@@ -324,7 +403,7 @@ async def rename_tags(kb_id):
 
 @manager.route('/<kb_id>/knowledge_graph', methods=['GET'])  # noqa: F821
 @login_required
-def knowledge_graph(kb_id):
+async def knowledge_graph(kb_id):
     if not KnowledgebaseService.accessible(kb_id, current_user.id):
         return get_json_result(
             data=False,
@@ -338,9 +417,9 @@ def knowledge_graph(kb_id):
     }
 
     obj = {"graph": {}, "mind_map": {}}
-    if not settings.docStoreConn.indexExist(search.index_name(kb.tenant_id), kb_id):
+    if not settings.docStoreConn.index_exist(search.index_name(kb.tenant_id), kb_id):
         return get_json_result(data=obj)
-    sres = settings.retriever.search(req, search.index_name(kb.tenant_id), [kb_id])
+    sres = await settings.retriever.search(req, search.index_name(kb.tenant_id), [kb_id])
     if not len(sres.ids):
         return get_json_result(data=obj)
 
@@ -428,7 +507,7 @@ async def list_pipeline_logs():
     if create_date_to > create_date_from:
         return get_data_error_result(message="Create data filter is abnormal.")
 
-    req = await request_json()
+    req = await get_request_json()
 
     operation_status = req.get("operation_status", [])
     if operation_status:
@@ -470,7 +549,7 @@ async def list_pipeline_dataset_logs():
     if create_date_to > create_date_from:
         return get_data_error_result(message="Create data filter is abnormal.")
 
-    req = await request_json()
+    req = await get_request_json()
 
     operation_status = req.get("operation_status", [])
     if operation_status:
@@ -492,7 +571,7 @@ async def delete_pipeline_logs():
     if not kb_id:
         return get_json_result(data=False, message='Lack of "KB ID"', code=RetCode.ARGUMENT_ERROR)
 
-    req = await request_json()
+    req = await get_request_json()
     log_ids = req.get("log_ids", [])
 
     PipelineOperationLogService.delete_by_ids(log_ids)
@@ -517,7 +596,7 @@ def pipeline_log_detail():
 @manager.route("/run_graphrag", methods=["POST"])  # noqa: F821
 @login_required
 async def run_graphrag():
-    req = await request_json()
+    req = await get_request_json()
 
     kb_id = req.get("kb_id", "")
     if not kb_id:
@@ -586,7 +665,7 @@ def trace_graphrag():
 @manager.route("/run_raptor", methods=["POST"])  # noqa: F821
 @login_required
 async def run_raptor():
-    req = await request_json()
+    req = await get_request_json()
 
     kb_id = req.get("kb_id", "")
     if not kb_id:
@@ -655,7 +734,7 @@ def trace_raptor():
 @manager.route("/run_mindmap", methods=["POST"])  # noqa: F821
 @login_required
 async def run_mindmap():
-    req = await request_json()
+    req = await get_request_json()
 
     kb_id = req.get("kb_id", "")
     if not kb_id:
@@ -810,11 +889,11 @@ async def check_embedding():
         index_nm = search.index_name(tenant_id)
 
         res0 = docStoreConn.search(
-            selectFields=[], highlightFields=[],
+            select_fields=[], highlight_fields=[],
             condition={"kb_id": kb_id, "available_int": 1},
-            matchExprs=[], orderBy=OrderByExpr(),
+            match_expressions=[], order_by=OrderByExpr(),
             offset=0, limit=1,
-            indexNames=index_nm, knowledgebaseIds=[kb_id]
+            index_names=index_nm, knowledgebase_ids=[kb_id]
         )
         total = docStoreConn.get_total(res0)
         if total <= 0:
@@ -826,14 +905,14 @@ async def check_embedding():
 
         for off in offsets:
             res1 = docStoreConn.search(
-                selectFields=list(base_fields),
-                highlightFields=[],
+                select_fields=list(base_fields),
+                highlight_fields=[],
                 condition={"kb_id": kb_id, "available_int": 1},
-                matchExprs=[], orderBy=OrderByExpr(),
+                match_expressions=[], order_by=OrderByExpr(),
                 offset=off, limit=1,
-                indexNames=index_nm, knowledgebaseIds=[kb_id]
+                index_names=index_nm, knowledgebase_ids=[kb_id]
             )
-            ids = docStoreConn.get_chunk_ids(res1)
+            ids = docStoreConn.get_doc_ids(res1)
             if not ids:
                 continue
 
@@ -857,11 +936,11 @@ async def check_embedding():
                 "question_kwd": full_doc.get("question_kwd") or []
             })
         return out
-    
+
     def _clean(s: str) -> str:
         s = re.sub(r"</?(table|td|caption|tr|th)( [^<>]{0,12})?>", " ", s or "")
         return s if s else "None"
-    req = await request_json()
+    req = await get_request_json()
     kb_id = req.get("kb_id", "")
     embd_id = req.get("embd_id", "")
     n = int(req.get("check_num", 5))
@@ -922,5 +1001,3 @@ async def check_embedding():
     if summary["avg_cos_sim"] > 0.9:
         return get_json_result(data={"summary": summary, "results": results})
     return get_json_result(code=RetCode.NOT_EFFECTIVE, message="Embedding model switch failed: the average similarity between old and new vectors is below 0.9, indicating incompatible vector spaces.", data={"summary": summary, "results": results})
-
-

api/apps/langfuse_app.py

@@ -15,28 +15,28 @@
 #
 
 
-from quart import request
 from api.apps import current_user, login_required
 from langfuse import Langfuse
 
 from api.db.db_models import DB
 from api.db.services.langfuse_service import TenantLangfuseService
-from api.utils.api_utils import get_error_data_result, get_json_result, server_error_response, validate_request
+from api.utils.api_utils import get_error_data_result, get_json_result, get_request_json, server_error_response, validate_request
 
 
 @manager.route("/api_key", methods=["POST", "PUT"])  # noqa: F821
 @login_required
 @validate_request("secret_key", "public_key", "host")
 async def set_api_key():
-    req = await request.get_json()
+    req = await get_request_json()
     secret_key = req.get("secret_key", "")
     public_key = req.get("public_key", "")
     host = req.get("host", "")
     if not all([secret_key, public_key, host]):
         return get_error_data_result(message="Missing required fields")
 
+    current_user_id = current_user.id
     langfuse_keys = dict(
-        tenant_id=current_user.id,
+        tenant_id=current_user_id,
         secret_key=secret_key,
         public_key=public_key,
         host=host,
@@ -46,23 +46,24 @@ async def set_api_key():
     if not langfuse.auth_check():
         return get_error_data_result(message="Invalid Langfuse keys")
 
-    langfuse_entry = TenantLangfuseService.filter_by_tenant(tenant_id=current_user.id)
+    langfuse_entry = TenantLangfuseService.filter_by_tenant(tenant_id=current_user_id)
     with DB.atomic():
         try:
             if not langfuse_entry:
                 TenantLangfuseService.save(**langfuse_keys)
             else:
-                TenantLangfuseService.update_by_tenant(tenant_id=current_user.id, langfuse_keys=langfuse_keys)
+                TenantLangfuseService.update_by_tenant(tenant_id=current_user_id, langfuse_keys=langfuse_keys)
             return get_json_result(data=langfuse_keys)
         except Exception as e:
-            server_error_response(e)
+            return server_error_response(e)
 
 
 @manager.route("/api_key", methods=["GET"])  # noqa: F821
 @login_required
 @validate_request()
 def get_api_key():
-    langfuse_entry = TenantLangfuseService.filter_by_tenant_with_info(tenant_id=current_user.id)
+    current_user_id = current_user.id
+    langfuse_entry = TenantLangfuseService.filter_by_tenant_with_info(tenant_id=current_user_id)
     if not langfuse_entry:
         return get_json_result(message="Have not record any Langfuse keys.")
 
@@ -73,7 +74,7 @@ def get_api_key():
     except langfuse.api.core.api_error.ApiError as api_err:
         return get_json_result(message=f"Error from Langfuse: {api_err}")
     except Exception as e:
-        server_error_response(e)
+        return server_error_response(e)
 
     langfuse_entry["project_id"] = langfuse.api.projects.get().dict()["data"][0]["id"]
     langfuse_entry["project_name"] = langfuse.api.projects.get().dict()["data"][0]["name"]
@@ -85,7 +86,8 @@ def get_api_key():
 @login_required
 @validate_request()
 def delete_api_key():
-    langfuse_entry = TenantLangfuseService.filter_by_tenant(tenant_id=current_user.id)
+    current_user_id = current_user.id
+    langfuse_entry = TenantLangfuseService.filter_by_tenant(tenant_id=current_user_id)
     if not langfuse_entry:
         return get_json_result(message="Have not record any Langfuse keys.")
 
@@ -94,4 +96,4 @@ def delete_api_key():
             TenantLangfuseService.delete_model(langfuse_entry)
             return get_json_result(data=True)
         except Exception as e:
-            server_error_response(e)
+            return server_error_response(e)

api/apps/llm_app.py

@@ -21,12 +21,11 @@ from quart import request
 from api.apps import login_required, current_user
 from api.db.services.tenant_llm_service import LLMFactoriesService, TenantLLMService
 from api.db.services.llm_service import LLMService
-from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
+from api.utils.api_utils import get_allowed_llm_factories, get_data_error_result, get_json_result, get_request_json, server_error_response, validate_request
 from common.constants import StatusEnum, LLMType
 from api.db.db_models import TenantLLM
-from api.utils.api_utils import get_json_result, get_allowed_llm_factories
 from rag.utils.base64_image import test_image
-from rag.llm import EmbeddingModel, ChatModel, RerankModel, CvModel, TTSModel
+from rag.llm import EmbeddingModel, ChatModel, RerankModel, CvModel, TTSModel, OcrModel, Seq2txtModel
 
 
 @manager.route("/factories", methods=["GET"])  # noqa: F821
@@ -44,7 +43,13 @@ def factories():
                 mdl_types[m.fid] = set([])
             mdl_types[m.fid].add(m.model_type)
         for f in fac:
-            f["model_types"] = list(mdl_types.get(f["name"], [LLMType.CHAT, LLMType.EMBEDDING, LLMType.RERANK, LLMType.IMAGE2TEXT, LLMType.SPEECH2TEXT, LLMType.TTS]))
+            f["model_types"] = list(
+                mdl_types.get(
+                    f["name"],
+                    [LLMType.CHAT, LLMType.EMBEDDING, LLMType.RERANK, LLMType.IMAGE2TEXT, LLMType.SPEECH2TEXT, LLMType.TTS, LLMType.OCR],
+                )
+            )
+
         return get_json_result(data=fac)
     except Exception as e:
         return server_error_response(e)
@@ -54,7 +59,7 @@ def factories():
 @login_required
 @validate_request("llm_factory", "api_key")
 async def set_api_key():
-    req = await request.json
+    req = await get_request_json()
     # test if api key works
     chat_passed, embd_passed, rerank_passed = False, False, False
     factory = req["llm_factory"]
@@ -75,7 +80,7 @@ async def set_api_key():
             assert factory in ChatModel, f"Chat model from {factory} is not supported yet."
             mdl = ChatModel[factory](req["api_key"], llm.llm_name, base_url=req.get("base_url"), **extra)
             try:
-                m, tc = mdl.chat(None, [{"role": "user", "content": "Hello! How are you doing!"}], {"temperature": 0.9, "max_tokens": 50})
+                m, tc = await mdl.async_chat(None, [{"role": "user", "content": "Hello! How are you doing!"}], {"temperature": 0.9, "max_tokens": 50})
                 if m.find("**ERROR**") >= 0:
                     raise Exception(m)
                 chat_passed = True
@@ -124,7 +129,7 @@ async def set_api_key():
 @login_required
 @validate_request("llm_factory")
 async def add_llm():
-    req = await request.json
+    req = await get_request_json()
     factory = req["llm_factory"]
     api_key = req.get("api_key", "x")
     llm_name = req.get("llm_name")
@@ -152,7 +157,7 @@ async def add_llm():
     elif factory == "Bedrock":
         # For Bedrock, due to its special authentication method
         # Assemble bedrock_ak, bedrock_sk, bedrock_region
-        api_key = apikey_json(["bedrock_ak", "bedrock_sk", "bedrock_region"])
+        api_key = apikey_json(["auth_mode", "bedrock_ak", "bedrock_sk", "bedrock_region", "aws_role_arn"])
 
     elif factory == "LocalAI":
         llm_name += "___LocalAI"
@@ -187,6 +192,12 @@ async def add_llm():
     elif factory == "OpenRouter":
         api_key = apikey_json(["api_key", "provider_order"])
 
+    elif factory == "MinerU":
+        api_key = apikey_json(["api_key", "provider_order"])
+
+    elif factory == "PaddleOCR":
+        api_key = apikey_json(["api_key", "provider_order"])
+
     llm = {
         "tenant_id": current_user.id,
         "llm_factory": factory,
@@ -200,61 +211,82 @@ async def add_llm():
     msg = ""
     mdl_nm = llm["llm_name"].split("___")[0]
     extra = {"provider": factory}
-    if llm["model_type"] == LLMType.EMBEDDING.value:
-        assert factory in EmbeddingModel, f"Embedding model from {factory} is not supported yet."
-        mdl = EmbeddingModel[factory](key=llm["api_key"], model_name=mdl_nm, base_url=llm["api_base"])
-        try:
-            arr, tc = mdl.encode(["Test if the api key is available"])
-            if len(arr[0]) == 0:
-                raise Exception("Fail")
-        except Exception as e:
-            msg += f"\nFail to access embedding model({mdl_nm})." + str(e)
-    elif llm["model_type"] == LLMType.CHAT.value:
-        assert factory in ChatModel, f"Chat model from {factory} is not supported yet."
-        mdl = ChatModel[factory](
-            key=llm["api_key"],
-            model_name=mdl_nm,
-            base_url=llm["api_base"],
-            **extra,
-        )
-        try:
-            m, tc = mdl.chat(None, [{"role": "user", "content": "Hello! How are you doing!"}], {"temperature": 0.9})
-            if not tc and m.find("**ERROR**:") >= 0:
-                raise Exception(m)
-        except Exception as e:
-            msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
-    elif llm["model_type"] == LLMType.RERANK:
-        assert factory in RerankModel, f"RE-rank model from {factory} is not supported yet."
-        try:
-            mdl = RerankModel[factory](key=llm["api_key"], model_name=mdl_nm, base_url=llm["api_base"])
-            arr, tc = mdl.similarity("Hello~ RAGFlower!", ["Hi, there!", "Ohh, my friend!"])
-            if len(arr) == 0:
-                raise Exception("Not known.")
-        except KeyError:
-            msg += f"{factory} dose not support this model({factory}/{mdl_nm})"
-        except Exception as e:
-            msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
-    elif llm["model_type"] == LLMType.IMAGE2TEXT.value:
-        assert factory in CvModel, f"Image to text model from {factory} is not supported yet."
-        mdl = CvModel[factory](key=llm["api_key"], model_name=mdl_nm, base_url=llm["api_base"])
-        try:
-            image_data = test_image
-            m, tc = mdl.describe(image_data)
-            if not tc and m.find("**ERROR**:") >= 0:
-                raise Exception(m)
-        except Exception as e:
-            msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
-    elif llm["model_type"] == LLMType.TTS:
-        assert factory in TTSModel, f"TTS model from {factory} is not supported yet."
-        mdl = TTSModel[factory](key=llm["api_key"], model_name=mdl_nm, base_url=llm["api_base"])
-        try:
-            for resp in mdl.tts("Hello~ RAGFlower!"):
-                pass
-        except RuntimeError as e:
-            msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
-    else:
-        # TODO: check other type of models
-        pass
+    model_type = llm["model_type"]
+    model_api_key = llm["api_key"]
+    model_base_url = llm.get("api_base", "")
+    match model_type:
+        case LLMType.EMBEDDING.value:
+            assert factory in EmbeddingModel, f"Embedding model from {factory} is not supported yet."
+            mdl = EmbeddingModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
+            try:
+                arr, tc = mdl.encode(["Test if the api key is available"])
+                if len(arr[0]) == 0:
+                    raise Exception("Fail")
+            except Exception as e:
+                msg += f"\nFail to access embedding model({mdl_nm})." + str(e)
+        case LLMType.CHAT.value:
+            assert factory in ChatModel, f"Chat model from {factory} is not supported yet."
+            mdl = ChatModel[factory](
+                key=model_api_key,
+                model_name=mdl_nm,
+                base_url=model_base_url,
+                **extra,
+            )
+            try:
+                m, tc = await mdl.async_chat(None, [{"role": "user", "content": "Hello! How are you doing!"}], {"temperature": 0.9})
+                if not tc and m.find("**ERROR**:") >= 0:
+                    raise Exception(m)
+            except Exception as e:
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+
+        case LLMType.RERANK.value:
+            assert factory in RerankModel, f"RE-rank model from {factory} is not supported yet."
+            try:
+                mdl = RerankModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
+                arr, tc = mdl.similarity("Hello~ RAGFlower!", ["Hi, there!", "Ohh, my friend!"])
+                if len(arr) == 0:
+                    raise Exception("Not known.")
+            except KeyError:
+                msg += f"{factory} dose not support this model({factory}/{mdl_nm})"
+            except Exception as e:
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+
+        case LLMType.IMAGE2TEXT.value:
+            assert factory in CvModel, f"Image to text model from {factory} is not supported yet."
+            mdl = CvModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
+            try:
+                image_data = test_image
+                m, tc = mdl.describe(image_data)
+                if not tc and m.find("**ERROR**:") >= 0:
+                    raise Exception(m)
+            except Exception as e:
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+        case LLMType.TTS.value:
+            assert factory in TTSModel, f"TTS model from {factory} is not supported yet."
+            mdl = TTSModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
+            try:
+                for resp in mdl.tts("Hello~ RAGFlower!"):
+                    pass
+            except RuntimeError as e:
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+        case LLMType.OCR.value:
+            assert factory in OcrModel, f"OCR model from {factory} is not supported yet."
+            try:
+                mdl = OcrModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
+                ok, reason = mdl.check_available()
+                if not ok:
+                    raise RuntimeError(reason or "Model not available")
+            except Exception as e:
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+        case LLMType.SPEECH2TEXT:
+            assert factory in Seq2txtModel, f"Speech model from {factory} is not supported yet."
+            try:
+                mdl = Seq2txtModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
+                # TODO: check the availability
+            except Exception as e:
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+        case _:
+            raise RuntimeError(f"Unknown model type: {model_type}")
 
     if msg:
         return get_data_error_result(message=msg)
@@ -269,7 +301,7 @@ async def add_llm():
 @login_required
 @validate_request("llm_factory", "llm_name")
 async def delete_llm():
-    req = await request.json
+    req = await get_request_json()
     TenantLLMService.filter_delete([TenantLLM.tenant_id == current_user.id, TenantLLM.llm_factory == req["llm_factory"], TenantLLM.llm_name == req["llm_name"]])
     return get_json_result(data=True)
 
@@ -278,7 +310,7 @@ async def delete_llm():
 @login_required
 @validate_request("llm_factory", "llm_name")
 async def enable_llm():
-    req = await request.json
+    req = await get_request_json()
     TenantLLMService.filter_update(
         [TenantLLM.tenant_id == current_user.id, TenantLLM.llm_factory == req["llm_factory"], TenantLLM.llm_name == req["llm_name"]], {"status": str(req.get("status", "1"))}
     )
@@ -289,7 +321,7 @@ async def enable_llm():
 @login_required
 @validate_request("llm_factory")
 async def delete_factory():
-    req = await request.json
+    req = await get_request_json()
     TenantLLMService.filter_delete([TenantLLM.tenant_id == current_user.id, TenantLLM.llm_factory == req["llm_factory"]])
     return get_json_result(data=True)
 
@@ -298,6 +330,7 @@ async def delete_factory():
 @login_required
 def my_llms():
     try:
+        TenantLLMService.ensure_mineru_from_env(current_user.id)
         include_details = request.args.get("include_details", "false").lower() == "true"
 
         if include_details:
@@ -340,16 +373,18 @@ def my_llms():
 
 @manager.route("/list", methods=["GET"])  # noqa: F821
 @login_required
-def list_app():
+async def list_app():
     self_deployed = ["FastEmbed", "Ollama", "Xinference", "LocalAI", "LM-Studio", "GPUStack"]
     weighted = []
     model_type = request.args.get("model_type")
+    tenant_id = current_user.id
     try:
-        objs = TenantLLMService.query(tenant_id=current_user.id)
+        TenantLLMService.ensure_mineru_from_env(tenant_id)
+        objs = TenantLLMService.query(tenant_id=tenant_id)
         facts = set([o.to_dict()["llm_factory"] for o in objs if o.api_key and o.status == StatusEnum.VALID.value])
         status = {(o.llm_name + "@" + o.llm_factory) for o in objs if o.status == StatusEnum.VALID.value}
         llms = LLMService.get_all()
-        llms = [m.to_dict() for m in llms if m.status == StatusEnum.VALID.value and m.fid not in weighted and (m.fid == 'Builtin' or (m.llm_name + "@" + m.fid) in status)]
+        llms = [m.to_dict() for m in llms if m.status == StatusEnum.VALID.value and m.fid not in weighted and (m.fid == "Builtin" or (m.llm_name + "@" + m.fid) in status)]
         for m in llms:
             m["available"] = m["fid"] in facts or m["llm_name"].lower() == "flag-embedding" or m["fid"] in self_deployed
             if "tei-" in os.getenv("COMPOSE_PROFILES", "") and m["model_type"] == LLMType.EMBEDDING and m["fid"] == "Builtin" and m["llm_name"] == os.getenv("TEI_MODEL", ""):

api/apps/mcp_server_app.py

@@ -21,13 +21,11 @@ from api.db.services.mcp_server_service import MCPServerService
 from api.db.services.user_service import TenantService
 from common.constants import RetCode, VALID_MCP_SERVER_TYPES
 
-from common.misc_utils import get_uuid
-from api.utils.api_utils import get_data_error_result, get_json_result, server_error_response, validate_request, \
-    get_mcp_tools
+from common.misc_utils import get_uuid, thread_pool_exec
+from api.utils.api_utils import get_data_error_result, get_json_result, get_mcp_tools, get_request_json, server_error_response, validate_request
 from api.utils.web_utils import get_float, safe_json_parse
 from common.mcp_tool_call_conn import MCPToolCallSession, close_multiple_mcp_toolcall_sessions
 
-
 @manager.route("/list", methods=["POST"])  # noqa: F821
 @login_required
 async def list_mcp() -> Response:
@@ -40,7 +38,7 @@ async def list_mcp() -> Response:
     else:
         desc = True
 
-    req = await request.get_json()
+    req = await get_request_json()
     mcp_ids = req.get("mcp_ids", [])
     try:
         servers = MCPServerService.get_servers(current_user.id, mcp_ids, 0, 0, orderby, desc, keywords) or []
@@ -73,7 +71,7 @@ def detail() -> Response:
 @login_required
 @validate_request("name", "url", "server_type")
 async def create() -> Response:
-    req = await request.get_json()
+    req = await get_request_json()
 
     server_type = req.get("server_type", "")
     if server_type not in VALID_MCP_SERVER_TYPES:
@@ -107,7 +105,7 @@ async def create() -> Response:
             return get_data_error_result(message="Tenant not found.")
 
         mcp_server = MCPServer(id=server_name, name=server_name, url=url, server_type=server_type, variables=variables, headers=headers)
-        server_tools, err_message = get_mcp_tools([mcp_server], timeout)
+        server_tools, err_message = await thread_pool_exec(get_mcp_tools, [mcp_server], timeout)
         if err_message:
             return get_data_error_result(err_message)
 
@@ -128,7 +126,7 @@ async def create() -> Response:
 @login_required
 @validate_request("mcp_id")
 async def update() -> Response:
-    req = await request.get_json()
+    req = await get_request_json()
 
     mcp_id = req.get("mcp_id", "")
     e, mcp_server = MCPServerService.get_by_id(mcp_id)
@@ -159,7 +157,7 @@ async def update() -> Response:
         req["id"] = mcp_id
 
         mcp_server = MCPServer(id=server_name, name=server_name, url=url, server_type=server_type, variables=variables, headers=headers)
-        server_tools, err_message = get_mcp_tools([mcp_server], timeout)
+        server_tools, err_message = await thread_pool_exec(get_mcp_tools, [mcp_server], timeout)
         if err_message:
             return get_data_error_result(err_message)
 
@@ -184,7 +182,7 @@ async def update() -> Response:
 @login_required
 @validate_request("mcp_ids")
 async def rm() -> Response:
-    req = await request.get_json()
+    req = await get_request_json()
     mcp_ids = req.get("mcp_ids", [])
 
     try:
@@ -202,7 +200,7 @@ async def rm() -> Response:
 @login_required
 @validate_request("mcpServers")
 async def import_multiple() -> Response:
-    req = await request.get_json()
+    req = await get_request_json()
     servers = req.get("mcpServers", {})
     if not servers:
         return get_data_error_result(message="No MCP servers provided.")
@@ -243,7 +241,7 @@ async def import_multiple() -> Response:
             headers = {"authorization_token": config["authorization_token"]} if "authorization_token" in config else {}
             variables = {k: v for k, v in config.items() if k not in {"type", "url", "headers"}}
             mcp_server = MCPServer(id=new_name, name=new_name, url=config["url"], server_type=config["type"], variables=variables, headers=headers)
-            server_tools, err_message = get_mcp_tools([mcp_server], timeout)
+            server_tools, err_message = await thread_pool_exec(get_mcp_tools, [mcp_server], timeout)
             if err_message:
                 results.append({"server": base_name, "success": False, "message": err_message})
                 continue
@@ -269,7 +267,7 @@ async def import_multiple() -> Response:
 @login_required
 @validate_request("mcp_ids")
 async def export_multiple() -> Response:
-    req = await request.get_json()
+    req = await get_request_json()
     mcp_ids = req.get("mcp_ids", [])
 
     if not mcp_ids:
@@ -301,7 +299,7 @@ async def export_multiple() -> Response:
 @login_required
 @validate_request("mcp_ids")
 async def list_tools() -> Response:
-    req = await request.get_json()
+    req = await get_request_json()
     mcp_ids = req.get("mcp_ids", [])
     if not mcp_ids:
         return get_data_error_result(message="No MCP server IDs provided.")
@@ -323,9 +321,8 @@ async def list_tools() -> Response:
                 tool_call_sessions.append(tool_call_session)
 
                 try:
-                    tools = tool_call_session.get_tools(timeout)
+                    tools = await thread_pool_exec(tool_call_session.get_tools, timeout)
                 except Exception as e:
-                    tools = []
                     return get_data_error_result(message=f"MCP list tools error: {e}")
 
                 results[server_key] = []
@@ -341,14 +338,14 @@ async def list_tools() -> Response:
         return server_error_response(e)
     finally:
         # PERF: blocking call to close sessions — consider moving to background thread or task queue
-        close_multiple_mcp_toolcall_sessions(tool_call_sessions)
+        await thread_pool_exec(close_multiple_mcp_toolcall_sessions, tool_call_sessions)
 
 
 @manager.route("/test_tool", methods=["POST"])  # noqa: F821
 @login_required
 @validate_request("mcp_id", "tool_name", "arguments")
 async def test_tool() -> Response:
-    req = await request.get_json()
+    req = await get_request_json()
     mcp_id = req.get("mcp_id", "")
     if not mcp_id:
         return get_data_error_result(message="No MCP server ID provided.")
@@ -368,10 +365,10 @@ async def test_tool() -> Response:
 
         tool_call_session = MCPToolCallSession(mcp_server, mcp_server.variables)
         tool_call_sessions.append(tool_call_session)
-        result = tool_call_session.tool_call(tool_name, arguments, timeout)
+        result = await thread_pool_exec(tool_call_session.tool_call, tool_name, arguments, timeout)
 
         # PERF: blocking call to close sessions — consider moving to background thread or task queue
-        close_multiple_mcp_toolcall_sessions(tool_call_sessions)
+        await thread_pool_exec(close_multiple_mcp_toolcall_sessions, tool_call_sessions)
         return get_json_result(data=result)
     except Exception as e:
         return server_error_response(e)
@@ -381,7 +378,7 @@ async def test_tool() -> Response:
 @login_required
 @validate_request("mcp_id", "tools")
 async def cache_tool() -> Response:
-    req = await request.get_json()
+    req = await get_request_json()
     mcp_id = req.get("mcp_id", "")
     if not mcp_id:
         return get_data_error_result(message="No MCP server ID provided.")
@@ -404,7 +401,7 @@ async def cache_tool() -> Response:
 @manager.route("/test_mcp", methods=["POST"])  # noqa: F821
 @validate_request("url", "server_type")
 async def test_mcp() -> Response:
-    req = await request.get_json()
+    req = await get_request_json()
 
     url = req.get("url", "")
     if not url:
@@ -425,13 +422,12 @@ async def test_mcp() -> Response:
         tool_call_session = MCPToolCallSession(mcp_server, mcp_server.variables)
 
         try:
-            tools = tool_call_session.get_tools(timeout)
+            tools = await thread_pool_exec(tool_call_session.get_tools, timeout)
         except Exception as e:
-            tools = []
             return get_data_error_result(message=f"Test MCP error: {e}")
         finally:
             # PERF: blocking call to close sessions — consider moving to background thread or task queue
-            close_multiple_mcp_toolcall_sessions([tool_call_session])
+            await thread_pool_exec(close_multiple_mcp_toolcall_sessions, [tool_call_session])
 
         for tool in tools:
             tool_dict = tool.model_dump()

api/apps/sdk/agents.py

@@ -14,20 +14,29 @@
 #  limitations under the License.
 #
 
+import asyncio
+import base64
+import hashlib
+import hmac
+import ipaddress
 import json
 import logging
 import time
 from typing import Any, cast
 
