Replaced twine with uv

### What problem does this PR solve?

Feat: Modify the style of the query variable dropdown list. #10866

### Type of change


- [x] New Feature (non-breaking change which adds functionality)

.github/workflows/release.yml

@@ -16,7 +16,7 @@ concurrency:
 
 jobs:
   release:
-    runs-on: [ "self-hosted", "overseas" ]
+    runs-on: [ "self-hosted", "ragflow-test" ]
     steps:
       - name: Ensure workspace ownership
         run: echo "chown -R $USER $GITHUB_WORKSPACE" && sudo chown -R $USER $GITHUB_WORKSPACE
@@ -75,62 +75,20 @@ jobs:
           # The body field does not support environment variable substitution directly.
           body_path: release_body.md
 
-      # https://github.com/marketplace/actions/docker-login
-      - name: Login to Docker Hub
-        uses: docker/login-action@v3
-        with:
-          username: infiniflow
-          password: ${{ secrets.DOCKERHUB_TOKEN }}
-
-      # https://github.com/marketplace/actions/build-and-push-docker-images
-      - name: Build and push full image
-        uses: docker/build-push-action@v6
-        with:
-          context: .
-          push: true
-          tags: |
-            infiniflow/ragflow:${{ env.RELEASE_TAG }}
-            infiniflow/ragflow:latest-full
-          file: Dockerfile
-          platforms: linux/amd64
-
-      # https://github.com/marketplace/actions/build-and-push-docker-images
-      - name: Build and push slim image
-        uses: docker/build-push-action@v6
-        with:
-          context: .
-          push: true
-          tags: |
-            infiniflow/ragflow:${{ env.RELEASE_TAG }}-slim
-            infiniflow/ragflow:latest-slim
-          file: Dockerfile
-          build-args: LIGHTEN=1
-          platforms: linux/amd64
-
-      - name: Build ragflow-sdk
+      - name: Build and push ragflow-sdk
         if: startsWith(github.ref, 'refs/tags/v')
         run: |
-          cd sdk/python && \
-          uv build
+          cd sdk/python && uv build && uv publish --token ${{ secrets.PYPI_API_TOKEN }}
 
-      - name: Publish package distributions to PyPI
-        if: startsWith(github.ref, 'refs/tags/v')
-        uses: pypa/gh-action-pypi-publish@release/v1
-        with:
-          packages-dir: sdk/python/dist/
-          password: ${{ secrets.PYPI_API_TOKEN }}
-          verbose: true
-
-      - name: Build ragflow-cli
+      - name: Build and push ragflow-cli
         if: startsWith(github.ref, 'refs/tags/v')
         run: |
-          cd admin/client && \
-          uv build
+          cd admin/client && uv build && uv publish --token ${{ secrets.PYPI_API_TOKEN }}
 
-      - name: Publish client package distributions to PyPI
-        if: startsWith(github.ref, 'refs/tags/v')
-        uses: pypa/gh-action-pypi-publish@release/v1
-        with:
-          packages-dir: admin/client/dist/
-          password: ${{ secrets.PYPI_API_TOKEN }}
-          verbose: true
+      - name: Build and push image
+        run: |
+          echo ${{ secrets.DOCKERHUB_TOKEN }} | sudo docker login --username infiniflow --password-stdin
+          sudo docker build --build-arg NEED_MIRROR=1 -t infiniflow/ragflow:${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

@@ -10,7 +10,7 @@ on:
       - '*.md'
       - '*.mdx'
   pull_request:
-    types: [ opened, synchronize, reopened, labeled ]
+    types: [ labeled, synchronize, reopened ]
     paths-ignore:
       - 'docs/**'
       - '*.md'
@@ -29,7 +29,7 @@ jobs:
     # 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' || contains(github.event.pull_request.labels.*.name, 'ci') }}
-    runs-on: [ "self-hosted", "debug" ]
+    runs-on: [ "self-hosted", "ragflow-test" ]
     steps:
       # https://github.com/hmarr/debug-action
       #- uses: hmarr/debug-action@v2
@@ -49,20 +49,20 @@ jobs:
       - name: Check workflow duplication
         if: ${{ !cancelled() && !failure() && (github.event_name != 'pull_request' || contains(github.event.pull_request.labels.*.name, 'ci')) }}
         run: |
-          if [[ ${{ github.event_name }} != 'pull_request' ]]; then
+          if [[ "$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 <<< "${{ secrets.GITHUB_TOKEN }}"
             PR_NUMBER=$(gh pr list --search ${HEAD} --state merged --json number --jq .[0].number)
             echo "HEAD=${HEAD}"
             echo "PR_NUMBER=${PR_NUMBER}"
-            if [[ -n ${PR_NUMBER} ]]; then
+            if [[ -n "${PR_NUMBER}" ]]; then
               PR_SHA_FP=${RUNNER_WORKSPACE_PREFIX}/artifacts/${GITHUB_REPOSITORY}/PR_${PR_NUMBER}
-              if [[ -f ${PR_SHA_FP} ]]; then
+              if [[ -f "${PR_SHA_FP}" ]]; then
                 read -r PR_SHA PR_RUN_ID < "${PR_SHA_FP}"
                 # Calculate the hash of the current workspace content
                 HEAD_SHA=$(git rev-parse HEAD^{tree})
-                if [[ ${HEAD_SHA} == ${PR_SHA} ]]; then
+                if [[ "${HEAD_SHA}" == "${PR_SHA}" ]]; then
                   echo "Cancel myself since the workspace content hash is the same with PR #${PR_NUMBER} merged. See ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${PR_RUN_ID} for details."
                   gh run cancel ${GITHUB_RUN_ID}
                   while true; do
@@ -91,122 +91,140 @@ jobs:
           version: ">=0.11.x"
           args: "check"
 
-      - name: Build ragflow:nightly-slim
-        run: |
-          RUNNER_WORKSPACE_PREFIX=${RUNNER_WORKSPACE_PREFIX:-$HOME}
-          sudo docker pull ubuntu:22.04
-          sudo DOCKER_BUILDKIT=1 docker build --build-arg LIGHTEN=1 --build-arg NEED_MIRROR=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .
-
       - name: Build ragflow:nightly
         run: |
-          sudo DOCKER_BUILDKIT=1 docker build --build-arg NEED_MIRROR=1 -f Dockerfile -t infiniflow/ragflow:nightly .
-
-      - name: Start ragflow:nightly-slim
-        run: |
-          sudo docker compose -f docker/docker-compose.yml down --volumes --remove-orphans
-          echo -e "\nRAGFLOW_IMAGE=infiniflow/ragflow:nightly-slim" >> docker/.env
-          sudo docker compose -f docker/docker-compose.yml up -d
-
-      - name: Stop ragflow:nightly-slim
-        if: always()  # always run this step even if previous steps failed
-        run: |
-          sudo docker compose -f docker/docker-compose.yml down -v
+          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 ${RAGFLOW_IMAGE} .
+          if [[ "$GITHUB_EVENT_NAME" == "schedule" ]]; then
+            export HTTP_API_TEST_LEVEL=p3
+          else
+            export HTTP_API_TEST_LEVEL=p2
+          fi
+          echo "HTTP_API_TEST_LEVEL=${HTTP_API_TEST_LEVEL}" >> $GITHUB_ENV
+          echo "RAGFLOW_CONTAINER=${GITHUB_RUN_ID}-ragflow-cpu-1" >> $GITHUB_ENV
 
       - name: Start ragflow:nightly
         run: |
-          echo -e "\nRAGFLOW_IMAGE=infiniflow/ragflow:nightly" >> docker/.env
-          sudo docker compose -f docker/docker-compose.yml up -d
+          # Determine runner number (default to 1 if not found)
+          RUNNER_NUM=$(sudo docker inspect $(hostname) --format '{{index .Config.Labels "com.docker.compose.container-number"}}' 2>/dev/null || true)
+          RUNNER_NUM=${RUNNER_NUM:-1}
+
+          # Compute port numbers using bash arithmetic
+          ES_PORT=$((1200 + RUNNER_NUM * 10))
+          OS_PORT=$((1201 + RUNNER_NUM * 10))
+          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))
+          MINIO_PORT=$((9000 + RUNNER_NUM * 10))
+          MINIO_CONSOLE_PORT=$((9001 + RUNNER_NUM * 10))
+          REDIS_PORT=$((6379 + RUNNER_NUM * 10))
+          TEI_PORT=$((6380 + RUNNER_NUM * 10))
+          KIBANA_PORT=$((6601 + RUNNER_NUM * 10))
+          SVR_HTTP_PORT=$((9380 + RUNNER_NUM * 10))
+          ADMIN_SVR_HTTP_PORT=$((9381 + RUNNER_NUM * 10))
+          SVR_MCP_PORT=$((9382 + RUNNER_NUM * 10))
+          SANDBOX_EXECUTOR_MANAGER_PORT=$((9385 + RUNNER_NUM * 10))
+          SVR_WEB_HTTP_PORT=$((80 + RUNNER_NUM * 10))
+          SVR_WEB_HTTPS_PORT=$((443 + RUNNER_NUM * 10))
+
+          # Persist computed ports into docker/.env so docker-compose uses the correct host bindings
+          echo "" >> docker/.env
+          echo -e "ES_PORT=${ES_PORT}" >> docker/.env
+          echo -e "OS_PORT=${OS_PORT}" >> docker/.env
+          echo -e "INFINITY_THRIFT_PORT=${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=${MYSQL_PORT}" >> docker/.env
+          echo -e "MINIO_PORT=${MINIO_PORT}" >> docker/.env
+          echo -e "MINIO_CONSOLE_PORT=${MINIO_CONSOLE_PORT}" >> docker/.env
+          echo -e "REDIS_PORT=${REDIS_PORT}" >> docker/.env
+          echo -e "TEI_PORT=${TEI_PORT}" >> docker/.env
+          echo -e "KIBANA_PORT=${KIBANA_PORT}" >> docker/.env
+          echo -e "SVR_HTTP_PORT=${SVR_HTTP_PORT}" >> docker/.env
+          echo -e "ADMIN_SVR_HTTP_PORT=${ADMIN_SVR_HTTP_PORT}" >> docker/.env
+          echo -e "SVR_MCP_PORT=${SVR_MCP_PORT}" >> docker/.env
+          echo -e "SANDBOX_EXECUTOR_MANAGER_PORT=${SANDBOX_EXECUTOR_MANAGER_PORT}" >> docker/.env
+          echo -e "SVR_WEB_HTTP_PORT=${SVR_WEB_HTTP_PORT}" >> docker/.env
+          echo -e "SVR_WEB_HTTPS_PORT=${SVR_WEB_HTTPS_PORT}" >> docker/.env
+
+          echo -e "COMPOSE_PROFILES=\${COMPOSE_PROFILES},tei-cpu" >> docker/.env
+          echo -e "TEI_MODEL=BAAI/bge-small-en-v1.5" >> docker/.env
+          echo -e "RAGFLOW_IMAGE=${RAGFLOW_IMAGE}" >> docker/.env
+          echo "HOST_ADDRESS=http://host.docker.internal:${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
 
       - 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=""
-          export HOST_ADDRESS=http://host.docker.internal:9380
-          until sudo docker exec ragflow-server curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          if [[ $GITHUB_EVENT_NAME == 'schedule' ]]; then
-            export HTTP_API_TEST_LEVEL=p3
-          else
-            export HTTP_API_TEST_LEVEL=p2
-          fi
-          UV_LINK_MODE=copy uv sync --python 3.10 --only-group test --no-default-groups --frozen && uv pip install sdk/python && uv run --only-group test --no-default-groups pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api
+          source .venv/bin/activate && pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api
 
       - name: Run frontend api tests against Elasticsearch
         run: |
           export http_proxy=""; export https_proxy=""; export no_proxy=""; export HTTP_PROXY=""; export HTTPS_PROXY=""; export NO_PROXY=""
-          export HOST_ADDRESS=http://host.docker.internal:9380
-          until sudo docker exec ragflow-server curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          cd sdk/python && UV_LINK_MODE=copy uv sync --python 3.10 --group test --frozen && source .venv/bin/activate && cd test/test_frontend_api && pytest -s --tb=short get_email.py test_dataset.py
+          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
           
       - 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=""
-          export HOST_ADDRESS=http://host.docker.internal:9380
-          until sudo docker exec ragflow-server curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          if [[ $GITHUB_EVENT_NAME == 'schedule' ]]; then
-            export HTTP_API_TEST_LEVEL=p3
-          else
-            export HTTP_API_TEST_LEVEL=p2
-          fi
-          UV_LINK_MODE=copy uv sync --python 3.10 --only-group test --no-default-groups --frozen && uv run --only-group test --no-default-groups pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api
+          source .venv/bin/activate && pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api
 
       - name: Stop ragflow:nightly
         if: always()  # always run this step even if previous steps failed
         run: |
-          sudo docker compose -f docker/docker-compose.yml down -v
+          sudo docker compose -f docker/docker-compose.yml -p ${GITHUB_RUN_ID} down -v
 
       - name: Start ragflow:nightly
         run: |
-          sudo DOC_ENGINE=infinity docker compose -f docker/docker-compose.yml up -d
+          sed -i '1i DOC_ENGINE=infinity' docker/.env
+          sudo docker compose -f docker/docker-compose.yml -p ${GITHUB_RUN_ID} up -d
 
       - 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=""
-          export HOST_ADDRESS=http://host.docker.internal:9380
-          until sudo docker exec ragflow-server curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          if [[ $GITHUB_EVENT_NAME == 'schedule' ]]; then
-            export HTTP_API_TEST_LEVEL=p3
-          else
-            export HTTP_API_TEST_LEVEL=p2
-          fi
-          UV_LINK_MODE=copy uv sync --python 3.10 --only-group test --no-default-groups --frozen && uv pip install sdk/python && DOC_ENGINE=infinity uv run --only-group test --no-default-groups pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api
+          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api
 
       - name: Run frontend api tests against Infinity
         run: |
           export http_proxy=""; export https_proxy=""; export no_proxy=""; export HTTP_PROXY=""; export HTTPS_PROXY=""; export NO_PROXY=""
-          export HOST_ADDRESS=http://host.docker.internal:9380
-          until sudo docker exec ragflow-server curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          cd sdk/python && UV_LINK_MODE=copy uv sync --python 3.10 --group test --frozen && source .venv/bin/activate && cd test/test_frontend_api && pytest -s --tb=short get_email.py test_dataset.py
+          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
 
       - 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=""
-          export HOST_ADDRESS=http://host.docker.internal:9380
-          until sudo docker exec ragflow-server curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
+          until sudo docker exec ${RAGFLOW_CONTAINER} curl -s --connect-timeout 5 ${HOST_ADDRESS} > /dev/null; do
             echo "Waiting for service to be available..."
             sleep 5
           done
-          if [[ $GITHUB_EVENT_NAME == 'schedule' ]]; then
-            export HTTP_API_TEST_LEVEL=p3
-          else
-            export HTTP_API_TEST_LEVEL=p2
-          fi
-          UV_LINK_MODE=copy uv sync --python 3.10 --only-group test --no-default-groups --frozen && DOC_ENGINE=infinity uv run --only-group test --no-default-groups pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api
+          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api
 
       - name: Stop ragflow:nightly
         if: always()  # always run this step even if previous steps failed
         run: |
-          sudo DOC_ENGINE=infinity docker compose -f docker/docker-compose.yml down -v
+          sudo docker compose -f docker/docker-compose.yml -p ${GITHUB_RUN_ID} down -v
+          sudo docker rmi -f ${RAGFLOW_IMAGE:-NO_IMAGE} || true

Dockerfile

@@ -4,8 +4,6 @@ USER root
 SHELL ["/bin/bash", "-c"]
 
 ARG NEED_MIRROR=0
-ARG LIGHTEN=0
-ENV LIGHTEN=${LIGHTEN}
 
 WORKDIR /ragflow
 
@@ -17,13 +15,6 @@ RUN --mount=type=bind,from=infiniflow/ragflow_deps:latest,source=/huggingface.co
         /huggingface.co/InfiniFlow/text_concat_xgb_v1.0 \
         /huggingface.co/InfiniFlow/deepdoc \
         | tar -xf - --strip-components=3 -C /ragflow/rag/res/deepdoc 
-RUN --mount=type=bind,from=infiniflow/ragflow_deps:latest,source=/huggingface.co,target=/huggingface.co \
-    if [ "$LIGHTEN" != "1" ]; then \
-        (tar -cf - \
-            /huggingface.co/BAAI/bge-large-zh-v1.5 \
-            /huggingface.co/maidalun1020/bce-embedding-base_v1 \
-            | tar -xf - --strip-components=2 -C /root/.ragflow) \
-    fi
 
 # 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.
@@ -63,11 +54,11 @@ RUN --mount=type=cache,id=ragflow_apt,target=/var/cache/apt,sharing=locked \
     apt install -y ghostscript
 
 RUN if [ "$NEED_MIRROR" == "1" ]; then \
-        pip3 config set global.index-url https://mirrors.aliyun.com/pypi/simple && \
-        pip3 config set global.trusted-host mirrors.aliyun.com; \
+        pip3 config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple && \
+        pip3 config set global.trusted-host pypi.tuna.tsinghua.edu.cn; \
         mkdir -p /etc/uv && \
         echo "[[index]]" > /etc/uv/uv.toml && \
-        echo 'url = "https://mirrors.aliyun.com/pypi/simple"' >> /etc/uv/uv.toml && \
+        echo 'url = "https://pypi.tuna.tsinghua.edu.cn/simple"' >> /etc/uv/uv.toml && \
         echo "default = true" >> /etc/uv/uv.toml; \
     fi; \
     pipx install uv
@@ -151,15 +142,11 @@ COPY pyproject.toml uv.lock ./
 # uv records index url into uv.lock but doesn't failover among multiple indexes
 RUN --mount=type=cache,id=ragflow_uv,target=/root/.cache/uv,sharing=locked \
     if [ "$NEED_MIRROR" == "1" ]; then \
-        sed -i 's|pypi.org|mirrors.aliyun.com/pypi|g' uv.lock; \
+        sed -i 's|pypi.org|pypi.tuna.tsinghua.edu.cn|g' uv.lock; \
     else \
-        sed -i 's|mirrors.aliyun.com/pypi|pypi.org|g' uv.lock; \
+        sed -i 's|pypi.tuna.tsinghua.edu.cn|pypi.org|g' uv.lock; \
     fi; \
-    if [ "$LIGHTEN" == "1" ]; then \
-        uv sync --python 3.10 --frozen; \
-    else \
-        uv sync --python 3.10 --frozen --all-extras; \
-    fi
+    uv sync --python 3.10 --frozen
 
 COPY web web
 COPY docs docs
@@ -169,11 +156,7 @@ RUN --mount=type=cache,id=ragflow_npm,target=/root/.npm,sharing=locked \
 COPY .git /ragflow/.git
 
 RUN version_info=$(git describe --tags --match=v* --first-parent --always); \
-    if [ "$LIGHTEN" == "1" ]; then \
-        version_info="$version_info slim"; \
-    else \
-        version_info="$version_info full"; \
-    fi; \
+    version_info="$version_info"; \
     echo "RAGFlow version: $version_info"; \
     echo $version_info > /ragflow/VERSION
 
@@ -202,6 +185,7 @@ COPY agentic_reasoning agentic_reasoning
 COPY pyproject.toml uv.lock ./
 COPY mcp mcp
 COPY plugin plugin
+COPY common common
 
 COPY docker/service_conf.yaml.template ./conf/service_conf.yaml.template
 COPY docker/entrypoint.sh ./

Dockerfile_tei

@@ -0,0 +1,14 @@
+FROM ghcr.io/huggingface/text-embeddings-inference:cpu-1.8
+
+# uv tool install huggingface_hub
+# hf download --local-dir tei_data/BAAI/bge-small-en-v1.5 BAAI/bge-small-en-v1.5
+# hf download --local-dir tei_data/BAAI/bge-m3 BAAI/bge-m3
+# hf download --local-dir tei_data/Qwen/Qwen3-Embedding-0.6B Qwen/Qwen3-Embedding-0.6B
+COPY tei_data /data
+
+# curl -X POST http://localhost:6380/embed -H "Content-Type: application/json" -d '{"inputs": "Hello, world! This is a test sentence."}'
+# curl -X POST http://tei:80/embed -H "Content-Type: application/json" -d '{"inputs": "Hello, world! This is a test sentence."}'
+# [[-0.058816575,0.019564206,0.026697718,...]]
+
+# curl -X POST http://localhost:6380/v1/embeddings -H "Content-Type: application/json" -d '{"input": "Hello, world! This is a test sentence."}'
+# {"object":"list","data":[{"object":"embedding","embedding":[-0.058816575,0.019564206,...],"index":0}],"model":"BAAI/bge-small-en-v1.5","usage":{"prompt_tokens":12,"total_tokens":12}}

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:v0.21.0">
+        <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:v0.21.1">
     </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">
@@ -43,7 +43,9 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
-#
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/ragflow-octoverse.png" width="1200"/>
+</div>
 
 <div align="center">
 <a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
@@ -84,6 +86,7 @@ Try our demo at [https://demo.ragflow.io](https://demo.ragflow.io).
 