+import jwt
+
 from agent.canvas import Canvas
 from api.db import CanvasCategory
 from api.db.services.canvas_service import UserCanvasService
+from api.db.services.file_service import FileService
 from api.db.services.user_canvas_version import UserCanvasVersionService
 from common.constants import RetCode
 from common.misc_utils import get_uuid
-from api.utils.api_utils import get_data_error_result, get_error_data_result, get_json_result, token_required
+from api.utils.api_utils import get_data_error_result, get_error_data_result, get_json_result, get_request_json, token_required
 from api.utils.api_utils import get_result
 from quart import request, Response
+from rag.utils.redis_conn import REDIS_CONN
 
 
 @manager.route('/agents', methods=['GET'])  # noqa: F821
@@ -42,7 +51,7 @@ def list_agents(tenant_id):
     page_number = int(request.args.get("page", 1))
     items_per_page = int(request.args.get("page_size", 30))
     order_by = request.args.get("orderby", "update_time")
-    if request.args.get("desc") == "False" or request.args.get("desc") == "false":
+    if str(request.args.get("desc","false")).lower() == "false":
         desc = False
     else:
         desc = True
@@ -53,7 +62,7 @@ def list_agents(tenant_id):
 @manager.route("/agents", methods=["POST"])  # noqa: F821
 @token_required
 async def create_agent(tenant_id: str):
-    req: dict[str, Any] = cast(dict[str, Any], await request.json)
+    req: dict[str, Any] = cast(dict[str, Any], await get_request_json())
     req["user_id"] = tenant_id
 
     if req.get("dsl") is not None:
@@ -90,7 +99,7 @@ async def create_agent(tenant_id: str):
 @manager.route("/agents/<agent_id>", methods=["PUT"])  # noqa: F821
 @token_required
 async def update_agent(tenant_id: str, agent_id: str):
-    req: dict[str, Any] = {k: v for k, v in cast(dict[str, Any], (await request.json)).items() if v is not None}
+    req: dict[str, Any] = {k: v for k, v in cast(dict[str, Any], (await get_request_json())).items() if v is not None}
     req["user_id"] = tenant_id
 
     if req.get("dsl") is not None:
@@ -132,48 +141,786 @@ def delete_agent(tenant_id: str, agent_id: str):
     UserCanvasService.delete_by_id(agent_id)
     return get_json_result(data=True)
 
+@manager.route("/webhook/<agent_id>", methods=["POST", "GET", "PUT", "PATCH", "DELETE", "HEAD"])  # noqa: F821
+@manager.route("/webhook_test/<agent_id>",methods=["POST", "GET", "PUT", "PATCH", "DELETE", "HEAD"],)  # noqa: F821
+async def webhook(agent_id: str):
+    is_test = request.path.startswith("/api/v1/webhook_test")
+    start_ts = time.time()
 
-@manager.route('/webhook/<agent_id>', methods=['POST'])  # noqa: F821
-@token_required
-async def webhook(tenant_id: str, agent_id: str):
-    req = await request.json
-    if not UserCanvasService.accessible(req["id"], tenant_id):
-        return get_json_result(
-            data=False, message='Only owner of canvas authorized for this operation.',
-            code=RetCode.OPERATING_ERROR)
-
-    e, cvs = UserCanvasService.get_by_id(req["id"])
-    if not e:
-        return get_data_error_result(message="canvas not found.")
-
-    if not isinstance(cvs.dsl, str):
-        cvs.dsl = json.dumps(cvs.dsl, ensure_ascii=False)
+    # 1. Fetch canvas by agent_id
+    exists, cvs = UserCanvasService.get_by_id(agent_id)
+    if not exists:
+        return get_data_error_result(code=RetCode.BAD_REQUEST,message="Canvas not found."),RetCode.BAD_REQUEST
 
+    # 2. Check canvas category
     if cvs.canvas_category == CanvasCategory.DataFlow:
-        return get_data_error_result(message="Dataflow can not be triggered by webhook.")
+        return get_data_error_result(code=RetCode.BAD_REQUEST,message="Dataflow can not be triggered by webhook."),RetCode.BAD_REQUEST
+
+    # 3. Load DSL from canvas
+    dsl = getattr(cvs, "dsl", None)
+    if not isinstance(dsl, dict):
+        return get_data_error_result(code=RetCode.BAD_REQUEST,message="Invalid DSL format."),RetCode.BAD_REQUEST
+
+    # 4. Check webhook configuration in DSL
+    webhook_cfg = {}
+    components = dsl.get("components", {})
+    for k, _ in components.items():
+        cpn_obj = components[k]["obj"]
+        if cpn_obj["component_name"].lower() == "begin" and cpn_obj["params"]["mode"] == "Webhook":
+            webhook_cfg = cpn_obj["params"]
+
+    if not webhook_cfg:
+        return get_data_error_result(code=RetCode.BAD_REQUEST,message="Webhook not configured for this agent."),RetCode.BAD_REQUEST
+
+    # 5. Validate request method against webhook_cfg.methods
+    allowed_methods = webhook_cfg.get("methods", [])
+    request_method = request.method.upper()
+    if allowed_methods and request_method not in allowed_methods:
+        return get_data_error_result(
+            code=RetCode.BAD_REQUEST,message=f"HTTP method '{request_method}' not allowed for this webhook."
+        ),RetCode.BAD_REQUEST
+
+    # 6. Validate webhook security
+    async def validate_webhook_security(security_cfg: dict):
+        """Validate webhook security rules based on security configuration."""
+
+        if not security_cfg:
+            return  # No security config → allowed by default
+
+        # 1. Validate max body size
+        await _validate_max_body_size(security_cfg)
+
+        # 2. Validate IP whitelist
+        _validate_ip_whitelist(security_cfg)
+
+        # # 3. Validate rate limiting
+        _validate_rate_limit(security_cfg)
+
+        # 4. Validate authentication
+        auth_type = security_cfg.get("auth_type", "none")
+
+        if auth_type == "none":
+            return
+
+        if auth_type == "token":
+            _validate_token_auth(security_cfg)
+
+        elif auth_type == "basic":
+            _validate_basic_auth(security_cfg)
+
+        elif auth_type == "jwt":
+            _validate_jwt_auth(security_cfg)
+
+        else:
+            raise Exception(f"Unsupported auth_type: {auth_type}")
+
+    async def _validate_max_body_size(security_cfg):
+        """Check request size does not exceed max_body_size."""
+        max_size = security_cfg.get("max_body_size")
+        if not max_size:
+            return
+
+        # Convert "10MB" → bytes
+        units = {"kb": 1024, "mb": 1024**2}
+        size_str = max_size.lower()
+
+        for suffix, factor in units.items():
+            if size_str.endswith(suffix):
+                limit = int(size_str.replace(suffix, "")) * factor
+                break
+        else:
+            raise Exception("Invalid max_body_size format")
+        MAX_LIMIT = 10 * 1024 * 1024  # 10MB
+        if limit > MAX_LIMIT:
+            raise Exception("max_body_size exceeds maximum allowed size (10MB)")
+
+        content_length = request.content_length or 0
+        if content_length > limit:
+            raise Exception(f"Request body too large: {content_length} > {limit}")
+
+    def _validate_ip_whitelist(security_cfg):
+        """Allow only IPs listed in ip_whitelist."""
+        whitelist = security_cfg.get("ip_whitelist", [])
+        if not whitelist:
+            return
+
+        client_ip = request.remote_addr
+
+
+        for rule in whitelist:
+            if "/" in rule:
+                # CIDR notation
+                if ipaddress.ip_address(client_ip) in ipaddress.ip_network(rule, strict=False):
+                    return
+            else:
+                # Single IP
+                if client_ip == rule:
+                    return
+
+        raise Exception(f"IP {client_ip} is not allowed by whitelist")
+
+    def _validate_rate_limit(security_cfg):
+        """Simple in-memory rate limiting."""
+        rl = security_cfg.get("rate_limit")
+        if not rl:
+            return
+
+        limit = int(rl.get("limit", 60))
+        if limit <= 0:
+            raise Exception("rate_limit.limit must be > 0")
+        per = rl.get("per", "minute")
+
+        window = {
+            "second": 1,
+            "minute": 60,
+            "hour": 3600,
+            "day": 86400,
+        }.get(per)
+
+        if not window:
+            raise Exception(f"Invalid rate_limit.per: {per}")
+
+        capacity = limit
+        rate = limit / window
+        cost = 1
+
+        key = f"rl:tb:{agent_id}"
+        now = time.time()
+
+        try:
+            res = REDIS_CONN.lua_token_bucket(
+                keys=[key],
+                args=[capacity, rate, now, cost],
+                client=REDIS_CONN.REDIS,
+            )
+
+            allowed = int(res[0])
+            if allowed != 1:
+                raise Exception("Too many requests (rate limit exceeded)")
+
+        except Exception as e:
+            raise Exception(f"Rate limit error: {e}")
+
+    def _validate_token_auth(security_cfg):
+        """Validate header-based token authentication."""
+        token_cfg = security_cfg.get("token",{})
+        header = token_cfg.get("token_header")
+        token_value = token_cfg.get("token_value")
+
+        provided = request.headers.get(header)
+        if provided != token_value:
+            raise Exception("Invalid token authentication")
+
+    def _validate_basic_auth(security_cfg):
+        """Validate HTTP Basic Auth credentials."""
+        auth_cfg = security_cfg.get("basic_auth", {})
+        username = auth_cfg.get("username")
+        password = auth_cfg.get("password")
+
+        auth = request.authorization
+        if not auth or auth.username != username or auth.password != password:
+            raise Exception("Invalid Basic Auth credentials")
+
+    def _validate_jwt_auth(security_cfg):
+        """Validate JWT token in Authorization header."""
+        jwt_cfg = security_cfg.get("jwt", {})
+        secret = jwt_cfg.get("secret")
+        if not secret:
+            raise Exception("JWT secret not configured")
+
+        auth_header = request.headers.get("Authorization", "")
+        if not auth_header.startswith("Bearer "):
+            raise Exception("Missing Bearer token")
+
+        token = auth_header[len("Bearer "):].strip()
+        if not token:
+            raise Exception("Empty Bearer token")
+
+        alg = (jwt_cfg.get("algorithm") or "HS256").upper()
+
+        decode_kwargs = {
+            "key": secret,
+            "algorithms": [alg],
+        }
+        options = {}
+        if jwt_cfg.get("audience"):
+            decode_kwargs["audience"] = jwt_cfg["audience"]
+            options["verify_aud"] = True
+        else:
+            options["verify_aud"] = False
+
+        if jwt_cfg.get("issuer"):
+            decode_kwargs["issuer"] = jwt_cfg["issuer"]
+            options["verify_iss"] = True
+        else:
+            options["verify_iss"] = False
+        try:
+            decoded = jwt.decode(
+                token,
+                options=options,
+                **decode_kwargs,
+            )
+        except Exception as e:
+            raise Exception(f"Invalid JWT: {str(e)}")
+
+        raw_required_claims = jwt_cfg.get("required_claims", [])
+        if isinstance(raw_required_claims, str):
+            required_claims = [raw_required_claims]
+        elif isinstance(raw_required_claims, (list, tuple, set)):
+            required_claims = list(raw_required_claims)
+        else:
+            required_claims = []
+
+        required_claims = [
+            c for c in required_claims
+            if isinstance(c, str) and c.strip()
+        ]
+
+        RESERVED_CLAIMS = {"exp", "sub", "aud", "iss", "nbf", "iat"}
+        for claim in required_claims:
+            if claim in RESERVED_CLAIMS:
+                raise Exception(f"Reserved JWT claim cannot be required: {claim}")
+
+        for claim in required_claims:
+            if claim not in decoded:
+                raise Exception(f"Missing JWT claim: {claim}")
+
+        return decoded
 
     try:
-        canvas = Canvas(cvs.dsl, tenant_id, agent_id)
+        security_config=webhook_cfg.get("security", {})
+        await validate_webhook_security(security_config)
     except Exception as e:
-        return get_json_result(
-            data=False, message=str(e),
-            code=RetCode.EXCEPTION_ERROR)
+        return get_data_error_result(code=RetCode.BAD_REQUEST,message=str(e)),RetCode.BAD_REQUEST
+    if not isinstance(cvs.dsl, str):
+        dsl = json.dumps(cvs.dsl, ensure_ascii=False)
+    try:
+        canvas = Canvas(dsl, cvs.user_id, agent_id, canvas_id=agent_id)
+    except Exception as e:
+        resp=get_data_error_result(code=RetCode.BAD_REQUEST,message=str(e))
+        resp.status_code = RetCode.BAD_REQUEST
+        return resp
+
+    # 7. Parse request body
+    async def parse_webhook_request(content_type):
+        """Parse request based on content-type and return structured data."""
+
+        # 1. Query
+        query_data = {k: v for k, v in request.args.items()}
+
+        # 2. Headers
+        header_data = {k: v for k, v in request.headers.items()}
+
+        # 3. Body
+        ctype = request.headers.get("Content-Type", "").split(";")[0].strip()
+        if ctype and ctype != content_type:
+            raise ValueError(
+                f"Invalid Content-Type: expect '{content_type}', got '{ctype}'"
+            )
+
+        body_data: dict = {}
 
-    async def sse():
-        nonlocal canvas
         try:
-            async for ans in canvas.run(query=req.get("query", ""), files=req.get("files", []), user_id=req.get("user_id", tenant_id), webhook_payload=req):
-                yield "data:" + json.dumps(ans, ensure_ascii=False) + "\n\n"
+            if ctype == "application/json":
+                body_data = await request.get_json() or {}
 
-            cvs.dsl = json.loads(str(canvas))
-            UserCanvasService.update_by_id(req["id"], cvs.to_dict())
-        except Exception as e:
-            logging.exception(e)
-            yield "data:" + json.dumps({"code": 500, "message": str(e), "data": False}, ensure_ascii=False) + "\n\n"
+            elif ctype == "multipart/form-data":
+                nonlocal canvas
+                form = await request.form
+                files = await request.files
 
-    resp = Response(sse(), mimetype="text/event-stream")
-    resp.headers.add_header("Cache-control", "no-cache")
-    resp.headers.add_header("Connection", "keep-alive")
-    resp.headers.add_header("X-Accel-Buffering", "no")
-    resp.headers.add_header("Content-Type", "text/event-stream; charset=utf-8")
-    return resp
+                body_data = {}
+
+                for key, value in form.items():
+                    body_data[key] = value
+
+                if len(files) > 10:
+                    raise Exception("Too many uploaded files")
+                for key, file in files.items():
+                    desc = FileService.upload_info(
+                        cvs.user_id,           # user
+                        file,              # FileStorage
+                        None                   # url (None for webhook)
+                    )
+                    file_parsed= await canvas.get_files_async([desc])
+                    body_data[key] = file_parsed
+
+            elif ctype == "application/x-www-form-urlencoded":
+                form = await request.form
+                body_data = dict(form)
+
+            else:
+                # text/plain / octet-stream / empty / unknown
+                raw = await request.get_data()
+                if raw:
+                    try:
+                        body_data = json.loads(raw.decode("utf-8"))
+                    except Exception:
+                        body_data = {}
+                else:
+                    body_data = {}
+
+        except Exception:
+            body_data = {}
+
+        return {
+            "query": query_data,
+            "headers": header_data,
+            "body": body_data,
+            "content_type": ctype,
+        }
+
+    def extract_by_schema(data, schema, name="section"):
+        """
+        Extract only fields defined in schema.
+        Required fields must exist.
+        Optional fields default to type-based default values.
+        Type validation included.
+        """
+        props = schema.get("properties", {})
+        required = schema.get("required", [])
+
+        extracted = {}
+
+        for field, field_schema in props.items():
+            field_type = field_schema.get("type")
+
+            # 1. Required field missing
+            if field in required and field not in data:
+                raise Exception(f"{name} missing required field: {field}")
+
+            # 2. Optional → default value
+            if field not in data:
+                extracted[field] = default_for_type(field_type)
+                continue
+
+            raw_value = data[field]
+
+            # 3. Auto convert value
+            try:
+                value = auto_cast_value(raw_value, field_type)
+            except Exception as e:
+                raise Exception(f"{name}.{field} auto-cast failed: {str(e)}")
+
+            # 4. Type validation
+            if not validate_type(value, field_type):
+                raise Exception(
+                    f"{name}.{field} type mismatch: expected {field_type}, got {type(value).__name__}"
+                )
+
+            extracted[field] = value
+
+        return extracted
+
+
+    def default_for_type(t):
+        """Return default value for the given schema type."""
+        if t == "file":
+            return []
+        if t == "object":
+            return {}
+        if t == "boolean":
+            return False
+        if t == "number":
+            return 0
+        if t == "string":
+            return ""
+        if t and t.startswith("array"):
+            return []
+        if t == "null":
+            return None
+        return None
+
+    def auto_cast_value(value, expected_type):
+        """Convert string values into schema type when possible."""
+
+        # Non-string values already good
+        if not isinstance(value, str):
+            return value
+
+        v = value.strip()
+
+        # Boolean
+        if expected_type == "boolean":
+            if v.lower() in ["true", "1"]:
+                return True
+            if v.lower() in ["false", "0"]:
+                return False
+            raise Exception(f"Cannot convert '{value}' to boolean")
+
+        # Number
+        if expected_type == "number":
+            # integer
+            if v.isdigit() or (v.startswith("-") and v[1:].isdigit()):
+                return int(v)
+
+            # float
+            try:
+                return float(v)
+            except Exception:
+                raise Exception(f"Cannot convert '{value}' to number")
+
+        # Object
+        if expected_type == "object":
+            try:
+                parsed = json.loads(v)
+                if isinstance(parsed, dict):
+                    return parsed
+                else:
+                    raise Exception("JSON is not an object")
+            except Exception:
+                raise Exception(f"Cannot convert '{value}' to object")
+
+        # Array <T>
+        if expected_type.startswith("array"):
+            try:
+                parsed = json.loads(v)
+                if isinstance(parsed, list):
+                    return parsed
+                else:
+                    raise Exception("JSON is not an array")
+            except Exception:
+                raise Exception(f"Cannot convert '{value}' to array")
+
+        # String (accept original)
+        if expected_type == "string":
+            return value
+
+        # File
+        if expected_type == "file":
+            return value
+        # Default: do nothing
+        return value
+
+
+    def validate_type(value, t):
+        """Validate value type against schema type t."""
+        if t == "file":
+            return isinstance(value, list)
+
+        if t == "string":
+            return isinstance(value, str)
+
+        if t == "number":
+            return isinstance(value, (int, float))
+
+        if t == "boolean":
+            return isinstance(value, bool)
+
+        if t == "object":
+            return isinstance(value, dict)
+
+        # array<string> / array<number> / array<object>
+        if t.startswith("array"):
+            if not isinstance(value, list):
+                return False
+
+            if "<" in t and ">" in t:
+                inner = t[t.find("<") + 1 : t.find(">")]
+
+                # Check each element type
+                for item in value:
+                    if not validate_type(item, inner):
+                        return False
+
+            return True
+
+        return True
+    parsed = await parse_webhook_request(webhook_cfg.get("content_types"))
+    SCHEMA = webhook_cfg.get("schema", {"query": {}, "headers": {}, "body": {}})
+
+    # Extract strictly by schema
+    try:
+        query_clean  = extract_by_schema(parsed["query"],   SCHEMA.get("query", {}),  name="query")
+        header_clean = extract_by_schema(parsed["headers"], SCHEMA.get("headers", {}), name="headers")
+        body_clean   = extract_by_schema(parsed["body"],    SCHEMA.get("body", {}),    name="body")
+    except Exception as e:
+        return get_data_error_result(code=RetCode.BAD_REQUEST,message=str(e)),RetCode.BAD_REQUEST
+
+    clean_request = {
+        "query": query_clean,
+        "headers": header_clean,
+        "body": body_clean,
+        "input": parsed
+    }
+
+    execution_mode = webhook_cfg.get("execution_mode", "Immediately")
+    response_cfg = webhook_cfg.get("response", {})
+
+    def append_webhook_trace(agent_id: str, start_ts: float,event: dict, ttl=600):
+        key = f"webhook-trace-{agent_id}-logs"
+
+        raw = REDIS_CONN.get(key)
+        obj = json.loads(raw) if raw else {"webhooks": {}}
+
+        ws = obj["webhooks"].setdefault(
+            str(start_ts),
+            {"start_ts": start_ts, "events": []}
+        )
+
+        ws["events"].append({
+            "ts": time.time(),
+            **event
+        })
+
+        REDIS_CONN.set_obj(key, obj, ttl)
+
+    if execution_mode == "Immediately":
+        status = response_cfg.get("status", 200)
+        try:
+            status = int(status)
+        except (TypeError, ValueError):
+            return get_data_error_result(code=RetCode.BAD_REQUEST,message=str(f"Invalid response status code: {status}")),RetCode.BAD_REQUEST
+
+        if not (200 <= status <= 399):
+            return get_data_error_result(code=RetCode.BAD_REQUEST,message=str(f"Invalid response status code: {status}, must be between 200 and 399")),RetCode.BAD_REQUEST
+
+        body_tpl = response_cfg.get("body_template", "")
+
+        def parse_body(body: str):
+            if not body:
+                return None, "application/json"
+
+            try:
+                parsed = json.loads(body)
+                return parsed, "application/json"
+            except (json.JSONDecodeError, TypeError):
+                return body, "text/plain"
+
+
+        body, content_type = parse_body(body_tpl)
+        resp = Response(
+            json.dumps(body, ensure_ascii=False) if content_type == "application/json" else body,
+            status=status,
+            content_type=content_type,
+        )
+
+        async def background_run():
+            try:
+                async for ans in canvas.run(
+                    query="",
+                    user_id=cvs.user_id,
+                    webhook_payload=clean_request
+                ):
+                    if is_test:
+                        append_webhook_trace(agent_id, start_ts, ans)
+
+                if is_test:
+                    append_webhook_trace(
+                        agent_id,
+                        start_ts,
+                        {
+                            "event": "finished",
+                            "elapsed_time": time.time() - start_ts,
+                            "success": True,
+                        }
+                    )
+
+                cvs.dsl = json.loads(str(canvas))
+                UserCanvasService.update_by_id(cvs.user_id, cvs.to_dict())
+
+            except Exception as e:
+                logging.exception("Webhook background run failed")
+                if is_test:
+                    try:
+                        append_webhook_trace(
+                            agent_id,
+                            start_ts,
+                            {
+                                "event": "error",
+                                "message": str(e),
+                                "error_type": type(e).__name__,
+                            }
+                        )
+                        append_webhook_trace(
+                            agent_id,
+                            start_ts,
+                            {
+                                "event": "finished",
+                                "elapsed_time": time.time() - start_ts,
+                                "success": False,
+                            }
+                        )
+                    except Exception:
+                        logging.exception("Failed to append webhook trace")
+
+        asyncio.create_task(background_run())
+        return resp
+    else:
+        async def sse():
+            nonlocal canvas
+            contents: list[str] = []
+            status = 200
+            try:
+                async for ans in canvas.run(
+                    query="",
+                    user_id=cvs.user_id,
+                    webhook_payload=clean_request,
+                ):
+                    if ans["event"] == "message":
+                        content = ans["data"]["content"]
+                        if ans["data"].get("start_to_think", False):
+                            content = "<think>"
+                        elif ans["data"].get("end_to_think", False):
+                            content = "</think>"
+                        if content:
+                            contents.append(content)
+                    if ans["event"] == "message_end":
+                        status = int(ans["data"].get("status", status))
+                    if is_test:
+                        append_webhook_trace(
+                            agent_id,
+                            start_ts,
+                            ans
+                        )
+                if is_test:
+                    append_webhook_trace(
+                        agent_id,
+                        start_ts,
+                        {
+                            "event": "finished",
+                            "elapsed_time": time.time() - start_ts,
+                            "success": True,
+                        }
+                    )
+                final_content = "".join(contents)
+                return {
+                    "message": final_content,
+                    "success": True,
+                    "code":  status,
+                }
+
+            except Exception as e:
+                if is_test:
+                    append_webhook_trace(
+                        agent_id,
+                        start_ts,
+                        {
+                            "event": "error",
+                            "message": str(e),
+                            "error_type": type(e).__name__,
+                        }
+                    )
+                    append_webhook_trace(
+                        agent_id,
+                        start_ts,
+                        {
+                            "event": "finished",
+                            "elapsed_time": time.time() - start_ts,
+                            "success": False,
+                        }
+                    )
+                return {"code": 400, "message": str(e),"success":False}
+
+        result = await sse()
+        return Response(
+            json.dumps(result),
+            status=result["code"],
+            mimetype="application/json",
+        )
+
+
+@manager.route("/webhook_trace/<agent_id>", methods=["GET"])  # noqa: F821
+async def webhook_trace(agent_id: str):
+    def encode_webhook_id(start_ts: str) -> str:
+        WEBHOOK_ID_SECRET = "webhook_id_secret"
+        sig = hmac.new(
+            WEBHOOK_ID_SECRET.encode("utf-8"),
+            start_ts.encode("utf-8"),
+            hashlib.sha256,
+        ).digest()
+        return base64.urlsafe_b64encode(sig).decode("utf-8").rstrip("=")
+
+    def decode_webhook_id(enc_id: str, webhooks: dict) -> str | None:
+        for ts in webhooks.keys():
+            if encode_webhook_id(ts) == enc_id:
+                return ts
+        return None
+    since_ts = request.args.get("since_ts", type=float)
+    webhook_id = request.args.get("webhook_id")
+
+    key = f"webhook-trace-{agent_id}-logs"
+    raw = REDIS_CONN.get(key)
+
+    if since_ts is None:
+        now = time.time()
+        return get_json_result(
+            data={
+                "webhook_id": None,
+                "events": [],
+                "next_since_ts": now,
+                "finished": False,
+            }
+        )
+
+    if not raw:
+        return get_json_result(
+            data={
+                "webhook_id": None,
+                "events": [],
+                "next_since_ts": since_ts,
+                "finished": False,
+            }
+        )
+
+    obj = json.loads(raw)
+    webhooks = obj.get("webhooks", {})
+
+    if webhook_id is None:
+        candidates = [
+            float(k) for k in webhooks.keys() if float(k) > since_ts
+        ]
+
+        if not candidates:
+            return get_json_result(
+                data={
+                    "webhook_id": None,
+                    "events": [],
+                    "next_since_ts": since_ts,
+                    "finished": False,
+                }
+            )
+
+        start_ts = min(candidates)
+        real_id = str(start_ts)
+        webhook_id = encode_webhook_id(real_id)
+
+        return get_json_result(
+            data={
+                "webhook_id": webhook_id,
+                "events": [],
+                "next_since_ts": start_ts,
+                "finished": False,
+            }
+        )
+
+    real_id = decode_webhook_id(webhook_id, webhooks)
+
+    if not real_id:
+        return get_json_result(
+            data={
+                "webhook_id": webhook_id,
+                "events": [],
+                "next_since_ts": since_ts,
+                "finished": True,
+            }
+        )
+
+    ws = webhooks.get(str(real_id))
+    events = ws.get("events", [])
+    new_events = [e for e in events if e.get("ts", 0) > since_ts]
+
+    next_ts = since_ts
+    for e in new_events:
+        next_ts = max(next_ts, e["ts"])
+
+    finished = any(e.get("event") == "finished" for e in new_events)
+
+    return get_json_result(
+        data={
+            "webhook_id": webhook_id,
+            "events": new_events,
+            "next_since_ts": next_ts,
+            "finished": finished,
+        }
+    )

api/apps/sdk/chat.py

@@ -21,13 +21,13 @@ from api.db.services.tenant_llm_service import TenantLLMService
 from api.db.services.user_service import TenantService
 from common.misc_utils import get_uuid
 from common.constants import RetCode, StatusEnum
-from api.utils.api_utils import check_duplicate_ids, get_error_data_result, get_result, token_required, request_json
+from api.utils.api_utils import check_duplicate_ids, get_error_data_result, get_result, token_required, get_request_json
 
 
 @manager.route("/chats", methods=["POST"])  # noqa: F821
 @token_required
 async def create(tenant_id):
-    req = await request_json()
+    req = await get_request_json()
     ids = [i for i in req.get("dataset_ids", []) if i]
     for kb_id in ids:
         kbs = KnowledgebaseService.accessible(kb_id=kb_id, user_id=tenant_id)
@@ -51,7 +51,9 @@ async def create(tenant_id):
             req["llm_id"] = llm.pop("model_name")
             if req.get("llm_id") is not None:
                 llm_name, llm_factory = TenantLLMService.split_model_name_and_factory(req["llm_id"])
-                if not TenantLLMService.query(tenant_id=tenant_id, llm_name=llm_name, llm_factory=llm_factory, model_type="chat"):
+                model_type = llm.get("model_type")
+                model_type = model_type if model_type in ["chat", "image2text"] else "chat"
+                if not TenantLLMService.query(tenant_id=tenant_id, llm_name=llm_name, llm_factory=llm_factory, model_type=model_type):
                     return get_error_data_result(f"`model_name` {req.get('llm_id')} doesn't exist")
         req["llm_setting"] = req.pop("llm")
     e, tenant = TenantService.get_by_id(tenant_id)
@@ -92,7 +94,7 @@ async def create(tenant_id):
     req["tenant_id"] = tenant_id
     # prompt more parameter
     default_prompt = {
-        "system": """You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence "The answer you are looking for is not found in the knowledge base!" Answers need to consider chat history.
+        "system": """You are an intelligent assistant. Please summarize the content of the dataset to answer the question. Please list the data in the dataset and answer in detail. When all dataset content is irrelevant to the question, your answer must include the sentence "The answer you are looking for is not found in the dataset!" Answers need to consider chat history.
       Here is the knowledge base:
       {knowledge}
       The above is the knowledge base.""",
@@ -146,7 +148,7 @@ async def create(tenant_id):
 async def update(tenant_id, chat_id):
     if not DialogService.query(tenant_id=tenant_id, id=chat_id, status=StatusEnum.VALID.value):
         return get_error_data_result(message="You do not own the chat")
-    req = await request_json()
+    req = await get_request_json()
     ids = req.get("dataset_ids", [])
     if "show_quotation" in req:
         req["do_refer"] = req.pop("show_quotation")
@@ -174,7 +176,9 @@ async def update(tenant_id, chat_id):
             req["llm_id"] = llm.pop("model_name")
             if req.get("llm_id") is not None:
                 llm_name, llm_factory = TenantLLMService.split_model_name_and_factory(req["llm_id"])
-                if not TenantLLMService.query(tenant_id=tenant_id, llm_name=llm_name, llm_factory=llm_factory, model_type="chat"):
+                model_type = llm.get("model_type")
+                model_type = model_type if model_type in ["chat", "image2text"] else "chat"
+                if not TenantLLMService.query(tenant_id=tenant_id, llm_name=llm_name, llm_factory=llm_factory, model_type=model_type):
                     return get_error_data_result(f"`model_name` {req.get('llm_id')} doesn't exist")
         req["llm_setting"] = req.pop("llm")
     e, tenant = TenantService.get_by_id(tenant_id)
@@ -229,7 +233,7 @@ async def update(tenant_id, chat_id):
 async def delete_chats(tenant_id):
     errors = []
     success_count = 0
-    req = await request_json()
+    req = await get_request_json()
     if not req:
         ids = None
     else:
@@ -250,7 +254,6 @@ async def delete_chats(tenant_id):
             continue
         temp_dict = {"status": StatusEnum.INVALID.value}
         success_count += DialogService.update_by_id(id, temp_dict)
-        print(success_count, "$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$", flush=True)
 
     if errors:
         if success_count > 0:
@@ -286,7 +289,7 @@ def list_chat(tenant_id):
     chats = DialogService.get_list(tenant_id, page_number, items_per_page, orderby, desc, id, name)
     if not chats:
         return get_result(data=[])
-    list_assts = []
+    list_assistants = []
     key_mapping = {
         "parameters": "variables",
         "prologue": "opener",
@@ -320,5 +323,5 @@ def list_chat(tenant_id):
         del res["kb_ids"]
         res["datasets"] = kb_list
         res["avatar"] = res.pop("icon")
-        list_assts.append(res)
-    return get_result(data=list_assts)
+        list_assistants.append(res)
+    return get_result(data=list_assistants)

api/apps/sdk/dataset.py

@@ -233,6 +233,15 @@ async def delete(tenant_id):
                 File2DocumentService.delete_by_document_id(doc.id)
             FileService.filter_delete(
                 [File.source_type == FileSource.KNOWLEDGEBASE, File.type == "folder", File.name == kb.name])
+
+            # Drop index for this dataset
+            try:
+                from rag.nlp import search
+                idxnm = search.index_name(kb.tenant_id)
+                settings.docStoreConn.delete_idx(idxnm, kb_id)
+            except Exception as e:
+                logging.warning(f"Failed to drop index for dataset {kb_id}: {e}")
+
             if not KnowledgebaseService.delete_by_id(kb_id):
                 errors.append(f"Delete dataset error for {kb_id}")
                 continue
@@ -481,7 +490,7 @@ def list_datasets(tenant_id):
 
 @manager.route('/datasets/<dataset_id>/knowledge_graph', methods=['GET'])  # noqa: F821
 @token_required
-def knowledge_graph(tenant_id, dataset_id):
+async def knowledge_graph(tenant_id, dataset_id):
     if not KnowledgebaseService.accessible(dataset_id, tenant_id):
         return get_result(
             data=False,
@@ -495,9 +504,9 @@ def knowledge_graph(tenant_id, dataset_id):
     }
 
     obj = {"graph": {}, "mind_map": {}}
-    if not settings.docStoreConn.indexExist(search.index_name(kb.tenant_id), dataset_id):
+    if not settings.docStoreConn.index_exist(search.index_name(kb.tenant_id), dataset_id):
         return get_result(data=obj)
-    sres = settings.retriever.search(req, search.index_name(kb.tenant_id), [dataset_id])
+    sres = await settings.retriever.search(req, search.index_name(kb.tenant_id), [dataset_id])
     if not len(sres.ids):
         return get_result(data=obj)
 

api/apps/sdk/dify_retrieval.py

@@ -15,14 +15,14 @@
 #
 import logging
 
-from quart import request, jsonify
+from quart import jsonify
 
 from api.db.services.document_service import DocumentService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMBundle
-from api.utils.api_utils import validate_request, build_error_result, apikey_required
+from common.metadata_utils import meta_filter, convert_conditions
+from api.utils.api_utils import apikey_required, build_error_result, get_request_json, validate_request
 from rag.app.tag import label_question
-from api.db.services.dialog_service import meta_filter, convert_conditions
 from common.constants import RetCode, LLMType
 from common import settings
 
@@ -113,7 +113,7 @@ async def retrieval(tenant_id):
       404:
         description: Knowledge base or document not found
     """
-    req = await request.json
+    req = await get_request_json()
     question = req["query"]
     kb_id = req["knowledge_id"]
     use_kg = req.get("use_kg", False)
@@ -135,7 +135,7 @@ async def retrieval(tenant_id):
             doc_ids.extend(meta_filter(metas, convert_conditions(metadata_condition), metadata_condition.get("logic", "and")))
         if not doc_ids and metadata_condition:
             doc_ids = ["-999"]
-        ranks = settings.retriever.retrieval(
+        ranks = await settings.retriever.retrieval(
             question,
             embd_mdl,
             kb.tenant_id,
@@ -150,7 +150,7 @@ async def retrieval(tenant_id):
         )
 
         if use_kg:
-            ck = settings.kg_retriever.retrieval(question,
+            ck = await settings.kg_retriever.retrieval(question,
                                                  [tenant_id],
                                                  [kb_id],
                                                  embd_mdl,

api/apps/sdk/doc.py

@@ -14,6 +14,7 @@
 #  limitations under the License.
 #
 import datetime
+import json
 import logging
 import pathlib
 import re
@@ -33,10 +34,10 @@ from api.db.services.file_service import FileService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMBundle
 from api.db.services.tenant_llm_service import TenantLLMService
-from api.db.services.task_service import TaskService, queue_tasks
-from api.db.services.dialog_service import meta_filter, convert_conditions
+from api.db.services.task_service import TaskService, queue_tasks, cancel_all_task_of
+from common.metadata_utils import meta_filter, convert_conditions
 from api.utils.api_utils import check_duplicate_ids, construct_json_result, get_error_data_result, get_parser_config, get_result, server_error_response, token_required, \
-    request_json
+    get_request_json
 from rag.app.qa import beAdoc, rmPrefix
 from rag.app.tag import label_question
 from rag.nlp import rag_tokenizer, search
@@ -231,12 +232,12 @@ async def update_doc(tenant_id, dataset_id, document_id):
         schema:
           type: object
     """
-    req = await request_json()
+    req = await get_request_json()
     if not KnowledgebaseService.query(id=dataset_id, tenant_id=tenant_id):
         return get_error_data_result(message="You don't own the dataset.")
     e, kb = KnowledgebaseService.get_by_id(dataset_id)
     if not e:
-        return get_error_data_result(message="Can't find this knowledgebase!")
+        return get_error_data_result(message="Can't find this dataset!")
     doc = DocumentService.query(kb_id=dataset_id, id=document_id)
     if not doc:
         return get_error_data_result(message="The dataset doesn't own the document.")
@@ -321,9 +322,7 @@ async def update_doc(tenant_id, dataset_id, document_id):
             try:
                 if not DocumentService.update_by_id(doc.id, {"status": str(status)}):
                     return get_error_data_result(message="Database error (Document update)!")
-
                 settings.docStoreConn.update({"doc_id": doc.id}, {"available_int": status}, search.index_name(kb.tenant_id), doc.kb_id)
-                return get_result(data=True)
             except Exception as e:
                 return server_error_response(e)
 
@@ -350,12 +349,10 @@ async def update_doc(tenant_id, dataset_id, document_id):
     }
     renamed_doc = {}
     for key, value in doc.to_dict().items():
-        if key == "run":
-            renamed_doc["run"] = run_mapping.get(str(value))
         new_key = key_mapping.get(key, key)
         renamed_doc[new_key] = value
         if key == "run":
-            renamed_doc["run"] = run_mapping.get(value)
+            renamed_doc["run"] = run_mapping.get(str(value))
 
     return get_result(data=renamed_doc)
 
@@ -536,7 +533,7 @@ def list_docs(dataset_id, tenant_id):
       return get_error_data_result(message=f"You don't own the dataset {dataset_id}. ")
 
     q = request.args
-    document_id = q.get("id")  
+    document_id = q.get("id")
     name        = q.get("name")
 
     if document_id and not DocumentService.query(id=document_id, kb_id=dataset_id):
@@ -545,23 +542,39 @@ def list_docs(dataset_id, tenant_id):
         return get_error_data_result(message=f"You don't own the document {name}.")
 
     page        = int(q.get("page", 1))
-    page_size   = int(q.get("page_size", 30))  
+    page_size   = int(q.get("page_size", 30))
     orderby     = q.get("orderby", "create_time")
     desc        = str(q.get("desc", "true")).strip().lower() != "false"
     keywords    = q.get("keywords", "")
 
     # filters - align with OpenAPI parameter names
-    suffix               = q.getlist("suffix") 
-    run_status           = q.getlist("run")   
-    create_time_from     = int(q.get("create_time_from", 0))  
-    create_time_to       = int(q.get("create_time_to", 0))    
+    suffix               = q.getlist("suffix")
+    run_status           = q.getlist("run")
+    create_time_from     = int(q.get("create_time_from", 0))
+    create_time_to       = int(q.get("create_time_to", 0))
+    metadata_condition_raw = q.get("metadata_condition")
+    metadata_condition = {}
+    if metadata_condition_raw:
+        try:
+            metadata_condition = json.loads(metadata_condition_raw)
+        except Exception:
+            return get_error_data_result(message="metadata_condition must be valid JSON.")
+    if metadata_condition and not isinstance(metadata_condition, dict):
+        return get_error_data_result(message="metadata_condition must be an object.")
 
-    # map run status (accept text or numeric) - align with API parameter
+    # map run status (text or numeric) - align with API parameter
     run_status_text_to_numeric = {"UNSTART": "0", "RUNNING": "1", "CANCEL": "2", "DONE": "3", "FAIL": "4"}
     run_status_converted = [run_status_text_to_numeric.get(v, v) for v in run_status]
 
+    doc_ids_filter = None
+    if metadata_condition:
+        metas = DocumentService.get_flatted_meta_by_kbs([dataset_id])
+        doc_ids_filter = meta_filter(metas, convert_conditions(metadata_condition), metadata_condition.get("logic", "and"))
+        if metadata_condition.get("conditions") and not doc_ids_filter:
+            return get_result(data={"total": 0, "docs": []})
+
     docs, total = DocumentService.get_list(
-        dataset_id, page, page_size, orderby, desc, keywords, document_id, name, suffix, run_status_converted
+        dataset_id, page, page_size, orderby, desc, keywords, document_id, name, suffix, run_status_converted, doc_ids_filter
     )
 
     # time range filter (0 means no bound)
@@ -575,7 +588,7 @@ def list_docs(dataset_id, tenant_id):
     # rename keys + map run status back to text for output
     key_mapping = {
         "chunk_num": "chunk_count",
-        "kb_id": "dataset_id", 
+        "kb_id": "dataset_id",
         "token_num": "token_count",
         "parser_id": "chunk_method",
     }
@@ -590,6 +603,70 @@ def list_docs(dataset_id, tenant_id):
 
     return get_result(data={"total": total, "docs": output_docs})
 
+
+@manager.route("/datasets/<dataset_id>/metadata/summary", methods=["GET"])  # noqa: F821
+@token_required
+async def metadata_summary(dataset_id, tenant_id):
+    if not KnowledgebaseService.accessible(kb_id=dataset_id, user_id=tenant_id):
+        return get_error_data_result(message=f"You don't own the dataset {dataset_id}. ")
+    req = await get_request_json()
+    try:
+        summary = DocumentService.get_metadata_summary(dataset_id, req.get("doc_ids"))
+        return get_result(data={"summary": summary})
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route("/datasets/<dataset_id>/metadata/update", methods=["POST"])  # noqa: F821
+@token_required
+async def metadata_batch_update(dataset_id, tenant_id):
+    if not KnowledgebaseService.accessible(kb_id=dataset_id, user_id=tenant_id):
+        return get_error_data_result(message=f"You don't own the dataset {dataset_id}. ")
+
+    req = await get_request_json()
+    selector = req.get("selector", {}) or {}
+    updates = req.get("updates", []) or []
+    deletes = req.get("deletes", []) or []
+
+    if not isinstance(selector, dict):
+        return get_error_data_result(message="selector must be an object.")
+    if not isinstance(updates, list) or not isinstance(deletes, list):
+        return get_error_data_result(message="updates and deletes must be lists.")
+
+    metadata_condition = selector.get("metadata_condition", {}) or {}
+    if metadata_condition and not isinstance(metadata_condition, dict):
+        return get_error_data_result(message="metadata_condition must be an object.")
+
+    document_ids = selector.get("document_ids", []) or []
+    if document_ids and not isinstance(document_ids, list):
+        return get_error_data_result(message="document_ids must be a list.")
+
+    for upd in updates:
+        if not isinstance(upd, dict) or not upd.get("key") or "value" not in upd:
+            return get_error_data_result(message="Each update requires key and value.")
+    for d in deletes:
+        if not isinstance(d, dict) or not d.get("key"):
+            return get_error_data_result(message="Each delete requires key.")
+   
+    if document_ids:
+        kb_doc_ids = KnowledgebaseService.list_documents_by_ids([dataset_id])
+        target_doc_ids = set(kb_doc_ids)
+        invalid_ids = set(document_ids) - set(kb_doc_ids)
+        if invalid_ids:
+            return get_error_data_result(message=f"These documents do not belong to dataset {dataset_id}: {', '.join(invalid_ids)}")
+        target_doc_ids = set(document_ids)
+
+    if metadata_condition:
+        metas = DocumentService.get_flatted_meta_by_kbs([dataset_id])
+        filtered_ids = set(meta_filter(metas, convert_conditions(metadata_condition), metadata_condition.get("logic", "and")))
+        target_doc_ids = target_doc_ids & filtered_ids
+        if metadata_condition.get("conditions") and not target_doc_ids:
+            return get_result(data={"updated": 0, "matched_docs": 0})
+
+    target_doc_ids = list(target_doc_ids)
+    updated = DocumentService.batch_update_metadata(dataset_id, target_doc_ids, updates, deletes)
+    return get_result(data={"updated": updated, "matched_docs": len(target_doc_ids)})
+
 @manager.route("/datasets/<dataset_id>/documents", methods=["DELETE"])  # noqa: F821
 @token_required
 async def delete(tenant_id, dataset_id):
@@ -631,7 +708,7 @@ async def delete(tenant_id, dataset_id):
     """
     if not KnowledgebaseService.accessible(kb_id=dataset_id, user_id=tenant_id):
         return get_error_data_result(message=f"You don't own the dataset {dataset_id}. ")
-    req = await request_json()
+    req = await get_request_json()
     if not req:
         doc_ids = None
     else:
@@ -741,7 +818,7 @@ async def parse(tenant_id, dataset_id):
     """
     if not KnowledgebaseService.accessible(kb_id=dataset_id, user_id=tenant_id):
         return get_error_data_result(message=f"You don't own the dataset {dataset_id}.")
-    req = await request_json()
+    req = await get_request_json()
     if not req.get("document_ids"):
         return get_error_data_result("`document_ids` is required")
     doc_list = req.get("document_ids")
@@ -824,7 +901,7 @@ async def stop_parsing(tenant_id, dataset_id):
     """
     if not KnowledgebaseService.accessible(kb_id=dataset_id, user_id=tenant_id):
         return get_error_data_result(message=f"You don't own the dataset {dataset_id}.")
-    req = await request_json()
+    req = await get_request_json()
 
     if not req.get("document_ids"):
         return get_error_data_result("`document_ids` is required")
@@ -839,6 +916,8 @@ async def stop_parsing(tenant_id, dataset_id):
             return get_error_data_result(message=f"You don't own the document {id}.")
         if int(doc[0].progress) == 1 or doc[0].progress == 0:
             return get_error_data_result("Can't stop parsing document with progress at 0 or 1")
+        # Send cancellation signal via Redis to stop background task
+        cancel_all_task_of(id)
         info = {"run": "2", "progress": 0, "chunk_num": 0}
         DocumentService.update_by_id(id, info)
         settings.docStoreConn.delete({"doc_id": doc[0].id}, search.index_name(tenant_id), dataset_id)
@@ -856,7 +935,7 @@ async def stop_parsing(tenant_id, dataset_id):
 
 @manager.route("/datasets/<dataset_id>/documents/<document_id>/chunks", methods=["GET"])  # noqa: F821
 @token_required
-def list_chunks(tenant_id, dataset_id, document_id):
+async def list_chunks(tenant_id, dataset_id, document_id):
     """
     List chunks of a document.
     ---
@@ -892,7 +971,7 @@ def list_chunks(tenant_id, dataset_id, document_id):
         type: string
         required: false
         default: ""
-        description: Chunk Id.
+        description: Chunk id.
       - in: header
         name: Authorization
         type: string
@@ -1001,8 +1080,8 @@ def list_chunks(tenant_id, dataset_id, document_id):
         res["chunks"].append(final_chunk)
         _ = Chunk(**final_chunk)
 
-    elif settings.docStoreConn.indexExist(search.index_name(tenant_id), dataset_id):
-        sres = settings.retriever.search(query, search.index_name(tenant_id), [dataset_id], emb_mdl=None, highlight=True)
+    elif settings.docStoreConn.index_exist(search.index_name(tenant_id), dataset_id):
+        sres = await settings.retriever.search(query, search.index_name(tenant_id), [dataset_id], emb_mdl=None, highlight=True)
         res["total"] = sres.total
         for id in sres.ids:
             d = {
@@ -1096,7 +1175,7 @@ async def add_chunk(tenant_id, dataset_id, document_id):
     if not doc:
         return get_error_data_result(message=f"You don't own the document {document_id}.")
     doc = doc[0]
-    req = await request_json()
+    req = await get_request_json()
     if not str(req.get("content", "")).strip():
         return get_error_data_result(message="`content` is required")
     if "important_keywords" in req:
@@ -1202,11 +1281,14 @@ async def rm_chunk(tenant_id, dataset_id, document_id):
     docs = DocumentService.get_by_ids([document_id])
     if not docs:
         raise LookupError(f"Can't find the document with ID {document_id}!")
-    req = await request_json()
+    req = await get_request_json()
     condition = {"doc_id": document_id}
     if "chunk_ids" in req:
         unique_chunk_ids, duplicate_messages = check_duplicate_ids(req["chunk_ids"], "chunk")
         condition["id"] = unique_chunk_ids
+    else:
+        unique_chunk_ids = []
+        duplicate_messages = []
     chunk_number = settings.docStoreConn.delete(condition, search.index_name(tenant_id), dataset_id)
     if chunk_number != 0:
         DocumentService.decrement_chunk_num(document_id, dataset_id, 1, chunk_number, 0)
@@ -1288,8 +1370,8 @@ async def update_chunk(tenant_id, dataset_id, document_id, chunk_id):
     if not doc:
         return get_error_data_result(message=f"You don't own the document {document_id}.")
     doc = doc[0]
-    req = await request_json()
-    if "content" in req:
+    req = await get_request_json()
+    if "content" in req and req["content"] is not None:
         content = req["content"]
     else:
         content = chunk.get("content_with_weight", "")
@@ -1411,7 +1493,7 @@ async def retrieval_test(tenant_id):
                     format: float
                     description: Similarity score.
     """
-    req = await request_json()
+    req = await get_request_json()
     if not req.get("dataset_ids"):
         return get_error_data_result("`dataset_ids` is required.")
     kb_ids = req["dataset_ids"]
@@ -1437,15 +1519,19 @@ async def retrieval_test(tenant_id):
     toc_enhance = req.get("toc_enhance", False)
     langs = req.get("cross_languages", [])
     if not isinstance(doc_ids, list):
-        return get_error_data_result("`documents` should be a list")
-    doc_ids_list = KnowledgebaseService.list_documents_by_ids(kb_ids)
-    for doc_id in doc_ids:
-        if doc_id not in doc_ids_list:
-            return get_error_data_result(f"The datasets don't own the document {doc_id}")
+        return get_error_data_result("`documents` should be a list")   
+    if doc_ids: 
+        doc_ids_list = KnowledgebaseService.list_documents_by_ids(kb_ids)
+        for doc_id in doc_ids:
+            if doc_id not in doc_ids_list:
+                return get_error_data_result(f"The datasets don't own the document {doc_id}")
     if not doc_ids:
         metadata_condition = req.get("metadata_condition", {}) or {}
         metas = DocumentService.get_meta_by_kbs(kb_ids)
         doc_ids = meta_filter(metas, convert_conditions(metadata_condition), metadata_condition.get("logic", "and"))
+        # If metadata_condition has conditions but no docs match, return empty result
+        if not doc_ids and metadata_condition.get("conditions"):
+            return get_result(data={"total": 0, "chunks": [], "doc_aggs": {}})
         if metadata_condition and not doc_ids:
             doc_ids = ["-999"]
     similarity_threshold = float(req.get("similarity_threshold", 0.2))
@@ -1467,13 +1553,13 @@ async def retrieval_test(tenant_id):
             rerank_mdl = LLMBundle(kb.tenant_id, LLMType.RERANK, llm_name=req["rerank_id"])
 
         if langs:
-            question = cross_languages(kb.tenant_id, None, question, langs)
+            question = await cross_languages(kb.tenant_id, None, question, langs)
 
         if req.get("keyword", False):
             chat_mdl = LLMBundle(kb.tenant_id, LLMType.CHAT)
-            question += keyword_extraction(chat_mdl, question)
+            question += await keyword_extraction(chat_mdl, question)
 
-        ranks = settings.retriever.retrieval(
+        ranks = await settings.retriever.retrieval(
             question,
             embd_mdl,
             tenant_ids,
@@ -1490,11 +1576,11 @@ async def retrieval_test(tenant_id):
         )
         if toc_enhance:
             chat_mdl = LLMBundle(kb.tenant_id, LLMType.CHAT)
-            cks = settings.retriever.retrieval_by_toc(question, ranks["chunks"], tenant_ids, chat_mdl, size)
+            cks = await settings.retriever.retrieval_by_toc(question, ranks["chunks"], tenant_ids, chat_mdl, size)
             if cks:
                 ranks["chunks"] = cks
         if use_kg:
-            ck = settings.kg_retriever.retrieval(question, [k.tenant_id for k in kbs], kb_ids, embd_mdl, LLMBundle(kb.tenant_id, LLMType.CHAT))
+            ck = await settings.kg_retriever.retrieval(question, [k.tenant_id for k in kbs], kb_ids, embd_mdl, LLMBundle(kb.tenant_id, LLMType.CHAT))
             if ck["content_with_weight"]:
                 ranks["chunks"].insert(0, ck)
 

api/apps/sdk/files.py

@@ -14,7 +14,6 @@
 #  limitations under the License.
 #
 
-
 import pathlib
 import re
 from quart import request, make_response
@@ -23,15 +22,15 @@ from pathlib import Path
 from api.db.services.document_service import DocumentService
 from api.db.services.file2document_service import File2DocumentService
 from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.utils.api_utils import server_error_response, token_required
-from common.misc_utils import get_uuid
+from api.utils.api_utils import get_json_result, get_request_json, server_error_response, token_required
+from common.misc_utils import get_uuid, thread_pool_exec
 from api.db import FileType
 from api.db.services import duplicate_name
 from api.db.services.file_service import FileService
-from api.utils.api_utils import get_json_result
 from api.utils.file_utils import filename_type
+from api.utils.web_utils import CONTENT_TYPE_MAP
 from common import settings
-
+from common.constants import RetCode
 
 @manager.route('/file/upload', methods=['POST'])  # noqa: F821
 @token_required
@@ -40,7 +39,7 @@ async def upload(tenant_id):
     Upload a file to the system.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     parameters:
@@ -86,19 +85,19 @@ async def upload(tenant_id):
         pf_id = root_folder["id"]
 
     if 'file' not in files:
-        return get_json_result(data=False, message='No file part!', code=400)
+        return get_json_result(data=False, message='No file part!', code=RetCode.BAD_REQUEST)
     file_objs = files.getlist('file')
 
     for file_obj in file_objs:
         if file_obj.filename == '':
-            return get_json_result(data=False, message='No selected file!', code=400)
+            return get_json_result(data=False, message='No selected file!', code=RetCode.BAD_REQUEST)
 
     file_res = []
 
     try:
         e, pf_folder = FileService.get_by_id(pf_id)
         if not e:
-            return get_json_result(data=False, message="Can't find this folder!", code=404)
+            return get_json_result(data=False, message="Can't find this folder!", code=RetCode.NOT_FOUND)
 
         for file_obj in file_objs:
             # Handle file path
@@ -114,13 +113,13 @@ async def upload(tenant_id):
             if file_len != len_id_list:
                 e, file = FileService.get_by_id(file_id_list[len_id_list - 1])
                 if not e:
-                    return get_json_result(data=False, message="Folder not found!", code=404)
+                    return get_json_result(data=False, message="Folder not found!", code=RetCode.NOT_FOUND)
                 last_folder = FileService.create_folder(file, file_id_list[len_id_list - 1], file_obj_names,
                                                         len_id_list)
             else:
                 e, file = FileService.get_by_id(file_id_list[len_id_list - 2])
                 if not e:
-                    return get_json_result(data=False, message="Folder not found!", code=404)
+                    return get_json_result(data=False, message="Folder not found!", code=RetCode.NOT_FOUND)
                 last_folder = FileService.create_folder(file, file_id_list[len_id_list - 2], file_obj_names,
                                                         len_id_list)
 
@@ -156,7 +155,7 @@ async def create(tenant_id):
     Create a new file or folder.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     parameters:
@@ -193,18 +192,19 @@ async def create(tenant_id):
                 type:
                   type: string
     """
-    req = await request.json
-    pf_id = await request.json.get("parent_id")
-    input_file_type = await request.json.get("type")
+    req = await get_request_json()
+    pf_id = req.get("parent_id")
+    input_file_type = req.get("type")
     if not pf_id:
         root_folder = FileService.get_root_folder(tenant_id)
         pf_id = root_folder["id"]
 
     try:
         if not FileService.is_parent_folder_exist(pf_id):
-            return get_json_result(data=False, message="Parent Folder Doesn't Exist!", code=400)
+            return get_json_result(data=False, message="Parent Folder Doesn't Exist!", code=RetCode.BAD_REQUEST)
         if FileService.query(name=req["name"], parent_id=pf_id):
-            return get_json_result(data=False, message="Duplicated folder name in the same folder.", code=409)
+            return get_json_result(data=False, message="Duplicated folder name in the same folder.",
+                                   code=RetCode.CONFLICT)
 
         if input_file_type == FileType.FOLDER.value:
             file_type = FileType.FOLDER.value
@@ -229,12 +229,12 @@ async def create(tenant_id):
 