 ## 🔥 Latest Updates
 
+- 2025-10-23 Supports MinerU & Docling as document parsing methods.
 - 2025-10-15 Supports orchestrable ingestion pipeline.
 - 2025-08-08 Supports OpenAI's latest GPT-5 series models.
 - 2025-08-01 Supports agentic workflow and MCP.
@@ -135,7 +138,7 @@ releases! 🌟
 ## 🔎 System Architecture
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/d6ac5664-c237-4200-a7c2-a4a00691b485" width="1000"/>
+<img src="https://github.com/user-attachments/assets/31b0dd6f-ca4f-445a-9457-70cb44a381b2" width="1000"/>
 </div>
 
 ## 🎬 Get Started
@@ -174,41 +177,42 @@ releases! 🌟
    > ```bash
    > vm.max_map_count=262144
    > ```
-
+   >
 2. Clone the repo:
 
    ```bash
    $ git clone https://github.com/infiniflow/ragflow.git
    ```
-
 3. Start up the server using the pre-built Docker images:
 
 > [!CAUTION]
 > 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 `v0.21.0-slim` edition of the RAGFlow Docker image. See the following table for descriptions of different RAGFlow editions. To download a RAGFlow edition different from `v0.21.0-slim`, update the `RAGFLOW_IMAGE` variable accordingly in **docker/.env** before using `docker compose` to start the server. For example: set `RAGFLOW_IMAGE=infiniflow/ragflow:v0.21.0` for the full edition `v0.21.0`.
+> The command below downloads the `v0.21.1-slim` edition of the RAGFlow Docker image. See the following table for descriptions of different RAGFlow editions. To download a RAGFlow edition different from `v0.21.1-slim`, update the `RAGFLOW_IMAGE` variable accordingly in **docker/.env** before using `docker compose` to start the server.
 
-   ```bash
+```bash
    $ cd ragflow/docker
    # Use CPU for embedding and DeepDoc tasks:
    $ docker compose -f docker-compose.yml up -d
 
    # To use GPU to accelerate embedding and DeepDoc tasks:
-   # docker compose -f docker-compose-gpu.yml up -d
-   ```
+   # sed -i '1i DEVICE=gpu' .env
+   # docker compose -f docker-compose.yml up -d
+```
 
-   | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-   |-------------------|-----------------|-----------------------|--------------------------|
-   | v0.21.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.21.0-slim      | &approx;2       | ❌                   | Stable release            |
-   | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
-   | nightly-slim      | &approx;2       | ❌                   | _Unstable_ nightly build  |
+| 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             |
+| nightly           | &approx;2       | ❌                    | _Unstable_ nightly build |
+
+> Note: Starting with `v0.22.0`, we ship only the slim edition and no longer append the **-slim** suffix to the image tag.
 
 4. Check the server status after having the server up and running:
 
    ```bash
-   $ docker logs -f ragflow-server
+   $ docker logs -f docker-ragflow-cpu-1
    ```
 
    _The following output confirms a successful launch of the system:_
@@ -226,14 +230,17 @@ releases! 🌟
 
    > If you skip this confirmation step and directly log in to RAGFlow, your browser may prompt a `network anormal`
    > 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.
+
    > With the default settings, you only need to enter `http://IP_OF_YOUR_MACHINE` (**sans** port number) as the default
    > HTTP serving port `80` can be omitted when using the default configurations.
+   >
 6. In [service_conf.yaml.template](./docker/service_conf.yaml.template), select the desired LLM factory in `user_default_llm` and update
    the `API_KEY` field with the corresponding API key.
 
    > See [llm_api_key_setup](https://ragflow.io/docs/dev/llm_api_key_setup) for more information.
+   >
 
    _The show is on!_
 
@@ -272,7 +279,6 @@ RAGFlow uses Elasticsearch by default for storing full text and vectors. To swit
 > `-v` will delete the docker container volumes, and the existing data will be cleared.
 
 2. Set `DOC_ENGINE` in **docker/.env** to `infinity`.
-
 3. Start the containers:
 
    ```bash
@@ -286,16 +292,6 @@ RAGFlow uses Elasticsearch by default for storing full text and vectors. To swit
 
 This image is approximately 2 GB in size and relies on external LLM and embedding services.
 
-```bash
-git clone https://github.com/infiniflow/ragflow.git
-cd ragflow/
-docker build --platform linux/amd64 --build-arg LIGHTEN=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .
-```
-
-## 🔧 Build a Docker image including embedding models
-
-This image is approximately 9 GB in size. As it includes embedding models, it relies on external LLM services only.
-
 ```bash
 git clone https://github.com/infiniflow/ragflow.git
 cd ragflow/
@@ -309,17 +305,15 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    pipx install uv pre-commit
    ```
-
 2. Clone the source code and install Python dependencies:
 
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 --all-extras # install RAGFlow dependent python modules
+   uv sync --python 3.10 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
-
 3. Launch the dependent services (MinIO, Elasticsearch, Redis, and MySQL) using Docker Compose:
 
    ```bash
@@ -331,13 +325,11 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```
    127.0.0.1       es01 infinity mysql minio redis sandbox-executor-manager
    ```
-
 4. If you cannot access HuggingFace, set the `HF_ENDPOINT` environment variable to use a mirror site:
 
    ```bash
    export HF_ENDPOINT=https://hf-mirror.com
    ```
-
 5. If your operating system does not have jemalloc, please install it as follows:
 
    ```bash
@@ -350,7 +342,6 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    # macOS
    sudo brew install jemalloc
    ```
-
 6. Launch backend service:
 
    ```bash
@@ -358,14 +349,12 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    export PYTHONPATH=$(pwd)
    bash docker/launch_backend_service.sh
    ```
-
 7. Install frontend dependencies:
 
    ```bash
    cd web
    npm install
    ```
-
 8. Launch frontend service:
 
    ```bash
@@ -375,14 +364,12 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    _The following output confirms a successful launch of the system:_
 
    ![](https://github.com/user-attachments/assets/0daf462c-a24d-4496-a66f-92533534e187)
-
 9. Stop RAGFlow front-end and back-end service after development is complete:
 
    ```bash
    pkill -f "ragflow_server.py|task_executor.py"
    ```
 
-
 ## 📚 Documentation
 
 - [Quickstart](https://ragflow.io/docs/dev/)

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:v0.21.0">
+        <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:v0.21.1">
     </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">
@@ -43,7 +43,13 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
-#
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/ragflow-octoverse.png" width="1200"/>
+</div>
+
+<div align="center">
+<a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</div>
 
 <details open>
 <summary><b>📕 Daftar Isi </b> </summary>
@@ -80,6 +86,7 @@ Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
 
 ## 🔥 Pembaruan Terbaru
 
+- 2025-10-23 Mendukung MinerU & Docling sebagai metode penguraian dokumen.
 - 2025-10-15 Dukungan untuk jalur data yang terorkestrasi.
 - 2025-08-08 Mendukung model seri GPT-5 terbaru dari OpenAI.
 - 2025-08-01 Mendukung alur kerja agen dan MCP.
@@ -129,7 +136,7 @@ Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
 ## 🔎 Arsitektur Sistem
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/d6ac5664-c237-4200-a7c2-a4a00691b485" width="1000"/>
+<img src="https://github.com/user-attachments/assets/31b0dd6f-ca4f-445a-9457-70cb44a381b2" width="1000"/>
 </div>
 
 ## 🎬 Mulai
@@ -168,41 +175,42 @@ Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
    > ```bash
    > vm.max_map_count=262144
    > ```
-
+   >
 2. Clone repositori:
 
    ```bash
    $ git clone https://github.com/infiniflow/ragflow.git
    ```
-
 3. Bangun image Docker pre-built dan jalankan server:
 
 > [!CAUTION]
 > 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 v0.21.0-slim dari gambar Docker RAGFlow. Silakan merujuk ke tabel berikut untuk deskripsi berbagai edisi RAGFlow. Untuk mengunduh edisi RAGFlow yang berbeda dari v0.21.0-slim, perbarui variabel RAGFLOW_IMAGE di docker/.env sebelum menggunakan docker compose untuk memulai server. Misalnya, atur RAGFLOW_IMAGE=infiniflow/ragflow:v0.21.0 untuk edisi lengkap v0.21.0.
+> Perintah di bawah ini mengunduh edisi v0.21.1 dari gambar Docker RAGFlow. Silakan merujuk ke tabel berikut untuk deskripsi berbagai edisi RAGFlow. Untuk mengunduh edisi RAGFlow yang berbeda dari v0.21.1, perbarui variabel RAGFLOW_IMAGE di docker/.env sebelum menggunakan docker compose untuk memulai server.
 
 ```bash
-$ cd ragflow/docker
-# Use CPU for embedding and DeepDoc tasks:
-$ docker compose -f docker-compose.yml up -d
+   $ cd ragflow/docker
+   # Use CPU for embedding and DeepDoc tasks:
+   $ docker compose -f docker-compose.yml up -d
 
-# To use GPU to accelerate embedding and DeepDoc tasks:
-# docker compose -f docker-compose-gpu.yml up -d
+   # To use GPU to accelerate embedding and DeepDoc tasks:
+   # sed -i '1i DEVICE=gpu' .env
+   # docker compose -f docker-compose.yml up -d
 ```
 
-| RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-| ----------------- | --------------- | --------------------- | ------------------------ |
-| v0.21.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-| v0.21.0-slim      | &approx;2       | ❌                    | Stable release           |
-| nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
-| nightly-slim      | &approx;2       | ❌                    | _Unstable_ nightly build |
+| 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             |
+| nightly           | &approx;2       | ❌                    | _Unstable_ nightly build |
+
+> Catatan: Mulai dari `v0.22.0`, kami hanya menyediakan edisi slim dan tidak lagi menambahkan akhiran **-slim** pada tag image.
 
 1. Periksa status server setelah server aktif dan berjalan:
 
    ```bash
-   $ docker logs -f ragflow-server
+   $ docker logs -f docker-ragflow-cpu-1
    ```
 
    _Output berikut menandakan bahwa sistem berhasil diluncurkan:_
@@ -220,14 +228,17 @@ $ docker compose -f docker-compose.yml up -d
 
    > Jika Anda melewatkan langkah ini dan langsung login ke RAGFlow, browser Anda mungkin menampilkan error `network anormal`
    > karena RAGFlow mungkin belum sepenuhnya siap.
-
+   >
 2. Buka browser web Anda, masukkan alamat IP server Anda, dan login ke RAGFlow.
+
    > Dengan pengaturan default, Anda hanya perlu memasukkan `http://IP_DEVICE_ANDA` (**tanpa** nomor port) karena
    > port HTTP default `80` bisa dihilangkan saat menggunakan konfigurasi default.
+   >
 3. Dalam [service_conf.yaml.template](./docker/service_conf.yaml.template), pilih LLM factory yang diinginkan di `user_default_llm` dan perbarui
    bidang `API_KEY` dengan kunci API yang sesuai.
 
    > Lihat [llm_api_key_setup](https://ragflow.io/docs/dev/llm_api_key_setup) untuk informasi lebih lanjut.
+   >
 
    _Sistem telah siap digunakan!_
 
@@ -253,16 +264,6 @@ Pembaruan konfigurasi ini memerlukan reboot semua kontainer agar efektif:
 
 Image ini berukuran sekitar 2 GB dan bergantung pada aplikasi LLM eksternal dan embedding.
 
-```bash
-git clone https://github.com/infiniflow/ragflow.git
-cd ragflow/
-docker build --platform linux/amd64 --build-arg LIGHTEN=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .
-```
-
-## 🔧 Membangun Docker Image Termasuk Model Embedding
-
-Image ini berukuran sekitar 9 GB. Karena sudah termasuk model embedding, ia hanya bergantung pada aplikasi LLM eksternal.
-
 ```bash
 git clone https://github.com/infiniflow/ragflow.git
 cd ragflow/
@@ -276,17 +277,15 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    pipx install uv pre-commit
    ```
-
 2. Clone kode sumber dan instal dependensi Python:
 
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 --all-extras # install RAGFlow dependent python modules
+   uv sync --python 3.10 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
-
 3. Jalankan aplikasi yang diperlukan (MinIO, Elasticsearch, Redis, dan MySQL) menggunakan Docker Compose:
 
    ```bash
@@ -298,13 +297,11 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```
    127.0.0.1       es01 infinity mysql minio redis sandbox-executor-manager
    ```
-
 4. Jika Anda tidak dapat mengakses HuggingFace, atur variabel lingkungan `HF_ENDPOINT` untuk menggunakan situs mirror:
 
    ```bash
    export HF_ENDPOINT=https://hf-mirror.com
    ```
-
 5. Jika sistem operasi Anda tidak memiliki jemalloc, instal sebagai berikut:
 
    ```bash
@@ -315,7 +312,6 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    # mac
    sudo brew install jemalloc
    ```
-
 6. Jalankan aplikasi backend:
 
    ```bash
@@ -323,14 +319,12 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    export PYTHONPATH=$(pwd)
    bash docker/launch_backend_service.sh
    ```
-
 7. Instal dependensi frontend:
 
    ```bash
    cd web
    npm install
    ```
-
 8. Jalankan aplikasi frontend:
 
    ```bash
@@ -340,15 +334,12 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    _Output berikut menandakan bahwa sistem berhasil diluncurkan:_
 
    ![](https://github.com/user-attachments/assets/0daf462c-a24d-4496-a66f-92533534e187)
-
-
 9. Hentikan layanan front-end dan back-end RAGFlow setelah pengembangan selesai:
 
    ```bash
    pkill -f "ragflow_server.py|task_executor.py"
    ```
 
-
 ## 📚 Dokumentasi
 
 - [Quickstart](https://ragflow.io/docs/dev/)

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:v0.21.0">
+        <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:v0.21.1">
     </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">
@@ -43,7 +43,13 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
-#
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/ragflow-octoverse.png" width="1200"/>
+</div>
+
+<div align="center">
+<a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</div>
 
 ## 💡 RAGFlow とは？
 
@@ -60,6 +66,7 @@
 
 ## 🔥 最新情報
 
+- 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。
 - 2025-08-08 OpenAI の最新 GPT-5 シリーズモデルをサポートします。
 - 2025-08-01 エージェントワークフローとMCPをサポート。
@@ -109,7 +116,7 @@
 ## 🔎 システム構成
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/d6ac5664-c237-4200-a7c2-a4a00691b485" width="1000"/>
+<img src="https://github.com/user-attachments/assets/31b0dd6f-ca4f-445a-9457-70cb44a381b2" width="1000"/>
 </div>
 
 ## 🎬 初期設定
@@ -147,41 +154,42 @@
    > ```bash
    > vm.max_map_count=262144
    > ```
-
+   >
 2. リポジトリをクローンする:
 
    ```bash
    $ git clone https://github.com/infiniflow/ragflow.git
    ```
-
 3. ビルド済みの Docker イメージをビルドし、サーバーを起動する:
 
 > [!CAUTION]
 > 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 v0.21.0-slim 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。v0.21.0-slim 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。例えば、完全版 v0.21.0 をダウンロードするには、RAGFLOW_IMAGE=infiniflow/ragflow:v0.21.0 と設定します。
+> 以下のコマンドは、RAGFlow Docker イメージの v0.21.1 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。v0.21.1 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
+```bash
    $ cd ragflow/docker
    # Use CPU for embedding and DeepDoc tasks:
    $ docker compose -f docker-compose.yml up -d
 