 @manager.route('/file/list', methods=['GET'])  # noqa: F821
 @token_required
-def list_files(tenant_id):
+async def list_files(tenant_id):
     """
     List files under a specific folder.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     parameters:
@@ -306,13 +306,13 @@ def list_files(tenant_id):
     try:
         e, file = FileService.get_by_id(pf_id)
         if not e:
-            return get_json_result(message="Folder not found!", code=404)
+            return get_json_result(message="Folder not found!", code=RetCode.NOT_FOUND)
 
         files, total = FileService.get_by_pf_id(tenant_id, pf_id, page_number, items_per_page, orderby, desc, keywords)
 
         parent_folder = FileService.get_parent_folder(pf_id)
         if not parent_folder:
-            return get_json_result(message="File not found!", code=404)
+            return get_json_result(message="File not found!", code=RetCode.NOT_FOUND)
 
         return get_json_result(data={"total": total, "files": files, "parent_folder": parent_folder.to_json()})
     except Exception as e:
@@ -321,12 +321,12 @@ def list_files(tenant_id):
 
 @manager.route('/file/root_folder', methods=['GET'])  # noqa: F821
 @token_required
-def get_root_folder(tenant_id):
+async def get_root_folder(tenant_id):
     """
     Get user's root folder.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     responses:
@@ -357,12 +357,12 @@ def get_root_folder(tenant_id):
 
 @manager.route('/file/parent_folder', methods=['GET'])  # noqa: F821
 @token_required
-def get_parent_folder():
+async def get_parent_folder():
     """
     Get parent folder info of a file.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     parameters:
@@ -392,7 +392,7 @@ def get_parent_folder():
     try:
         e, file = FileService.get_by_id(file_id)
         if not e:
-            return get_json_result(message="Folder not found!", code=404)
+            return get_json_result(message="Folder not found!", code=RetCode.NOT_FOUND)
 
         parent_folder = FileService.get_parent_folder(file_id)
         return get_json_result(data={"parent_folder": parent_folder.to_json()})
@@ -402,12 +402,12 @@ def get_parent_folder():
 
 @manager.route('/file/all_parent_folder', methods=['GET'])  # noqa: F821
 @token_required
-def get_all_parent_folders(tenant_id):
+async def get_all_parent_folders(tenant_id):
     """
     Get all parent folders of a file.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     parameters:
@@ -439,7 +439,7 @@ def get_all_parent_folders(tenant_id):
     try:
         e, file = FileService.get_by_id(file_id)
         if not e:
-            return get_json_result(message="Folder not found!", code=404)
+            return get_json_result(message="Folder not found!", code=RetCode.NOT_FOUND)
 
         parent_folders = FileService.get_all_parent_folders(file_id)
         parent_folders_res = [folder.to_json() for folder in parent_folders]
@@ -455,7 +455,7 @@ async def rm(tenant_id):
     Delete one or multiple files/folders.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     parameters:
@@ -481,40 +481,40 @@ async def rm(tenant_id):
               type: boolean
               example: true
     """
-    req = await request.json
+    req = await get_request_json()
     file_ids = req["file_ids"]
     try:
         for file_id in file_ids:
             e, file = FileService.get_by_id(file_id)
             if not e:
-                return get_json_result(message="File or Folder not found!", code=404)
+                return get_json_result(message="File or Folder not found!", code=RetCode.NOT_FOUND)
             if not file.tenant_id:
-                return get_json_result(message="Tenant not found!", code=404)
+                return get_json_result(message="Tenant not found!", code=RetCode.NOT_FOUND)
 
             if file.type == FileType.FOLDER.value:
                 file_id_list = FileService.get_all_innermost_file_ids(file_id, [])
                 for inner_file_id in file_id_list:
                     e, file = FileService.get_by_id(inner_file_id)
                     if not e:
-                        return get_json_result(message="File not found!", code=404)
+                        return get_json_result(message="File not found!", code=RetCode.NOT_FOUND)
                     settings.STORAGE_IMPL.rm(file.parent_id, file.location)
                 FileService.delete_folder_by_pf_id(tenant_id, file_id)
             else:
                 settings.STORAGE_IMPL.rm(file.parent_id, file.location)
                 if not FileService.delete(file):
-                    return get_json_result(message="Database error (File removal)!", code=500)
+                    return get_json_result(message="Database error (File removal)!", code=RetCode.SERVER_ERROR)
 
             informs = File2DocumentService.get_by_file_id(file_id)
             for inform in informs:
                 doc_id = inform.document_id
                 e, doc = DocumentService.get_by_id(doc_id)
                 if not e:
-                    return get_json_result(message="Document not found!", code=404)
+                    return get_json_result(message="Document not found!", code=RetCode.NOT_FOUND)
                 tenant_id = DocumentService.get_tenant_id(doc_id)
                 if not tenant_id:
-                    return get_json_result(message="Tenant not found!", code=404)
+                    return get_json_result(message="Tenant not found!", code=RetCode.NOT_FOUND)
                 if not DocumentService.remove_document(doc, tenant_id):
-                    return get_json_result(message="Database error (Document removal)!", code=500)
+                    return get_json_result(message="Database error (Document removal)!", code=RetCode.SERVER_ERROR)
             File2DocumentService.delete_by_file_id(file_id)
 
         return get_json_result(data=True)
@@ -529,7 +529,7 @@ async def rename(tenant_id):
     Rename a file.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     parameters:
@@ -556,27 +556,29 @@ async def rename(tenant_id):
               type: boolean
               example: true
     """
-    req = await request.json
+    req = await get_request_json()
     try:
         e, file = FileService.get_by_id(req["file_id"])
         if not e:
-            return get_json_result(message="File not found!", code=404)
+            return get_json_result(message="File not found!", code=RetCode.NOT_FOUND)
 
         if file.type != FileType.FOLDER.value and pathlib.Path(req["name"].lower()).suffix != pathlib.Path(
                 file.name.lower()).suffix:
-            return get_json_result(data=False, message="The extension of file can't be changed", code=400)
+            return get_json_result(data=False, message="The extension of file can't be changed",
+                                   code=RetCode.BAD_REQUEST)
 
         for existing_file in FileService.query(name=req["name"], pf_id=file.parent_id):
             if existing_file.name == req["name"]:
-                return get_json_result(data=False, message="Duplicated file name in the same folder.", code=409)
+                return get_json_result(data=False, message="Duplicated file name in the same folder.",
+                                       code=RetCode.CONFLICT)
 
         if not FileService.update_by_id(req["file_id"], {"name": req["name"]}):
-            return get_json_result(message="Database error (File rename)!", code=500)
+            return get_json_result(message="Database error (File rename)!", code=RetCode.SERVER_ERROR)
 
         informs = File2DocumentService.get_by_file_id(req["file_id"])
         if informs:
             if not DocumentService.update_by_id(informs[0].document_id, {"name": req["name"]}):
-                return get_json_result(message="Database error (Document rename)!", code=500)
+                return get_json_result(message="Database error (Document rename)!", code=RetCode.SERVER_ERROR)
 
         return get_json_result(data=True)
     except Exception as e:
@@ -590,7 +592,7 @@ async def get(tenant_id, file_id):
     Download a file.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     produces:
@@ -606,13 +608,13 @@ async def get(tenant_id, file_id):
         description: File stream
         schema:
           type: file
-      404:
+      RetCode.NOT_FOUND:
         description: File not found
     """
     try:
         e, file = FileService.get_by_id(file_id)
         if not e:
-            return get_json_result(message="Document not found!", code=404)
+            return get_json_result(message="Document not found!", code=RetCode.NOT_FOUND)
 
         blob = settings.STORAGE_IMPL.get(file.parent_id, file.location)
         if not blob:
@@ -631,6 +633,21 @@ async def get(tenant_id, file_id):
         return server_error_response(e)
 
 
+@manager.route("/file/download/<attachment_id>", methods=["GET"])  # noqa: F821
+@token_required
+async def download_attachment(tenant_id, attachment_id):
+    try:
+        ext = request.args.get("ext", "markdown")
+        data = await thread_pool_exec(settings.STORAGE_IMPL.get, tenant_id, attachment_id)
+        response = await make_response(data)
+        response.headers.set("Content-Type", CONTENT_TYPE_MAP.get(ext, f"application/{ext}"))
+
+        return response
+
+    except Exception as e:
+        return server_error_response(e)
+
+
 @manager.route('/file/mv', methods=['POST'])  # noqa: F821
 @token_required
 async def move(tenant_id):
@@ -638,7 +655,7 @@ async def move(tenant_id):
     Move one or multiple files to another folder.
     ---
     tags:
-      - File Management
+      - File
     security:
       - ApiKeyAuth: []
     parameters:
@@ -667,7 +684,7 @@ async def move(tenant_id):
               type: boolean
               example: true
     """
-    req = await request.json
+    req = await get_request_json()
     try:
         file_ids = req["src_file_ids"]
         parent_id = req["dest_file_id"]
@@ -677,13 +694,13 @@ async def move(tenant_id):
         for file_id in file_ids:
             file = files_dict[file_id]
             if not file:
-                return get_json_result(message="File or Folder not found!", code=404)
+                return get_json_result(message="File or Folder not found!", code=RetCode.NOT_FOUND)
             if not file.tenant_id:
-                return get_json_result(message="Tenant not found!", code=404)
+                return get_json_result(message="Tenant not found!", code=RetCode.NOT_FOUND)
 
         fe, _ = FileService.get_by_id(parent_id)
         if not fe:
-            return get_json_result(message="Parent Folder not found!", code=404)
+            return get_json_result(message="Parent Folder not found!", code=RetCode.NOT_FOUND)
 
         FileService.move_file(file_ids, parent_id)
         return get_json_result(data=True)
@@ -694,7 +711,7 @@ async def move(tenant_id):
 @manager.route('/file/convert', methods=['POST'])  # noqa: F821
 @token_required
 async def convert(tenant_id):
-    req = await request.json
+    req = await get_request_json()
     kb_ids = req["kb_ids"]
     file_ids = req["file_ids"]
     file2documents = []
@@ -705,7 +722,7 @@ async def convert(tenant_id):
         for file_id in file_ids:
             file = files_set[file_id]
             if not file:
-                return get_json_result(message="File not found!", code=404)
+                return get_json_result(message="File not found!", code=RetCode.NOT_FOUND)
             file_ids_list = [file_id]
             if file.type == FileType.FOLDER.value:
                 file_ids_list = FileService.get_all_innermost_file_ids(file_id, [])
@@ -716,13 +733,13 @@ async def convert(tenant_id):
                     doc_id = inform.document_id
                     e, doc = DocumentService.get_by_id(doc_id)
                     if not e:
-                        return get_json_result(message="Document not found!", code=404)
+                        return get_json_result(message="Document not found!", code=RetCode.NOT_FOUND)
                     tenant_id = DocumentService.get_tenant_id(doc_id)
                     if not tenant_id:
-                        return get_json_result(message="Tenant not found!", code=404)
+                        return get_json_result(message="Tenant not found!", code=RetCode.NOT_FOUND)
                     if not DocumentService.remove_document(doc, tenant_id):
                         return get_json_result(
-                            message="Database error (Document removal)!", code=404)
+                            message="Database error (Document removal)!", code=RetCode.NOT_FOUND)
                 File2DocumentService.delete_by_file_id(id)
 
                 # insert
@@ -730,11 +747,11 @@ async def convert(tenant_id):
                     e, kb = KnowledgebaseService.get_by_id(kb_id)
                     if not e:
                         return get_json_result(
-                            message="Can't find this knowledgebase!", code=404)
+                            message="Can't find this dataset!", code=RetCode.NOT_FOUND)
                     e, file = FileService.get_by_id(id)
                     if not e:
                         return get_json_result(
-                            message="Can't find this file!", code=404)
+                            message="Can't find this file!", code=RetCode.NOT_FOUND)
 
                     doc = DocumentService.insert({
                         "id": get_uuid(),

api/apps/sdk/memories.py

@@ -0,0 +1,291 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import logging
+import os
+import time
+
+from quart import request
+from api.apps import login_required, current_user
+from api.db import TenantPermission
+from api.db.services.memory_service import MemoryService
+from api.db.services.user_service import UserTenantService
+from api.db.services.canvas_service import UserCanvasService
+from api.db.services.task_service import TaskService
+from api.db.joint_services.memory_message_service import get_memory_size_cache, judge_system_prompt_is_default
+from api.utils.api_utils import validate_request, get_request_json, get_error_argument_result, get_json_result
+from api.utils.memory_utils import format_ret_data_from_memory, get_memory_type_human
+from api.constants import MEMORY_NAME_LIMIT, MEMORY_SIZE_LIMIT
+from memory.services.messages import MessageService
+from memory.utils.prompt_util import PromptAssembler
+from common.constants import MemoryType, RetCode, ForgettingPolicy
+
+
+@manager.route("/memories", methods=["POST"])  # noqa: F821
+@login_required
+@validate_request("name", "memory_type", "embd_id", "llm_id")
+async def create_memory():
+    timing_enabled = os.getenv("RAGFLOW_API_TIMING")
+    t_start = time.perf_counter() if timing_enabled else None
+    req = await get_request_json()
+    t_parsed = time.perf_counter() if timing_enabled else None
+    # check name length
+    name = req["name"]
+    memory_name = name.strip()
+    if len(memory_name) == 0:
+        if timing_enabled:
+            logging.info(
+                "api_timing create_memory invalid_name parse_ms=%.2f total_ms=%.2f path=%s",
+                (t_parsed - t_start) * 1000,
+                (time.perf_counter() - t_start) * 1000,
+                request.path,
+            )
+        return get_error_argument_result("Memory name cannot be empty or whitespace.")
+    if len(memory_name) > MEMORY_NAME_LIMIT:
+        if timing_enabled:
+            logging.info(
+                "api_timing create_memory invalid_name parse_ms=%.2f total_ms=%.2f path=%s",
+                (t_parsed - t_start) * 1000,
+                (time.perf_counter() - t_start) * 1000,
+                request.path,
+            )
+        return get_error_argument_result(f"Memory name '{memory_name}' exceeds limit of {MEMORY_NAME_LIMIT}.")
+    # check memory_type valid
+    if not isinstance(req["memory_type"], list):
+        if timing_enabled:
+            logging.info(
+                "api_timing create_memory invalid_memory_type parse_ms=%.2f total_ms=%.2f path=%s",
+                (t_parsed - t_start) * 1000,
+                (time.perf_counter() - t_start) * 1000,
+                request.path,
+            )
+        return get_error_argument_result("Memory type must be a list.")
+    memory_type = set(req["memory_type"])
+    invalid_type = memory_type - {e.name.lower() for e in MemoryType}
+    if invalid_type:
+        if timing_enabled:
+            logging.info(
+                "api_timing create_memory invalid_memory_type parse_ms=%.2f total_ms=%.2f path=%s",
+                (t_parsed - t_start) * 1000,
+                (time.perf_counter() - t_start) * 1000,
+                request.path,
+            )
+        return get_error_argument_result(f"Memory type '{invalid_type}' is not supported.")
+    memory_type = list(memory_type)
+
+    try:
+        t_before_db = time.perf_counter() if timing_enabled else None
+        res, memory = MemoryService.create_memory(
+            tenant_id=current_user.id,
+            name=memory_name,
+            memory_type=memory_type,
+            embd_id=req["embd_id"],
+            llm_id=req["llm_id"]
+        )
+        if timing_enabled:
+            logging.info(
+                "api_timing create_memory parse_ms=%.2f validate_ms=%.2f db_ms=%.2f total_ms=%.2f path=%s",
+                (t_parsed - t_start) * 1000,
+                (t_before_db - t_parsed) * 1000,
+                (time.perf_counter() - t_before_db) * 1000,
+                (time.perf_counter() - t_start) * 1000,
+                request.path,
+            )
+
+        if res:
+            return get_json_result(message=True, data=format_ret_data_from_memory(memory))
+        else:
+            return get_json_result(message=memory, code=RetCode.SERVER_ERROR)
+
+    except Exception as e:
+        return get_json_result(message=str(e), code=RetCode.SERVER_ERROR)
+
+
+@manager.route("/memories/<memory_id>", methods=["PUT"])  # noqa: F821
+@login_required
+async def update_memory(memory_id):
+    req = await get_request_json()
+    update_dict = {}
+    # check name length
+    if "name" in req:
+        name = req["name"]
+        memory_name = name.strip()
+        if len(memory_name) == 0:
+            return get_error_argument_result("Memory name cannot be empty or whitespace.")
+        if len(memory_name) > MEMORY_NAME_LIMIT:
+            return get_error_argument_result(f"Memory name '{memory_name}' exceeds limit of {MEMORY_NAME_LIMIT}.")
+        update_dict["name"] = memory_name
+    # check permissions valid
+    if req.get("permissions"):
+        if req["permissions"] not in [e.value for e in TenantPermission]:
+            return get_error_argument_result(f"Unknown permission '{req['permissions']}'.")
+        update_dict["permissions"] = req["permissions"]
+    if req.get("llm_id"):
+        update_dict["llm_id"] = req["llm_id"]
+    if req.get("embd_id"):
+        update_dict["embd_id"] = req["embd_id"]
+    if req.get("memory_type"):
+        memory_type = set(req["memory_type"])
+        invalid_type = memory_type - {e.name.lower() for e in MemoryType}
+        if invalid_type:
+            return get_error_argument_result(f"Memory type '{invalid_type}' is not supported.")
+        update_dict["memory_type"] = list(memory_type)
+    # check memory_size valid
+    if req.get("memory_size"):
+        if not 0 < int(req["memory_size"]) <= MEMORY_SIZE_LIMIT:
+            return get_error_argument_result(f"Memory size should be in range (0, {MEMORY_SIZE_LIMIT}] Bytes.")
+        update_dict["memory_size"] = req["memory_size"]
+    # check forgetting_policy valid
+    if req.get("forgetting_policy"):
+        if req["forgetting_policy"] not in [e.value for e in ForgettingPolicy]:
+            return get_error_argument_result(f"Forgetting policy '{req['forgetting_policy']}' is not supported.")
+        update_dict["forgetting_policy"] = req["forgetting_policy"]
+    # check temperature valid
+    if "temperature" in req:
+        temperature = float(req["temperature"])
+        if not 0 <= temperature <= 1:
+            return get_error_argument_result("Temperature should be in range [0, 1].")
+        update_dict["temperature"] = temperature
+    # allow update to empty fields
+    for field in ["avatar", "description", "system_prompt", "user_prompt"]:
+        if field in req:
+            update_dict[field] = req[field]
+    current_memory = MemoryService.get_by_memory_id(memory_id)
+    if not current_memory:
+        return get_json_result(code=RetCode.NOT_FOUND, message=f"Memory '{memory_id}' not found.")
+
+    memory_dict = current_memory.to_dict()
+    memory_dict.update({"memory_type": get_memory_type_human(current_memory.memory_type)})
+    to_update = {}
+    for k, v in update_dict.items():
+        if isinstance(v, list) and set(memory_dict[k]) != set(v):
+            to_update[k] = v
+        elif memory_dict[k] != v:
+            to_update[k] = v
+
+    if not to_update:
+        return get_json_result(message=True, data=memory_dict)
+    # check memory empty when update embd_id, memory_type
+    memory_size = get_memory_size_cache(memory_id, current_memory.tenant_id)
+    not_allowed_update = [f for f in ["embd_id", "memory_type"] if f in to_update and memory_size > 0]
+    if not_allowed_update:
+        return get_error_argument_result(f"Can't update {not_allowed_update} when memory isn't empty.")
+    if "memory_type" in to_update:
+        if "system_prompt" not in to_update and judge_system_prompt_is_default(current_memory.system_prompt, current_memory.memory_type):
+            # update old default prompt, assemble a new one
+            to_update["system_prompt"] = PromptAssembler.assemble_system_prompt({"memory_type": to_update["memory_type"]})
+
+    try:
+        MemoryService.update_memory(current_memory.tenant_id, memory_id, to_update)
+        updated_memory = MemoryService.get_by_memory_id(memory_id)
+        return get_json_result(message=True, data=format_ret_data_from_memory(updated_memory))
+
+    except Exception as e:
+        logging.error(e)
+        return get_json_result(message=str(e), code=RetCode.SERVER_ERROR)
+
+
+@manager.route("/memories/<memory_id>", methods=["DELETE"])  # noqa: F821
+@login_required
+async def delete_memory(memory_id):
+    memory = MemoryService.get_by_memory_id(memory_id)
+    if not memory:
+        return get_json_result(message=True, code=RetCode.NOT_FOUND)
+    try:
+        MemoryService.delete_memory(memory_id)
+        if MessageService.has_index(memory.tenant_id, memory_id):
+            MessageService.delete_message({"memory_id": memory_id}, memory.tenant_id, memory_id)
+        return get_json_result(message=True)
+    except Exception as e:
+        logging.error(e)
+        return get_json_result(message=str(e), code=RetCode.SERVER_ERROR)
+
+
+@manager.route("/memories", methods=["GET"])  # noqa: F821
+@login_required
+async def list_memory():
+    args = request.args
+    try:
+        tenant_ids = args.getlist("tenant_id")
+        memory_types = args.getlist("memory_type")
+        storage_type = args.get("storage_type")
+        keywords = args.get("keywords", "")
+        page = int(args.get("page", 1))
+        page_size = int(args.get("page_size", 50))
+        # make filter dict
+        filter_dict: dict = {"storage_type": storage_type}
+        if not tenant_ids:
+            # restrict to current user's tenants
+            user_tenants = UserTenantService.get_user_tenant_relation_by_user_id(current_user.id)
+            filter_dict["tenant_id"] = [tenant["tenant_id"] for tenant in user_tenants]
+        else:
+            if len(tenant_ids) == 1 and ',' in tenant_ids[0]:
+                tenant_ids = tenant_ids[0].split(',')
+            filter_dict["tenant_id"] = tenant_ids
+        if memory_types and len(memory_types) == 1 and ',' in memory_types[0]:
+            memory_types = memory_types[0].split(',')
+        filter_dict["memory_type"] = memory_types
+
+        memory_list, count = MemoryService.get_by_filter(filter_dict, keywords, page, page_size)
+        [memory.update({"memory_type": get_memory_type_human(memory["memory_type"])}) for memory in memory_list]
+        return get_json_result(message=True, data={"memory_list": memory_list, "total_count": count})
+
+    except Exception as e:
+        logging.error(e)
+        return get_json_result(message=str(e), code=RetCode.SERVER_ERROR)
+
+
+@manager.route("/memories/<memory_id>/config", methods=["GET"])  # noqa: F821
+@login_required
+async def get_memory_config(memory_id):
+    memory = MemoryService.get_with_owner_name_by_id(memory_id)
+    if not memory:
+        return get_json_result(code=RetCode.NOT_FOUND, message=f"Memory '{memory_id}' not found.")
+    return get_json_result(message=True, data=format_ret_data_from_memory(memory))
+
+
+@manager.route("/memories/<memory_id>", methods=["GET"])  # noqa: F821
+@login_required
+async def get_memory_detail(memory_id):
+    args = request.args
+    agent_ids = args.getlist("agent_id")
+    if len(agent_ids) == 1 and ',' in agent_ids[0]:
+        agent_ids = agent_ids[0].split(',')
+    keywords = args.get("keywords", "")
+    keywords = keywords.strip()
+    page = int(args.get("page", 1))
+    page_size = int(args.get("page_size", 50))
+    memory = MemoryService.get_by_memory_id(memory_id)
+    if not memory:
+        return get_json_result(code=RetCode.NOT_FOUND, message=f"Memory '{memory_id}' not found.")
+    messages = MessageService.list_message(
+        memory.tenant_id, memory_id, agent_ids, keywords, page, page_size)
+    agent_name_mapping = {}
+    extract_task_mapping = {}
+    if messages["message_list"]:
+        agent_list = UserCanvasService.get_basic_info_by_canvas_ids([message["agent_id"] for message in messages["message_list"]])
+        agent_name_mapping = {agent["id"]: agent["title"] for agent in agent_list}
+        task_list = TaskService.get_tasks_progress_by_doc_ids([memory_id])
+        if task_list:
+            task_list.sort(key=lambda t: t["create_time"]) # asc, use newer when exist more than one task
+            for task in task_list:
+                # the 'digest' field carries the source_id when a task is created, so use 'digest' as key
+                extract_task_mapping.update({int(task["digest"]): task})
+    for message in messages["message_list"]:
+        message["agent_name"] = agent_name_mapping.get(message["agent_id"], "Unknown")
+        message["task"] = extract_task_mapping.get(message["message_id"], {})
+        for extract_msg in message["extract"]:
+            extract_msg["agent_name"] = agent_name_mapping.get(extract_msg["agent_id"], "Unknown")
+    return get_json_result(data={"messages": messages, "storage_type": memory.storage_type}, message=True)

api/apps/sdk/messages.py

@@ -0,0 +1,158 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+from quart import request
+from api.apps import login_required
+from api.db.services.memory_service import MemoryService
+from common.time_utils import current_timestamp, timestamp_to_date
+
+from memory.services.messages import MessageService
+from api.db.joint_services import memory_message_service
+from api.utils.api_utils import validate_request, get_request_json, get_error_argument_result, get_json_result
+from common.constants import RetCode
+
+
+@manager.route("/messages", methods=["POST"]) # noqa: F821
+@login_required
+@validate_request("memory_id", "agent_id", "session_id", "user_input", "agent_response")
+async def add_message():
+
+    req = await get_request_json()
+    memory_ids = req["memory_id"]
+
+    message_dict = {
+        "user_id": req.get("user_id"),
+        "agent_id": req["agent_id"],
+        "session_id": req["session_id"],
+        "user_input": req["user_input"],
+        "agent_response": req["agent_response"],
+    }
+
+    res, msg = await memory_message_service.queue_save_to_memory_task(memory_ids, message_dict)
+
+    if res:
+        return get_json_result(message=msg)
+
+    return get_json_result(code=RetCode.SERVER_ERROR, message="Some messages failed to add. Detail:" + msg)
+
+
+@manager.route("/messages/<memory_id>:<message_id>", methods=["DELETE"]) # noqa: F821
+@login_required
+async def forget_message(memory_id: str, message_id: int):
+
+    memory = MemoryService.get_by_memory_id(memory_id)
+    if not memory:
+        return get_json_result(code=RetCode.NOT_FOUND, message=f"Memory '{memory_id}' not found.")
+
+    forget_time = timestamp_to_date(current_timestamp())
+    update_succeed = MessageService.update_message(
+        {"memory_id": memory_id, "message_id": int(message_id)},
+        {"forget_at": forget_time},
+        memory.tenant_id, memory_id)
+    if update_succeed:
+        return get_json_result(message=update_succeed)
+    else:
+        return get_json_result(code=RetCode.SERVER_ERROR, message=f"Failed to forget message '{message_id}' in memory '{memory_id}'.")
+
+
+@manager.route("/messages/<memory_id>:<message_id>", methods=["PUT"]) # noqa: F821
+@login_required
+@validate_request("status")
+async def update_message(memory_id: str, message_id: int):
+    req = await get_request_json()
+    status = req["status"]
+    if not isinstance(status, bool):
+        return get_error_argument_result("Status must be a boolean.")
+
+    memory = MemoryService.get_by_memory_id(memory_id)
+    if not memory:
+        return get_json_result(code=RetCode.NOT_FOUND, message=f"Memory '{memory_id}' not found.")
+
+    update_succeed = MessageService.update_message({"memory_id": memory_id, "message_id": int(message_id)}, {"status": status}, memory.tenant_id, memory_id)
+    if update_succeed:
+        return get_json_result(message=update_succeed)
+    else:
+        return get_json_result(code=RetCode.SERVER_ERROR, message=f"Failed to set status for message '{message_id}' in memory '{memory_id}'.")
+
+
+@manager.route("/messages/search", methods=["GET"]) # noqa: F821
+@login_required
+async def search_message():
+    args = request.args
+    empty_fields = [f for f in ["memory_id", "query"] if not args.get(f)]
+    if empty_fields:
+        return get_error_argument_result(f"{', '.join(empty_fields)} can't be empty.")
+
+    memory_ids = args.getlist("memory_id")
+    if len(memory_ids) == 1 and ',' in memory_ids[0]:
+        memory_ids = memory_ids[0].split(',')
+    query = args.get("query")
+    similarity_threshold = float(args.get("similarity_threshold", 0.2))
+    keywords_similarity_weight = float(args.get("keywords_similarity_weight", 0.7))
+    top_n = int(args.get("top_n", 5))
+    agent_id = args.get("agent_id", "")
+    session_id = args.get("session_id", "")
+
+    filter_dict = {
+        "memory_id": memory_ids,
+        "agent_id": agent_id,
+        "session_id": session_id
+    }
+    params = {
+        "query": query,
+        "similarity_threshold": similarity_threshold,
+        "keywords_similarity_weight": keywords_similarity_weight,
+        "top_n": top_n
+    }
+    res = memory_message_service.query_message(filter_dict, params)
+    return get_json_result(message=True, data=res)
+
+
+@manager.route("/messages", methods=["GET"]) # noqa: F821
+@login_required
+async def get_messages():
+    args = request.args
+    memory_ids = args.getlist("memory_id")
+    if len(memory_ids) == 1 and ',' in memory_ids[0]:
+        memory_ids = memory_ids[0].split(',')
+    agent_id = args.get("agent_id", "")
+    session_id = args.get("session_id", "")
+    limit = int(args.get("limit", 10))
+    if not memory_ids:
+        return get_error_argument_result("memory_ids is required.")
+    memory_list = MemoryService.get_by_ids(memory_ids)
+    uids = [memory.tenant_id for memory in memory_list]
+    res = MessageService.get_recent_messages(
+        uids,
+        memory_ids,
+        agent_id,
+        session_id,
+        limit
+    )
+    return get_json_result(message=True, data=res)
+
+
+@manager.route("/messages/<memory_id>:<message_id>/content", methods=["GET"]) # noqa: F821
+@login_required
+async def get_message_content(memory_id:str, message_id: int):
+    memory = MemoryService.get_by_memory_id(memory_id)
+    if not memory:
+        return get_json_result(code=RetCode.NOT_FOUND, message=f"Memory '{memory_id}' not found.")
+
+    res = MessageService.get_by_message_id(memory_id, message_id, memory.tenant_id)
+    if res:
+        return get_json_result(message=True, data=res)
+    else:
+        return get_json_result(code=RetCode.NOT_FOUND, message=f"Message '{message_id}' in memory '{memory_id}' not found.")