    # To use GPU to accelerate embedding and DeepDoc tasks:
-   # docker compose -f docker-compose-gpu.yml up -d
-   ```
+   # sed -i '1i DEVICE=gpu' .env
+   # docker compose -f docker-compose.yml up -d
+```
 
-   | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-   | ----------------- | --------------- | --------------------- | ------------------------ |
-   | v0.21.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.21.0-slim      | &approx;2       | ❌                    | Stable release           |
-   | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
-   | nightly-slim      | &approx;2       | ❌                     | _Unstable_ nightly build |
+| 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             |
+| nightly           | &approx;2       | ❌                    | _Unstable_ nightly build |
+
+> 注意：`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:
 
    ```bash
-   $ docker logs -f ragflow-server
+   $ docker logs -f docker-ragflow-cpu-1
    ```
 
    _以下の出力は、システムが正常に起動したことを確認するものです:_
@@ -197,12 +205,15 @@
    ```
 
    > もし確認ステップをスキップして直接 RAGFlow にログインした場合、その時点で RAGFlow が完全に初期化されていない可能性があるため、ブラウザーがネットワーク異常エラーを表示するかもしれません。
-
+   >
 2. ウェブブラウザで、プロンプトに従ってサーバーの IP アドレスを入力し、RAGFlow にログインします。
+
    > デフォルトの設定を使用する場合、デフォルトの HTTP サービングポート `80` は省略できるので、与えられたシナリオでは、`http://IP_OF_YOUR_MACHINE`（ポート番号は省略）だけを入力すればよい。
+   >
 3. [service_conf.yaml.template](./docker/service_conf.yaml.template) で、`user_default_llm` で希望の LLM ファクトリを選択し、`API_KEY` フィールドを対応する API キーで更新する。
 
    > 詳しくは [llm_api_key_setup](https://ragflow.io/docs/dev/llm_api_key_setup) を参照してください。
+   >
 
    _これで初期設定完了！ショーの開幕です！_
 
@@ -231,33 +242,27 @@
 RAGFlow はデフォルトで Elasticsearch を使用して全文とベクトルを保存します。［Infinity］に切り替え（https://github.com/infiniflow/infinity/)、次の手順に従います。
 
 1. 実行中のすべてのコンテナを停止するには：
+
    ```bash
    $ docker compose -f docker/docker-compose.yml down -v
    ```
+
    Note: `-v` は docker コンテナのボリュームを削除し、既存のデータをクリアします。
 2. **docker/.env** の「DOC \_ ENGINE」を「infinity」に設定します。
-
 3. 起動コンテナ：
+
    ```bash
    $ docker compose -f docker-compose.yml up -d
    ```
+
    > [!WARNING]
    > Linux/arm64 マシンでの Infinity への切り替えは正式にサポートされていません。
+   >
 
 ## 🔧 ソースコードで Docker イメージを作成（埋め込みモデルなし）
 
 この Docker イメージのサイズは約 1GB で、外部の大モデルと埋め込みサービスに依存しています。
 
-```bash
-git clone https://github.com/infiniflow/ragflow.git
-cd ragflow/
-docker build --platform linux/amd64 --build-arg LIGHTEN=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .
-```
-
-## 🔧 ソースコードをコンパイルした Docker イメージ（埋め込みモデルを含む）
-
-この Docker のサイズは約 9GB で、埋め込みモデルを含むため、外部の大モデルサービスのみが必要です。
-
 ```bash
 git clone https://github.com/infiniflow/ragflow.git
 cd ragflow/
@@ -271,17 +276,15 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    pipx install uv pre-commit
    ```
-
 2. ソースコードをクローンし、Python の依存関係をインストールする:
 
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 --all-extras # install RAGFlow dependent python modules
+   uv sync --python 3.10 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
-
 3. Docker Compose を使用して依存サービス（MinIO、Elasticsearch、Redis、MySQL）を起動する:
 
    ```bash
@@ -293,13 +296,11 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```
    127.0.0.1       es01 infinity mysql minio redis sandbox-executor-manager
    ```
-
 4. HuggingFace にアクセスできない場合は、`HF_ENDPOINT` 環境変数を設定してミラーサイトを使用してください:
 
    ```bash
    export HF_ENDPOINT=https://hf-mirror.com
    ```
-
 5. オペレーティングシステムにjemallocがない場合は、次のようにインストールします:
 
    ```bash
@@ -310,7 +311,6 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    # mac
    sudo brew install jemalloc
    ```
-
 6. バックエンドサービスを起動する:
 
    ```bash
@@ -318,14 +318,12 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    export PYTHONPATH=$(pwd)
    bash docker/launch_backend_service.sh
    ```
-
 7. フロントエンドの依存関係をインストールする:
 
    ```bash
    cd web
    npm install
    ```
-
 8. フロントエンドサービスを起動する:
 
    ```bash
@@ -335,14 +333,12 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    _以下の画面で、システムが正常に起動したことを示します:_
 
    ![](https://github.com/user-attachments/assets/0daf462c-a24d-4496-a66f-92533534e187)
-
 9. 開発が完了したら、RAGFlow のフロントエンド サービスとバックエンド サービスを停止します:
 
    ```bash
    pkill -f "ragflow_server.py|task_executor.py"
    ```
 
-
 ## 📚 ドキュメンテーション
 
 - [Quickstart](https://ragflow.io/docs/dev/)

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:v0.21.0">
+        <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:v0.21.1">
     </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">
@@ -43,7 +43,14 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
-#
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/ragflow-octoverse.png" width="1200"/>
+</div>
+
+<div align="center">
+<a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</div>
+
 
 ## 💡 RAGFlow란?
 
@@ -60,6 +67,7 @@
 
 ## 🔥 업데이트
 
+- 2025-10-23 문서 파싱 방법으로 MinerU 및 Docling을 지원합니다.
 - 2025-10-15 조정된 데이터 파이프라인 지원.
 - 2025-08-08 OpenAI의 최신 GPT-5 시리즈 모델을 지원합니다.
 - 2025-08-01 에이전트 워크플로우와 MCP를 지원합니다.
@@ -109,7 +117,7 @@
 ## 🔎 시스템 아키텍처
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/d6ac5664-c237-4200-a7c2-a4a00691b485" width="1000"/>
+<img src="https://github.com/user-attachments/assets/31b0dd6f-ca4f-445a-9457-70cb44a381b2" width="1000"/>
 </div>
 
 ## 🎬 시작하기
@@ -160,7 +168,7 @@
 > 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 v0.21.0-slim 버전을 다운로드합니다. 다양한 RAGFlow 버전에 대한 설명은 다음 표를 참조하십시오. v0.21.0-slim과 다른 RAGFlow 버전을 다운로드하려면, docker/.env 파일에서 RAGFLOW_IMAGE 변수를 적절히 업데이트한 후 docker compose를 사용하여 서버를 시작하십시오. 예를 들어, 전체 버전인 v0.21.0을 다운로드하려면 RAGFLOW_IMAGE=infiniflow/ragflow:v0.21.0로 설정합니다.
+   > 아래 명령어는 RAGFlow Docker 이미지의 v0.21.1 버전을 다운로드합니다. 다양한 RAGFlow 버전에 대한 설명은 다음 표를 참조하십시오. v0.21.1과 다른 RAGFlow 버전을 다운로드하려면, docker/.env 파일에서 RAGFLOW_IMAGE 변수를 적절히 업데이트한 후 docker compose를 사용하여 서버를 시작하십시오.
 
    ```bash
    $ cd ragflow/docker
@@ -168,20 +176,22 @@
    $ docker compose -f docker-compose.yml up -d
 
    # To use GPU to accelerate embedding and DeepDoc tasks:
-   # docker compose -f docker-compose-gpu.yml up -d
+   # sed -i '1i DEVICE=gpu' .env
+   # docker compose -f docker-compose.yml up -d
    ```
 
    | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
    | ----------------- | --------------- | --------------------- | ------------------------ |
-   | v0.21.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.21.0-slim      | &approx;2       | ❌                    | Stable release           |
-   | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
-   | nightly-slim      | &approx;2       | ❌                     | _Unstable_ nightly build |
+   | v0.21.1           | &approx;9       | ✔️                    | Stable release           |
+   | v0.21.1-slim      | &approx;2       | ❌                    | Stable release           |
+   | nightly           | &approx;2       | ❌                    | _Unstable_ nightly build |
+
+   > 참고: `v0.22.0`부터는 slim 에디션만 배포하며 이미지 태그에 **-slim** 접미사를 더 이상 붙이지 않습니다.
 
 1. 서버가 시작된 후 서버 상태를 확인하세요:
 
    ```bash
-   $ docker logs -f ragflow-server
+   $ docker logs -f docker-ragflow-cpu-1
    ```
 
    _다음 출력 결과로 시스템이 성공적으로 시작되었음을 확인합니다:_
@@ -247,16 +257,6 @@ RAGFlow 는 기본적으로 Elasticsearch 를 사용하여 전체 텍스트 및
 
 이 Docker 이미지의 크기는 약 1GB이며, 외부 대형 모델과 임베딩 서비스에 의존합니다.
 
-```bash
-git clone https://github.com/infiniflow/ragflow.git
-cd ragflow/
-docker build --platform linux/amd64 --build-arg LIGHTEN=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .
-```
-
-## 🔧 소스 코드로 Docker 이미지를 컴파일합니다(임베딩 모델 포함)
-
-이 Docker의 크기는 약 9GB이며, 이미 임베딩 모델을 포함하고 있으므로 외부 대형 모델 서비스에만 의존하면 됩니다.
-
 ```bash
 git clone https://github.com/infiniflow/ragflow.git
 cd ragflow/
@@ -276,7 +276,7 @@ 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 --all-extras # install RAGFlow dependent python modules
+   uv sync --python 3.10 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```

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:v0.21.0">
+        <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:v0.21.1">
     </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">
@@ -43,7 +43,13 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
-#
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/ragflow-octoverse.png" width="1200"/>
+</div>
+
+<div align="center">
+<a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</div>
 
 <details open>
 <summary><b>📕 Índice</b></summary>
@@ -80,7 +86,8 @@ Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
 
 ## 🔥 Últimas Atualizações
 
-- 10-15-2025 Suporte para pipelines de dados orquestrados.
+- 23-10-2025 Suporta MinerU e Docling como métodos de análise de documentos.
+- 15-10-2025 Suporte para pipelines de dados orquestrados.
 - 08-08-2025 Suporta a mais recente série GPT-5 da OpenAI.
 - 01-08-2025 Suporta fluxo de trabalho agente e MCP.
 - 23-05-2025 Adicione o componente executor de código Python/JS ao Agente.
@@ -129,7 +136,7 @@ Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
 ## 🔎 Arquitetura do Sistema
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/d6ac5664-c237-4200-a7c2-a4a00691b485" width="1000"/>
+<img src="https://github.com/user-attachments/assets/31b0dd6f-ca4f-445a-9457-70cb44a381b2" width="1000"/>
 </div>
 
 ## 🎬 Primeiros Passos
@@ -147,84 +154,86 @@ Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
 
 ### 🚀 Iniciar o servidor
 
-1.  Certifique-se de que `vm.max_map_count` >= 262144:
+1. Certifique-se de que `vm.max_map_count` >= 262144:
 
-    > Para verificar o valor de `vm.max_map_count`:
-    >
-    > ```bash
-    > $ sysctl vm.max_map_count
-    > ```
-    >
-    > Se necessário, redefina `vm.max_map_count` para um valor de pelo menos 262144:
-    >
-    > ```bash
-    > # Neste caso, defina para 262144:
-    > $ sudo sysctl -w vm.max_map_count=262144
-    > ```
-    >
-    > Essa mudança será resetada após a reinicialização do sistema. Para garantir que a alteração permaneça permanente, adicione ou atualize o valor de `vm.max_map_count` em **/etc/sysctl.conf**:
-    >
-    > ```bash
-    > vm.max_map_count=262144
-    > ```
+   > Para verificar o valor de `vm.max_map_count`:
+   >
+   > ```bash
+   > $ sysctl vm.max_map_count
+   > ```
+   >
+   > Se necessário, redefina `vm.max_map_count` para um valor de pelo menos 262144:
+   >
+   > ```bash
+   > # Neste caso, defina para 262144:
+   > $ sudo sysctl -w vm.max_map_count=262144
+   > ```
+   >
+   > Essa mudança será resetada após a reinicialização do sistema. Para garantir que a alteração permaneça permanente, adicione ou atualize o valor de `vm.max_map_count` em **/etc/sysctl.conf**:
+   >
+   > ```bash
+   > vm.max_map_count=262144
+   > ```
+   >
+2. Clone o repositório:
 
-2.  Clone o repositório:
-
-    ```bash
-    $ git clone https://github.com/infiniflow/ragflow.git
-    ```
-
-3.  Inicie o servidor usando as imagens Docker pré-compiladas:
+   ```bash
+   $ git clone https://github.com/infiniflow/ragflow.git
+   ```
+3. Inicie o servidor usando as imagens Docker pré-compiladas:
 
 > [!CAUTION]
 > 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 `v0.21.0-slim` 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 `v0.21.0-slim`, atualize a variável `RAGFLOW_IMAGE` conforme necessário no **docker/.env** antes de usar `docker compose` para iniciar o servidor. Por exemplo: defina `RAGFLOW_IMAGE=infiniflow/ragflow:v0.21.0` para a edição completa `v0.21.0`.
+    > O comando abaixo baixa a edição`v0.21.1` 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 `v0.21.1`, atualize a variável `RAGFLOW_IMAGE` conforme necessário no **docker/.env** antes de usar `docker compose` para iniciar o servidor.
 
-    ```bash
-    $ cd ragflow/docker
-    # Use CPU for embedding and DeepDoc tasks:
-    $ docker compose -f docker-compose.yml up -d
+```bash
+   $ cd ragflow/docker
+   # Use CPU for embedding and DeepDoc tasks:
+   $ docker compose -f docker-compose.yml up -d
 
-    # To use GPU to accelerate embedding and DeepDoc tasks:
-    # docker compose -f docker-compose-gpu.yml up -d
-    ```
+   # To use GPU to accelerate embedding and DeepDoc tasks:
+   # sed -i '1i DEVICE=gpu' .env
+   # docker compose -f docker-compose.yml up -d
+```
 
-    | Tag da imagem RAGFlow | Tamanho da imagem (GB) | Possui modelos de incorporação? | Estável?                 |
-    | --------------------- | ---------------------- | ------------------------------- | ------------------------ |
-    | v0.21.0               | ~9                     | :heavy_check_mark:              | Lançamento estável       |
-    | v0.21.0-slim          | ~2                     | ❌                              | Lançamento estável       |
-    | nightly               | ~9                     | :heavy_check_mark:              | _Instável_ build noturno |
-    | nightly-slim          | ~2                     | ❌                               | _Instável_ build noturno |
+| Tag da imagem RAGFlow | Tamanho da imagem (GB) | Possui modelos de incorporação? | Estável?                      |
+| --------------------- | ---------------------- | --------------------------------- | ------------------------------ |
+| v0.21.1               | &approx;9              | ✔️                              | Lançamento estável           |
+| v0.21.1-slim          | &approx;2              | ❌                                | Lançamento estável           |
+| nightly               | &approx;2              | ❌                                | Construção noturna instável |
 
-4.  Verifique o status do servidor após tê-lo iniciado:
+    > Observação: A partir da`v0.22.0`, distribuímos apenas a edição slim e não adicionamos mais o sufixo **-slim** às tags das imagens.
 
-    ```bash
-    $ docker logs -f ragflow-server
-    ```
+4. Verifique o status do servidor após tê-lo iniciado:
 
-    _O seguinte resultado confirma o lançamento bem-sucedido do sistema:_
+   ```bash
+   $ docker logs -f docker-ragflow-cpu-1
+   ```
 
-    ```bash
-         ____   ___    ______ ______ __
-        / __ \ /   |  / ____// ____// /____  _      __
-       / /_/ // /| | / / __ / /_   / // __ \| | /| / /
-      / _, _// ___ |/ /_/ // __/  / // /_/ /| |/ |/ /
-     /_/ |_|/_/  |_|\____//_/    /_/ \____/ |__/|__/
+   _O seguinte resultado confirma o lançamento bem-sucedido do sistema:_
 
-     * Rodando em todos os endereços (0.0.0.0)
-    ```
+   ```bash
+        ____   ___    ______ ______ __
+       / __ \ /   |  / ____// ____// /____  _      __
+      / /_/ // /| | / / __ / /_   / // __ \| | /| / /
+     / _, _// ___ |/ /_/ // __/  / // /_/ /| |/ |/ /
+    /_/ |_|/_/  |_|\____//_/    /_/ \____/ |__/|__/
 
-    > 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.
+    * Rodando em todos os endereços (0.0.0.0)
+   ```
 
-5.  No seu navegador, insira o endereço IP do seu servidor e faça login no RAGFlow.
+   > 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.
+   >
+5. No seu navegador, insira o endereço IP do seu servidor e faça login no RAGFlow.
 
-    > Com as configurações padrão, você só precisa digitar `http://IP_DO_SEU_MÁQUINA` (**sem** o número da porta), pois a porta HTTP padrão `80` pode ser omitida ao usar as configurações padrão.
+   > Com as configurações padrão, você só precisa digitar `http://IP_DO_SEU_MÁQUINA` (**sem** o número da porta), pois a porta HTTP padrão `80` pode ser omitida ao usar as configurações padrão.
+   >
+6. Em [service_conf.yaml.template](./docker/service_conf.yaml.template), selecione a fábrica LLM desejada em `user_default_llm` e atualize o campo `API_KEY` com a chave de API correspondente.
 
-6.  Em [service_conf.yaml.template](./docker/service_conf.yaml.template), selecione a fábrica LLM desejada em `user_default_llm` e atualize o campo `API_KEY` com a chave de API correspondente.
-
-    > Consulte [llm_api_key_setup](https://ragflow.io/docs/dev/llm_api_key_setup) para mais informações.
+   > Consulte [llm_api_key_setup](https://ragflow.io/docs/dev/llm_api_key_setup) para mais informações.
+   >
 
 _O show está no ar!_
 
@@ -255,9 +264,9 @@ O RAGFlow usa o Elasticsearch por padrão para armazenar texto completo e vetore
    ```bash
    $ docker compose -f docker/docker-compose.yml down -v
    ```
+
    Note: `-v` irá deletar os volumes do contêiner, e os dados existentes serão apagados.
 2. Defina `DOC_ENGINE` no **docker/.env** para `infinity`.
-
 3. Inicie os contêineres:
 
    ```bash
@@ -271,16 +280,6 @@ O RAGFlow usa o Elasticsearch por padrão para armazenar texto completo e vetore
 
 Esta imagem tem cerca de 2 GB de tamanho e depende de serviços externos de LLM e incorporação.
 
-```bash
-git clone https://github.com/infiniflow/ragflow.git
-cd ragflow/
-docker build --platform linux/amd64 --build-arg LIGHTEN=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .
-```
-
-## 🔧 Criar uma imagem Docker incluindo modelos de incorporação
-
-Esta imagem tem cerca de 9 GB de tamanho. Como inclui modelos de incorporação, depende apenas de serviços externos de LLM.
-
 ```bash
 git clone https://github.com/infiniflow/ragflow.git
 cd ragflow/
@@ -294,17 +293,15 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```bash
    pipx install uv pre-commit
    ```
-
 2. Clone o código-fonte e instale as dependências Python:
 