api/apps/sdk/session.py

@@ -14,10 +14,15 @@
 #  limitations under the License.
 #
 import json
+import copy
 import re
 import time
 
 import tiktoken
+import os
+import tempfile
+import logging
+
 from quart import Response, jsonify, request
 
 from agent.canvas import Canvas
@@ -25,27 +30,30 @@ from api.db.db_models import APIToken
 from api.db.services.api_service import API4ConversationService
 from api.db.services.canvas_service import UserCanvasService, completion_openai
 from api.db.services.canvas_service import completion as agent_completion
-from api.db.services.conversation_service import ConversationService, iframe_completion
-from api.db.services.conversation_service import completion as rag_completion
-from api.db.services.dialog_service import DialogService, ask, chat, gen_mindmap, meta_filter
+from api.db.services.conversation_service import ConversationService
+from api.db.services.conversation_service import async_iframe_completion as iframe_completion
+from api.db.services.conversation_service import async_completion as rag_completion
+from api.db.services.dialog_service import DialogService, async_ask, async_chat, gen_mindmap
 from api.db.services.document_service import DocumentService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMBundle
+from common.metadata_utils import apply_meta_data_filter, convert_conditions, meta_filter
 from api.db.services.search_service import SearchService
-from api.db.services.user_service import UserTenantService
+from api.db.services.user_service import TenantService,UserTenantService
 from common.misc_utils import get_uuid
 from api.utils.api_utils import check_duplicate_ids, get_data_openai, get_error_data_result, get_json_result, \
-    get_result, server_error_response, token_required, validate_request
+    get_result, get_request_json, server_error_response, token_required, validate_request
 from rag.app.tag import label_question
 from rag.prompts.template import load_prompt
-from rag.prompts.generator import cross_languages, gen_meta_filter, keyword_extraction, chunks_format
+from rag.prompts.generator import cross_languages, keyword_extraction, chunks_format
 from common.constants import RetCode, LLMType, StatusEnum
 from common import settings
 
+
 @manager.route("/chats/<chat_id>/sessions", methods=["POST"])  # noqa: F821
 @token_required
 async def create(tenant_id, chat_id):
-    req = await request.json
+    req = await get_request_json()
     req["dialog_id"] = chat_id
     dia = DialogService.query(tenant_id=tenant_id, id=req["dialog_id"], status=StatusEnum.VALID.value)
     if not dia:
@@ -56,7 +64,7 @@ async def create(tenant_id, chat_id):
         "name": req.get("name", "New session"),
         "message": [{"role": "assistant", "content": dia[0].prompt_config.get("prologue")}],
         "user_id": req.get("user_id", ""),
-        "reference": [{}],
+        "reference": [],
     }
     if not conv.get("name"):
         return get_error_data_result(message="`name` can not be empty.")
@@ -73,7 +81,7 @@ async def create(tenant_id, chat_id):
 
 @manager.route("/agents/<agent_id>/sessions", methods=["POST"])  # noqa: F821
 @token_required
-def create_agent_session(tenant_id, agent_id):
+async def create_agent_session(tenant_id, agent_id):
     user_id = request.args.get("user_id", tenant_id)
     e, cvs = UserCanvasService.get_by_id(agent_id)
     if not e:
@@ -84,7 +92,7 @@ def create_agent_session(tenant_id, agent_id):
         cvs.dsl = json.dumps(cvs.dsl, ensure_ascii=False)
 
     session_id = get_uuid()
-    canvas = Canvas(cvs.dsl, tenant_id, agent_id)
+    canvas = Canvas(cvs.dsl, tenant_id, agent_id, canvas_id=cvs.id)
     canvas.reset()
 
     cvs.dsl = json.loads(str(canvas))
@@ -98,7 +106,7 @@ def create_agent_session(tenant_id, agent_id):
 @manager.route("/chats/<chat_id>/sessions/<session_id>", methods=["PUT"])  # noqa: F821
 @token_required
 async def update(tenant_id, chat_id, session_id):
-    req = await request.json
+    req = await get_request_json()
     req["dialog_id"] = chat_id
     conv_id = session_id
     conv = ConversationService.query(id=conv_id, dialog_id=chat_id)
@@ -120,16 +128,38 @@ async def update(tenant_id, chat_id, session_id):
 @manager.route("/chats/<chat_id>/completions", methods=["POST"])  # noqa: F821
 @token_required
 async def chat_completion(tenant_id, chat_id):
-    req = await request.json
+    req = await get_request_json()
     if not req:
         req = {"question": ""}
     if not req.get("session_id"):
         req["question"] = ""
-    if not DialogService.query(tenant_id=tenant_id, id=chat_id, status=StatusEnum.VALID.value):
+    dia = DialogService.query(tenant_id=tenant_id, id=chat_id, status=StatusEnum.VALID.value)
+    if not dia:
         return get_error_data_result(f"You don't own the chat {chat_id}")
+    dia = dia[0]
     if req.get("session_id"):
         if not ConversationService.query(id=req["session_id"], dialog_id=chat_id):
             return get_error_data_result(f"You don't own the session {req['session_id']}")
+
+    metadata_condition = req.get("metadata_condition") or {}
+    if metadata_condition and not isinstance(metadata_condition, dict):
+        return get_error_data_result(message="metadata_condition must be an object.")
+
+    if metadata_condition and req.get("question"):
+        metas = DocumentService.get_meta_by_kbs(dia.kb_ids or [])
+        filtered_doc_ids = meta_filter(
+            metas,
+            convert_conditions(metadata_condition),
+            metadata_condition.get("logic", "and"),
+        )
+        if metadata_condition.get("conditions") and not filtered_doc_ids:
+            filtered_doc_ids = ["-999"]
+
+        if filtered_doc_ids:
+            req["doc_ids"] = ",".join(filtered_doc_ids)
+        else:
+            req.pop("doc_ids", None)
+
     if req.get("stream", True):
         resp = Response(rag_completion(tenant_id, chat_id, **req), mimetype="text/event-stream")
         resp.headers.add_header("Cache-control", "no-cache")
@@ -140,7 +170,7 @@ async def chat_completion(tenant_id, chat_id):
         return resp
     else:
         answer = None
-        for ans in rag_completion(tenant_id, chat_id, **req):
+        async for ans in rag_completion(tenant_id, chat_id, **req):
             answer = ans
             break
         return get_result(data=answer)
@@ -192,7 +222,19 @@ async def chat_completion_openai_like(tenant_id, chat_id):
             {"role": "user", "content": "Can you tell me how to install neovim"},
         ],
         stream=stream,
-        extra_body={"reference": reference}
+        extra_body={
+            "reference": reference,
+            "metadata_condition": {
+                "logic": "and",
+                "conditions": [
+                    {
+                        "name": "author",
+                        "comparison_operator": "is",
+                        "value": "bob"
+                    }
+                ]
+            }
+        }
     )
 
     if stream:
@@ -206,9 +248,13 @@ async def chat_completion_openai_like(tenant_id, chat_id):
         if reference:
             print(completion.choices[0].message.reference)
     """
-    req = await request.get_json()
+    req = await get_request_json()
 
-    need_reference = bool(req.get("reference", False))
+    extra_body = req.get("extra_body") or {}
+    if extra_body and not isinstance(extra_body, dict):
+        return get_error_data_result("extra_body must be an object.")
+
+    need_reference = bool(extra_body.get("reference", False))
 
     messages = req.get("messages", [])
     # To prevent empty [] input
@@ -226,6 +272,22 @@ async def chat_completion_openai_like(tenant_id, chat_id):
         return get_error_data_result(f"You don't own the chat {chat_id}")
     dia = dia[0]
 
+    metadata_condition = extra_body.get("metadata_condition") or {}
+    if metadata_condition and not isinstance(metadata_condition, dict):
+        return get_error_data_result(message="metadata_condition must be an object.")
+
+    doc_ids_str = None
+    if metadata_condition:
+        metas = DocumentService.get_meta_by_kbs(dia.kb_ids or [])
+        filtered_doc_ids = meta_filter(
+            metas,
+            convert_conditions(metadata_condition),
+            metadata_condition.get("logic", "and"),
+        )
+        if metadata_condition.get("conditions") and not filtered_doc_ids:
+            filtered_doc_ids = ["-999"]
+        doc_ids_str = ",".join(filtered_doc_ids) if filtered_doc_ids else None
+
     # Filter system and non-sense assistant messages
     msg = []
     for m in messages:
@@ -244,11 +306,14 @@ async def chat_completion_openai_like(tenant_id, chat_id):
         # The value for the usage field on all chunks except for the last one will be null.
         # The usage field on the last chunk contains token usage statistics for the entire request.
         # The choices field on the last chunk will always be an empty array [].
-        def streamed_response_generator(chat_id, dia, msg):
+        async def streamed_response_generator(chat_id, dia, msg):
             token_used = 0
-            answer_cache = ""
-            reasoning_cache = ""
             last_ans = {}
+            full_content = ""
+            full_reasoning = ""
+            final_answer = None
+            final_reference = None
+            in_think = False
             response = {
                 "id": f"chatcmpl-{chat_id}",
                 "choices": [
@@ -273,49 +338,35 @@ async def chat_completion_openai_like(tenant_id, chat_id):
             }
 
             try:
-                for ans in chat(dia, msg, True, toolcall_session=toolcall_session, tools=tools, quote=need_reference):
+                chat_kwargs = {"toolcall_session": toolcall_session, "tools": tools, "quote": need_reference}
+                if doc_ids_str:
+                    chat_kwargs["doc_ids"] = doc_ids_str
+                async for ans in async_chat(dia, msg, True, **chat_kwargs):
                     last_ans = ans
-                    answer = ans["answer"]
-
-                    reasoning_match = re.search(r"<think>(.*?)</think>", answer, flags=re.DOTALL)
-                    if reasoning_match:
-                        reasoning_part = reasoning_match.group(1)
-                        content_part = answer[reasoning_match.end():]
-                    else:
-                        reasoning_part = ""
-                        content_part = answer
-
-                    reasoning_incremental = ""
-                    if reasoning_part:
-                        if reasoning_part.startswith(reasoning_cache):
-                            reasoning_incremental = reasoning_part.replace(reasoning_cache, "", 1)
-                        else:
-                            reasoning_incremental = reasoning_part
-                        reasoning_cache = reasoning_part
-
-                    content_incremental = ""
-                    if content_part:
-                        if content_part.startswith(answer_cache):
-                            content_incremental = content_part.replace(answer_cache, "", 1)
-                    else:
-                        content_incremental = content_part
-                    answer_cache = content_part
-
-                    token_used += len(reasoning_incremental) + len(content_incremental)
-
-                    if not any([reasoning_incremental, content_incremental]):
+                    if ans.get("final"):
+                        if ans.get("answer"):
+                            full_content = ans["answer"]
+                        final_answer = ans.get("answer") or full_content
+                        final_reference = ans.get("reference", {})
                         continue
-
-                    if reasoning_incremental:
-                        response["choices"][0]["delta"]["reasoning_content"] = reasoning_incremental
-                    else:
-                        response["choices"][0]["delta"]["reasoning_content"] = None
-
-                    if content_incremental:
-                        response["choices"][0]["delta"]["content"] = content_incremental
-                    else:
+                    if ans.get("start_to_think"):
+                        in_think = True
+                        continue
+                    if ans.get("end_to_think"):
+                        in_think = False
+                        continue
+                    delta = ans.get("answer") or ""
+                    if not delta:
+                        continue
+                    token_used += len(delta)
+                    if in_think:
+                        full_reasoning += delta
+                        response["choices"][0]["delta"]["reasoning_content"] = delta
                         response["choices"][0]["delta"]["content"] = None
-
+                    else:
+                        full_content += delta
+                        response["choices"][0]["delta"]["content"] = delta
+                        response["choices"][0]["delta"]["reasoning_content"] = None
                     yield f"data:{json.dumps(response, ensure_ascii=False)}\n\n"
             except Exception as e:
                 response["choices"][0]["delta"]["content"] = "**ERROR**: " + str(e)
@@ -325,11 +376,11 @@ async def chat_completion_openai_like(tenant_id, chat_id):
             response["choices"][0]["delta"]["content"] = None
             response["choices"][0]["delta"]["reasoning_content"] = None
             response["choices"][0]["finish_reason"] = "stop"
-            response["usage"] = {"prompt_tokens": len(prompt), "completion_tokens": token_used,
-                                 "total_tokens": len(prompt) + token_used}
+            response["usage"] = {"prompt_tokens": len(prompt), "completion_tokens": token_used, "total_tokens": len(prompt) + token_used}
             if need_reference:
-                response["choices"][0]["delta"]["reference"] = chunks_format(last_ans.get("reference", []))
-                response["choices"][0]["delta"]["final_content"] = last_ans.get("answer", "")
+                reference_payload = final_reference if final_reference is not None else last_ans.get("reference", [])
+                response["choices"][0]["delta"]["reference"] = chunks_format(reference_payload)
+                response["choices"][0]["delta"]["final_content"] = final_answer if final_answer is not None else full_content
             yield f"data:{json.dumps(response, ensure_ascii=False)}\n\n"
             yield "data:[DONE]\n\n"
 
@@ -341,7 +392,10 @@ async def chat_completion_openai_like(tenant_id, chat_id):
         return resp
     else:
         answer = None
-        for ans in chat(dia, msg, False, toolcall_session=toolcall_session, tools=tools, quote=need_reference):
+        chat_kwargs = {"toolcall_session": toolcall_session, "tools": tools, "quote": need_reference}
+        if doc_ids_str:
+            chat_kwargs["doc_ids"] = doc_ids_str
+        async for ans in async_chat(dia, msg, False, **chat_kwargs):
             # focus answer content only
             answer = ans
             break
@@ -375,7 +429,7 @@ async def chat_completion_openai_like(tenant_id, chat_id):
             ],
         }
         if need_reference:
-            response["choices"][0]["message"]["reference"] = chunks_format(answer.get("reference", []))
+            response["choices"][0]["message"]["reference"] = chunks_format(answer.get("reference", {}))
 
         return jsonify(response)
 
@@ -384,8 +438,8 @@ async def chat_completion_openai_like(tenant_id, chat_id):
 @validate_request("model", "messages")  # noqa: F821
 @token_required
 async def agents_completion_openai_compatibility(tenant_id, agent_id):
-    req = await request.json
-    tiktokenenc = tiktoken.get_encoding("cl100k_base")
+    req = await get_request_json()
+    tiktoken_encode = tiktoken.get_encoding("cl100k_base")
     messages = req.get("messages", [])
     if not messages:
         return get_error_data_result("You must provide at least one message.")
@@ -393,7 +447,7 @@ async def agents_completion_openai_compatibility(tenant_id, agent_id):
         return get_error_data_result(f"You don't own the agent {agent_id}")
 
     filtered_messages = [m for m in messages if m["role"] in ["user", "assistant"]]
-    prompt_tokens = sum(len(tiktokenenc.encode(m["content"])) for m in filtered_messages)
+    prompt_tokens = sum(len(tiktoken_encode.encode(m["content"])) for m in filtered_messages)
     if not filtered_messages:
         return jsonify(
             get_data_openai(
@@ -401,7 +455,7 @@ async def agents_completion_openai_compatibility(tenant_id, agent_id):
                 content="No valid messages found (user or assistant).",
                 finish_reason="stop",
                 model=req.get("model", ""),
-                completion_tokens=len(tiktokenenc.encode("No valid messages found (user or assistant).")),
+                completion_tokens=len(tiktoken_encode.encode("No valid messages found (user or assistant).")),
                 prompt_tokens=prompt_tokens,
             )
         )
@@ -438,15 +492,19 @@ async def agents_completion_openai_compatibility(tenant_id, agent_id):
             ):
             return jsonify(response)
 
+        return None
+
 
 @manager.route("/agents/<agent_id>/completions", methods=["POST"])  # noqa: F821
 @token_required
 async def agent_completions(tenant_id, agent_id):
-    req = await request.json
+    req = await get_request_json()
+    return_trace = bool(req.get("return_trace", False))
 
     if req.get("stream", True):
 
         async def generate():
+            trace_items = []
             async for answer in agent_completion(tenant_id=tenant_id, agent_id=agent_id, **req):
                 if isinstance(answer, str):
                     try:
@@ -454,7 +512,21 @@ async def agent_completions(tenant_id, agent_id):
                     except Exception:
                         continue
 
-                if ans.get("event") not in ["message", "message_end"]:
+                event = ans.get("event")
+                if event == "node_finished":
+                    if return_trace:
+                        data = ans.get("data", {})
+                        trace_items.append(
+                            {
+                                "component_id": data.get("component_id"),
+                                "trace": [copy.deepcopy(data)],
+                            }
+                        )
+                        ans.setdefault("data", {})["trace"] = trace_items
+                        answer = "data:" + json.dumps(ans, ensure_ascii=False) + "\n\n"
+                    yield answer
+
+                if event not in ["message", "message_end"]:
                     continue
 
                 yield answer
@@ -471,6 +543,7 @@ async def agent_completions(tenant_id, agent_id):
     full_content = ""
     reference = {}
     final_ans = ""
+    trace_items = []
     async for answer in agent_completion(tenant_id=tenant_id, agent_id=agent_id, **req):
         try:
             ans = json.loads(answer[5:])
@@ -481,17 +554,28 @@ async def agent_completions(tenant_id, agent_id):
             if ans.get("data", {}).get("reference", None):
                 reference.update(ans["data"]["reference"])
 
+            if return_trace and ans.get("event") == "node_finished":
+                data = ans.get("data", {})
+                trace_items.append(
+                    {
+                        "component_id": data.get("component_id"),
+                        "trace": [copy.deepcopy(data)],
+                    }
+                )
+
             final_ans = ans
         except Exception as e:
             return get_result(data=f"**ERROR**: {str(e)}")
     final_ans["data"]["content"] = full_content
     final_ans["data"]["reference"] = reference
+    if return_trace and final_ans:
+        final_ans["data"]["trace"] = trace_items
     return get_result(data=final_ans)
 
 
 @manager.route("/chats/<chat_id>/sessions", methods=["GET"])  # noqa: F821
 @token_required
-def list_session(tenant_id, chat_id):
+async def list_session(tenant_id, chat_id):
     if not DialogService.query(tenant_id=tenant_id, id=chat_id, status=StatusEnum.VALID.value):
         return get_error_data_result(message=f"You don't own the assistant {chat_id}.")
     id = request.args.get("id")
@@ -545,7 +629,7 @@ def list_session(tenant_id, chat_id):
 
 @manager.route("/agents/<agent_id>/sessions", methods=["GET"])  # noqa: F821
 @token_required
-def list_agent_session(tenant_id, agent_id):
+async def list_agent_session(tenant_id, agent_id):
     if not UserCanvasService.query(user_id=tenant_id, id=agent_id):
         return get_error_data_result(message=f"You don't own the agent {agent_id}.")
     id = request.args.get("id")
@@ -614,7 +698,7 @@ async def delete(tenant_id, chat_id):
 
     errors = []
     success_count = 0
-    req = await request.json
+    req = await get_request_json()
     convs = ConversationService.query(dialog_id=chat_id)
     if not req:
         ids = None
@@ -662,7 +746,7 @@ async def delete(tenant_id, chat_id):
 async def delete_agent_session(tenant_id, agent_id):
     errors = []
     success_count = 0
-    req = await request.json
+    req = await get_request_json()
     cvs = UserCanvasService.query(user_id=tenant_id, id=agent_id)
     if not cvs:
         return get_error_data_result(f"You don't own the agent {agent_id}")
@@ -715,7 +799,7 @@ async def delete_agent_session(tenant_id, agent_id):
 @manager.route("/sessions/ask", methods=["POST"])  # noqa: F821
 @token_required
 async def ask_about(tenant_id):
-    req = await request.json
+    req = await get_request_json()
     if not req.get("question"):
         return get_error_data_result("`question` is required.")
     if not req.get("dataset_ids"):
@@ -732,10 +816,10 @@ async def ask_about(tenant_id):
             return get_error_data_result(f"The dataset {kb_id} doesn't own parsed file")
     uid = tenant_id
 
-    def stream():
+    async def stream():
         nonlocal req, uid
         try:
-            for ans in ask(req["question"], req["kb_ids"], uid):
+            async for ans in async_ask(req["question"], req["kb_ids"], uid):
                 yield "data:" + json.dumps({"code": 0, "message": "", "data": ans}, ensure_ascii=False) + "\n\n"
         except Exception as e:
             yield "data:" + json.dumps(
@@ -754,7 +838,7 @@ async def ask_about(tenant_id):
 @manager.route("/sessions/related_questions", methods=["POST"])  # noqa: F821
 @token_required
 async def related_questions(tenant_id):
-    req = await request.json
+    req = await get_request_json()
     if not req.get("question"):
         return get_error_data_result("`question` is required.")
     question = req["question"]
@@ -787,7 +871,7 @@ Reason:
  - At the same time, related terms can also help search engines better understand user needs and return more accurate search results.
 
 """
-    ans = chat_mdl.chat(
+    ans = await chat_mdl.async_chat(
         prompt,
         [
             {
@@ -805,7 +889,7 @@ Related search terms:
 
 @manager.route("/chatbots/<dialog_id>/completions", methods=["POST"])  # noqa: F821
 async def chatbot_completions(dialog_id):
-    req = await request.json
+    req = await get_request_json()
 
     token = request.headers.get("Authorization").split()
     if len(token) != 2:
@@ -826,12 +910,13 @@ async def chatbot_completions(dialog_id):
         resp.headers.add_header("Content-Type", "text/event-stream; charset=utf-8")
         return resp
 
-    for answer in iframe_completion(dialog_id, **req):
+    async for answer in iframe_completion(dialog_id, **req):
         return get_result(data=answer)
 
+    return None
 
 @manager.route("/chatbots/<dialog_id>/info", methods=["GET"])  # noqa: F821
-def chatbots_inputs(dialog_id):
+async def chatbots_inputs(dialog_id):
     token = request.headers.get("Authorization").split()
     if len(token) != 2:
         return get_error_data_result(message='Authorization is not valid!"')
@@ -855,7 +940,7 @@ def chatbots_inputs(dialog_id):
 
 @manager.route("/agentbots/<agent_id>/completions", methods=["POST"])  # noqa: F821
 async def agent_bot_completions(agent_id):
-    req = await request.json
+    req = await get_request_json()
 
     token = request.headers.get("Authorization").split()
     if len(token) != 2:
@@ -876,9 +961,10 @@ async def agent_bot_completions(agent_id):
     async for answer in agent_completion(objs[0].tenant_id, agent_id, **req):
         return get_result(data=answer)
 
+    return None
 
 @manager.route("/agentbots/<agent_id>/inputs", methods=["GET"])  # noqa: F821
-def begin_inputs(agent_id):
+async def begin_inputs(agent_id):
     token = request.headers.get("Authorization").split()
     if len(token) != 2:
         return get_error_data_result(message='Authorization is not valid!"')
@@ -891,7 +977,7 @@ def begin_inputs(agent_id):
     if not e:
         return get_error_data_result(f"Can't find agent by ID: {agent_id}")
 
-    canvas = Canvas(json.dumps(cvs.dsl), objs[0].tenant_id)
+    canvas = Canvas(json.dumps(cvs.dsl), objs[0].tenant_id, canvas_id=cvs.id)
     return get_result(
         data={"title": cvs.title, "avatar": cvs.avatar, "inputs": canvas.get_component_input_form("begin"),
               "prologue": canvas.get_prologue(), "mode": canvas.get_mode()})
@@ -908,7 +994,7 @@ async def ask_about_embedded():
     if not objs:
         return get_error_data_result(message='Authentication error: API key is invalid!"')
 
-    req = await request.json
+    req = await get_request_json()
     uid = objs[0].tenant_id
 
     search_id = req.get("search_id", "")
@@ -917,10 +1003,10 @@ async def ask_about_embedded():
         if search_app := SearchService.get_detail(search_id):
             search_config = search_app.get("search_config", {})
 
-    def stream():
+    async def stream():
         nonlocal req, uid
         try:
-            for ans in ask(req["question"], req["kb_ids"], uid, search_config=search_config):
+            async for ans in async_ask(req["question"], req["kb_ids"], uid, search_config=search_config):
                 yield "data:" + json.dumps({"code": 0, "message": "", "data": ans}, ensure_ascii=False) + "\n\n"
         except Exception as e:
             yield "data:" + json.dumps(
@@ -947,7 +1033,7 @@ async def retrieval_test_embedded():
     if not objs:
         return get_error_data_result(message='Authentication error: API key is invalid!"')
 
-    req = await request.json
+    req = await get_request_json()
     page = int(req.get("page", 1))
     size = int(req.get("size", 30))
     question = req["question"]
@@ -963,28 +1049,42 @@ async def retrieval_test_embedded():
     use_kg = req.get("use_kg", False)
     top = int(req.get("top_k", 1024))
     langs = req.get("cross_languages", [])
-    tenant_ids = []
-
+    rerank_id = req.get("rerank_id", "")
     tenant_id = objs[0].tenant_id
     if not tenant_id:
         return get_error_data_result(message="permission denined.")
 
-    if req.get("search_id", ""):
-        search_config = SearchService.get_detail(req.get("search_id", "")).get("search_config", {})
-        meta_data_filter = search_config.get("meta_data_filter", {})
-        metas = DocumentService.get_meta_by_kbs(kb_ids)
-        if meta_data_filter.get("method") == "auto":
-            chat_mdl = LLMBundle(tenant_id, LLMType.CHAT, llm_name=search_config.get("chat_id", ""))
-            filters: dict = gen_meta_filter(chat_mdl, metas, question)
-            doc_ids.extend(meta_filter(metas, filters["conditions"], filters.get("logic", "and")))
-            if not doc_ids:
-                doc_ids = None
-        elif meta_data_filter.get("method") == "manual":
-            doc_ids.extend(meta_filter(metas, meta_data_filter["manual"], meta_data_filter.get("logic", "and")))
-            if meta_data_filter["manual"] and not doc_ids:
-                doc_ids = ["-999"]
+    async def _retrieval():
+        nonlocal similarity_threshold, vector_similarity_weight, top, rerank_id
+        local_doc_ids = list(doc_ids) if doc_ids else []
+        tenant_ids = []
+        _question = question
+
+        meta_data_filter = {}
+        chat_mdl = None
+        if req.get("search_id", ""):
+            search_config = SearchService.get_detail(req.get("search_id", "")).get("search_config", {})
+            meta_data_filter = search_config.get("meta_data_filter", {})
+            if meta_data_filter.get("method") in ["auto", "semi_auto"]:
+                chat_mdl = LLMBundle(tenant_id, LLMType.CHAT, llm_name=search_config.get("chat_id", ""))
+            # Apply search_config settings if not explicitly provided in request
+            if not req.get("similarity_threshold"):
+                similarity_threshold = float(search_config.get("similarity_threshold", similarity_threshold))
+            if not req.get("vector_similarity_weight"):
+                vector_similarity_weight = float(search_config.get("vector_similarity_weight", vector_similarity_weight))
+            if not req.get("top_k"):
+                top = int(search_config.get("top_k", top))
+            if not req.get("rerank_id"):
+                rerank_id = search_config.get("rerank_id", "")
+        else:
+            meta_data_filter = req.get("meta_data_filter") or {}
+            if meta_data_filter.get("method") in ["auto", "semi_auto"]:
+                chat_mdl = LLMBundle(tenant_id, LLMType.CHAT)
+
+        if meta_data_filter:
+            metas = DocumentService.get_meta_by_kbs(kb_ids)
+            local_doc_ids = await apply_meta_data_filter(meta_data_filter, metas, _question, chat_mdl, local_doc_ids)
 
-    try:
         tenants = UserTenantService.query(user_id=tenant_id)
         for kb_id in kb_ids:
             for tenant in tenants:
@@ -992,7 +1092,7 @@ async def retrieval_test_embedded():
                     tenant_ids.append(tenant.tenant_id)
                     break
             else:
-                return get_json_result(data=False, message="Only owner of knowledgebase authorized for this operation.",
+                return get_json_result(data=False, message="Only owner of dataset authorized for this operation.",
                                        code=RetCode.OPERATING_ERROR)
 
         e, kb = KnowledgebaseService.get_by_id(kb_ids[0])
@@ -1000,25 +1100,25 @@ async def retrieval_test_embedded():
             return get_error_data_result(message="Knowledgebase not found!")
 
         if langs:
-            question = cross_languages(kb.tenant_id, None, question, langs)
+            _question = await cross_languages(kb.tenant_id, None, _question, langs)
 
         embd_mdl = LLMBundle(kb.tenant_id, LLMType.EMBEDDING.value, llm_name=kb.embd_id)
 
         rerank_mdl = None
-        if req.get("rerank_id"):
-            rerank_mdl = LLMBundle(kb.tenant_id, LLMType.RERANK.value, llm_name=req["rerank_id"])
+        if rerank_id:
+            rerank_mdl = LLMBundle(kb.tenant_id, LLMType.RERANK.value, llm_name=rerank_id)
 
         if req.get("keyword", False):
             chat_mdl = LLMBundle(kb.tenant_id, LLMType.CHAT)
-            question += keyword_extraction(chat_mdl, question)
+            _question += await keyword_extraction(chat_mdl, _question)
 
-        labels = label_question(question, [kb])
-        ranks = settings.retriever.retrieval(
-            question, embd_mdl, tenant_ids, kb_ids, page, size, similarity_threshold, vector_similarity_weight, top,
-            doc_ids, rerank_mdl=rerank_mdl, highlight=req.get("highlight"), rank_feature=labels
+        labels = label_question(_question, [kb])
+        ranks = await settings.retriever.retrieval(
+            _question, embd_mdl, tenant_ids, kb_ids, page, size, similarity_threshold, vector_similarity_weight, top,
+            local_doc_ids, rerank_mdl=rerank_mdl, highlight=req.get("highlight"), rank_feature=labels
         )
         if use_kg:
-            ck = settings.kg_retriever.retrieval(question, tenant_ids, kb_ids, embd_mdl,
+            ck = await settings.kg_retriever.retrieval(_question, tenant_ids, kb_ids, embd_mdl,
                                                  LLMBundle(kb.tenant_id, LLMType.CHAT))
             if ck["content_with_weight"]:
                 ranks["chunks"].insert(0, ck)
@@ -1028,6 +1128,9 @@ async def retrieval_test_embedded():
         ranks["labels"] = labels
 
         return get_json_result(data=ranks)
+
+    try:
+        return await _retrieval()
     except Exception as e:
         if str(e).find("not_found") > 0:
             return get_json_result(data=False, message="No chunk found! Check the chunk status please!",
@@ -1046,7 +1149,7 @@ async def related_questions_embedded():
     if not objs:
         return get_error_data_result(message='Authentication error: API key is invalid!"')
 
-    req = await request.json
+    req = await get_request_json()
     tenant_id = objs[0].tenant_id
     if not tenant_id:
         return get_error_data_result(message="permission denined.")
@@ -1064,7 +1167,7 @@ async def related_questions_embedded():
 
     gen_conf = search_config.get("llm_setting", {"temperature": 0.9})
     prompt = load_prompt("related_question")
-    ans = chat_mdl.chat(
+    ans = await chat_mdl.async_chat(
         prompt,
         [
             {
@@ -1081,7 +1184,7 @@ Related search terms:
 
 
 @manager.route("/searchbots/detail", methods=["GET"])  # noqa: F821
-def detail_share_embedded():
+async def detail_share_embedded():
     token = request.headers.get("Authorization").split()
     if len(token) != 2:
         return get_error_data_result(message='Authorization is not valid!"')
@@ -1123,12 +1226,102 @@ async def mindmap():
         return get_error_data_result(message='Authentication error: API key is invalid!"')
 
     tenant_id = objs[0].tenant_id
-    req = await request.json
+    req = await get_request_json()
 
     search_id = req.get("search_id", "")
     search_app = SearchService.get_detail(search_id) if search_id else {}
 
-    mind_map = gen_mindmap(req["question"], req["kb_ids"], tenant_id, search_app.get("search_config", {}))
+    mind_map =await gen_mindmap(req["question"], req["kb_ids"], tenant_id, search_app.get("search_config", {}))
     if "error" in mind_map:
         return server_error_response(Exception(mind_map["error"]))
     return get_json_result(data=mind_map)
+
+@manager.route("/sequence2txt", methods=["POST"])  # noqa: F821
+@token_required
+async def sequence2txt(tenant_id):
+    req = await request.form
+    stream_mode = req.get("stream", "false").lower() == "true"
+    files = await request.files
+    if "file" not in files:
+        return get_error_data_result(message="Missing 'file' in multipart form-data")
+
+    uploaded = files["file"]
+
+    ALLOWED_EXTS = {
+        ".wav", ".mp3", ".m4a", ".aac",
+        ".flac", ".ogg", ".webm",
+        ".opus", ".wma"
+    }
+
+    filename = uploaded.filename or ""
+    suffix = os.path.splitext(filename)[-1].lower()
+    if suffix not in ALLOWED_EXTS:
+        return get_error_data_result(message=
+            f"Unsupported audio format: {suffix}. "
+            f"Allowed: {', '.join(sorted(ALLOWED_EXTS))}"
+        )
+    fd, temp_audio_path = tempfile.mkstemp(suffix=suffix)
+    os.close(fd)
+    await uploaded.save(temp_audio_path)
+
+    tenants = TenantService.get_info_by(tenant_id)
+    if not tenants:
+        return get_error_data_result(message="Tenant not found!")
+
+    asr_id = tenants[0]["asr_id"]
+    if not asr_id:
+        return get_error_data_result(message="No default ASR model is set")
+
+    asr_mdl=LLMBundle(tenants[0]["tenant_id"], LLMType.SPEECH2TEXT, asr_id)
+    if not stream_mode:
+        text = asr_mdl.transcription(temp_audio_path)
+        try:
+            os.remove(temp_audio_path)
+        except Exception as e:
+            logging.error(f"Failed to remove temp audio file: {str(e)}")
+        return get_json_result(data={"text": text})
+    async def event_stream():
+        try:
+            for evt in asr_mdl.stream_transcription(temp_audio_path):
+                yield f"data: {json.dumps(evt, ensure_ascii=False)}\n\n"
+        except Exception as e:
+            err = {"event": "error", "text": str(e)}
+            yield f"data: {json.dumps(err, ensure_ascii=False)}\n\n"
+        finally:
+            try:
+                os.remove(temp_audio_path)
+            except Exception as e:
+                logging.error(f"Failed to remove temp audio file: {str(e)}")
+
+    return Response(event_stream(), content_type="text/event-stream")
+
+@manager.route("/tts", methods=["POST"])  # noqa: F821
+@token_required
+async def tts(tenant_id):
+    req = await get_request_json()
+    text = req["text"]
+
+    tenants = TenantService.get_info_by(tenant_id)
+    if not tenants:
+        return get_error_data_result(message="Tenant not found!")
+
+    tts_id = tenants[0]["tts_id"]
+    if not tts_id:
+        return get_error_data_result(message="No default TTS model is set")
+
+    tts_mdl = LLMBundle(tenants[0]["tenant_id"], LLMType.TTS, tts_id)
+
+    def stream_audio():
+        try:
+            for txt in re.split(r"[，。/《》？；：！\n\r:;]+", text):
+                for chunk in tts_mdl.tts(txt):
+                    yield chunk
+        except Exception as e:
+            yield ("data:" + json.dumps({"code": 500, "message": str(e), "data": {"answer": "**ERROR**: " + str(e)}}, ensure_ascii=False)).encode("utf-8")
+
+    resp = Response(stream_audio(), mimetype="audio/mpeg")
+    resp.headers.add_header("Cache-Control", "no-cache")
+    resp.headers.add_header("Connection", "keep-alive")
+    resp.headers.add_header("X-Accel-Buffering", "no")
+
+    return resp

api/apps/search_app.py

@@ -24,14 +24,14 @@ from api.db.services.search_service import SearchService
 from api.db.services.user_service import TenantService, UserTenantService
 from common.misc_utils import get_uuid
 from common.constants import RetCode, StatusEnum
-from api.utils.api_utils import get_data_error_result, get_json_result, not_allowed_parameters, server_error_response, validate_request
+from api.utils.api_utils import get_data_error_result, get_json_result, not_allowed_parameters, get_request_json, server_error_response, validate_request
 
 
 @manager.route("/create", methods=["post"])  # noqa: F821
 @login_required
 @validate_request("name")
 async def create():
-    req = await request.get_json()
+    req = await get_request_json()
     search_name = req["name"]
     description = req.get("description", "")
     if not isinstance(search_name, str):
@@ -66,7 +66,7 @@ async def create():
 @validate_request("search_id", "name", "search_config", "tenant_id")
 @not_allowed_parameters("id", "created_by", "create_time", "update_time", "create_date", "update_date", "created_by")
 async def update():
-    req = await request.get_json()
+    req = await get_request_json()
     if not isinstance(req["name"], str):
         return get_data_error_result(message="Search name must be string.")
     if req["name"].strip() == "":
@@ -150,7 +150,7 @@ async def list_search_app():
     else:
         desc = True
 
-    req = await request.get_json()
+    req = await get_request_json()
     owner_ids = req.get("owner_ids", [])
     try:
         if not owner_ids:
@@ -174,7 +174,7 @@ async def list_search_app():
 @login_required
 @validate_request("search_id")
 async def rm():
-    req = await request.get_json()
+    req = await get_request_json()
     search_id = req["search_id"]
     if not SearchService.accessible4deletion(search_id, current_user.id):
         return get_json_result(data=False, message="No authorization.", code=RetCode.AUTHENTICATION_ERROR)

api/apps/system_app.py

@@ -177,8 +177,8 @@ def healthz():
     return jsonify(result), (200 if all_ok else 500)
 
 
-@manager.route("/ping", methods=["GET"]) # noqa: F821
-def ping():
+@manager.route("/ping", methods=["GET"])  # noqa: F821
+async def ping():
     return "pong", 200
 
 
@@ -213,7 +213,7 @@ def new_token():
         if not tenants:
             return get_data_error_result(message="Tenant not found!")
 
-        tenant_id = [tenant for tenant in tenants if tenant.role == 'owner'][0].tenant_id
+        tenant_id = [tenant for tenant in tenants if tenant.role == "owner"][0].tenant_id
         obj = {
             "tenant_id": tenant_id,
             "token": generate_confirmation_token(),
@@ -268,13 +268,12 @@ def token_list():
         if not tenants:
             return get_data_error_result(message="Tenant not found!")
 
-        tenant_id = [tenant for tenant in tenants if tenant.role == 'owner'][0].tenant_id
+        tenant_id = [tenant for tenant in tenants if tenant.role == "owner"][0].tenant_id
         objs = APITokenService.query(tenant_id=tenant_id)
         objs = [o.to_dict() for o in objs]
         for o in objs:
             if not o["beta"]:
-                o["beta"] = generate_confirmation_token().replace(
-                    "ragflow-", "")[:32]
+                o["beta"] = generate_confirmation_token().replace("ragflow-", "")[:32]
                 APITokenService.filter_update([APIToken.tenant_id == tenant_id, APIToken.token == o["token"]], o)
         return get_json_result(data=objs)
     except Exception as e:
@@ -307,13 +306,19 @@ def rm(token):
               type: boolean
               description: Deletion status.
     """
-    APITokenService.filter_delete(
-        [APIToken.tenant_id == current_user.id, APIToken.token == token]
-    )
-    return get_json_result(data=True)
+    try:
+        tenants = UserTenantService.query(user_id=current_user.id)
+        if not tenants:
+            return get_data_error_result(message="Tenant not found!")
+
+        tenant_id = tenants[0].tenant_id
+        APITokenService.filter_delete([APIToken.tenant_id == tenant_id, APIToken.token == token])
+        return get_json_result(data=True)
+    except Exception as e:
+        return server_error_response(e)
 
 
-@manager.route('/config', methods=['GET'])  # noqa: F821
+@manager.route("/config", methods=["GET"])  # noqa: F821
 def get_config():
     """
     Get system configuration.
@@ -330,6 +335,4 @@ def get_config():
                         type: integer 0 means disabled, 1 means enabled
                         description: Whether user registration is enabled
     """
-    return get_json_result(data={
-        "registerEnabled": settings.REGISTER_ENABLED
-    })
+    return get_json_result(data={"registerEnabled": settings.REGISTER_ENABLED})

api/apps/tenant_app.py

@@ -13,8 +13,8 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-
-from quart import request
+import logging
+import asyncio
 from api.db import UserTenantRole
 from api.db.db_models import UserTenant
 from api.db.services.user_service import UserTenantService, UserService
@@ -22,10 +22,10 @@ from api.db.services.user_service import UserTenantService, UserService
 from common.constants import RetCode, StatusEnum
 from common.misc_utils import get_uuid
 from common.time_utils import delta_seconds
-from api.utils.api_utils import get_json_result, validate_request, server_error_response, get_data_error_result
+from api.utils.api_utils import get_data_error_result, get_json_result, get_request_json, server_error_response, validate_request
 from api.utils.web_utils import send_invite_email
 from common import settings
-from api.apps import smtp_mail_server, login_required, current_user
+from api.apps import login_required, current_user
 
 
 @manager.route("/<tenant_id>/user/list", methods=["GET"])  # noqa: F821
@@ -56,7 +56,7 @@ async def create(tenant_id):
             message='No authorization.',
             code=RetCode.AUTHENTICATION_ERROR)
 
-    req = await request.json
+    req = await get_request_json()
     invite_user_email = req["email"]
     invite_users = UserService.query(email=invite_user_email)
     if not invite_users:
@@ -81,20 +81,24 @@ async def create(tenant_id):
         role=UserTenantRole.INVITE,
         status=StatusEnum.VALID.value)
 
-    if smtp_mail_server and settings.SMTP_CONF:
-        from threading import Thread
+    try:
 
         user_name = ""
         _, user = UserService.get_by_id(current_user.id)
         if user:
             user_name = user.nickname
 
-        Thread(
-            target=send_invite_email,
-            args=(invite_user_email, settings.MAIL_FRONTEND_URL, tenant_id, user_name or current_user.email),
-            daemon=True
-        ).start()
-
+        asyncio.create_task(
+            send_invite_email(
+                to_email=invite_user_email,
+                invite_url=settings.MAIL_FRONTEND_URL,
+                tenant_id=tenant_id,
+                inviter=user_name or current_user.email
+            )
+        )
+    except Exception as e:
+        logging.exception(f"Failed to send invite email to {invite_user_email}: {e}")
+        return get_json_result(data=False, message="Failed to send invite email.", code=RetCode.SERVER_ERROR)
     usr = invite_users[0].to_dict()
     usr = {k: v for k, v in usr.items() if k in ["id", "avatar", "email", "nickname"]}
 

api/apps/user_app.py

@@ -21,8 +21,9 @@ import re
 import secrets
 import time
 from datetime import datetime
+import base64
 
-from quart import redirect, request, session, make_response
+from quart import make_response, redirect, request, session
 from werkzeug.security import check_password_hash, generate_password_hash
 
 from api.apps.auth import get_auth_client
@@ -39,12 +40,13 @@ from common.connection_utils import construct_response
 from api.utils.api_utils import (
     get_data_error_result,
     get_json_result,
+    get_request_json,
     server_error_response,
     validate_request,
 )
 from api.utils.crypt import decrypt
 from rag.utils.redis_conn import REDIS_CONN
-from api.apps import smtp_mail_server, login_required, current_user, login_user, logout_user
+from api.apps import login_required, current_user, login_user, logout_user
 from api.utils.web_utils import (
     send_email_html,
     OTP_LENGTH,
@@ -57,6 +59,7 @@ from api.utils.web_utils import (
     captcha_key,
 )
 from common import settings
+from common.http_client import async_request
 
 
 @manager.route("/login", methods=["POST", "GET"])  # noqa: F821
@@ -90,11 +93,14 @@ async def login():
         schema:
           type: object
     """
-    json_body = await request.json
+    json_body = await get_request_json()
     if not json_body:
         return get_json_result(data=False, code=RetCode.AUTHENTICATION_ERROR, message="Unauthorized!")
 
     email = json_body.get("email", "")
+    if email == "admin@ragflow.io":
+        return get_json_result(data=False, code=RetCode.AUTHENTICATION_ERROR, message="Default admin account cannot be used to login normal services!")
+    
     users = UserService.query(email=email)
     if not users:
         return get_json_result(
@@ -121,8 +127,8 @@ async def login():
         response_data = user.to_json()
         user.access_token = get_uuid()
         login_user(user)
-        user.update_time = (current_timestamp(),)
-        user.update_date = (datetime_format(datetime.now()),)
+        user.update_time = current_timestamp()
+        user.update_date = datetime_format(datetime.now())
         user.save()
         msg = "Welcome back!"
 
@@ -136,7 +142,7 @@ async def login():
 
 
 @manager.route("/login/channels", methods=["GET"])  # noqa: F821
-def get_login_channels():
+async def get_login_channels():
     """
     Get all supported authentication channels.
     """
@@ -157,7 +163,7 @@ def get_login_channels():
 
 
 @manager.route("/login/<channel>", methods=["GET"])  # noqa: F821
-def oauth_login(channel):
+async def oauth_login(channel):
     channel_config = settings.OAUTH_CONFIG.get(channel)
     if not channel_config:
         raise ValueError(f"Invalid channel name: {channel}")
@@ -170,7 +176,7 @@ def oauth_login(channel):
 
 
 @manager.route("/oauth/callback/<channel>", methods=["GET"])  # noqa: F821
-def oauth_callback(channel):
+async def oauth_callback(channel):
     """
     Handle the OAuth/OIDC callback for various channels dynamically.
     """
@@ -192,7 +198,10 @@ def oauth_callback(channel):
             return redirect("/?error=missing_code")
 
         # Exchange authorization code for access token
-        token_info = auth_cli.exchange_code_for_token(code)
+        if hasattr(auth_cli, "async_exchange_code_for_token"):
+            token_info = await auth_cli.async_exchange_code_for_token(code)
+        else:
+            token_info = auth_cli.exchange_code_for_token(code)
         access_token = token_info.get("access_token")
         if not access_token:
             return redirect("/?error=token_failed")
@@ -200,7 +209,10 @@ def oauth_callback(channel):
         id_token = token_info.get("id_token")
 
         # Fetch user info
-        user_info = auth_cli.fetch_user_info(access_token, id_token=id_token)
+        if hasattr(auth_cli, "async_fetch_user_info"):
+            user_info = await auth_cli.async_fetch_user_info(access_token, id_token=id_token)
+        else:
+            user_info = auth_cli.fetch_user_info(access_token, id_token=id_token)
         if not user_info.email:
             return redirect("/?error=email_missing")
 
@@ -259,7 +271,7 @@ def oauth_callback(channel):
 
 
 @manager.route("/github_callback", methods=["GET"])  # noqa: F821
-def github_callback():
+async def github_callback():
     """
     **Deprecated**, Use `/oauth/callback/<channel>` instead.
 
@@ -279,9 +291,8 @@ def github_callback():
         schema:
           type: object
     """
-    import requests
-
-    res = requests.post(
+    res = await async_request(
+        "POST",
         settings.GITHUB_OAUTH.get("url"),
         data={
             "client_id": settings.GITHUB_OAUTH.get("client_id"),
@@ -299,7 +310,7 @@ def github_callback():
 
     session["access_token"] = res["access_token"]
     session["access_token_from"] = "github"
-    user_info = user_info_from_github(session["access_token"])
+    user_info = await user_info_from_github(session["access_token"])
     email_address = user_info["email"]
     users = UserService.query(email=email_address)
     user_id = get_uuid()
@@ -348,7 +359,7 @@ def github_callback():
 
 
 @manager.route("/feishu_callback", methods=["GET"])  # noqa: F821
-def feishu_callback():
+async def feishu_callback():
     """
     Feishu OAuth callback endpoint.
     ---
@@ -366,9 +377,8 @@ def feishu_callback():
         schema:
           type: object
     """
-    import requests
-
-    app_access_token_res = requests.post(
+    app_access_token_res = await async_request(
+        "POST",
         settings.FEISHU_OAUTH.get("app_access_token_url"),
         data=json.dumps(
             {
@@ -382,7 +392,8 @@ def feishu_callback():
     if app_access_token_res["code"] != 0:
         return redirect("/?error=%s" % app_access_token_res)
 
-    res = requests.post(
+    res = await async_request(
+        "POST",
         settings.FEISHU_OAUTH.get("user_access_token_url"),
         data=json.dumps(
             {
@@ -403,7 +414,7 @@ def feishu_callback():
         return redirect("/?error=contact:user.email:readonly not in scope")
     session["access_token"] = res["data"]["access_token"]
     session["access_token_from"] = "feishu"
-    user_info = user_info_from_feishu(session["access_token"])
+    user_info = await user_info_from_feishu(session["access_token"])
     email_address = user_info["email"]
     users = UserService.query(email=email_address)
     user_id = get_uuid()
@@ -451,36 +462,34 @@ def feishu_callback():
     return redirect("/?auth=%s" % user.get_id())
 
 
-def user_info_from_feishu(access_token):
-    import requests
-
+async def user_info_from_feishu(access_token):
     headers = {
         "Content-Type": "application/json; charset=utf-8",
         "Authorization": f"Bearer {access_token}",
     }
-    res = requests.get("https://open.feishu.cn/open-apis/authen/v1/user_info", headers=headers)
+    res = await async_request("GET", "https://open.feishu.cn/open-apis/authen/v1/user_info", headers=headers)
     user_info = res.json()["data"]
     user_info["email"] = None if user_info.get("email") == "" else user_info["email"]
     return user_info
 
 
-def user_info_from_github(access_token):
-    import requests
-
+async def user_info_from_github(access_token):
     headers = {"Accept": "application/json", "Authorization": f"token {access_token}"}
-    res = requests.get(f"https://api.github.com/user?access_token={access_token}", headers=headers)
+    res = await async_request("GET", f"https://api.github.com/user?access_token={access_token}", headers=headers)
     user_info = res.json()
-    email_info = requests.get(
+    email_info_response = await async_request(
+        "GET",
         f"https://api.github.com/user/emails?access_token={access_token}",
         headers=headers,
-    ).json()
+    )
+    email_info = email_info_response.json()
     user_info["email"] = next((email for email in email_info if email["primary"]), None)["email"]
     return user_info
 
 
 @manager.route("/logout", methods=["GET"])  # noqa: F821
 @login_required
-def log_out():
+async def log_out():
     """
     User logout endpoint.
     ---
@@ -531,7 +540,7 @@ async def setting_user():
           type: object
     """
     update_dict = {}
-    request_data = await request.json
+    request_data = await get_request_json()
     if request_data.get("password"):
         new_password = request_data.get("new_password")
         if not check_password_hash(current_user.password, decrypt(request_data["password"])):
@@ -570,7 +579,7 @@ async def setting_user():
 
 @manager.route("/info", methods=["GET"])  # noqa: F821
 @login_required
-def user_profile():
+async def user_profile():
     """
     Get user profile information.
     ---
@@ -651,7 +660,7 @@ def user_register(user_id, user):
     tenant_llm = get_init_tenant_llm(user_id)
 
     if not UserService.save(**user):
-        return
+        return None
     TenantService.insert(**tenant)
     UserTenantService.insert(**usr_tenant)
     TenantLLMService.insert_many(tenant_llm)
@@ -698,7 +707,7 @@ async def user_add():
             code=RetCode.OPERATING_ERROR,
         )
 
-    req = await request.json
+    req = await get_request_json()
     email_address = req["email"]
 
     # Validate the email address
@@ -755,7 +764,7 @@ async def user_add():
 
 @manager.route("/tenant_info", methods=["GET"])  # noqa: F821
 @login_required
-def tenant_info():
+async def tenant_info():
     """
     Get tenant information.
     ---
@@ -831,14 +840,14 @@ async def set_tenant_info():
         schema:
           type: object
     """
-    req = await request.json
+    req = await get_request_json()
     try:
         tid = req.pop("tenant_id")
         TenantService.update_by_id(tid, req)
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)
-        
+
 
 @manager.route("/forget/captcha", methods=["GET"])  # noqa: F821
 async def forget_get_captcha():
@@ -875,7 +884,7 @@ async def forget_send_otp():
     - Verify the image captcha stored at captcha:{email} (case-insensitive).
     - On success, generate an email OTP (A–Z with length = OTP_LENGTH), store hash + salt (and timestamp) in Redis with TTL, reset attempts and cooldown, and send the OTP via email.
     """
-    req = await request.get_json()
+    req = await get_request_json()
     email = req.get("email") or ""
     captcha = (req.get("captcha") or "").strip()
 
@@ -918,47 +927,45 @@ async def forget_send_otp():
 
     ttl_min = OTP_TTL_SECONDS // 60
 
-    if not smtp_mail_server:
-        logging.warning("SMTP mail server not initialized; skip sending email.")
-    else:
-        try:
-            send_email_html(
-                subject="Your Password Reset Code",
-                to_email=email,
-                template_key="reset_code",
-                code=otp,
-                ttl_min=ttl_min,
-            )
-        except Exception:
-            return get_json_result(data=False, code=RetCode.SERVER_ERROR, message="failed to send email")
-        
+    try:
+        await send_email_html(
+            subject="Your Password Reset Code",
+            to_email=email,
+            template_key="reset_code",
+            code=otp,
+            ttl_min=ttl_min,
+        )
+
+    except Exception as e:
+        logging.exception(e)
+        return get_json_result(data=False, code=RetCode.SERVER_ERROR, message="failed to send email")
+
     return get_json_result(data=True, code=RetCode.SUCCESS, message="verification passed, email sent")
 
 
-@manager.route("/forget", methods=["POST"])  # noqa: F821
-async def forget():
+def _verified_key(email: str) -> str:
+    return f"otp:verified:{email}"
+
+
+@manager.route("/forget/verify-otp", methods=["POST"])  # noqa: F821
+async def forget_verify_otp():
     """
-    POST: Verify email + OTP and reset password, then log the user in.
-    Request JSON: { email, otp, new_password, confirm_new_password }
+    Verify email + OTP only. On success:
+    - consume the OTP and attempt counters
+    - set a short-lived verified flag in Redis for the email
+    Request JSON: { email, otp }
     """
-    req = await request.get_json()
+    req = await get_request_json()
     email = req.get("email") or ""
     otp = (req.get("otp") or "").strip()
-    new_pwd = req.get("new_password")
-    new_pwd2 = req.get("confirm_new_password")
 
-    if not all([email, otp, new_pwd, new_pwd2]):
-        return get_json_result(data=False, code=RetCode.ARGUMENT_ERROR, message="email, otp and passwords are required")
-
-    # For reset, passwords are provided as-is (no decrypt needed)
-    if new_pwd != new_pwd2:
-        return get_json_result(data=False, code=RetCode.ARGUMENT_ERROR, message="passwords do not match")
+    if not all([email, otp]):
+        return get_json_result(data=False, code=RetCode.ARGUMENT_ERROR, message="email and otp are required")
 
     users = UserService.query(email=email)
     if not users:
         return get_json_result(data=False, code=RetCode.DATA_ERROR, message="invalid email")
 
-    user = users[0]
     # Verify OTP from Redis
     k_code, k_attempts, k_last, k_lock = otp_keys(email)
     if REDIS_CONN.get(k_lock):
@@ -974,7 +981,6 @@ async def forget():
     except Exception:
         return get_json_result(data=False, code=RetCode.EXCEPTION_ERROR, message="otp storage corrupted")
 
-    # Case-insensitive verification: OTP generated uppercase
     calc = hash_code(otp.upper(), salt)
     if calc != stored_hash:
         # bump attempts
@@ -987,23 +993,70 @@ async def forget():
             REDIS_CONN.set(k_lock, int(time.time()), ATTEMPT_LOCK_SECONDS)
         return get_json_result(data=False, code=RetCode.AUTHENTICATION_ERROR, message="expired otp")
 
-    # Success: consume OTP and reset password
+    # Success: consume OTP and attempts; mark verified
     REDIS_CONN.delete(k_code)
     REDIS_CONN.delete(k_attempts)
     REDIS_CONN.delete(k_last)
     REDIS_CONN.delete(k_lock)
 