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 --all-extras # instala os módulos Python dependentes do RAGFlow
+   uv sync --python 3.10 # instala os módulos Python dependentes do RAGFlow
    uv run download_deps.py
    pre-commit install
    ```
-
 3. Inicie os serviços dependentes (MinIO, Elasticsearch, Redis e MySQL) usando Docker Compose:
 
    ```bash
@@ -316,24 +313,21 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    ```
    127.0.0.1       es01 infinity mysql minio redis sandbox-executor-manager
    ```
-
 4. Se não conseguir acessar o HuggingFace, defina a variável de ambiente `HF_ENDPOINT` para usar um site espelho:
 
    ```bash
    export HF_ENDPOINT=https://hf-mirror.com
    ```
-
 5. Se o seu sistema operacional não tiver jemalloc, instale-o da seguinte maneira:
 
-    ```bash
-    # ubuntu
-    sudo apt-get install libjemalloc-dev
-    # centos
-    sudo yum instalar jemalloc
-    # mac
-    sudo brew install jemalloc
-    ```
-
+   ```bash
+   # ubuntu
+   sudo apt-get install libjemalloc-dev
+   # centos
+   sudo yum instalar jemalloc
+   # mac
+   sudo brew install jemalloc
+   ```
 6. Lance o serviço de back-end:
 
    ```bash
@@ -341,14 +335,12 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    export PYTHONPATH=$(pwd)
    bash docker/launch_backend_service.sh
    ```
-
 7. Instale as dependências do front-end:
 
    ```bash
    cd web
    npm install
    ```
-
 8. Lance o serviço de front-end:
 
    ```bash
@@ -358,13 +350,11 @@ docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly
    _O seguinte resultado confirma o lançamento bem-sucedido do sistema:_
 
    ![](https://github.com/user-attachments/assets/0daf462c-a24d-4496-a66f-92533534e187)
-
 9. Pare os serviços de front-end e back-end do RAGFlow após a conclusão do desenvolvimento:
 
-    ```bash
-    pkill -f "ragflow_server.py|task_executor.py"
-    ```
-
+   ```bash
+   pkill -f "ragflow_server.py|task_executor.py"
+   ```
 
 ## 📚 Documentação
 

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:v0.21.0">
+        <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:v0.21.1">
     </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">
@@ -43,7 +43,9 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
-#
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/ragflow-octoverse.png" width="1200"/>
+</div>
 
 <div align="center">
 <a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
@@ -83,6 +85,7 @@
 
 ## 🔥 近期更新
 
+- 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
@@ -132,7 +135,7 @@
 ## 🔎 系統架構
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/d6ac5664-c237-4200-a7c2-a4a00691b485" width="1000"/>
+<img src="https://github.com/user-attachments/assets/31b0dd6f-ca4f-445a-9457-70cb44a381b2" width="1000"/>
 </div>
 
 ## 🎬 快速開始
@@ -170,47 +173,48 @@
    > ```bash
    > vm.max_map_count=262144
    > ```
-
+   >
 2. 克隆倉庫：
 
    ```bash
    $ git clone https://github.com/infiniflow/ragflow.git
    ```
-
 3. 進入 **docker** 資料夾，利用事先編譯好的 Docker 映像啟動伺服器：
 
 > [!CAUTION]
 > 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 slim Docker 映像 `v0.21.0-slim`。請參考下表查看不同 Docker 發行版的說明。如需下載不同於 `v0.21.0-slim` 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。例如，你可以透過設定 `RAGFLOW_IMAGE=infiniflow/ragflow:v0.21.0` 來下載 RAGFlow 鏡像的 `v0.21.0` 完整發行版。
+> 執行以下指令會自動下載 RAGFlow slim Docker 映像 `v0.21.1`。請參考下表查看不同 Docker 發行版的說明。如需下載不同於 `v0.21.1` 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
+```bash
    $ cd ragflow/docker
    # Use CPU for embedding and DeepDoc tasks:
    $ docker compose -f docker-compose.yml up -d
 