+    # set verified flag with limited TTL, reuse OTP_TTL_SECONDS or smaller window
     try:
-        UserService.update_user_password(user.id, new_pwd)
+        REDIS_CONN.set(_verified_key(email), "1", OTP_TTL_SECONDS)
+    except Exception:
+        return get_json_result(data=False, code=RetCode.SERVER_ERROR, message="failed to set verification state")
+
+    return get_json_result(data=True, code=RetCode.SUCCESS, message="otp verified")
+
+
+@manager.route("/forget/reset-password", methods=["POST"])  # noqa: F821
+async def forget_reset_password():
+    """
+    Reset password after successful OTP verification.
+    Requires: { email, new_password, confirm_new_password }
+    Steps:
+    - check verified flag in Redis
+    - update user password
+    - auto login
+    - clear verified flag
+    """
+    
+    req = await get_request_json()
+    email = req.get("email") or ""
+    new_pwd = req.get("new_password")
+    new_pwd2 = req.get("confirm_new_password")
+
+    new_pwd_base64 = decrypt(new_pwd)
+    new_pwd_string = base64.b64decode(new_pwd_base64).decode('utf-8')
+    new_pwd2_string = base64.b64decode(decrypt(new_pwd2)).decode('utf-8')
+
+    REDIS_CONN.get(_verified_key(email))
+    if not REDIS_CONN.get(_verified_key(email)):
+        return get_json_result(data=False, code=RetCode.AUTHENTICATION_ERROR, message="email not verified")
+
+    if not all([email, new_pwd, new_pwd2]):
+        return get_json_result(data=False, code=RetCode.ARGUMENT_ERROR, message="email and passwords are required")
+
+    if new_pwd_string != new_pwd2_string:
+        return get_json_result(data=False, code=RetCode.ARGUMENT_ERROR, message="passwords do not match")
+
+    users = UserService.query_user_by_email(email=email)
+    if not users:
+        return get_json_result(data=False, code=RetCode.DATA_ERROR, message="invalid email")
+    
+    user = users[0]
+    try:
+        UserService.update_user_password(user.id, new_pwd_base64)
     except Exception as e:
         logging.exception(e)
         return get_json_result(data=False, code=RetCode.EXCEPTION_ERROR, message="failed to reset password")
 
-    # Auto login (reuse login flow)
-    user.access_token = get_uuid()
-    login_user(user)
-    user.update_time = (current_timestamp(),)
-    user.update_date = (datetime_format(datetime.now()),)
-    user.save()
+    # clear verified flag
+    try:
+        REDIS_CONN.delete(_verified_key(email))
+    except Exception:
+        pass
+
     msg = "Password reset successful. Logged in."
-    return construct_response(data=user.to_json(), auth=user.get_id(), message=msg)
+    return await construct_response(data=user.to_json(), auth=user.get_id(), message=msg)
+
+

api/constants.py

@@ -24,3 +24,5 @@ REQUEST_MAX_WAIT_SEC = 300
 
 DATASET_NAME_LIMIT = 128
 FILE_NAME_LEN_LIMIT = 255
+MEMORY_NAME_LIMIT = 128
+MEMORY_SIZE_LIMIT = 10*1024*1024 # Byte

api/db/db_models.py

@@ -281,7 +281,11 @@ class RetryingPooledMySQLDatabase(PooledMySQLDatabase):
         except Exception as e:
             logging.error(f"Failed to reconnect: {e}")
             time.sleep(0.1)
-            self.connect()
+            try:
+                self.connect()
+            except Exception as e2:
+                logging.error(f"Failed to reconnect on second attempt: {e2}")
+                raise
 
     def begin(self):
         for attempt in range(self.max_retries + 1):
@@ -352,7 +356,11 @@ class RetryingPooledPostgresqlDatabase(PooledPostgresqlDatabase):
         except Exception as e:
             logging.error(f"Failed to reconnect to PostgreSQL: {e}")
             time.sleep(0.1)
-            self.connect()
+            try:
+                self.connect()
+            except Exception as e2:
+                logging.error(f"Failed to reconnect to PostgreSQL on second attempt: {e2}")
+                raise
 
     def begin(self):
         for attempt in range(self.max_retries + 1):
@@ -749,7 +757,7 @@ class Knowledgebase(DataBaseModel):
 
     parser_id = CharField(max_length=32, null=False, help_text="default parser ID", default=ParserType.NAIVE.value, index=True)
     pipeline_id = CharField(max_length=32, null=True, help_text="Pipeline ID", index=True)
-    parser_config = JSONField(null=False, default={"pages": [[1, 1000000]]})
+    parser_config = JSONField(null=False, default={"pages": [[1, 1000000]], "table_context_size": 0, "image_context_size": 0})
     pagerank = IntegerField(default=0, index=False)
 
     graphrag_task_id = CharField(max_length=32, null=True, help_text="Graph RAG task ID", index=True)
@@ -774,7 +782,7 @@ class Document(DataBaseModel):
     kb_id = CharField(max_length=256, null=False, index=True)
     parser_id = CharField(max_length=32, null=False, help_text="default parser ID", index=True)
     pipeline_id = CharField(max_length=32, null=True, help_text="pipeline ID", index=True)
-    parser_config = JSONField(null=False, default={"pages": [[1, 1000000]]})
+    parser_config = JSONField(null=False, default={"pages": [[1, 1000000]], "table_context_size": 0, "image_context_size": 0})
     source_type = CharField(max_length=128, null=False, default="local", help_text="where dose this document come from", index=True)
     type = CharField(max_length=32, null=False, help_text="file extension", index=True)
     created_by = CharField(max_length=32, null=False, help_text="who created it", index=True)
@@ -1113,184 +1121,177 @@ class SyncLogs(DataBaseModel):
         db_table = "sync_logs"
 
 
+class EvaluationDataset(DataBaseModel):
+    """Ground truth dataset for RAG evaluation"""
+    id = CharField(max_length=32, primary_key=True)
+    tenant_id = CharField(max_length=32, null=False, index=True, help_text="tenant ID")
+    name = CharField(max_length=255, null=False, index=True, help_text="dataset name")
+    description = TextField(null=True, help_text="dataset description")
+    kb_ids = JSONField(null=False, help_text="knowledge base IDs to evaluate against")
+    created_by = CharField(max_length=32, null=False, index=True, help_text="creator user ID")
+    create_time = BigIntegerField(null=False, index=True, help_text="creation timestamp")
+    update_time = BigIntegerField(null=False, help_text="last update timestamp")
+    status = IntegerField(null=False, default=1, help_text="1=valid, 0=invalid")
+
+    class Meta:
+        db_table = "evaluation_datasets"
+
+
+class EvaluationCase(DataBaseModel):
+    """Individual test case in an evaluation dataset"""
+    id = CharField(max_length=32, primary_key=True)
+    dataset_id = CharField(max_length=32, null=False, index=True, help_text="FK to evaluation_datasets")
+    question = TextField(null=False, help_text="test question")
+    reference_answer = TextField(null=True, help_text="optional ground truth answer")
+    relevant_doc_ids = JSONField(null=True, help_text="expected relevant document IDs")
+    relevant_chunk_ids = JSONField(null=True, help_text="expected relevant chunk IDs")
+    metadata = JSONField(null=True, help_text="additional context/tags")
+    create_time = BigIntegerField(null=False, help_text="creation timestamp")
+
+    class Meta:
+        db_table = "evaluation_cases"
+
+
+class EvaluationRun(DataBaseModel):
+    """A single evaluation run"""
+    id = CharField(max_length=32, primary_key=True)
+    dataset_id = CharField(max_length=32, null=False, index=True, help_text="FK to evaluation_datasets")
+    dialog_id = CharField(max_length=32, null=False, index=True, help_text="dialog configuration being evaluated")
+    name = CharField(max_length=255, null=False, help_text="run name")
+    config_snapshot = JSONField(null=False, help_text="dialog config at time of evaluation")
+    metrics_summary = JSONField(null=True, help_text="aggregated metrics")
+    status = CharField(max_length=32, null=False, default="PENDING", help_text="PENDING/RUNNING/COMPLETED/FAILED")
+    created_by = CharField(max_length=32, null=False, index=True, help_text="user who started the run")
+    create_time = BigIntegerField(null=False, index=True, help_text="creation timestamp")
+    complete_time = BigIntegerField(null=True, help_text="completion timestamp")
+
+    class Meta:
+        db_table = "evaluation_runs"
+
+
+class EvaluationResult(DataBaseModel):
+    """Result for a single test case in an evaluation run"""
+    id = CharField(max_length=32, primary_key=True)
+    run_id = CharField(max_length=32, null=False, index=True, help_text="FK to evaluation_runs")
+    case_id = CharField(max_length=32, null=False, index=True, help_text="FK to evaluation_cases")
+    generated_answer = TextField(null=False, help_text="generated answer")
+    retrieved_chunks = JSONField(null=False, help_text="chunks that were retrieved")
+    metrics = JSONField(null=False, help_text="all computed metrics")
+    execution_time = FloatField(null=False, help_text="response time in seconds")
+    token_usage = JSONField(null=True, help_text="prompt/completion tokens")
+    create_time = BigIntegerField(null=False, help_text="creation timestamp")
+
+    class Meta:
+        db_table = "evaluation_results"
+
+
+class Memory(DataBaseModel):
+    id = CharField(max_length=32, primary_key=True)
+    name = CharField(max_length=128, null=False, index=False, help_text="Memory name")
+    avatar = TextField(null=True, help_text="avatar base64 string")
+    tenant_id = CharField(max_length=32, null=False, index=True)
+    memory_type = IntegerField(null=False, default=1, index=True, help_text="Bit flags (LSB->MSB): 1=raw, 2=semantic, 4=episodic, 8=procedural. E.g., 5 enables raw + episodic.")
+    storage_type = CharField(max_length=32, default='table', null=False, index=True, help_text="table|graph")
+    embd_id = CharField(max_length=128, null=False, index=False, help_text="embedding model ID")
+    llm_id = CharField(max_length=128, null=False, index=False, help_text="chat model ID")
+    permissions = CharField(max_length=16, null=False, index=True, help_text="me|team", default="me")
+    description = TextField(null=True, help_text="description")
+    memory_size = IntegerField(default=5242880, null=False, index=False)
+    forgetting_policy = CharField(max_length=32, null=False, default="FIFO", index=False, help_text="LRU|FIFO")
+    temperature = FloatField(default=0.5, index=False)
+    system_prompt = TextField(null=True, help_text="system prompt", index=False)
+    user_prompt = TextField(null=True, help_text="user prompt", index=False)
+
+    class Meta:
+        db_table = "memory"
+
+class SystemSettings(DataBaseModel):
+    name = CharField(max_length=128, primary_key=True)
+    source = CharField(max_length=32, null=False, index=False)
+    data_type = CharField(max_length=32, null=False, index=False)
+    value = CharField(max_length=1024, null=False, index=False)
+    class Meta:
+        db_table = "system_settings"
+
+def alter_db_add_column(migrator, table_name, column_name, column_type):
+    try:
+        migrate(migrator.add_column(table_name, column_name, column_type))
+    except OperationalError as ex:
+        error_codes = [1060]
+        error_messages = ['Duplicate column name']
+
+        should_skip_error = (
+                (hasattr(ex, 'args') and ex.args and ex.args[0] in error_codes) or
+                (str(ex) in error_messages)
+        )
+
+        if not should_skip_error:
+            logging.critical(f"Failed to add {settings.DATABASE_TYPE.upper()}.{table_name} column {column_name}, operation error: {ex}")
+
+    except Exception as ex:
+        logging.critical(f"Failed to add {settings.DATABASE_TYPE.upper()}.{table_name} column {column_name}, error: {ex}")
+        pass
+
+def alter_db_column_type(migrator, table_name, column_name, new_column_type):
+    try:
+        migrate(migrator.alter_column_type(table_name, column_name, new_column_type))
+    except Exception as ex:
+        logging.critical(f"Failed to alter {settings.DATABASE_TYPE.upper()}.{table_name} column {column_name} type, error: {ex}")
+        pass
+
+def alter_db_rename_column(migrator, table_name, old_column_name, new_column_name):
+    try:
+        migrate(migrator.rename_column(table_name, old_column_name, new_column_name))
+    except Exception:
+        # rename fail will lead to a weired error.
+        # logging.critical(f"Failed to rename {settings.DATABASE_TYPE.upper()}.{table_name} column {old_column_name} to {new_column_name}, error: {ex}")
+        pass
+
 def migrate_db():
     logging.disable(logging.ERROR)
     migrator = DatabaseMigrator[settings.DATABASE_TYPE.upper()].value(DB)
-    try:
-        migrate(migrator.add_column("file", "source_type", CharField(max_length=128, null=False, default="", help_text="where dose this document come from", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("tenant", "rerank_id", CharField(max_length=128, null=False, default="BAAI/bge-reranker-v2-m3", help_text="default rerank model ID")))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("dialog", "rerank_id", CharField(max_length=128, null=False, default="", help_text="default rerank model ID")))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("dialog", "top_k", IntegerField(default=1024)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.alter_column_type("tenant_llm", "api_key", CharField(max_length=2048, null=True, help_text="API KEY", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("api_token", "source", CharField(max_length=16, null=True, help_text="none|agent|dialog", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("tenant", "tts_id", CharField(max_length=256, null=True, help_text="default tts model ID", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("api_4_conversation", "source", CharField(max_length=16, null=True, help_text="none|agent|dialog", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("task", "retry_count", IntegerField(default=0)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.alter_column_type("api_token", "dialog_id", CharField(max_length=32, null=True, index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("tenant_llm", "max_tokens", IntegerField(default=8192, index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("api_4_conversation", "dsl", JSONField(null=True, default={})))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("knowledgebase", "pagerank", IntegerField(default=0, index=False)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("api_token", "beta", CharField(max_length=255, null=True, index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("task", "digest", TextField(null=True, help_text="task digest", default="")))
-    except Exception:
-        pass
-
-    try:
-        migrate(migrator.add_column("task", "chunk_ids", LongTextField(null=True, help_text="chunk ids", default="")))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("conversation", "user_id", CharField(max_length=255, null=True, help_text="user_id", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("document", "meta_fields", JSONField(null=True, default={})))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("task", "task_type", CharField(max_length=32, null=False, default="")))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("task", "priority", IntegerField(default=0)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("user_canvas", "permission", CharField(max_length=16, null=False, help_text="me|team", default="me", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("llm", "is_tools", BooleanField(null=False, help_text="support tools", default=False)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("mcp_server", "variables", JSONField(null=True, help_text="MCP Server variables", default=dict)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.rename_column("task", "process_duation", "process_duration"))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.rename_column("document", "process_duation", "process_duration"))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("document", "suffix", CharField(max_length=32, null=False, default="", help_text="The real file extension suffix", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("api_4_conversation", "errors", TextField(null=True, help_text="errors")))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("dialog", "meta_data_filter", JSONField(null=True, default={})))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.alter_column_type("canvas_template", "title", JSONField(null=True, default=dict, help_text="Canvas title")))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.alter_column_type("canvas_template", "description", JSONField(null=True, default=dict, help_text="Canvas description")))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("user_canvas", "canvas_category", CharField(max_length=32, null=False, default="agent_canvas", help_text="agent_canvas|dataflow_canvas", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("canvas_template", "canvas_category", CharField(max_length=32, null=False, default="agent_canvas", help_text="agent_canvas|dataflow_canvas", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("knowledgebase", "pipeline_id", CharField(max_length=32, null=True, help_text="Pipeline ID", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("document", "pipeline_id", CharField(max_length=32, null=True, help_text="Pipeline ID", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("knowledgebase", "graphrag_task_id", CharField(max_length=32, null=True, help_text="Gragh RAG task ID", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("knowledgebase", "raptor_task_id", CharField(max_length=32, null=True, help_text="RAPTOR task ID", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("knowledgebase", "graphrag_task_finish_at", DateTimeField(null=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("knowledgebase", "raptor_task_finish_at", CharField(null=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("knowledgebase", "mindmap_task_id", CharField(max_length=32, null=True, help_text="Mindmap task ID", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("knowledgebase", "mindmap_task_finish_at", CharField(null=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.alter_column_type("tenant_llm", "api_key", TextField(null=True, help_text="API KEY")))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("tenant_llm", "status", CharField(max_length=1, null=False, help_text="is it validate(0: wasted, 1: validate)", default="1", index=True)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("connector2kb", "auto_parse", CharField(max_length=1, null=False, default="1", index=False)))
-    except Exception:
-        pass
-    try:
-        migrate(migrator.add_column("llm_factories", "rank", IntegerField(default=0, index=False)))
-    except Exception:
-        pass
+    alter_db_add_column(migrator, "file", "source_type", CharField(max_length=128, null=False, default="", help_text="where dose this document come from", index=True))
+    alter_db_add_column(migrator, "tenant", "rerank_id", CharField(max_length=128, null=False, default="BAAI/bge-reranker-v2-m3", help_text="default rerank model ID"))
+    alter_db_add_column(migrator, "dialog", "rerank_id", CharField(max_length=128, null=False, default="", help_text="default rerank model ID"))
+    alter_db_column_type(migrator, "dialog", "top_k", IntegerField(default=1024))
+    alter_db_add_column(migrator, "tenant_llm", "api_key", CharField(max_length=2048, null=True, help_text="API KEY", index=True))
+    alter_db_add_column(migrator, "api_token", "source", CharField(max_length=16, null=True, help_text="none|agent|dialog", index=True))
+    alter_db_add_column(migrator, "tenant", "tts_id", CharField(max_length=256, null=True, help_text="default tts model ID", index=True))
+    alter_db_add_column(migrator, "api_4_conversation", "source", CharField(max_length=16, null=True, help_text="none|agent|dialog", index=True))
+    alter_db_add_column(migrator, "task", "retry_count", IntegerField(default=0))
+    alter_db_column_type(migrator, "api_token", "dialog_id", CharField(max_length=32, null=True, index=True))
+    alter_db_add_column(migrator, "tenant_llm", "max_tokens", IntegerField(default=8192, index=True))
+    alter_db_add_column(migrator, "api_4_conversation", "dsl", JSONField(null=True, default={}))
+    alter_db_add_column(migrator, "knowledgebase", "pagerank", IntegerField(default=0, index=False))
+    alter_db_add_column(migrator, "api_token", "beta", CharField(max_length=255, null=True, index=True))
+    alter_db_add_column(migrator, "task", "digest", TextField(null=True, help_text="task digest", default=""))
+    alter_db_add_column(migrator, "task", "chunk_ids", LongTextField(null=True, help_text="chunk ids", default=""))
+    alter_db_add_column(migrator, "conversation", "user_id", CharField(max_length=255, null=True, help_text="user_id", index=True))
+    alter_db_add_column(migrator, "document", "meta_fields", JSONField(null=True, default={}))
+    alter_db_add_column(migrator, "task", "task_type", CharField(max_length=32, null=False, default=""))
+    alter_db_add_column(migrator, "task", "priority", IntegerField(default=0))
+    alter_db_add_column(migrator, "user_canvas", "permission", CharField(max_length=16, null=False, help_text="me|team", default="me", index=True))
+    alter_db_add_column(migrator, "llm", "is_tools", BooleanField(null=False, help_text="support tools", default=False))
+    alter_db_add_column(migrator, "mcp_server", "variables", JSONField(null=True, help_text="MCP Server variables", default=dict))
+    alter_db_rename_column(migrator, "task", "process_duation", "process_duration")
+    alter_db_rename_column(migrator, "document", "process_duation", "process_duration")
+    alter_db_add_column(migrator, "document", "suffix", CharField(max_length=32, null=False, default="", help_text="The real file extension suffix", index=True))
+    alter_db_add_column(migrator, "api_4_conversation", "errors", TextField(null=True, help_text="errors"))
+    alter_db_add_column(migrator, "dialog", "meta_data_filter", JSONField(null=True, default={}))
+    alter_db_column_type(migrator, "canvas_template", "title", JSONField(null=True, default=dict, help_text="Canvas title"))
+    alter_db_column_type(migrator, "canvas_template", "description", JSONField(null=True, default=dict, help_text="Canvas description"))
+    alter_db_add_column(migrator, "user_canvas", "canvas_category", CharField(max_length=32, null=False, default="agent_canvas", help_text="agent_canvas|dataflow_canvas", index=True))
+    alter_db_add_column(migrator, "canvas_template", "canvas_category", CharField(max_length=32, null=False, default="agent_canvas", help_text="agent_canvas|dataflow_canvas", index=True))
+    alter_db_add_column(migrator, "knowledgebase", "pipeline_id", CharField(max_length=32, null=True, help_text="Pipeline ID", index=True))
+    alter_db_add_column(migrator, "document", "pipeline_id", CharField(max_length=32, null=True, help_text="Pipeline ID", index=True))
+    alter_db_add_column(migrator, "knowledgebase", "graphrag_task_id", CharField(max_length=32, null=True, help_text="Gragh RAG task ID", index=True))
+    alter_db_add_column(migrator, "knowledgebase", "raptor_task_id", CharField(max_length=32, null=True, help_text="RAPTOR task ID", index=True))
+    alter_db_add_column(migrator, "knowledgebase", "graphrag_task_finish_at", DateTimeField(null=True))
+    alter_db_add_column(migrator, "knowledgebase", "raptor_task_finish_at", CharField(null=True))
+    alter_db_add_column(migrator, "knowledgebase", "mindmap_task_id", CharField(max_length=32, null=True, help_text="Mindmap task ID", index=True))
+    alter_db_add_column(migrator, "knowledgebase", "mindmap_task_finish_at", CharField(null=True))
+    alter_db_column_type(migrator, "tenant_llm", "api_key", TextField(null=True, help_text="API KEY"))
+    alter_db_add_column(migrator, "tenant_llm", "status", CharField(max_length=1, null=False, help_text="is it validate(0: wasted, 1: validate)", default="1", index=True))
+    alter_db_add_column(migrator, "connector2kb", "auto_parse", CharField(max_length=1, null=False, default="1", index=False))
+    alter_db_add_column(migrator, "llm_factories", "rank", IntegerField(default=0, index=False))
     logging.disable(logging.NOTSET)

api/db/init_data.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import asyncio
 import logging
 import json
 import os
@@ -29,19 +30,24 @@ from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.tenant_llm_service import LLMFactoriesService, TenantLLMService
 from api.db.services.llm_service import LLMService, LLMBundle, get_init_tenant_llm
 from api.db.services.user_service import TenantService, UserTenantService
+from api.db.services.system_settings_service import SystemSettingsService
+from api.db.joint_services.memory_message_service import init_message_id_sequence, init_memory_size_cache
 from common.constants import LLMType
 from common.file_utils import get_project_base_directory
 from common import settings
 from api.common.base64 import encode_to_base64
 
+DEFAULT_SUPERUSER_NICKNAME = os.getenv("DEFAULT_SUPERUSER_NICKNAME", "admin")
+DEFAULT_SUPERUSER_EMAIL = os.getenv("DEFAULT_SUPERUSER_EMAIL", "admin@ragflow.io")
+DEFAULT_SUPERUSER_PASSWORD = os.getenv("DEFAULT_SUPERUSER_PASSWORD", "admin")
 
-def init_superuser():
+def init_superuser(nickname=DEFAULT_SUPERUSER_NICKNAME, email=DEFAULT_SUPERUSER_EMAIL, password=DEFAULT_SUPERUSER_PASSWORD, role=UserTenantRole.OWNER):
     user_info = {
         "id": uuid.uuid1().hex,
-        "password": encode_to_base64("admin"),
-        "nickname": "admin",
+        "password": encode_to_base64(password),
+        "nickname": nickname,
         "is_superuser": True,
-        "email": "admin@ragflow.io",
+        "email": email,
         "creator": "system",
         "status": "1",
     }
@@ -58,7 +64,7 @@ def init_superuser():
         "tenant_id": user_info["id"],
         "user_id": user_info["id"],
         "invited_by": user_info["id"],
-        "role": UserTenantRole.OWNER
+        "role": role
     }
 
     tenant_llm = get_init_tenant_llm(user_info["id"])
@@ -70,11 +76,10 @@ def init_superuser():
     UserTenantService.insert(**usr_tenant)
     TenantLLMService.insert_many(tenant_llm)
     logging.info(
-        "Super user initialized. email: admin@ragflow.io, password: admin. Changing the password after login is strongly recommended.")
+        f"Super user initialized. email: {email},A default password has been set; changing the password after login is strongly recommended.")
 
     chat_mdl = LLMBundle(tenant["id"], LLMType.CHAT, tenant["llm_id"])
-    msg = chat_mdl.chat(system="", history=[
-        {"role": "user", "content": "Hello!"}], gen_conf={})
+    msg = asyncio.run(chat_mdl.async_chat(system="", history=[{"role": "user", "content": "Hello!"}], gen_conf={}))
     if msg.find("ERROR: ") == 0:
         logging.error(
             "'{}' doesn't work. {}".format(
@@ -154,20 +159,49 @@ def add_graph_templates():
                 CanvasTemplateService.save(**cnvs)
             except Exception:
                 CanvasTemplateService.update_by_id(cnvs["id"], cnvs)
-        except Exception:
-            logging.exception("Add agent templates error: ")
+        except Exception as e:
+            logging.exception(f"Add agent templates error: {e}")
 
 
 def init_web_data():
     start_time = time.time()
 
+    init_table()
+
     init_llm_factory()
     # if not UserService.get_all().count():
     #    init_superuser()
 
     add_graph_templates()
+    init_message_id_sequence()
+    init_memory_size_cache()
     logging.info("init web data success:{}".format(time.time() - start_time))
 