    # To use GPU to accelerate embedding and DeepDoc tasks:
-   # docker compose -f docker-compose-gpu.yml up -d
-   ```
+   # sed -i '1i DEVICE=gpu' .env
+   # docker compose -f docker-compose.yml up -d
+```
 
-   | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
-   | ----------------- | --------------- | --------------------- | ------------------------ |
-   | v0.21.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.21.0-slim      | &approx;2       | ❌                    | Stable release           |
-   | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
-   | nightly-slim      | &approx;2       | ❌                     | _Unstable_ nightly build |
+| 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             |
+| nightly           | &approx;2       | ❌                    | _Unstable_ nightly build |
 
-   > [!TIP]
-   > 如果你遇到 Docker 映像檔拉不下來的問題，可以在 **docker/.env** 檔案內根據變數 `RAGFLOW_IMAGE` 的註解提示選擇華為雲或阿里雲的對應映像。
-   >
-   > - 華為雲鏡像名：`swr.cn-north-4.myhuaweicloud.com/infiniflow/ragflow`
-   > - 阿里雲鏡像名：`registry.cn-hangzhou.aliyuncs.com/infiniflow/ragflow`
+> 注意：自 `v0.22.0` 起，我們僅發佈 slim 版本，並且不再在映像標籤後附加 **-slim** 後綴。
+
+> [!TIP]
+> 如果你遇到 Docker 映像檔拉不下來的問題，可以在 **docker/.env** 檔案內根據變數 `RAGFLOW_IMAGE` 的註解提示選擇華為雲或阿里雲的對應映像。
+>
+> - 華為雲鏡像名：`swr.cn-north-4.myhuaweicloud.com/infiniflow/ragflow`
+> - 阿里雲鏡像名：`registry.cn-hangzhou.aliyuncs.com/infiniflow/ragflow`
 
 4. 伺服器啟動成功後再次確認伺服器狀態：
 
    ```bash
-   $ docker logs -f ragflow-server
+   $ docker logs -f docker-ragflow-cpu-1
    ```
 
    _出現以下介面提示說明伺服器啟動成功：_
@@ -226,12 +230,15 @@
    ```
 
    > 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。
-
+   >
 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。
+
    > 上面這個範例中，您只需輸入 http://IP_OF_YOUR_MACHINE 即可：未改動過設定則無需輸入連接埠（預設的 HTTP 服務連接埠 80）。
+   >
 6. 在 [service_conf.yaml.template](./docker/service_conf.yaml.template) 檔案的 `user_default_llm` 欄位設定 LLM factory，並在 `API_KEY` 欄填入和你選擇的大模型相對應的 API key。
 
    > 詳見 [llm_api_key_setup](https://ragflow.io/docs/dev/llm_api_key_setup)。
+   >
 
    _好戲開始，接著奏樂接著舞！ _
 
@@ -249,7 +256,7 @@
 
 > [./docker/README](./docker/README.md) 解釋了 [service_conf.yaml.template](./docker/service_conf.yaml.template) 用到的環境變數設定和服務配置。
 
-如需更新預設的 HTTP 服務連接埠(80), 可以在[docker-compose.yml](./docker/docker-compose.yml) 檔案中將配置`80:80` 改為`<YOUR_SERVING_PORT>:80` 。
+如需更新預設的 HTTP 服務連接埠(80), 可以在[docker-compose.yml](./docker/docker-compose.yml) 檔案中將配置 `80:80` 改為 `<YOUR_SERVING_PORT>:80` 。
 
 > 所有系統配置都需要透過系統重新啟動生效：
 >
@@ -266,10 +273,9 @@ RAGFlow 預設使用 Elasticsearch 儲存文字和向量資料. 如果要切換
    ```bash
    $ docker compose -f docker/docker-compose.yml down -v
    ```
+
    Note: `-v` 將會刪除 docker 容器的 volumes，已有的資料會被清空。
-
 2. 設定 **docker/.env** 目錄中的 `DOC_ENGINE` 為 `infinity`.
-
 3. 啟動容器:
 
    ```bash
@@ -286,17 +292,7 @@ RAGFlow 預設使用 Elasticsearch 儲存文字和向量資料. 如果要切換
 ```bash
 git clone https://github.com/infiniflow/ragflow.git
 cd ragflow/
-docker build --platform linux/amd64 --build-arg LIGHTEN=1 --build-arg NEED_MIRROR=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .
-```
-
-## 🔧 原始碼編譯 Docker 映像（包含 embedding 模型）
-
-本 Docker 大小約 9 GB 左右。由於已包含 embedding 模型，所以只需依賴外部的大模型服務即可。
-
-```bash
-git clone https://github.com/infiniflow/ragflow.git
-cd ragflow/
-docker build --platform linux/amd64 --build-arg NEED_MIRROR=1 -f Dockerfile -t infiniflow/ragflow:nightly .
+docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly .
 ```
 
 ## 🔨 以原始碼啟動服務
@@ -307,17 +303,15 @@ docker build --platform linux/amd64 --build-arg NEED_MIRROR=1 -f Dockerfile -t i
    pipx install uv pre-commit
    export UV_INDEX=https://mirrors.aliyun.com/pypi/simple
    ```
-
 2. 下載原始碼並安裝 Python 依賴：
 
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 --all-extras # install RAGFlow dependent python modules
+   uv sync --python 3.10 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```
-
 3. 透過 Docker Compose 啟動依賴的服務（MinIO, Elasticsearch, Redis, and MySQL）：
 
    ```bash
@@ -329,13 +323,11 @@ docker build --platform linux/amd64 --build-arg NEED_MIRROR=1 -f Dockerfile -t i
    ```
    127.0.0.1       es01 infinity mysql minio redis sandbox-executor-manager
    ```
-
 4. 如果無法存取 HuggingFace，可以把環境變數 `HF_ENDPOINT` 設為對應的鏡像網站：
 
    ```bash
    export HF_ENDPOINT=https://hf-mirror.com
    ```
-
 5. 如果你的操作系统没有 jemalloc，请按照如下方式安装：
 
    ```bash
@@ -346,7 +338,6 @@ docker build --platform linux/amd64 --build-arg NEED_MIRROR=1 -f Dockerfile -t i
    # mac
    sudo brew install jemalloc
    ```
-
 6. 啟動後端服務：
 
    ```bash
@@ -354,14 +345,12 @@ docker build --platform linux/amd64 --build-arg NEED_MIRROR=1 -f Dockerfile -t i
    export PYTHONPATH=$(pwd)
    bash docker/launch_backend_service.sh
    ```
-
 7. 安裝前端依賴：
 
    ```bash
    cd web
    npm install
    ```
-
 8. 啟動前端服務：
 
    ```bash
@@ -371,15 +360,16 @@ docker build --platform linux/amd64 --build-arg NEED_MIRROR=1 -f Dockerfile -t i
    以下界面說明系統已成功啟動：_
 
    ![](https://github.com/user-attachments/assets/0daf462c-a24d-4496-a66f-92533534e187)
+
    ```
 
+   ```
 9. 開發完成後停止 RAGFlow 前端和後端服務：
 
    ```bash
    pkill -f "ragflow_server.py|task_executor.py"
    ```
 
-
 ## 📚 技術文檔
 
 - [Quickstart](https://ragflow.io/docs/dev/)

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:v0.21.0">
+        <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:v0.21.1">
     </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">
@@ -43,7 +43,9 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
-#
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/ragflow-octoverse.png" width="1200"/>
+</div>
 
 <div align="center">
 <a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
@@ -83,6 +85,7 @@
 
 ## 🔥 近期更新
 
+- 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。
 - 2025-08-08 支持 OpenAI 最新的 GPT-5 系列模型。
 - 2025-08-01 支持 agentic workflow 和 MCP。
@@ -132,7 +135,7 @@
 ## 🔎 系统架构
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/d6ac5664-c237-4200-a7c2-a4a00691b485" width="1000"/>
+<img src="https://github.com/user-attachments/assets/31b0dd6f-ca4f-445a-9457-70cb44a381b2" width="1000"/>
 </div>
 
 ## 🎬 快速开始
@@ -183,7 +186,7 @@
 > 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 slim Docker 镜像 `v0.21.0-slim`。请参考下表查看不同 Docker 发行版的描述。如需下载不同于 `v0.21.0-slim` 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。比如，你可以通过设置 `RAGFLOW_IMAGE=infiniflow/ragflow:v0.21.0` 来下载 RAGFlow 镜像的 `v0.21.0` 完整发行版。
+   > 运行以下命令会自动下载 RAGFlow slim Docker 镜像 `v0.21.1`。请参考下表查看不同 Docker 发行版的描述。如需下载不同于 `v0.21.1` 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
@@ -191,15 +194,17 @@
    $ docker compose -f docker-compose.yml up -d
 
    # To use GPU to accelerate embedding and DeepDoc tasks:
-   # docker compose -f docker-compose-gpu.yml up -d
+   # sed -i '1i DEVICE=gpu' .env
+   # docker compose -f docker-compose.yml up -d
    ```
 
    | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
    | ----------------- | --------------- | --------------------- | ------------------------ |
-   | v0.21.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.21.0-slim      | &approx;2       | ❌                    | Stable release           |
-   | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
-   | nightly-slim      | &approx;2       | ❌                     | _Unstable_ nightly build |
+   | v0.21.1           | &approx;9       | ✔️                    | Stable release           |
+   | v0.21.1-slim      | &approx;2       | ❌                    | Stable release           |
+   | nightly           | &approx;2       | ❌                    | _Unstable_ nightly build |
+
+   > 注意：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。
 
    > [!TIP]
    > 如果你遇到 Docker 镜像拉不下来的问题，可以在 **docker/.env** 文件内根据变量 `RAGFLOW_IMAGE` 的注释提示选择华为云或者阿里云的相应镜像。
@@ -210,7 +215,7 @@
 4. 服务器启动成功后再次确认服务器状态：
 
    ```bash
-   $ docker logs -f ragflow-server
+   $ docker logs -f docker-ragflow-cpu-1
    ```
 
    _出现以下界面提示说明服务器启动成功：_
@@ -286,17 +291,7 @@ RAGFlow 默认使用 Elasticsearch 存储文本和向量数据. 如果要切换
 ```bash
 git clone https://github.com/infiniflow/ragflow.git
 cd ragflow/
-docker build --platform linux/amd64 --build-arg LIGHTEN=1 --build-arg NEED_MIRROR=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .
-```
-
-## 🔧 源码编译 Docker 镜像（包含 embedding 模型）
-
-本 Docker 大小约 9 GB 左右。由于已包含 embedding 模型，所以只需依赖外部的大模型服务即可。
-
-```bash
-git clone https://github.com/infiniflow/ragflow.git
-cd ragflow/
-docker build --platform linux/amd64 --build-arg NEED_MIRROR=1 -f Dockerfile -t infiniflow/ragflow:nightly .
+docker build --platform linux/amd64 -f Dockerfile -t infiniflow/ragflow:nightly .
 ```
 
 ## 🔨 以源代码启动服务
@@ -313,7 +308,7 @@ docker build --platform linux/amd64 --build-arg NEED_MIRROR=1 -f Dockerfile -t i
    ```bash
    git clone https://github.com/infiniflow/ragflow.git
    cd ragflow/
-   uv sync --python 3.10 --all-extras # install RAGFlow dependent python modules
+   uv sync --python 3.10 # install RAGFlow dependent python modules
    uv run download_deps.py
    pre-commit install
    ```

admin/client/README.md

@@ -48,13 +48,20 @@ 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
+    pip install ragflow-cli==0.21.1
     ```
 3.  Launch the CLI client:
     ```bash
-    ragflow-cli -h 0.0.0.0 -p 9381
+    ragflow-cli -h 127.0.0.1 -p 9381
     ```
-	Enter superuser's password to login. Default password is `admin`.
+    You will be prompted to enter the superuser's password to log in.
+    The default password is admin.
+
+    **Parameters:**
+    
+    - -h: RAGFlow admin server host address
+    
+    - -p: RAGFlow admin server port
 
 
 

admin/client/admin_client.py

@@ -21,9 +21,8 @@ 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, Token
+from lark import Lark, Transformer, Tree
 import requests
-from requests.auth import HTTPBasicAuth
 
 GRAMMAR = r"""
 start: command
@@ -43,6 +42,15 @@ sql_command: list_services
            | 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
 
 // meta command definition
 meta_command: "\\" meta_command_name [meta_args]
@@ -71,6 +79,19 @@ 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
 
 list_services: LIST SERVICES ";"
 show_service: SHOW SERVICE NUMBER ";"
@@ -88,6 +109,19 @@ 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 ";"
+
+action_list: identifier ("," identifier)*
+
 identifier: WORD
 quoted_string: QUOTED_STRING
 status: WORD
@@ -134,34 +168,86 @@ class AdminTransformer(Transformer):
 
     def show_user(self, items):
         user_name = items[2]
-        return {"type": "show_user", "username": user_name}
+        return {"type": "show_user", "user_name": user_name}
 
     def drop_user(self, items):
         user_name = items[2]
-        return {"type": "drop_user", "username": user_name}
+        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", "username": user_name, "password": new_password}
+        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", "username": user_name, "password": password, "role": "user"}
+        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, "username": user_name}
+        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", "username": user_name}
+        return {"type": "list_datasets", "user_name": user_name}
 
     def list_agents(self, items):
         user_name = items[3]
-        return {"type": "list_agents", "username": user_name}
+        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 action_list(self, items):
+        return items
 
     def meta_command(self, items):
         command_name = str(items[0]).lower()
@@ -205,6 +291,8 @@ class AdminCLI(Cmd):
         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
 
@@ -213,12 +301,8 @@ class AdminCLI(Cmd):
 
     def onecmd(self, command: str) -> bool:
         try:
-            # print(f"command: {command}")
             result = self.parse_command(command)
 
-            # if 'type' in result and result.get('type') == 'empty':
-            #     return False
-
             if isinstance(result, dict):
                 if 'type' in result and result.get('type') == 'empty':
                     return False
@@ -244,7 +328,7 @@ class AdminCLI(Cmd):
     def default(self, line: str) -> bool:
         return self.onecmd(line)
 
-    def parse_command(self, command_str: str) -> dict[str, str] | Tree[Token]:
+    def parse_command(self, command_str: str) -> dict[str, str]:
         if not command_str.strip():
             return {'type': 'empty'}
 
@@ -256,32 +340,38 @@ class AdminCLI(Cmd):
         except Exception as e:
             return {'type': 'error', 'message': f'Parse error: {str(e)}'}
 
-    def verify_admin(self, args):
-
-        conn_info = self._parse_connection_args(args)
-        if 'error' in conn_info:
-            print(f"Error: {conn_info['error']}")
-            return
-
-        self.host = conn_info['host']
-        self.port = conn_info['port']
+    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/auth'
+        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 > 3:
+            if try_count > attempt_count:
                 return False
 
-            admin_passwd = input(f"password for {self.admin_account}: ").strip()
+            if single_command:
+                admin_passwd = arguments['password']
+            else:
+                admin_passwd = input(f"password for {self.admin_account}: ").strip()
             try:
-                self.admin_password = encode_to_base64(admin_passwd)
-                response = requests.get(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password))
+                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/0.21.1'
+                        })
                         print("Authentication successful.")
                         return True
                     else:
@@ -289,8 +379,9 @@ class AdminCLI(Cmd):
                         print(f"Authentication failed: {error_message}, try again")
                         continue
                 else:
-                    print(f"Bad response，status: {response.status_code}, try again")
-            except Exception:
+                    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 _print_table_simple(self, data):
@@ -341,9 +432,9 @@ class AdminCLI(Cmd):
             row = "|"
             for col in columns:
                 value = str(item.get(col, ''))
-                if len(value) > col_widths[col]:
+                if get_string_width(value) > col_widths[col]:
                     value = value[:col_widths[col] - 3] + "..."
-                row += f" {value:<{col_widths[col]}} |"
+                row += f" {value:<{col_widths[col] - (get_string_width(value) - len(value))}} |"
             print(row)
 
         print(separator)
@@ -375,23 +466,31 @@ class AdminCLI(Cmd):
                 print("\nGoodbye!")
                 break
 
-    def run_single_command(self, args):
-        conn_info = self._parse_connection_args(args)
-        if 'error' in conn_info:
-            print(f"Error: {conn_info['error']}")
-            return
+    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]:
+    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=8080, help='Admin service port')
-
+        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)
-            return {
-                'host': parsed_args.host,
-                'port': parsed_args.port,
-            }
+            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'}
 
@@ -438,6 +537,24 @@ class AdminCLI(Cmd):
                 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 'meta':
                 self._handle_meta_command(command_dict)
             case _:
@@ -447,30 +564,30 @@ class AdminCLI(Cmd):
         print("Listing all services")
 
         url = f'http://{self.host}:{self.port}/api/v1/admin/services'
-        response = requests.get(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password))
+        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']}")
+            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 = requests.get(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password))
+        response = self.session.get(url)
         res_json = response.json()
         if response.status_code == 200:
             res_data = res_json['data']
-            if res_data['alive']:
-                print(f"Service {res_data['service_name']} is alive. Detail:")
+            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:
                     self._print_table_simple(res_data['message'])
             else:
-                print(f"Service {res_data['service_name']} is down. Detail: {res_data['message']}")
+                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']}")
 
@@ -490,7 +607,7 @@ class AdminCLI(Cmd):
         print("Listing all users")
 
         url = f'http://{self.host}:{self.port}/api/v1/admin/users'
-        response = requests.get(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password))
+        response = self.session.get(url)
         res_json = response.json()
         if response.status_code == 200:
             self._print_table_simple(res_json['data'])
@@ -498,23 +615,23 @@ class AdminCLI(Cmd):
             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['username']
-        username: str = username_tree.children[0].strip("'\"")
-        print(f"Showing user: {username}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{username}'
-        response = requests.get(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password))
+        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:
             self._print_table_simple(res_json['data'])
         else:
-            print(f"Fail to get user {username}, code: {res_json['code']}, message: {res_json['message']}")
+            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['username']
-        username: str = username_tree.children[0].strip("'\"")
-        print(f"Drop user: {username}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{username}'
-        response = requests.delete(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password))
+        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"])
@@ -522,14 +639,13 @@ class AdminCLI(Cmd):
             print(f"Fail to drop user, code: {res_json['code']}, message: {res_json['message']}")
 
     def _handle_alter_user(self, command):
-        username_tree: Tree = command['username']
-        username: str = username_tree.children[0].strip("'\"")
+        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: {username}, password: {password}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{username}/password'
-        response = requests.put(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password),
-                                json={'new_password': encrypt(password)})
+        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"])
@@ -537,34 +653,32 @@ class AdminCLI(Cmd):
             print(f"Fail to alter password, code: {res_json['code']}, message: {res_json['message']}")
 
     def _handle_create_user(self, command):
-        username_tree: Tree = command['username']
-        username: str = username_tree.children[0].strip("'\"")
+        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: {username}, password: {password}, role: {role}")
+        print(f"Create user: {user_name}, password: {password}, role: {role}")
         url = f'http://{self.host}:{self.port}/api/v1/admin/users'
-        response = requests.post(
+        response = self.session.post(
             url,
-            auth=HTTPBasicAuth(self.admin_account, self.admin_password),
-            json={'username': username, 'password': encrypt(password), 'role': role}
+            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 {username}, code: {res_json['code']}, message: {res_json['message']}")
+            print(f"Fail to create user {user_name}, code: {res_json['code']}, message: {res_json['message']}")
 
     def _handle_activate_user(self, command):
-        username_tree: Tree = command['username']
-        username: str = username_tree.children[0].strip("'\"")
+        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 {username} activate status, turn {activate_status.lower()}.")
-            url = f'http://{self.host}:{self.port}/api/v1/admin/users/{username}/activate'
-            response = requests.put(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password),
-                                    json={'activate_status': activate_status})
+            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"])
@@ -574,28 +688,178 @@ class AdminCLI(Cmd):
             print(f"Unknown activate status: {activate_status}.")
 
     def _handle_list_datasets(self, command):
-        username_tree: Tree = command['username']
-        username: str = username_tree.children[0].strip("'\"")
-        print(f"Listing all datasets of user: {username}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{username}/datasets'
-        response = requests.get(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password))
+        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:
             self._print_table_simple(res_json['data'])
         else:
-            print(f"Fail to get all datasets of {username}, code: {res_json['code']}, message: {res_json['message']}")
+            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['username']
-        username: str = username_tree.children[0].strip("'\"")
-        print(f"Listing all agents of user: {username}")
-        url = f'http://{self.host}:{self.port}/api/v1/admin/users/{username}/agents'
-        response = requests.get(url, auth=HTTPBasicAuth(self.admin_account, self.admin_password))
+        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:
             self._print_table_simple(res_json['data'])
         else:
-            print(f"Fail to get all agents of {username}, code: {res_json['code']}, message: {res_json['message']}")
+            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 _handle_meta_command(self, command):
         meta_command = command['command']
@@ -638,27 +902,29 @@ def main():
 
     cli = AdminCLI()
 
-    if len(sys.argv) == 1 or (len(sys.argv) > 1 and sys.argv[1] == '-'):
-        print(r"""