+def init_table():
+    # init system_settings
+    with open(os.path.join(get_project_base_directory(), "conf", "system_settings.json"), "r") as f:
+        records_from_file = json.load(f)["system_settings"]
+
+    record_index = {}
+    records_from_db = SystemSettingsService.get_all()
+    for index, record in enumerate(records_from_db):
+        record_index[record.name] = index
+
+    to_save = []
+    for record in records_from_file:
+        setting_name = record["name"]
+        if setting_name not in record_index:
+            to_save.append(record)
+
+    len_to_save = len(to_save)
+    if len_to_save > 0:
+        # not initialized
+        try:
+            SystemSettingsService.insert_many(to_save, len_to_save)
+        except Exception as e:
+            logging.exception("System settings init error: {}".format(e))
+            raise e
+
 
 if __name__ == '__main__':
     init_web_db()

api/db/joint_services/memory_message_service.py

@@ -0,0 +1,420 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import logging
+from typing import List
+
+from common import settings
+from common.time_utils import current_timestamp, timestamp_to_date, format_iso_8601_to_ymd_hms
+from common.constants import MemoryType, LLMType
+from common.doc_store.doc_store_base import FusionExpr
+from common.misc_utils import get_uuid
+from api.db.db_utils import bulk_insert_into_db
+from api.db.db_models import Task
+from api.db.services.task_service import TaskService
+from api.db.services.memory_service import MemoryService
+from api.db.services.tenant_llm_service import TenantLLMService
+from api.db.services.llm_service import LLMBundle
+from api.utils.memory_utils import get_memory_type_human
+from memory.services.messages import MessageService
+from memory.services.query import MsgTextQuery, get_vector
+from memory.utils.prompt_util import PromptAssembler
+from memory.utils.msg_util import get_json_result_from_llm_response
+from rag.utils.redis_conn import REDIS_CONN
+
+
+async def save_to_memory(memory_id: str, message_dict: dict):
+    """
+    :param memory_id:
+    :param message_dict: {
+        "user_id": str,
+        "agent_id": str,
+        "session_id": str,
+        "user_input": str,
+        "agent_response": str
+    }
+    """
+    memory = MemoryService.get_by_memory_id(memory_id)
+    if not memory:
+        return False, f"Memory '{memory_id}' not found."
+
+    tenant_id = memory.tenant_id
+    extracted_content = await extract_by_llm(
+        tenant_id,
+        memory.llm_id,
+        {"temperature": memory.temperature},
+        get_memory_type_human(memory.memory_type),
+        message_dict.get("user_input", ""),
+        message_dict.get("agent_response", "")
+    ) if memory.memory_type != MemoryType.RAW.value else []  # if only RAW, no need to extract
+    raw_message_id = REDIS_CONN.generate_auto_increment_id(namespace="memory")
+    message_list = [{
+        "message_id": raw_message_id,
+        "message_type": MemoryType.RAW.name.lower(),
+        "source_id": 0,
+        "memory_id": memory_id,
+        "user_id": "",
+        "agent_id": message_dict["agent_id"],
+        "session_id": message_dict["session_id"],
+        "content": f"User Input: {message_dict.get('user_input')}\nAgent Response: {message_dict.get('agent_response')}",
+        "valid_at": timestamp_to_date(current_timestamp()),
+        "invalid_at": None,
+        "forget_at": None,
+        "status": True
+    }, *[{
+        "message_id": REDIS_CONN.generate_auto_increment_id(namespace="memory"),
+        "message_type": content["message_type"],
+        "source_id": raw_message_id,
+        "memory_id": memory_id,
+        "user_id": "",
+        "agent_id": message_dict["agent_id"],
+        "session_id": message_dict["session_id"],
+        "content": content["content"],
+        "valid_at": content["valid_at"],
+        "invalid_at": content["invalid_at"] if content["invalid_at"] else None,
+        "forget_at": None,
+        "status": True
+    } for content in extracted_content]]
+    return await embed_and_save(memory, message_list)
+
+
+async def save_extracted_to_memory_only(memory_id: str, message_dict, source_message_id: int, task_id: str=None):
+    memory = MemoryService.get_by_memory_id(memory_id)
+    if not memory:
+        msg = f"Memory '{memory_id}' not found."
+        if task_id:
+            TaskService.update_progress(task_id, {"progress": -1, "progress_msg": timestamp_to_date(current_timestamp())+ " " + msg})
+        return False, msg
+
+    if memory.memory_type == MemoryType.RAW.value:
+        msg = f"Memory '{memory_id}' don't need to extract."
+        if task_id:
+            TaskService.update_progress(task_id, {"progress": 1.0, "progress_msg": timestamp_to_date(current_timestamp())+ " " + msg})
+        return True, msg
+
+    tenant_id = memory.tenant_id
+    extracted_content = await extract_by_llm(
+        tenant_id,
+        memory.llm_id,
+        {"temperature": memory.temperature},
+        get_memory_type_human(memory.memory_type),
+        message_dict.get("user_input", ""),
+        message_dict.get("agent_response", ""),
+        task_id=task_id
+    )
+    message_list = [{
+        "message_id": REDIS_CONN.generate_auto_increment_id(namespace="memory"),
+        "message_type": content["message_type"],
+        "source_id": source_message_id,
+        "memory_id": memory_id,
+        "user_id": "",
+        "agent_id": message_dict["agent_id"],
+        "session_id": message_dict["session_id"],
+        "content": content["content"],
+        "valid_at": content["valid_at"],
+        "invalid_at": content["invalid_at"] if content["invalid_at"] else None,
+        "forget_at": None,
+        "status": True
+    } for content in extracted_content]
+    if not message_list:
+        msg = "No memory extracted from raw message."
+        if task_id:
+            TaskService.update_progress(task_id, {"progress": 1.0, "progress_msg": timestamp_to_date(current_timestamp())+ " " + msg})
+        return True, msg
+
+    if task_id:
+        TaskService.update_progress(task_id, {"progress": 0.5, "progress_msg": timestamp_to_date(current_timestamp())+ " " + f"Extracted {len(message_list)} messages from raw dialogue."})
+    return await embed_and_save(memory, message_list, task_id)
+
+
+async def extract_by_llm(tenant_id: str, llm_id: str, extract_conf: dict, memory_type: List[str], user_input: str,
+                         agent_response: str, system_prompt: str = "", user_prompt: str="", task_id: str=None) -> List[dict]:
+    llm_type = TenantLLMService.llm_id2llm_type(llm_id)
+    if not llm_type:
+        raise RuntimeError(f"Unknown type of LLM '{llm_id}'")
+    if not system_prompt:
+        system_prompt = PromptAssembler.assemble_system_prompt({"memory_type": memory_type})
+    conversation_content = f"User Input: {user_input}\nAgent Response: {agent_response}"
+    conversation_time = timestamp_to_date(current_timestamp())
+    user_prompts = []
+    if user_prompt:
+        user_prompts.append({"role": "user", "content": user_prompt})
+        user_prompts.append({"role": "user", "content": f"Conversation: {conversation_content}\nConversation Time: {conversation_time}\nCurrent Time: {conversation_time}"})
+    else:
+        user_prompts.append({"role": "user", "content": PromptAssembler.assemble_user_prompt(conversation_content, conversation_time, conversation_time)})
+    llm = LLMBundle(tenant_id, llm_type, llm_id)
+    if task_id:
+        TaskService.update_progress(task_id, {"progress": 0.15, "progress_msg": timestamp_to_date(current_timestamp())+ " " + "Prepared prompts and LLM."})
+    res = await llm.async_chat(system_prompt, user_prompts, extract_conf)
+    res_json = get_json_result_from_llm_response(res)
+    if task_id:
+        TaskService.update_progress(task_id, {"progress": 0.35, "progress_msg": timestamp_to_date(current_timestamp())+ " " + "Get extracted result from LLM."})
+    return [{
+        "content": extracted_content["content"],
+        "valid_at": format_iso_8601_to_ymd_hms(extracted_content["valid_at"]),
+        "invalid_at": format_iso_8601_to_ymd_hms(extracted_content["invalid_at"]) if extracted_content.get("invalid_at") else "",
+        "message_type": message_type
+    } for message_type, extracted_content_list in res_json.items() for extracted_content in extracted_content_list]
+
+
+async def embed_and_save(memory, message_list: list[dict], task_id: str=None):
+    embedding_model = LLMBundle(memory.tenant_id, llm_type=LLMType.EMBEDDING, llm_name=memory.embd_id)
+    if task_id:
+        TaskService.update_progress(task_id, {"progress": 0.65, "progress_msg": timestamp_to_date(current_timestamp())+ " " + "Prepared embedding model."})
+    vector_list, _ = embedding_model.encode([msg["content"] for msg in message_list])
+    for idx, msg in enumerate(message_list):
+        msg["content_embed"] = vector_list[idx]
+    if task_id:
+        TaskService.update_progress(task_id, {"progress": 0.85, "progress_msg": timestamp_to_date(current_timestamp())+ " " + "Embedded extracted content."})
+    vector_dimension = len(vector_list[0])
+    if not MessageService.has_index(memory.tenant_id, memory.id):
+        created = MessageService.create_index(memory.tenant_id, memory.id, vector_size=vector_dimension)
+        if not created:
+            error_msg = "Failed to create message index."
+            if task_id:
+                TaskService.update_progress(task_id, {"progress": -1, "progress_msg": timestamp_to_date(current_timestamp())+ " " + error_msg})
+            return False, error_msg
+
+    new_msg_size = sum([MessageService.calculate_message_size(m) for m in message_list])
+    current_memory_size = get_memory_size_cache(memory.tenant_id, memory.id)
+    if new_msg_size + current_memory_size > memory.memory_size:
+        size_to_delete = current_memory_size + new_msg_size - memory.memory_size
+        if memory.forgetting_policy == "FIFO":
+            message_ids_to_delete, delete_size = MessageService.pick_messages_to_delete_by_fifo(memory.id, memory.tenant_id,
+                                                                                                size_to_delete)
+            MessageService.delete_message({"message_id": message_ids_to_delete}, memory.tenant_id, memory.id)
+            decrease_memory_size_cache(memory.id, delete_size)
+        else:
+            error_msg = "Failed to insert message into memory. Memory size reached limit and cannot decide which to delete."
+            if task_id:
+                TaskService.update_progress(task_id, {"progress": -1, "progress_msg": timestamp_to_date(current_timestamp())+ " " + error_msg})
+            return False, error_msg
+    fail_cases = MessageService.insert_message(message_list, memory.tenant_id, memory.id)
+    if fail_cases:
+        error_msg = "Failed to insert message into memory. Details: " + "; ".join(fail_cases)
+        if task_id:
+            TaskService.update_progress(task_id, {"progress": -1, "progress_msg": timestamp_to_date(current_timestamp())+ " " + error_msg})
+        return False, error_msg
+
+    if task_id:
+        TaskService.update_progress(task_id, {"progress": 0.95, "progress_msg": timestamp_to_date(current_timestamp())+ " " + "Saved messages to storage."})
+    increase_memory_size_cache(memory.id, new_msg_size)
+    return True, "Message saved successfully."
+
+
+def query_message(filter_dict: dict, params: dict):
+    """
+    :param filter_dict: {
+        "memory_id": List[str],
+        "agent_id": optional
+        "session_id": optional
+    }
+    :param params: {
+        "query": question str,
+        "similarity_threshold": float,
+        "keywords_similarity_weight": float,
+        "top_n": int
+    }
+    """
+    memory_ids = filter_dict["memory_id"]
+    memory_list = MemoryService.get_by_ids(memory_ids)
+    if not memory_list:
+        return []
+
+    condition_dict = {k: v for k, v in filter_dict.items() if v}
+    uids = [memory.tenant_id for memory in memory_list]
+
+    question = params["query"]
+    question = question.strip()
+    memory = memory_list[0]
+    embd_model = LLMBundle(memory.tenant_id, llm_type=LLMType.EMBEDDING, llm_name=memory.embd_id)
+    match_dense = get_vector(question, embd_model, similarity=params["similarity_threshold"])
+    match_text, _ = MsgTextQuery().question(question, min_match=params["similarity_threshold"])
+    keywords_similarity_weight = params.get("keywords_similarity_weight", 0.7)
+    fusion_expr = FusionExpr("weighted_sum", params["top_n"], {"weights": ",".join([str(1 - keywords_similarity_weight), str(keywords_similarity_weight)])})
+
+    return MessageService.search_message(memory_ids, condition_dict, uids, [match_text, match_dense, fusion_expr], params["top_n"])
+
+
+def init_message_id_sequence():
+    message_id_redis_key = "id_generator:memory"
+    if REDIS_CONN.exist(message_id_redis_key):
+        current_max_id = REDIS_CONN.get(message_id_redis_key)
+        logging.info(f"No need to init message_id sequence, current max id is {current_max_id}.")
+    else:
+        max_id = 1
+        exist_memory_list = MemoryService.get_all_memory()
+        if not exist_memory_list:
+            REDIS_CONN.set(message_id_redis_key, max_id)
+        else:
+            max_id = MessageService.get_max_message_id(
+                uid_list=[m.tenant_id for m in exist_memory_list],
+                memory_ids=[m.id for m in exist_memory_list]
+            )
+            REDIS_CONN.set(message_id_redis_key, max_id)
+        logging.info(f"Init message_id sequence done, current max id is {max_id}.")
+
+
+def get_memory_size_cache(memory_id: str, uid: str):
+    redis_key = f"memory_{memory_id}"
+    if REDIS_CONN.exist(redis_key):
+        return int(REDIS_CONN.get(redis_key))
+    else:
+        memory_size_map = MessageService.calculate_memory_size(
+            [memory_id],
+            [uid]
+        )
+        memory_size = memory_size_map.get(memory_id, 0)
+        set_memory_size_cache(memory_id, memory_size)
+        return memory_size
+
+
+def set_memory_size_cache(memory_id: str, size: int):
+    redis_key = f"memory_{memory_id}"
+    return REDIS_CONN.set(redis_key, size)
+
+
+def increase_memory_size_cache(memory_id: str, size: int):
+    redis_key = f"memory_{memory_id}"
+    return REDIS_CONN.incrby(redis_key, size)
+
+
+def decrease_memory_size_cache(memory_id: str, size: int):
+    redis_key = f"memory_{memory_id}"
+    return REDIS_CONN.decrby(redis_key, size)
+
+
+def init_memory_size_cache():
+    memory_list = MemoryService.get_all_memory()
+    if not memory_list:
+        logging.info("No memory found, no need to init memory size.")
+    else:
+        for m in memory_list:
+            get_memory_size_cache(m.id, m.tenant_id)
+        logging.info("Memory size cache init done.")
+
+
+def judge_system_prompt_is_default(system_prompt: str, memory_type: int|list[str]):
+    memory_type_list = memory_type if isinstance(memory_type, list) else get_memory_type_human(memory_type)
+    return system_prompt == PromptAssembler.assemble_system_prompt({"memory_type": memory_type_list})
+
+
+async def queue_save_to_memory_task(memory_ids: list[str], message_dict: dict):
+    """
+    :param memory_ids:
+    :param message_dict: {
+        "user_id": str,
+        "agent_id": str,
+        "session_id": str,
+        "user_input": str,
+        "agent_response": str
+    }
+    """
+    def new_task(_memory_id: str, _source_id: int):
+        return {
+            "id": get_uuid(),
+            "doc_id": _memory_id,
+            "task_type": "memory",
+            "progress": 0.0,
+            "digest": str(_source_id)
+        }
+
+    not_found_memory = []
+    failed_memory = []
+    for memory_id in memory_ids:
+        memory = MemoryService.get_by_memory_id(memory_id)
+        if not memory:
+            not_found_memory.append(memory_id)
+            continue
+
+        raw_message_id = REDIS_CONN.generate_auto_increment_id(namespace="memory")
+        raw_message = {
+            "message_id": raw_message_id,
+            "message_type": MemoryType.RAW.name.lower(),
+            "source_id": 0,
+            "memory_id": memory_id,
+            "user_id": "",
+            "agent_id": message_dict["agent_id"],
+            "session_id": message_dict["session_id"],
+            "content": f"User Input: {message_dict.get('user_input')}\nAgent Response: {message_dict.get('agent_response')}",
+            "valid_at": timestamp_to_date(current_timestamp()),
+            "invalid_at": None,
+            "forget_at": None,
+            "status": True
+        }
+        res, msg = await embed_and_save(memory, [raw_message])
+        if not res:
+            failed_memory.append({"memory_id": memory_id, "fail_msg": msg})
+            continue
+
+        task = new_task(memory_id, raw_message_id)
+        bulk_insert_into_db(Task, [task], replace_on_conflict=True)
+        task_message = {
+            "id": task["id"],
+            "task_id": task["id"],
+            "task_type": task["task_type"],
+            "memory_id": memory_id,
+            "source_id": raw_message_id,
+            "message_dict": message_dict
+        }
+        if not REDIS_CONN.queue_product(settings.get_svr_queue_name(priority=0), message=task_message):
+            failed_memory.append({"memory_id": memory_id, "fail_msg": "Can't access Redis."})
+
+    error_msg = ""
+    if not_found_memory:
+        error_msg = f"Memory {not_found_memory} not found."
+    if failed_memory:
+        error_msg += "".join([f"Memory {fm['memory_id']} failed. Detail: {fm['fail_msg']}" for fm in failed_memory])
+
+    if error_msg:
+        return False, error_msg
+
+    return True, "All add to task."
+
+
+async def handle_save_to_memory_task(task_param: dict):
+    """
+    :param task_param: {
+        "id": task_id
+        "memory_id": id
+        "source_id": id
+        "message_dict": {
+            "user_id": str,
+            "agent_id": str,
+            "session_id": str,
+            "user_input": str,
+            "agent_response": str
+        }
+    }
+    """
+    _, task = TaskService.get_by_id(task_param["id"])
+    if not task:
+        return False, f"Task {task_param['id']} is not found."
+    if task.progress == -1:
+        return False, f"Task {task_param['id']} is already failed."
+    now_time = current_timestamp()
+    TaskService.update_by_id(task_param["id"], {"begin_at": timestamp_to_date(now_time)})
+
+    memory_id = task_param["memory_id"]
+    source_id = task_param["source_id"]
+    message_dict = task_param["message_dict"]
+    success, msg = await save_extracted_to_memory_only(memory_id, message_dict, source_id, task.id)
+    if success:
+        TaskService.update_progress(task.id, {"progress": 1.0,  "progress_msg": timestamp_to_date(current_timestamp())+ " " + msg})
+        return True, msg
+
+    logging.error(msg)
+    TaskService.update_progress(task.id, {"progress": -1, "progress_msg": timestamp_to_date(current_timestamp())+ " " + msg})
+    return False, msg

api/db/joint_services/user_account_service.py

@@ -34,6 +34,8 @@ from api.db.services.task_service import TaskService
 from api.db.services.tenant_llm_service import TenantLLMService
 from api.db.services.user_canvas_version import UserCanvasVersionService
 from api.db.services.user_service import TenantService, UserService, UserTenantService
+from api.db.services.memory_service import MemoryService
+from memory.services.messages import MessageService
 from rag.nlp import search
 from common.constants import ActiveEnum
 from common import settings
@@ -153,7 +155,7 @@ def delete_user_data(user_id: str) -> dict:
             done_msg += "Start to delete owned tenant.\n"
             tenant_id = owned_tenant[0]["tenant_id"]
             kb_ids = KnowledgebaseService.get_kb_ids(usr.id)
-            # step1.1 delete knowledgebase related file and info
+            # step1.1 delete dataset related file and info
             if kb_ids:
                 # step1.1.1 delete files in storage, remove bucket
                 for kb_id in kb_ids:
@@ -182,7 +184,7 @@ def delete_user_data(user_id: str) -> dict:
                                          search.index_name(tenant_id), kb_ids)
                 done_msg += f"- Deleted {r} chunk records.\n"
                 kb_delete_res = KnowledgebaseService.delete_by_ids(kb_ids)
-                done_msg += f"- Deleted {kb_delete_res} knowledgebase records.\n"
+                done_msg += f"- Deleted {kb_delete_res} dataset records.\n"
                 # step1.1.4 delete agents
                 agent_delete_res = delete_user_agents(usr.id)
                 done_msg += f"- Deleted {agent_delete_res['agents_deleted_count']} agent, {agent_delete_res['version_deleted_count']} versions records.\n"
@@ -200,7 +202,16 @@ def delete_user_data(user_id: str) -> dict:
             done_msg += f"- Deleted {llm_delete_res} tenant-LLM records.\n"
             langfuse_delete_res = TenantLangfuseService.delete_ty_tenant_id(tenant_id)
             done_msg += f"- Deleted {langfuse_delete_res} langfuse records.\n"
-            # step1.3 delete own tenant
+            # step1.3 delete memory and messages
+            user_memory = MemoryService.get_by_tenant_id(tenant_id)
+            if user_memory:
+                for memory in user_memory:
+                    if MessageService.has_index(tenant_id, memory.id):
+                        MessageService.delete_index(tenant_id, memory.id)
+                done_msg += " Deleted memory index."
+                memory_delete_res = MemoryService.delete_by_ids([m.id for m in user_memory])
+                done_msg += f"Deleted {memory_delete_res} memory datasets."
+            # step1.4 delete own tenant
             tenant_delete_res = TenantService.delete_by_id(tenant_id)
             done_msg += f"- Deleted {tenant_delete_res} tenant.\n"
         # step2 delete user-tenant relation
@@ -258,7 +269,7 @@ def delete_user_data(user_id: str) -> dict:
                     # step2.1.5 delete document record
                     doc_delete_res = DocumentService.delete_by_ids([d['id'] for d in created_documents])
                     done_msg += f"- Deleted {doc_delete_res} documents.\n"
-                    # step2.1.6 update knowledge base doc&chunk&token cnt
+                    # step2.1.6 update dataset doc&chunk&token cnt
                     for kb_id, doc_num in kb_doc_info.items():
                         KnowledgebaseService.decrease_document_num_in_delete(kb_id, doc_num)
 
@@ -273,7 +284,7 @@ def delete_user_data(user_id: str) -> dict:
 
     except Exception as e:
         logging.exception(e)
-        return {"success": False, "message": f"Error: {str(e)}. Already done:\n{done_msg}"}
+        return {"success": False, "message": "An internal error occurred during user deletion. Some operations may have completed.","details": done_msg}
 
 
 def delete_user_agents(user_id: str) -> dict:

api/db/services/canvas_service.py

@@ -123,6 +123,19 @@ class UserCanvasService(CommonService):
             logging.exception(e)
             return False, None
 
+    @classmethod
+    @DB.connection_context()
+    def get_basic_info_by_canvas_ids(cls, canvas_id):
+        fields = [
+            cls.model.id,
+            cls.model.avatar,
+            cls.model.user_id,
+            cls.model.title,
+            cls.model.permission,
+            cls.model.canvas_category
+        ]
+        return cls.model.select(*fields).where(cls.model.id.in_(canvas_id)).dicts()
+
     @classmethod
     @DB.connection_context()
     def get_by_tenant_ids(cls, joined_tenant_ids, user_id,
@@ -198,7 +211,7 @@ async def completion(tenant_id, agent_id, session_id=None, **kwargs):
         if not isinstance(cvs.dsl, str):
             cvs.dsl = json.dumps(cvs.dsl, ensure_ascii=False)
         session_id=get_uuid()
-        canvas = Canvas(cvs.dsl, tenant_id, agent_id)
+        canvas = Canvas(cvs.dsl, tenant_id, agent_id, canvas_id=cvs.id)
         canvas.reset()
         conv = {
             "id": session_id,

api/db/services/common_service.py

@@ -169,10 +169,12 @@ class CommonService:
         """
         if "id" not in kwargs:
             kwargs["id"] = get_uuid()
-        kwargs["create_time"] = current_timestamp()
-        kwargs["create_date"] = datetime_format(datetime.now())
-        kwargs["update_time"] = current_timestamp()
-        kwargs["update_date"] = datetime_format(datetime.now())
+        timestamp = current_timestamp()
+        cur_datetime = datetime_format(datetime.now())
+        kwargs["create_time"] = timestamp
+        kwargs["create_date"] = cur_datetime
+        kwargs["update_time"] = timestamp
+        kwargs["update_date"] = cur_datetime
         sample_obj = cls.model(**kwargs).save(force_insert=True)
         return sample_obj
 
@@ -188,10 +190,15 @@ class CommonService:
             data_list (list): List of dictionaries containing record data to insert.
             batch_size (int, optional): Number of records to insert in each batch. Defaults to 100.
         """
+        current_ts = current_timestamp()
+        current_datetime = datetime_format(datetime.now())
         with DB.atomic():
             for d in data_list:
-                d["create_time"] = current_timestamp()
-                d["create_date"] = datetime_format(datetime.now())
+                d["create_time"] = current_ts
+                d["create_date"] = current_datetime
+                d["update_time"] = current_ts
+                d["update_date"] = current_datetime
+
             for i in range(0, len(data_list), batch_size):
                 cls.model.insert_many(data_list[i : i + batch_size]).execute()
 
@@ -207,10 +214,14 @@ class CommonService:
             data_list (list): List of dictionaries containing record data to update.
                              Each dictionary must include an 'id' field.
         """
+
+        timestamp = current_timestamp()
+        cur_datetime = datetime_format(datetime.now())
+        for data in data_list:
+            data["update_time"] = timestamp
+            data["update_date"] = cur_datetime
         with DB.atomic():
             for data in data_list:
-                data["update_time"] = current_timestamp()
-                data["update_date"] = datetime_format(datetime.now())
                 cls.model.update(data).where(cls.model.id == data["id"]).execute()
 
     @classmethod

api/db/services/connector_service.py

@@ -29,7 +29,6 @@ from common.misc_utils import get_uuid
 from common.constants import TaskStatus
 from common.time_utils import current_timestamp, timestamp_to_date
 
-
 class ConnectorService(CommonService):
     model = Connector
 
@@ -202,6 +201,7 @@ class SyncLogsService(CommonService):
             return None
 
         class FileObj(BaseModel):
+            id: str
             filename: str
             blob: bytes
 
@@ -209,14 +209,26 @@ class SyncLogsService(CommonService):
                 return self.blob
 
         errs = []
-        files = [FileObj(filename=d["semantic_identifier"]+(f"{d['extension']}" if d["semantic_identifier"][::-1].find(d['extension'][::-1])<0 else ""), blob=d["blob"]) for d in docs]
+        files = [FileObj(id=d["id"], filename=d["semantic_identifier"]+(f"{d['extension']}" if d["semantic_identifier"][::-1].find(d['extension'][::-1])<0 else ""), blob=d["blob"]) for d in docs]
         doc_ids = []
         err, doc_blob_pairs = FileService.upload_document(kb, files, tenant_id, src)
         errs.extend(err)
 
+        # Create a mapping from filename to metadata for later use
+        metadata_map = {}
+        for d in docs:
+            if d.get("metadata"):
+                filename = d["semantic_identifier"]+(f"{d['extension']}" if d["semantic_identifier"][::-1].find(d['extension'][::-1])<0 else "")
+                metadata_map[filename] = d["metadata"]
+
         kb_table_num_map = {}
         for doc, _ in doc_blob_pairs:
             doc_ids.append(doc["id"])
+            
+            # Set metadata if available for this document
+            if doc["name"] in metadata_map:
+                DocumentService.update_by_id(doc["id"], {"meta_fields": metadata_map[doc["name"]]})
+            
             if not auto_parse or auto_parse == "0":
                 continue
             DocumentService.run(tenant_id, doc, kb_table_num_map)

api/db/services/conversation_service.py

@@ -19,7 +19,7 @@ from common.constants import StatusEnum
 from api.db.db_models import Conversation, DB
 from api.db.services.api_service import API4ConversationService
 from api.db.services.common_service import CommonService
-from api.db.services.dialog_service import DialogService, chat
+from api.db.services.dialog_service import DialogService, async_chat
 from common.misc_utils import get_uuid
 import json
 
@@ -64,11 +64,13 @@ class ConversationService(CommonService):
             offset += limit
         return res
 
+
 def structure_answer(conv, ans, message_id, session_id):
     reference = ans["reference"]
     if not isinstance(reference, dict):
         reference = {}
         ans["reference"] = {}
+    is_final = ans.get("final", True)
 
     chunk_list = chunks_format(reference)
 
@@ -81,16 +83,33 @@ def structure_answer(conv, ans, message_id, session_id):
 
     if not conv.message:
         conv.message = []
+    content = ans["answer"]
+    if ans.get("start_to_think"):
+        content = "<think>"
+    elif ans.get("end_to_think"):
+        content = "</think>"
+
     if not conv.message or conv.message[-1].get("role", "") != "assistant":
-        conv.message.append({"role": "assistant", "content": ans["answer"], "created_at": time.time(), "id": message_id})
+        conv.message.append({"role": "assistant", "content": content, "created_at": time.time(), "id": message_id})
     else:
-        conv.message[-1] = {"role": "assistant", "content": ans["answer"], "created_at": time.time(), "id": message_id}
+        if is_final:
+            if ans.get("answer"):
+                conv.message[-1] = {"role": "assistant", "content": ans["answer"], "created_at": time.time(), "id": message_id}
+            else:
+                conv.message[-1]["created_at"] = time.time()
+                conv.message[-1]["id"] = message_id
+        else:
+            conv.message[-1]["content"] = (conv.message[-1].get("content") or "") + content
+            conv.message[-1]["created_at"] = time.time()
+            conv.message[-1]["id"] = message_id
     if conv.reference:
-        conv.reference[-1] = reference
+        should_update_reference = is_final or bool(reference.get("chunks")) or bool(reference.get("doc_aggs"))
+        if should_update_reference:
+            conv.reference[-1] = reference
     return ans
 
 
-def completion(tenant_id, chat_id, question, name="New session", session_id=None, stream=True, **kwargs):
+async def async_completion(tenant_id, chat_id, question, name="New session", session_id=None, stream=True, **kwargs):
     assert name, "`name` can not be empty."
     dia = DialogService.query(id=chat_id, tenant_id=tenant_id, status=StatusEnum.VALID.value)
     assert dia, "You do not own the chat."
@@ -112,11 +131,21 @@ def completion(tenant_id, chat_id, question, name="New session", session_id=None
                                             "reference": {},
                                             "audio_binary": None,
                                             "id": None,
-                                            "session_id": session_id
+                                        "session_id": session_id
                                         }},
                                     ensure_ascii=False) + "\n\n"
             yield "data:" + json.dumps({"code": 0, "message": "", "data": True}, ensure_ascii=False) + "\n\n"
             return
+        else:
+            answer = {
+                "answer": conv["message"][0]["content"],
+                "reference": {},
+                "audio_binary": None,
+                "id": None,
+                "session_id": session_id
+            }
+            yield answer
+            return
 
     conv = ConversationService.query(id=session_id, dialog_id=chat_id)
     if not conv:
@@ -148,7 +177,7 @@ def completion(tenant_id, chat_id, question, name="New session", session_id=None
 
     if stream:
         try:
-            for ans in chat(dia, msg, True, **kwargs):
+            async for ans in async_chat(dia, msg, True, **kwargs):
                 ans = structure_answer(conv, ans, message_id, session_id)
                 yield "data:" + json.dumps({"code": 0, "data": ans}, ensure_ascii=False) + "\n\n"
             ConversationService.update_by_id(conv.id, conv.to_dict())
@@ -160,14 +189,13 @@ def completion(tenant_id, chat_id, question, name="New session", session_id=None
 
     else:
         answer = None
-        for ans in chat(dia, msg, False, **kwargs):
+        async for ans in async_chat(dia, msg, False, **kwargs):
             answer = structure_answer(conv, ans, message_id, session_id)
             ConversationService.update_by_id(conv.id, conv.to_dict())
             break
         yield answer
 
-
-def iframe_completion(dialog_id, question, session_id=None, stream=True, **kwargs):
+async def async_iframe_completion(dialog_id, question, session_id=None, stream=True, **kwargs):
     e, dia = DialogService.get_by_id(dialog_id)
     assert e, "Dialog not found"
     if not session_id:
@@ -222,7 +250,7 @@ def iframe_completion(dialog_id, question, session_id=None, stream=True, **kwarg
 
     if stream:
         try:
-            for ans in chat(dia, msg, True, **kwargs):
+            async for ans in async_chat(dia, msg, True, **kwargs):
                 ans = structure_answer(conv, ans, message_id, session_id)
                 yield "data:" + json.dumps({"code": 0, "message": "", "data": ans},
                                            ensure_ascii=False) + "\n\n"
@@ -235,7 +263,7 @@ def iframe_completion(dialog_id, question, session_id=None, stream=True, **kwarg
 
     else:
         answer = None
-        for ans in chat(dia, msg, False, **kwargs):
+        async for ans in async_chat(dia, msg, False, **kwargs):
             answer = structure_answer(conv, ans, message_id, session_id)
             API4ConversationService.append_message(conv.id, conv.to_dict())
             break