-            ____  ___   ______________                 ___       __          _     
-           / __ \/   | / ____/ ____/ /___ _      __   /   | ____/ /___ ___  (_)___ 
-          / /_/ / /| |/ / __/ /_  / / __ \ | /| / /  / /| |/ __  / __ `__ \/ / __ \
-         / _, _/ ___ / /_/ / __/ / / /_/ / |/ |/ /  / ___ / /_/ / / / / / / / / / /
-        /_/ |_/_/  |_\____/_/   /_/\____/|__/|__/  /_/  |_\__,_/_/ /_/ /_/_/_/ /_/ 
-        """)
-        if cli.verify_admin(sys.argv):
-            cli.cmdloop()
+    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:
-        print(r"""
-            ____  ___   ______________                 ___       __          _     
-           / __ \/   | / ____/ ____/ /___ _      __   /   | ____/ /___ ___  (_)___ 
-          / /_/ / /| |/ / __/ /_  / / __ \ | /| / /  / /| |/ __  / __ `__ \/ / __ \
-         / _, _/ ___ / /_/ / __/ / / /_/ / |/ |/ /  / ___ / /_/ / / / / / / / / / /
-        /_/ |_/_/  |_\____/_/   /_/\____/|__/|__/  /_/  |_\__,_/_/ /_/ /_/_/_/ /_/ 
-        """)
-        if cli.verify_admin(sys.argv):
+        if cli.verify_admin(args, single_command=False):
+            print(r"""
+                ____  ___   ______________                 ___       __          _     
+               / __ \/   | / ____/ ____/ /___ _      __   /   | ____/ /___ ___  (_)___ 
+              / /_/ / /| |/ / __/ /_  / / __ \ | /| / /  / /| |/ __  / __ `__ \/ / __ \
+             / _, _/ ___ / /_/ / __/ / / /_/ / |/ |/ /  / ___ / /_/ / / / / / / / / / /
+            /_/ |_/_/  |_\____/_/   /_/\____/|__/|__/  /_/  |_\__,_/_/ /_/ /_/_/_/ /_/ 
+            """)
             cli.cmdloop()
-            # cli.run_single_command(sys.argv[1:])
 
 
 if __name__ == '__main__':

admin/client/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ragflow-cli"
-version = "0.21.0.dev5"
+version = "0.21.1"
 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" }

admin/pyproject.toml

@@ -1,24 +0,0 @@
-[project]
-name = "ragflow-cli"
-version = "0.21.0.dev2"
-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"
-dependencies = [
-    "requests>=2.30.0,<3.0.0",
-    "beartype>=0.18.5,<0.19.0",
-    "pycryptodomex>=3.10.0",
-    "lark>=1.1.0",
-]
-
-[dependency-groups]
-test = [
-    "pytest>=8.3.5",
-    "requests>=2.32.3",
-    "requests-toolbelt>=1.0.0",
-]
-
-[project.scripts]
-ragflow-cli = "ragflow_cli.admin_client:main"

admin/server/admin_server.py

@@ -26,7 +26,10 @@ from routes import admin_bp
 from api.utils.log_utils import init_root_logger
 from api.constants import SERVICE_CONF
 from api import settings
-from admin.server.config import load_configurations, SERVICE_CONFIGS
+from config import load_configurations, SERVICE_CONFIGS
+from auth import init_default_admin, setup_auth
+from flask_session import Session
+from flask_login import LoginManager
 
 stop_event = threading.Event()
 
@@ -42,7 +45,17 @@ if __name__ == '__main__':
 
     app = Flask(__name__)
     app.register_blueprint(admin_bp)
+    app.config["SESSION_PERMANENT"] = False
+    app.config["SESSION_TYPE"] = "filesystem"
+    app.config["MAX_CONTENT_LENGTH"] = int(
+        os.environ.get("MAX_CONTENT_LENGTH", 1024 * 1024 * 1024)
+    )
+    Session(app)
+    login_manager = LoginManager()
+    login_manager.init_app(app)
     settings.init_settings()
+    setup_auth(login_manager)
+    init_default_admin()
     SERVICE_CONFIGS.configs = load_configurations(SERVICE_CONF)
 
     try:

admin/server/auth.py

@@ -18,11 +18,120 @@
 import logging
 import uuid
 from functools import wraps
+from datetime import datetime
 from flask import request, jsonify
+from flask_login import current_user, login_user
+from itsdangerous.url_safe import URLSafeTimedSerializer as Serializer
 
-from api.common.exceptions import AdminException
+from api import settings
+from api.common.exceptions import AdminException, UserNotFoundError
 from api.db.init_data import encode_to_base64
 from api.db.services import UserService
+from api.db import ActiveEnum, StatusEnum
+from api.utils.crypt import decrypt
+from api.utils import get_uuid
+from common.time_utils import current_timestamp, datetime_format, get_format_time
+from api.utils.api_utils import (
+    construct_response,
+)
+
+
+def setup_auth(login_manager):
+    @login_manager.request_loader
+    def load_user(web_request):
+        jwt = Serializer(secret_key=settings.SECRET_KEY)
+        authorization = web_request.headers.get("Authorization")
+        if authorization:
+            try:
+                access_token = str(jwt.loads(authorization))
+
+                if not access_token or not access_token.strip():
+                    logging.warning("Authentication attempt with empty access token")
+                    return None
+
+                # Access tokens should be UUIDs (32 hex characters)
+                if len(access_token.strip()) < 32:
+                    logging.warning(f"Authentication attempt with invalid token format: {len(access_token)} chars")
+                    return None
+
+                user = UserService.query(
+                    access_token=access_token, 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
+                    return user[0]
+                else:
+                    return None
+            except Exception as e:
+                logging.warning(f"load_user got exception {e}")
+                return None
+        else:
+            return None
+
+
+def init_default_admin():
+    # Verify that at least one active admin user exists. If not, create a default one.
+    users = UserService.query(is_superuser=True)
+    if not users:
+        default_admin = {
+            "id": uuid.uuid1().hex,
+            "password": encode_to_base64("admin"),
+            "nickname": "admin",
+            "is_superuser": True,
+            "email": "admin@ragflow.io",
+            "creator": "system",
+            "status": "1",
+        }
+        if not UserService.save(**default_admin):
+            raise AdminException("Can't init admin.", 500)
+    elif not any([u.is_active == ActiveEnum.ACTIVE.value for u in users]):
+        raise AdminException("No active admin. Please update 'is_active' in db manually.", 500)
+
+
+def check_admin_auth(func):
+    @wraps(func)
+    def wrapper(*args, **kwargs):
+        user = UserService.filter_by_id(current_user.id)
+        if not user:
+            raise UserNotFoundError(current_user.email)
+        if not user.is_superuser:
+            raise AdminException("Not admin", 403)
+        if user.is_active == ActiveEnum.INACTIVE.value:
+            raise AdminException(f"User {current_user.email} inactive", 403)
+
+        return func(*args, **kwargs)
+
+    return wrapper
+
+
+def login_admin(email: str, password: str):
+    """
+    :param email: admin email
+    :param password: string before decrypt
+    """
+    users = UserService.query(email=email)
+    if not users:
+        raise UserNotFoundError(email)
+    psw = decrypt(password)
+    user = UserService.query_user(email, psw)
+    if not user:
+        raise AdminException("Email and password do not match!")
+    if not user.is_superuser:
+        raise AdminException("Not admin", 403)
+    if user.is_active == ActiveEnum.INACTIVE.value:
+        raise AdminException(f"User {email} inactive", 403)
+
+    resp = user.to_json()
+    user.access_token = get_uuid()
+    login_user(user)
+    user.update_time = (current_timestamp(),)
+    user.update_date = (datetime_format(datetime.now()),)
+    user.last_login_time = get_format_time()
+    user.save()
+    msg = "Welcome back!"
+    return construct_response(data=resp, auth=user.get_id(), message=msg)
 
 
 def check_admin(username: str, password: str):
@@ -61,12 +170,18 @@ def login_verify(f):
 
         username = auth.parameters['username']
         password = auth.parameters['password']
-        # TODO: to check the username and password from DB
-        if check_admin(username, password) is False:
+        try:
+            if check_admin(username, password) is False:
+                return jsonify({
+                    "code": 500,
+                    "message": "Access denied",
+                    "data": None
+                }), 200
+        except Exception as e:
+            error_msg = str(e)
             return jsonify({
-                "code": 403,
-                "message": "Access denied",
-                "data": None
+                "code": 500,
+                "message": error_msg
             }), 200
 
         return f(*args, **kwargs)

admin/server/config.py

@@ -26,6 +26,8 @@ from urllib.parse import urlparse
 
 
 class ServiceConfigs:
+    configs = dict
+
     def __init__(self):
         self.configs = []
         self.lock = threading.Lock()
@@ -229,7 +231,8 @@ def load_configurations(config_path: str) -> list[BaseConfig]:
                 host: str = v['host']
                 http_port: int = v['http_port']
                 config = RAGFlowServerConfig(id=id_count, name=name, host=host, port=http_port,
-                                             service_type="ragflow_server", detail_func_name="check_ragflow_server_alive")
+                                             service_type="ragflow_server",
+                                             detail_func_name="check_ragflow_server_alive")
                 configurations.append(config)
                 id_count += 1
             case "es":
@@ -254,7 +257,8 @@ def load_configurations(config_path: str) -> list[BaseConfig]:
                 host = parts[0]
                 port = int(parts[1])
                 database: str = v.get('db_name', 'default_db')
-                config = InfinityConfig(id=id_count, name=name, host=host, port=port, service_type="retrieval", retrieval_type="infinity",
+                config = InfinityConfig(id=id_count, name=name, host=host, port=port, service_type="retrieval",
+                                        retrieval_type="infinity",
                                         db_name=database, detail_func_name="get_infinity_status")
                 configurations.append(config)
                 id_count += 1
@@ -266,7 +270,8 @@ def load_configurations(config_path: str) -> list[BaseConfig]:
                 port = int(parts[1])
                 user = v.get('user')
                 password = v.get('password')
-                config = MinioConfig(id=id_count, name=name, host=host, port=port, user=user, password=password, service_type="file_store",
+                config = MinioConfig(id=id_count, name=name, host=host, port=port, user=user, password=password,
+                                     service_type="file_store",
                                      store_type="minio", detail_func_name="check_minio_alive")
                 configurations.append(config)
                 id_count += 1

admin/server/roles.py

@@ -0,0 +1,76 @@
+#
+#  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 Dict, Any
+
+from api.common.exceptions import AdminException
+
+
+class RoleMgr:
+    @staticmethod
+    def create_role(role_name: str, description: str):
+        error_msg = f"not implement: create role: {role_name}, description: {description}"
+        logging.error(error_msg)
+        raise AdminException(error_msg)
+
+    @staticmethod
+    def update_role_description(role_name: str, description: str) -> Dict[str, Any]:
+        error_msg = f"not implement: update role: {role_name} with description: {description}"
+        logging.error(error_msg)
+        raise AdminException(error_msg)
+
+    @staticmethod
+    def delete_role(role_name: str) -> Dict[str, Any]:
+        error_msg = f"not implement: drop role: {role_name}"
+        logging.error(error_msg)
+        raise AdminException(error_msg)
+
+    @staticmethod
+    def list_roles() -> Dict[str, Any]:
+        error_msg = "not implement: list roles"
+        logging.error(error_msg)
+        raise AdminException(error_msg)
+
+    @staticmethod
+    def get_role_permission(role_name: str) -> Dict[str, Any]:
+        error_msg = f"not implement: show role {role_name}"
+        logging.error(error_msg)
+        raise AdminException(error_msg)
+
+    @staticmethod
+    def grant_role_permission(role_name: str, actions: list, resource: str) -> Dict[str, Any]:
+        error_msg = f"not implement: grant role {role_name} actions: {actions} on {resource}"
+        logging.error(error_msg)
+        raise AdminException(error_msg)
+
+    @staticmethod
+    def revoke_role_permission(role_name: str, actions: list, resource: str) -> Dict[str, Any]:
+        error_msg = f"not implement: revoke role {role_name} actions: {actions} on {resource}"
+        logging.error(error_msg)
+        raise AdminException(error_msg)
+
+    @staticmethod
+    def update_user_role(user_name: str, role_name: str) -> Dict[str, Any]:
+        error_msg = f"not implement: update user role: {user_name} to role {role_name}"
+        logging.error(error_msg)
+        raise AdminException(error_msg)
+
+    @staticmethod
+    def get_user_permission(user_name: str) -> Dict[str, Any]:
+        error_msg = f"not implement: get user permission: {user_name}"
+        logging.error(error_msg)
+        raise AdminException(error_msg)

admin/server/routes.py

@@ -14,17 +14,44 @@
 #  limitations under the License.
 #
 
+import secrets
 
 from flask import Blueprint, request
+from flask_login import current_user, logout_user, login_required
 
-from admin.server.auth import login_verify
+from auth import login_verify, login_admin, check_admin_auth
 from responses import success_response, error_response
 from services import UserMgr, ServiceMgr, UserServiceMgr
+from roles import RoleMgr
 from api.common.exceptions import AdminException
 
 admin_bp = Blueprint('admin', __name__, url_prefix='/api/v1/admin')
 
 
+@admin_bp.route('/login', methods=['POST'])
+def login():
+    if not request.json:
+        return error_response('Authorize admin failed.' ,400)
+    try:
+        email = request.json.get("email", "")
+        password = request.json.get("password", "")
+        return login_admin(email, password)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route('/logout', methods=['GET'])
+@login_required
+def logout():
+    try:
+        current_user.access_token = f"INVALID_{secrets.token_hex(16)}"
+        current_user.save()
+        logout_user()
+        return success_response(True)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
 @admin_bp.route('/auth', methods=['GET'])
 @login_verify
 def auth_admin():
@@ -35,7 +62,8 @@ def auth_admin():
 
 
 @admin_bp.route('/users', methods=['GET'])
-@login_verify
+@login_required
+@check_admin_auth
 def list_users():
     try:
         users = UserMgr.get_all_users()
@@ -45,7 +73,8 @@ def list_users():
 
 
 @admin_bp.route('/users', methods=['POST'])
-@login_verify
+@login_required
+@check_admin_auth
 def create_user():
     try:
         data = request.get_json()
@@ -71,7 +100,8 @@ def create_user():
 
 
 @admin_bp.route('/users/<username>', methods=['DELETE'])
-@login_verify
+@login_required
+@check_admin_auth
 def delete_user(username):
     try:
         res = UserMgr.delete_user(username)
@@ -87,7 +117,8 @@ def delete_user(username):
 
 
 @admin_bp.route('/users/<username>/password', methods=['PUT'])
-@login_verify
+@login_required
+@check_admin_auth
 def change_password(username):
     try:
         data = request.get_json()
@@ -105,7 +136,8 @@ def change_password(username):
 
 
 @admin_bp.route('/users/<username>/activate', methods=['PUT'])
-@login_verify
+@login_required
+@check_admin_auth
 def alter_user_activate_status(username):
     try:
         data = request.get_json()
@@ -121,7 +153,8 @@ def alter_user_activate_status(username):
 
 
 @admin_bp.route('/users/<username>', methods=['GET'])
-@login_verify
+@login_required
+@check_admin_auth
 def get_user_details(username):
     try:
         user_details = UserMgr.get_user_details(username)
@@ -134,7 +167,8 @@ def get_user_details(username):
 
 
 @admin_bp.route('/users/<username>/datasets', methods=['GET'])
-@login_verify
+@login_required
+@check_admin_auth
 def get_user_datasets(username):
     try:
         datasets_list = UserServiceMgr.get_user_datasets(username)
@@ -147,7 +181,8 @@ def get_user_datasets(username):
 
 
 @admin_bp.route('/users/<username>/agents', methods=['GET'])
-@login_verify
+@login_required
+@check_admin_auth
 def get_user_agents(username):
     try:
         agents_list = UserServiceMgr.get_user_agents(username)
@@ -160,7 +195,8 @@ def get_user_agents(username):
 
 
 @admin_bp.route('/services', methods=['GET'])
-@login_verify
+@login_required
+@check_admin_auth
 def get_services():
     try:
         services = ServiceMgr.get_all_services()
@@ -170,7 +206,8 @@ def get_services():
 
 
 @admin_bp.route('/service_types/<service_type>', methods=['GET'])
-@login_verify
+@login_required
+@check_admin_auth
 def get_services_by_type(service_type_str):
     try:
         services = ServiceMgr.get_services_by_type(service_type_str)
@@ -180,7 +217,8 @@ def get_services_by_type(service_type_str):
 
 
 @admin_bp.route('/services/<service_id>', methods=['GET'])
-@login_verify
+@login_required
+@check_admin_auth
 def get_service(service_id):
     try:
         services = ServiceMgr.get_service_details(service_id)
@@ -190,7 +228,8 @@ def get_service(service_id):
 
 
 @admin_bp.route('/services/<service_id>', methods=['DELETE'])
-@login_verify
+@login_required
+@check_admin_auth
 def shutdown_service(service_id):
     try:
         services = ServiceMgr.shutdown_service(service_id)
@@ -200,10 +239,133 @@ def shutdown_service(service_id):
 
 
 @admin_bp.route('/services/<service_id>', methods=['PUT'])
-@login_verify
+@login_required
+@check_admin_auth
 def restart_service(service_id):
     try:
         services = ServiceMgr.restart_service(service_id)
         return success_response(services)
     except Exception as e:
         return error_response(str(e), 500)
+
+
+@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:
+            return error_response("Role name is required", 400)
+        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'])
+@login_required
+@check_admin_auth
+def update_role(role_name: str):
+    try:
+        data = request.get_json()
+        if not data or 'description' not in data:
+            return error_response("Role description is required", 400)
+        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'])
+@login_required
+@check_admin_auth
+def delete_role(role_name: str):
+    try:
+        res = RoleMgr.delete_role(role_name)
+        return success_response(res)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route('/roles', methods=['GET'])
+@login_required
+@check_admin_auth
+def list_roles():
+    try:
+        res = RoleMgr.list_roles()
+        return success_response(res)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@admin_bp.route('/roles/<role_name>/permission', methods=['GET'])
+@login_required
+@check_admin_auth
+def get_role_permission(role_name: str):
+    try:
+        res = RoleMgr.get_role_permission(role_name)
+        return success_response(res)
+    except Exception as e:
+        return error_response(str(e), 500)
+
+
+@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:
+            return error_response("Permission is required", 400)
+        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'])
+@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:
+            return error_response("Permission is required", 400)
+        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'])
+@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:
+            return error_response("Role name is required", 400)
+        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'])
+@login_required
+@check_admin_auth
+def get_user_permission(user_name: str):
+    try:
+        res = RoleMgr.get_user_permission(user_name)
+        return success_response(res)
+    except Exception as e:
+        return error_response(str(e), 500)

admin/server/services.py

@@ -27,7 +27,7 @@ from api.utils.crypt import decrypt
 from api.utils import health_utils
 
 from api.common.exceptions import AdminException, UserAlreadyExistsError, UserNotFoundError
-from admin.server.config import SERVICE_CONFIGS
+from config import SERVICE_CONFIGS
 
 
 class UserMgr:
@@ -36,8 +36,13 @@ 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})
+            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
@@ -50,7 +55,6 @@ class UserMgr:
                 'email': user.email,
                 'language': user.language,
                 'last_login_time': user.last_login_time,
-                'is_authenticated': user.is_authenticated,
                 'is_active': user.is_active,
                 'is_anonymous': user.is_anonymous,
                 'login_channel': user.login_channel,
@@ -166,8 +170,7 @@ class UserServiceMgr:
         return [{
             'title': r['title'],
             'permission': r['permission'],
-            'canvas_type': r['canvas_type'],
-            'canvas_category': r['canvas_category']
+            'canvas_category': r['canvas_category'].split('_')[0]
         } for r in res]
 
 
@@ -181,12 +184,12 @@ class ServiceMgr:
             config_dict = config.to_dict()
             try:
                 service_detail = ServiceMgr.get_service_details(service_id)
-                if service_detail['alive']:
-                    config_dict['status'] = 'Alive'
+                if "status" in service_detail:
+                    config_dict['status'] = service_detail['status']
                 else:
-                    config_dict['status'] = 'Timeout'
+                    config_dict['status'] = 'timeout'
             except Exception:
-                config_dict['status'] = 'Timeout'
+                config_dict['status'] = 'timeout'
             result.append(config_dict)
         return result
 
@@ -206,7 +209,7 @@ class ServiceMgr:
         }
         service_info = service_config_mapping.get(service_id, {})
         if not service_info:
-            raise AdminException(f"Invalid service_id: {service_id}")
+            raise AdminException(f"invalid service_id: {service_id}")
 
         detail_func = getattr(health_utils, service_info.get('detail_func_name'))
         res = detail_func()

agent/templates/advanced_ingestion_pipeline.json

@@ -2,10 +2,12 @@
     "id": 23,
     "title": {
         "en": "Advanced Ingestion Pipeline",
+        "de": "Erweiterte Ingestion Pipeline",
         "zh": "编排复杂的 Ingestion Pipeline"
     },
     "description": {
         "en": "This template demonstrates how to use an LLM to generate summaries, keywords, Q&A, and metadata for each chunk to support diverse retrieval needs.",
+        "de": "Diese Vorlage demonstriert, wie ein LLM verwendet wird, um Zusammenfassungen, Schlüsselwörter, Fragen & Antworten und Metadaten für jedes Segment zu generieren, um vielfältige Abrufanforderungen zu unterstützen.",
         "zh": "此模板演示如何利用大模型为切片生成摘要、关键词、问答及元数据，以满足多样化的召回需求。"
     },
     "canvas_type": "Ingestion Pipeline",

agent/templates/chunk_summary.json

@@ -2,10 +2,12 @@
     "id": 24,
     "title": {
         "en": "Chunk Summary",
+        "de": "Segmentzusammenfassung",
         "zh": "总结切片"
     },
     "description": {
         "en": "This template uses an LLM to generate chunk summaries for building text and vector indexes. During retrieval, summaries enhance matching, and the original chunks are returned as results.",
+        "de": "Diese Vorlage nutzt ein LLM zur Generierung von Segmentzusammenfassungen für den Aufbau von Text- und Vektorindizes. Bei der Abfrage verbessern die Zusammenfassungen die Übereinstimmung, während die ursprünglichen Segmente als Ergebnisse zurückgegeben werden.",
         "zh": "此模板利用大模型生成切片摘要，并据此建立全文索引与向量。检索时以摘要提升匹配效果，最终召回对应的原文切片。"
     },
     "canvas_type": "Ingestion Pipeline",
@@ -350,7 +352,7 @@
                             ]
                         },
                         "label": "Tokenizer",
-                        "name": "Tokenizer"
+                        "name": "Indexer"
                     },
                     "dragging": false,
                     "id": "Tokenizer:EightRocketsAppear",

agent/templates/generate_SEO_blog.json

@@ -2,9 +2,11 @@
     "id": 8,
     "title": {
         "en": "Generate SEO Blog",
+        "de": "SEO Blog generieren",
         "zh": "生成SEO博客"},
     "description": {
         "en": "This is a multi-agent version of the SEO blog generation workflow. It simulates a small team of AI “writers”, where each agent plays a specialized role — just like a real editorial team.",
+        "de": "Dies ist eine Multi-Agenten-Version des Workflows zur Erstellung von SEO-Blogs. Sie simuliert ein kleines Team von KI-„Autoren“, in dem jeder Agent eine spezielle Rolle übernimmt – genau wie in einem echten Redaktionsteam.",
         "zh": "多智能体架构可根据简单的用户输入自动生成完整的SEO博客文章。模拟小型“作家”团队，其中每个智能体扮演一个专业角色——就像真正的编辑团队。"},
     "canvas_type": "Agent",
     "dsl": {

agent/templates/knowledge_base_report.json

@@ -2,9 +2,11 @@
     "id": 20,
     "title": {
         "en": "Report Agent Using Knowledge Base",
+        "de": "Berichtsagent mit Wissensdatenbank",
         "zh": "知识库检索智能体"},
     "description": {
         "en": "A report generation assistant using local knowledge base, with advanced capabilities in task planning, reasoning, and reflective analysis. Recommended for academic research paper Q&A",
+        "de": "Ein Berichtsgenerierungsassistent, der eine lokale Wissensdatenbank nutzt, mit erweiterten Fähigkeiten in Aufgabenplanung, Schlussfolgerung und reflektierender Analyse. Empfohlen für akademische Forschungspapier-Fragen und -Antworten.",
         "zh": "一个使用本地知识库的报告生成助手，具备高级能力，包括任务规划、推理和反思性分析。推荐用于学术研究论文问答。"},
     "canvas_type": "Agent",
     "dsl": {

agent/templates/knowledge_base_report_r.json

@@ -1,10 +1,12 @@
 {
     "id": 21,
     "title": {
-        "en": "Report Agent Using Knowledge Base", 
+        "en": "Report Agent Using Knowledge Base",
+        "de": "Berichtsagent mit Wissensdatenbank",
         "zh": "知识库检索智能体"},
     "description": {
         "en": "A report generation assistant using local knowledge base, with advanced capabilities in task planning, reasoning, and reflective analysis. Recommended for academic research paper Q&A",
+        "de": "Ein Berichtsgenerierungsassistent, der eine lokale Wissensdatenbank nutzt, mit erweiterten Fähigkeiten in Aufgabenplanung, Schlussfolgerung und reflektierender Analyse. Empfohlen für akademische Forschungspapier-Fragen und -Antworten.",
         "zh": "一个使用本地知识库的报告生成助手，具备高级能力，包括任务规划、推理和反思性分析。推荐用于学术研究论文问答。"},
     "canvas_type": "Recommended",
     "dsl": {

agent/templates/market_generate_seo_blog.json

@@ -2,9 +2,11 @@
     "id": 12,
     "title": {
         "en": "Generate SEO Blog",
+        "de": "SEO Blog generieren",
         "zh": "生成SEO博客"},
     "description": {
-        "en": "This workflow automatically generates a complete SEO-optimized blog article based on a simple user input. You don’t need any writing experience. Just provide a topic or short request — the system will handle the rest.",
+        "en": "This workflow automatically generates a complete SEO-optimized blog article based on a simple user input. You don't need any writing experience. Just provide a topic or short request — the system will handle the rest.",
+        "de": "Dieser Workflow generiert automatisch einen vollständigen SEO-optimierten Blogartikel basierend auf einer einfachen Benutzereingabe. Sie benötigen keine Schreiberfahrung. Geben Sie einfach ein Thema oder eine kurze Anfrage ein – das System übernimmt den Rest.",
         "zh": "此工作流根据简单的用户输入自动生成完整的SEO博客文章。你无需任何写作经验，只需提供一个主题或简短请求，系统将处理其余部分。"},
     "canvas_type": "Marketing",
     "dsl": {
@@ -916,4 +918,4 @@
             "retrieval": []
         },
     "avatar": "data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQABAAD/4gHYSUNDX1BST0ZJTEUAAQEAAAHIAAAAAAQwAABtbnRyUkdCIFhZWiAH4AABAAEAAAAAAABhY3NwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAQAA9tYAAQAAAADTLQAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAlkZXNjAAAA8AAAACRyWFlaAAABFAAAABRnWFlaAAABKAAAABRiWFlaAAABPAAAABR3dHB0AAABUAAAABRyVFJDAAABZAAAAChnVFJDAAABZAAAAChiVFJDAAABZAAAAChjcHJ0AAABjAAAADxtbHVjAAAAAAAAAAEAAAAMZW5VUwAAAAgAAAAcAHMAUgBHAEJYWVogAAAAAAAAb6IAADj1AAADkFhZWiAAAAAAAABimQAAt4UAABjaWFlaIAAAAAAAACSgAAAPhAAAts9YWVogAAAAAAAA9tYAAQAAAADTLXBhcmEAAAAAAAQAAAACZmYAAPKnAAANWQAAE9AAAApbAAAAAAAAAABtbHVjAAAAAAAAAAEAAAAMZW5VUwAAACAAAAAcAEcAbwBvAGcAbABlACAASQBuAGMALgAgADIAMAAxADb/2wBDAAEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQH/2wBDAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQH/wAARCAAwADADASIAAhEBAxEB/8QAGQAAAwEBAQAAAAAAAAAAAAAABgkKBwUI/8QAMBAAAAYCAQIEBQQCAwAAAAAAAQIDBAUGBxEhCAkAEjFBFFFhcaETFiKRFyOx8PH/xAAaAQACAwEBAAAAAAAAAAAAAAACAwABBgQF/8QALBEAAgIBAgUCBAcAAAAAAAAAAQIDBBEFEgATITFRIkEGIzJhFBUWgaGx8P/aAAwDAQACEQMRAD8AfF2hez9089t7pvxgQMa1Gb6qZ6oQE9m/NEvCIStyPfJSOF/M1epzMugo/qtMqbiRc1mJjoJKCLMNIxKcsLJedfO1Ct9cI63x9fx6CA/19t+oh4LFA5HfuAgP/A8eOIsnsTBrkBHXA7+v53+Q+ficTgJft9gIgA+/P9/1r342O/YA8A8k3/if+IbAN7+2/f8AAiI6H19PGoPyESTMZQPKUAHkQEN+3r9dh78/YPGUTk2wb/qAZZIugH1OHH5DjkdfbnWw2DsOxPj+xjrnx2H39unBopJGBn9s+PHv1HXjPJtH+J+B40O9a16h/wB/92j/ALrPa/wR104UyAobHlXhuo2HrEtK4qy3CwjKOuJLRHJLSkXWrFKs/gVrJVrE8TUiH8bPrP20UEu8m4hNpMJJuTOfnbUw/kUqyZgMHGjAO9+mtDsQ53sdcB6eMhnpEjhNQxRKICAgHy5+/roOdjr7c+J6O4x07dx484/n7nzw1gexBGfIPkZ/3t39uGpqc6+fP5/Ht8vGFZCzJjWpWuBxvO2yPjrtclUUK7BqmUI4fuASeyhG5FzFI0Bw4aQ0iZNoDgzvRW4qtyFkI4XmwyEk2YNnDp0sVBu3IUyy5iqH8gqKERSIRNIii67hddRJs1at01Xbx2sgzZoLu10UFJR+4V1A5cxF3FqNcLvjwcno43uuLrOxZYjujaClcb4QQfxEizpFiQyM9olcueRnjC2ZMt9iY06zL0qytrMSqSOVGsfHMaGhZ3l4lSRI2MqE74zJvRTveNFWWIh3RWw+XCAM5icKQLrCH57T17FhErSlRXnWvyZXKQwWJ3eraD14p5YuZCFgacskK2oGkVuKO5GYTHzf7DaD12cBD3DgPOIDrWw9PnrXPgDkpVsUDGMG+DD6E9gHXIjrYjwUPQTCXYgHPhIV974+F6E1hpC14Yzmzj56YaQEeZhXsayD1zLPW7pygxaMf81Nzu1iJsnIuDIKnaJAkPldqrHaoORZ73tMVEbFdSXT9nVgRQgnBq6j8e/HCIEATpAnH5KlmRVkFRFJwks/bqImSXJ5VFyA3N6Ikh3bCW3YHp5cowOmCfTgA+xJCnrjtwHKcLvJj2ZGcTRFj19kEhckdzgEjKnABGSSzdc1Fe5byXXGNjKdvRcw5NxvLidNZFFCxUa62KrzMaChw8hhYScFJtROAgmuLByq1MsgkZYPaVVuDe0wraRaqAdJwgRQo+YR8xTlAQNx6b49w41vXiJpCalLh1jZhyrTqRM4+jstdRmYryNkydLQRWg1LNGcWd5jIFFvCythlIySa0mNu74sKRQtaWsTmupqPItw0lE52ufpyYzrSkx6cw5bLmBEpkTsz+dt8P5QFuCRtAIkBH9MuwKHICIaDQhnojMs9mKaeGcrMxXlQtAYkdVljimRrE5MqI4zL8oSqQ6wxjodBqK05qdK3Vo3aCSVkBW7bjuC1NFJJBPaqyx6fp6pWkliYLXK2XrukkRu2CCVoSWMgsdMyySKwoLFcIGWSTUMg4IBgTcICoBhRcplMcpFkhIqQp1ClMBTmA0Zfe1zpjvHfXff65bZlzXpB3jjGTgiirmPjAfs16PHqHeQ75Wbj3xxZpOEkV3LRJJSPdomUBZISJLncV2k+8D07dxXp7xsYuTapA9UkJUYWIzNhadnWEZeCXGLQQiJi1ViHfhHL2unWh+mlORsrW0JFpEFnGVfm1mU4kq0FY3eD6corJncv6dr5NLSMNXVaTUksjTiMnaq8uFfSVuDyiJ1iZpy0LOJtpa3YfkcQ5fdozyxI2m5qqcrHN61YYmHsh6v3o9ParYmYJEtlhIx6+gUbjgD23M6oqg92YL0JyF6Bps+qDValVA9h9Lj5SZI3SHXdEQlj1wiQtLLIe6pGzjO3BlBkK1hxpblLVH5wdW0BcFKf/JwRtjsot2z8omaSdxbzzk1iEjsE0AM9rrRZNRIrVyo7dGO6E+oh8axLlJ5H5VaJKx7ePRGFbW6vUeFfHQIWPTI9Tm7HHfuhqY7E6C7JFqUzM6iZXIoncNxX7+bIVdJnTT48x3OQU1krIDW3UeixVhyISzYz6cadY5Xph6TseRNTRsTElzzBn9Vlly0TAERsdgnMYyLROjyFbg5R4ZlsGaMT4yNi2Zlq1GwjZB3jq0PsaJfA3t0jL0W0Y9xf1V41lpWckXMLaZiwxuKYPqc6LlHdkeRF+Qxswx5ASDqBVrsL+2A/N6SiCbYymV2BywJiMZj3GRRMTnL+lVyHCll3R7Szv0vqXMtQ74T+HijljIScLaEpkKCB3rqMBIi0jPs5JeOKTZMZEi5VVnouzy0k3jXjWSMlY6UcVGDxlKMVDqx91SILWSi3D2KdgYy3kP8E9X/AE1SnRXBNdNRMlefT6g7aY6giK+cPLGNg0bY68rcnpsNh9PqIBve/EcPQ3WIq2dR93xpSgk5SAZ9R6MLAOZFUkpLSUDXp6/KPpGUkmTdswlnKnwbl5ITMdGwcXJi7LKsqzUmT5tWYmkXuF9wjBvb76b7dHheazJ9RElUJOCxViuMlUJC0Gtz6PKyjLBY4qMWUe12r1xZ6lOyT6XPEBKN2CkTDOlZd02TBdTMt7Upx2knrkdCv1UKjDKn1A7XBYH6SCOOrWn5Oi/DtRiu+GleRthDL8rXdVjZlcfWrSIxVlGGGCOnH//Z"
-}
+}

agent/templates/seo_blog.json

@@ -2,9 +2,11 @@
     "id": 4,
     "title": {
         "en": "Generate SEO Blog",
+        "de": "SEO Blog generieren",
         "zh": "生成SEO博客"},
     "description": {
-        "en": "This workflow automatically generates a complete SEO-optimized blog article based on a simple user input. You don’t need any writing experience. Just provide a topic or short request — the system will handle the rest.",
+        "en": "This workflow automatically generates a complete SEO-optimized blog article based on a simple user input. You don't need any writing experience. Just provide a topic or short request — the system will handle the rest.",
+        "de": "Dieser Workflow generiert automatisch einen vollständigen SEO-optimierten Blogartikel basierend auf einer einfachen Benutzereingabe. Sie benötigen keine Schreiberfahrung. Geben Sie einfach ein Thema oder eine kurze Anfrage ein – das System übernimmt den Rest.",
         "zh": "此工作流根据简单的用户输入自动生成完整的SEO博客文章。你无需任何写作经验，只需提供一个主题或简短请求，系统将处理其余部分。"},
     "canvas_type": "Recommended",
     "dsl": {
@@ -916,4 +918,4 @@
             "retrieval": []
         },
     "avatar": "data:image/jpeg;base64,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"
-}
+}

agent/templates/sql_assistant.json

@@ -2,10 +2,12 @@
     "id": 17,
     "title": {
         "en": "SQL Assistant",
+        "de": "SQL Assistent",
         "zh": "SQL助理"},
     "description": {
-        "en": "SQL Assistant is an AI-powered tool that lets business users turn plain-English questions into fully formed SQL queries. Simply type your question (e.g., “Show me last quarter’s top 10 products by revenue”) and SQL Assistant generates the exact SQL, runs it against your database, and returns the results in seconds. ",
-        "zh": "用户能够将简单文本问题转化为完整的SQL查询并输出结果。只需输入您的问题（例如，“展示上个季度前十名按收入排序的产品”），SQL助理就会生成精确的SQL语句，对其运行您的数据库，并几秒钟内返回结果。"},
+        "en": "SQL Assistant is an AI-powered tool that lets business users turn plain-English questions into fully formed SQL queries. Simply type your question (e.g., 'Show me last quarter's top 10 products by revenue') and SQL Assistant generates the exact SQL, runs it against your database, and returns the results in seconds. ",
+        "de": "SQL-Assistent ist ein KI-gestütztes Tool, mit dem Geschäftsanwender einfache englische Fragen in vollständige SQL-Abfragen umwandeln können. Geben Sie einfach Ihre Frage ein (z.B. 'Zeige mir die Top 10 Produkte des letzten Quartals nach Umsatz') und der SQL-Assistent generiert das exakte SQL, führt es gegen Ihre Datenbank aus und liefert die Ergebnisse in Sekunden.",
+        "zh": "用户能够将简单文本问题转化为完整的SQL查询并输出结果。只需输入您的问题（例如，展示上个季度前十名按收入排序的产品），SQL助理就会生成精确的SQL语句，对其运行您的数据库，并几秒钟内返回结果。"},
     "canvas_type": "Marketing",
     "dsl": {
             "components": {
@@ -713,4 +715,4 @@
             "retrieval": []
         },
     "avatar": "data:image/jpeg;base64,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"
-}
+}

agent/templates/title_chunker.json

@@ -2,10 +2,12 @@
     "id": 25,
     "title": {
         "en": "Title Chunker",
+        "de": "Titel basierte Segmentierung",
         "zh": "标题切片"
     },
     "description": {
         "en": "This template slices the parsed file based on its title structure. It is ideal for documents with well-defined headings, such as product manuals, legal contracts, research reports, and academic papers.",
+        "de": "Diese Vorlage segmentiert die geparste Datei basierend auf ihrer Titelstruktur. Sie eignet sich ideal für Dokumente mit klar definierten Überschriften, wie Produkthandbücher, Verträge, Forschungsberichte und wissenschaftliche Arbeiten.",
         "zh": "此模板将解析后的文件按标题结构进行切片，适用于具有清晰标题层级的文档类型，如产品手册、合同法规、研究报告和学术论文等。"
     },
     "canvas_type": "Ingestion Pipeline",

agent/tools/retrieval.py

@@ -18,12 +18,14 @@ import re
 from abc import ABC
 from agent.tools.base import ToolParamBase, ToolBase, ToolMeta
 from api.db import LLMType
+from api.db.services.document_service import DocumentService
+from api.db.services.dialog_service import meta_filter
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMBundle
 from api import settings
 from api.utils.api_utils import timeout
 from rag.app.tag import label_question
-from rag.prompts.generator import cross_languages, kb_prompt
+from rag.prompts.generator import cross_languages, kb_prompt, gen_meta_filter
 
 
 class RetrievalParam(ToolParamBase):
@@ -58,6 +60,7 @@ class RetrievalParam(ToolParamBase):
         self.use_kg = False
         self.cross_languages = []
         self.toc_enhance = False
+        self.meta_data_filter={}
 
     def check(self):
         self.check_decimal_float(self.similarity_threshold, "[Retrieval] Similarity threshold")
@@ -117,6 +120,21 @@ class Retrieval(ToolBase, ABC):
         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)
+        
+        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":
+                chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT)
+                filters = gen_meta_filter(chat_mdl, metas, query)
+                doc_ids.extend(meta_filter(metas, filters))
+                if not doc_ids:
+                    doc_ids = None
+            elif self._param.meta_data_filter.get("method") == "manual":
+                doc_ids.extend(meta_filter(metas, self._param.meta_data_filter["manual"]))
+                if not doc_ids:
+                    doc_ids = None
+
         if self._param.cross_languages:
             query = cross_languages(kbs[0].tenant_id, None, query, self._param.cross_languages)
 
@@ -131,6 +149,7 @@ class Retrieval(ToolBase, ABC):
                 self._param.top_n,
                 self._param.similarity_threshold,
                 1 - self._param.keywords_similarity_weight,
+                doc_ids=doc_ids,
                 aggs=False,
                 rerank_mdl=rerank_mdl,
                 rank_feature=label_question(query, kbs),

api/apps/__init__.py

@@ -27,7 +27,7 @@ from itsdangerous.url_safe import URLSafeTimedSerializer as Serializer
 from api.db import StatusEnum
 from api.db.db_models import close_connection
 from api.db.services import UserService
-from api.utils.json import CustomJSONEncoder
+from api.utils.json_encode import CustomJSONEncoder
 from api.utils import commands
 
 from flask_mail import Mail

api/apps/api_app.py

@@ -33,7 +33,7 @@ from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.task_service import queue_tasks, TaskService
 from api.db.services.user_service import UserTenantService
 from api import settings
-from api.utils import get_uuid, current_timestamp, datetime_format
+from api.utils import get_uuid
 from api.utils.api_utils import server_error_response, get_data_error_result, get_json_result, validate_request, \
     generate_confirmation_token
 
@@ -41,6 +41,7 @@ from api.utils.file_utils import filename_type, thumbnail
 from rag.app.tag import label_question
 from rag.prompts.generator import keyword_extraction
 from rag.utils.storage_factory import STORAGE_IMPL
+from common.time_utils import current_timestamp, datetime_format
 
 from api.db.services.canvas_service import UserCanvasService
 from agent.canvas import Canvas
@@ -58,7 +59,7 @@ def new_token():
             return get_data_error_result(message="Tenant not found!")
 
         tenant_id = tenants[0].tenant_id
-        obj = {"tenant_id": tenant_id, "token": generate_confirmation_token(tenant_id),
+        obj = {"tenant_id": tenant_id, "token": generate_confirmation_token(),
                "create_time": current_timestamp(),
                "create_date": datetime_format(datetime.now()),
                "update_time": None,
@@ -867,7 +868,7 @@ def retrieval():
     similarity_threshold = float(req.get("similarity_threshold", 0.2))
     vector_similarity_weight = float(req.get("vector_similarity_weight", 0.3))
     top = int(req.get("top_k", 1024))
-    highlight = bool(req.get("highlight", False)) 
+    highlight = bool(req.get("highlight", False))
 
     try:
         kbs = KnowledgebaseService.get_by_ids(kb_ids)

api/apps/chunk_app.py

@@ -35,7 +35,7 @@ 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.settings import PAGERANK_FLD
-from rag.utils import rmSpace
+from common.string_utils import remove_redundant_spaces
 
 
 @manager.route('/list', methods=['POST'])  # noqa: F821
@@ -60,12 +60,12 @@ 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=True)
+        sres = 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 = {
                 "chunk_id": id,
-                "content_with_weight": rmSpace(sres.highlight[id]) if question and id in sres.highlight else sres.field[
+                "content_with_weight": remove_redundant_spaces(sres.highlight[id]) if question and id in sres.highlight else sres.field[
                     id].get(
                     "content_with_weight", ""),
                 "doc_id": sres.field[id]["doc_id"],
@@ -350,7 +350,8 @@ def retrieval_test():
                                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"),
+                               doc_ids, rerank_mdl=rerank_mdl,
+                                             highlight=req.get("highlight", False),
                                rank_feature=labels
                                )
         if use_kg:

api/apps/document_app.py

@@ -45,7 +45,7 @@ from api.utils.api_utils import (
 from api.utils.file_utils import filename_type, get_project_base_directory, thumbnail
 from api.utils.web_utils import CONTENT_TYPE_MAP, html2pdf, is_valid_url
 from deepdoc.parser.html_parser import RAGFlowHtmlParser
-from rag.nlp import search
+from rag.nlp import search, rag_tokenizer
 from rag.utils.storage_factory import STORAGE_IMPL
 
 
@@ -524,6 +524,21 @@ def rename():
             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.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,
+            )
+
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)

api/apps/file_app.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License
 #
+import logging
 import os
 import pathlib
 import re
@@ -234,54 +235,63 @@ def get_all_parent_folders():
         return server_error_response(e)
 
 
-@manager.route('/rm', methods=['POST'])  # noqa: F821
+@manager.route("/rm", methods=["POST"])  # noqa: F821
 @login_required
 @validate_request("file_ids")
 def rm():
     req = request.json
     file_ids = req["file_ids"]
+
+    def _delete_single_file(file):
+        try:
+            if file.location:
+                STORAGE_IMPL.rm(file.parent_id, file.location)
+        except Exception:
+            logging.exception(f"Fail to remove object: {file.parent_id}/{file.location}")
+
+        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)
+
     try:
         for file_id in file_ids:
             e, file = FileService.get_by_id(file_id)
-            if not e:
+            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=settings.RetCode.AUTHENTICATION_ERROR)
+                return get_json_result(data=False, message="No authorization.", code=settings.RetCode.AUTHENTICATION_ERROR)
+
             if file.source_type == FileSource.KNOWLEDGEBASE:
                 continue
 
             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_data_error_result(message="File not found!")
-                    STORAGE_IMPL.rm(file.parent_id, file.location)
-                FileService.delete_folder_by_pf_id(current_user.id, file_id)
-            else:
-                STORAGE_IMPL.rm(file.parent_id, file.location)
-                if not FileService.delete(file):
-                    return get_data_error_result(
-                        message="Database error (File removal)!")
+                _delete_folder_recursive(file, current_user.id)
+                continue
 
-            # delete file2document
-            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_data_error_result(message="Document not found!")
-                tenant_id = DocumentService.get_tenant_id(doc_id)
-                if not tenant_id:
-                    return get_data_error_result(message="Tenant not found!")
-                if not DocumentService.remove_document(doc, tenant_id):
-                    return get_data_error_result(
-                        message="Database error (Document removal)!")
-            File2DocumentService.delete_by_file_id(file_id)
+            _delete_single_file(file)
 
         return get_json_result(data=True)
+
     except Exception as e:
         return server_error_response(e)
 
@@ -355,31 +365,89 @@ def get(file_id):
         return server_error_response(e)
 
 
-@manager.route('/mv', methods=['POST'])  # noqa: F821
+@manager.route("/mv", methods=["POST"])  # noqa: F821
 @login_required
 @validate_request("src_file_ids", "dest_file_id")
 def move():
     req = request.json
     try:
         file_ids = req["src_file_ids"]
-        parent_id = req["dest_file_id"]
+        dest_parent_id = req["dest_file_id"]
+
+        ok, dest_folder = FileService.get_by_id(dest_parent_id)
+        if not ok or not dest_folder:
+            return get_data_error_result(message="Parent Folder not found!")
+
         files = FileService.get_by_ids(file_ids)
-        files_dict = {}
-        for file in files:
-            files_dict[file.id] = file
+        if not files:
+            return get_data_error_result(message="Source files not found!")
+
+        files_dict = {f.id: f for f in files}
 
         for file_id in file_ids:
-            file = files_dict[file_id]
+            file = files_dict.get(file_id)
             if 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=settings.RetCode.AUTHENTICATION_ERROR)
-        fe, _ = FileService.get_by_id(parent_id)
-        if not fe:
-            return get_data_error_result(message="Parent Folder not found!")
-        FileService.move_file(file_ids, parent_id)
+                return get_json_result(
+                    data=False,
+                    message="No authorization.",
+                    code=settings.RetCode.AUTHENTICATION_ERROR,
+                )
+
+        def _move_entry_recursive(source_file_entry, dest_folder):
+            if source_file_entry.type == FileType.FOLDER.value:
+                existing_folder = FileService.query(name=source_file_entry.name, parent_id=dest_folder.id)
+                if existing_folder:
+                    new_folder = existing_folder[0]
+                else:
+                    new_folder = FileService.insert(
+                        {
+                            "id": get_uuid(),
+                            "parent_id": dest_folder.id,
+                            "tenant_id": source_file_entry.tenant_id,
+                            "created_by": current_user.id,
+                            "name": source_file_entry.name,
+                            "location": "",
+                            "size": 0,
+                            "type": FileType.FOLDER.value,
+                        }
+                    )
+
+                sub_files = FileService.list_all_files_by_parent_id(source_file_entry.id)
+                for sub_file in sub_files:
+                    _move_entry_recursive(sub_file, new_folder)
+
+                FileService.delete_by_id(source_file_entry.id)
+                return
+
+            old_parent_id = source_file_entry.parent_id
+            old_location = source_file_entry.location
+            filename = source_file_entry.name
+
+            new_location = filename
+            while STORAGE_IMPL.obj_exist(dest_folder.id, new_location):
+                new_location += "_"
+
+            try:
+                STORAGE_IMPL.move(old_parent_id, old_location, dest_folder.id, new_location)
+            except Exception as storage_err:
+                raise RuntimeError(f"Move file failed at storage layer: {str(storage_err)}")
+
+            FileService.update_by_id(
+                source_file_entry.id,
+                {
+                    "parent_id": dest_folder.id,
+                    "location": new_location,
+                },
+            )
+
+        for file in files:
+            _move_entry_recursive(file, dest_folder)
+
         return get_json_result(data=True)
+
     except Exception as e:
         return server_error_response(e)

api/apps/kb_app.py

@@ -15,11 +15,15 @@
 #
 import json
 import logging
+import random
 
 from flask import request
 from flask_login import login_required, current_user
+import numpy as np
 
+from api.db import LLMType
 from api.db.services import duplicate_name
+from api.db.services.llm_service import LLMBundle
 from api.db.services.document_service import DocumentService, queue_raptor_o_graphrag_tasks
 from api.db.services.file2document_service import File2DocumentService
 from api.db.services.file_service import FileService
@@ -38,6 +42,7 @@ from api.constants import DATASET_NAME_LIMIT
 from rag.settings import PAGERANK_FLD
 from rag.utils.redis_conn import REDIS_CONN
 from rag.utils.storage_factory import STORAGE_IMPL
+from rag.utils.doc_store_conn import OrderByExpr  
 
 
 @manager.route('/create', methods=['post'])  # noqa: F821
@@ -70,6 +75,7 @@ def create():
         e, t = TenantService.get_by_id(current_user.id)
         if not e:
             return get_data_error_result(message="Tenant not found.")
+
         req["parser_config"] = {
             "layout_recognize": "DeepDOC",
             "chunk_token_num": 512,
@@ -579,7 +585,7 @@ def run_graphrag():
     sample_document = documents[0]
     document_ids = [document["id"] for document in documents]
 
-    task_id = queue_raptor_o_graphrag_tasks(doc=sample_document, ty="graphrag", priority=0, fake_doc_id=GRAPH_RAPTOR_FAKE_DOC_ID, doc_ids=list(document_ids))
+    task_id = queue_raptor_o_graphrag_tasks(sample_doc_id=sample_document, ty="graphrag", priority=0, fake_doc_id=GRAPH_RAPTOR_FAKE_DOC_ID, doc_ids=list(document_ids))
 
     if not KnowledgebaseService.update_by_id(kb.id, {"graphrag_task_id": task_id}):
         logging.warning(f"Cannot save graphrag_task_id for kb {kb_id}")
@@ -648,7 +654,7 @@ def run_raptor():
     sample_document = documents[0]
     document_ids = [document["id"] for document in documents]
 
-    task_id = queue_raptor_o_graphrag_tasks(doc=sample_document, ty="raptor", priority=0, fake_doc_id=GRAPH_RAPTOR_FAKE_DOC_ID, doc_ids=list(document_ids))
+    task_id = queue_raptor_o_graphrag_tasks(sample_doc_id=sample_document, ty="raptor", priority=0, fake_doc_id=GRAPH_RAPTOR_FAKE_DOC_ID, doc_ids=list(document_ids))
 
     if not KnowledgebaseService.update_by_id(kb.id, {"raptor_task_id": task_id}):
         logging.warning(f"Cannot save raptor_task_id for kb {kb_id}")
@@ -717,7 +723,7 @@ def run_mindmap():
     sample_document = documents[0]
     document_ids = [document["id"] for document in documents]
 
-    task_id = queue_raptor_o_graphrag_tasks(doc=sample_document, ty="mindmap", priority=0, fake_doc_id=GRAPH_RAPTOR_FAKE_DOC_ID, doc_ids=list(document_ids))
+    task_id = queue_raptor_o_graphrag_tasks(sample_doc_id=sample_document, ty="mindmap", priority=0, fake_doc_id=GRAPH_RAPTOR_FAKE_DOC_ID, doc_ids=list(document_ids))
 
     if not KnowledgebaseService.update_by_id(kb.id, {"mindmap_task_id": task_id}):
         logging.warning(f"Cannot save mindmap_task_id for kb {kb_id}")
@@ -787,3 +793,141 @@ def delete_kb_task():
         return server_error_response(f"Internal error: cannot delete task {pipeline_task_type}")
 
     return get_json_result(data=True)
+
+@manager.route("/check_embedding", methods=["post"])  # noqa: F821
+@login_required
+def check_embedding():
+
+    def _guess_vec_field(src: dict) -> str | None:
+        for k in src or {}:
+            if k.endswith("_vec"):
+                return k
+        return None
+
+    def _as_float_vec(v):
+        if v is None:
+            return []
+        if isinstance(v, str):
+            return [float(x) for x in v.split("\t") if x != ""]
+        if isinstance(v, (list, tuple, np.ndarray)):
+            return [float(x) for x in v]
+        return []
+
+    def _to_1d(x):
+        a = np.asarray(x, dtype=np.float32)
+        return a.reshape(-1)  
+
+    def _cos_sim(a, b, eps=1e-12):
+        a = _to_1d(a)
+        b = _to_1d(b)
+        na = np.linalg.norm(a)
+        nb = np.linalg.norm(b)
+        if na < eps or nb < eps: 
+            return 0.0
+        return float(np.dot(a, b) / (na * nb))
+
+    def sample_random_chunks_with_vectors(
+        docStoreConn,
+        tenant_id: str,
+        kb_id: str,
+        n: int = 5,
+        base_fields=("docnm_kwd","doc_id","content_with_weight","page_num_int","position_int","top_int"),
+    ):
+        index_nm = search.index_name(tenant_id)
+
+        res0 = docStoreConn.search(
+            selectFields=[], highlightFields=[],
+            condition={"kb_id": kb_id, "available_int": 1},
+            matchExprs=[], orderBy=OrderByExpr(),
+            offset=0, limit=1,
+            indexNames=index_nm, knowledgebaseIds=[kb_id]
+        )
+        total = docStoreConn.getTotal(res0)
+        if total <= 0:
+            return []
+
+        n = min(n, total)
+        offsets = sorted(random.sample(range(total), n))
+        out = []
+
+        for off in offsets:
+            res1 = docStoreConn.search(
+                selectFields=list(base_fields),
+                highlightFields=[],
+                condition={"kb_id": kb_id, "available_int": 1},
+                matchExprs=[], orderBy=OrderByExpr(),
+                offset=off, limit=1,
+                indexNames=index_nm, knowledgebaseIds=[kb_id]
+            )
+            ids = docStoreConn.getChunkIds(res1)
+            if not ids: 
+                continue
+
+            cid = ids[0]
+            full_doc = docStoreConn.get(cid, index_nm, [kb_id]) or {}
+            vec_field = _guess_vec_field(full_doc)
+            vec = _as_float_vec(full_doc.get(vec_field))
+
+            out.append({
+                "chunk_id": cid,
+                "kb_id": kb_id,
+                "doc_id": full_doc.get("doc_id"),
+                "doc_name": full_doc.get("docnm_kwd"),
+                "vector_field": vec_field,
+                "vector_dim": len(vec),
+                "vector": vec,
+                "page_num_int": full_doc.get("page_num_int"),
+                "position_int": full_doc.get("position_int"),
+                "top_int": full_doc.get("top_int"),
+                "content_with_weight": full_doc.get("content_with_weight") or "",
+            })
+        return out
+    req = request.json
+    kb_id = req.get("kb_id", "")
+    embd_id = req.get("embd_id", "")
+    n = int(req.get("check_num", 5))
+    _, kb = KnowledgebaseService.get_by_id(kb_id)
+    tenant_id = kb.tenant_id
+
+    emb_mdl = LLMBundle(tenant_id, LLMType.EMBEDDING, embd_id)
+    samples = sample_random_chunks_with_vectors(settings.docStoreConn, tenant_id=tenant_id, kb_id=kb_id, n=n)
+
+    results, eff_sims = [], []
+    for ck in samples:
+        txt = (ck.get("content_with_weight") or "").strip()
+        if not txt:
+            results.append({"chunk_id": ck["chunk_id"], "reason": "no_text"})
+            continue
+
+        if not ck.get("vector"):
+            results.append({"chunk_id": ck["chunk_id"], "reason": "no_stored_vector"})
+            continue
+
+        try:
+            qv, _ = emb_mdl.encode_queries(txt)  
+            sim = _cos_sim(qv, ck["vector"])
+        except Exception:
+            return get_error_data_result(message="embedding failure")
+
+        eff_sims.append(sim)
+        results.append({
+            "chunk_id": ck["chunk_id"],
+            "doc_id": ck["doc_id"],
+            "doc_name": ck["doc_name"],
+            "vector_field": ck["vector_field"],
+            "vector_dim": ck["vector_dim"],
+            "cos_sim": round(sim, 6),
+        })
+
+    summary = {
+        "kb_id": kb_id,
+        "model": embd_id,
+        "sampled": len(samples),
+        "valid": len(eff_sims),
+        "avg_cos_sim": round(float(np.mean(eff_sims)) if eff_sims else 0.0, 6),
+        "min_cos_sim": round(float(np.min(eff_sims)) if eff_sims else 0.0, 6),
+        "max_cos_sim": round(float(np.max(eff_sims)) if eff_sims else 0.0, 6),
+    }
+    if summary["avg_cos_sim"] > 0.99:
+        return get_json_result(data={"summary": summary, "results": results})
+    return get_json_result(code=settings.RetCode.NOT_EFFECTIVE, message="failed", data={"summary": summary, "results": results})

api/apps/llm_app.py

@@ -15,11 +15,11 @@
 #
 import logging
 import json
+import os
 from flask import request
 from flask_login 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 import settings
 from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
 from api.db import StatusEnum, LLMType
 from api.db.db_models import TenantLLM
@@ -194,6 +194,9 @@ def add_llm():
     elif factory == "Azure-OpenAI":
         api_key = apikey_json(["api_key", "api_version"])
 
+    elif factory == "OpenRouter":
+        api_key = apikey_json(["api_key", "provider_order"])
+
     llm = {
         "tenant_id": current_user.id,
         "llm_factory": factory,
@@ -212,7 +215,7 @@ def add_llm():
         mdl = EmbeddingModel[factory](
             key=llm['api_key'],
             model_name=mdl_nm,
-            base_url=llm["api_base"])
+            base_url=llm["api_base"])   
         try:
             arr, tc = mdl.encode(["Test if the api key is available"])
             if len(arr[0]) == 0:
@@ -261,7 +264,7 @@ def add_llm():
         try:
             image_data = test_image
             m, tc = mdl.describe(image_data)
-            if not m and not tc:
+            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)
@@ -365,8 +368,8 @@ def my_llms():
 @manager.route('/list', methods=['GET'])  # noqa: F821
 @login_required
 def list_app():
-    self_deployed = ["Youdao", "FastEmbed", "BAAI", "Ollama", "Xinference", "LocalAI", "LM-Studio", "GPUStack"]
-    weighted = ["Youdao", "FastEmbed", "BAAI"] if settings.LIGHTEN != 0 else []
+    self_deployed = ["FastEmbed", "Ollama", "Xinference", "LocalAI", "LM-Studio", "GPUStack"]
+    weighted = []
     model_type = request.args.get("model_type")
     try:
         objs = TenantLLMService.query(tenant_id=current_user.id)
@@ -376,6 +379,8 @@ def list_app():
                 for m in llms if m.status == StatusEnum.VALID.value and m.fid not in weighted]
         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', ''):
+                m["available"] = True
 
         llm_set = set([m["llm_name"] + "@" + m["fid"] for m in llms])
         for o in objs:

api/apps/sdk/chat.py

@@ -169,6 +169,8 @@ def update(tenant_id, chat_id):
         if len(embd_count) > 1:
             return get_result(message='Datasets use different embedding models."', code=settings.RetCode.AUTHENTICATION_ERROR)
         req["kb_ids"] = ids
+    else:
+        req["kb_ids"] = []
     llm = req.get("llm")
     if llm:
         if "model_name" in llm:

api/apps/sdk/doc.py

@@ -41,8 +41,8 @@ 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 cross_languages, keyword_extraction
-from rag.utils import rmSpace
 from rag.utils.storage_factory import STORAGE_IMPL
+from common.string_utils import remove_redundant_spaces
 
 MAXIMUM_OF_UPLOADING_FILES = 256
 
@@ -470,6 +470,20 @@ def list_docs(dataset_id, tenant_id):
         required: false
         default: 0
         description: Unix timestamp for filtering documents created before this time. 0 means no filter.
+      - in: query
+        name: suffix
+        type: array
+        items:
+          type: string
+        required: false
+        description: Filter by file suffix (e.g., ["pdf", "txt", "docx"]).
+      - in: query
+        name: run
+        type: array
+        items:
+          type: string
+        required: false
+        description: Filter by document run status. Supports both numeric ("0", "1", "2", "3", "4") and text formats ("UNSTART", "RUNNING", "CANCEL", "DONE", "FAIL").
       - in: header
         name: Authorization
         type: string
@@ -512,63 +526,62 @@ def list_docs(dataset_id, tenant_id):
                     description: Processing status.
     """
     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}. ")
-    id = request.args.get("id")
-    name = request.args.get("name")
+      return get_error_data_result(message=f"You don't own the dataset {dataset_id}. ")
 
-    if id and not DocumentService.query(id=id, kb_id=dataset_id):
-        return get_error_data_result(message=f"You don't own the document {id}.")
+    q = request.args
+    document_id = q.get("id")  
+    name        = q.get("name")
+
+    if document_id and not DocumentService.query(id=document_id, kb_id=dataset_id):
+        return get_error_data_result(message=f"You don't own the document {document_id}.")
     if name and not DocumentService.query(name=name, kb_id=dataset_id):
         return get_error_data_result(message=f"You don't own the document {name}.")
 
-    page = int(request.args.get("page", 1))
-    keywords = request.args.get("keywords", "")
-    page_size = int(request.args.get("page_size", 30))
-    orderby = request.args.get("orderby", "create_time")
-    if request.args.get("desc") == "False":
-        desc = False
-    else:
-        desc = True
-    docs, tol = DocumentService.get_list(dataset_id, page, page_size, orderby, desc, keywords, id, name)
+    page        = int(q.get("page", 1))
+    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", "")
 
-    create_time_from = int(request.args.get("create_time_from", 0))
-    create_time_to = int(request.args.get("create_time_to", 0))
+    # 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))    
 
+    # map run status (accept 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]
+
+    docs, total = DocumentService.get_list(
+        dataset_id, page, page_size, orderby, desc, keywords, document_id, name, suffix, run_status_converted
+    )
+
+    # time range filter (0 means no bound)
     if create_time_from or create_time_to:
-        filtered_docs = []
-        for doc in docs:
-            doc_create_time = doc.get("create_time", 0)
-            if (create_time_from == 0 or doc_create_time >= create_time_from) and (create_time_to == 0 or doc_create_time <= create_time_to):
-                filtered_docs.append(doc)
-        docs = filtered_docs
+        docs = [
+            d for d in docs
+            if (create_time_from == 0 or d.get("create_time", 0) >= create_time_from)
+            and (create_time_to == 0 or d.get("create_time", 0) <= create_time_to)
+        ]
 
-    # rename key's name
-    renamed_doc_list = []
+    # 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",
     }
-    run_mapping = {
-        "0": "UNSTART",
-        "1": "RUNNING",
-        "2": "CANCEL",
-        "3": "DONE",
-        "4": "FAIL",
-    }
-    for doc in docs:
-        renamed_doc = {}
-        for key, value in doc.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_list.append(renamed_doc)
-    return get_result(data={"total": tol, "docs": renamed_doc_list})
+    run_status_numeric_to_text = {"0": "UNSTART", "1": "RUNNING", "2": "CANCEL", "3": "DONE", "4": "FAIL"}
 
+    output_docs = []
+    for d in docs:
+        renamed_doc = {key_mapping.get(k, k): v for k, v in d.items()}
+        if "run" in d:
+            renamed_doc["run"] = run_status_numeric_to_text.get(str(d["run"]), d["run"])
+        output_docs.append(renamed_doc)
+
+    return get_result(data={"total": total, "docs": output_docs})
 
 @manager.route("/datasets/<dataset_id>/documents", methods=["DELETE"])  # noqa: F821
 @token_required
@@ -987,7 +1000,7 @@ def list_chunks(tenant_id, dataset_id, document_id):
         for id in sres.ids:
             d = {
                 "id": id,
-                "content": (rmSpace(sres.highlight[id]) if question and id in sres.highlight else sres.field[id].get("content_with_weight", "")),
+                "content": (remove_redundant_spaces(sres.highlight[id]) if question and id in sres.highlight else sres.field[id].get("content_with_weight", "")),
                 "document_id": sres.field[id]["doc_id"],
                 "docnm_kwd": sres.field[id]["docnm_kwd"],
                 "important_keywords": sres.field[id].get("important_kwd", []),

api/apps/system_app.py

@@ -24,7 +24,6 @@ from api.db.services.api_service import APITokenService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.user_service import UserTenantService
 from api import settings
-from api.utils import current_timestamp, datetime_format
 from api.utils.api_utils import (
     get_json_result,
     get_data_error_result,
@@ -32,6 +31,7 @@ from api.utils.api_utils import (
     generate_confirmation_token,
 )
 from api.versions import get_ragflow_version
+from common.time_utils import current_timestamp, datetime_format
 from rag.utils.storage_factory import STORAGE_IMPL, STORAGE_IMPL_TYPE
 from timeit import default_timer as timer
 
@@ -217,8 +217,8 @@ def new_token():
         tenant_id = [tenant for tenant in tenants if tenant.role == 'owner'][0].tenant_id
         obj = {
             "tenant_id": tenant_id,
-            "token": generate_confirmation_token(tenant_id),
-            "beta": generate_confirmation_token(generate_confirmation_token(tenant_id)).replace("ragflow-", "")[:32],
+            "token": generate_confirmation_token(),
+            "beta": generate_confirmation_token().replace("ragflow-", "")[:32],
             "create_time": current_timestamp(),
             "create_date": datetime_format(datetime.now()),
             "update_time": None,
@@ -274,7 +274,7 @@ def token_list():
         objs = [o.to_dict() for o in objs]
         for o in objs:
             if not o["beta"]:
-                o["beta"] = generate_confirmation_token(generate_confirmation_token(tenants[0].tenant_id)).replace(
+                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)

api/apps/tenant_app.py

@@ -23,7 +23,8 @@ from api.db import UserTenantRole, StatusEnum
 from api.db.db_models import UserTenant
 from api.db.services.user_service import UserTenantService, UserService
 
-from api.utils import get_uuid, delta_seconds
+from api.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.web_utils import send_invite_email
 

api/apps/user_app.py

@@ -15,11 +15,14 @@
 #
 import json
 import logging
+import string
+import os
 import re
 import secrets
+import time
 from datetime import datetime
 
-from flask import redirect, request, session
+from flask import redirect, request, session, make_response
 from flask_login import current_user, login_required, login_user, logout_user
 from werkzeug.security import check_password_hash, generate_password_hash
 
@@ -31,13 +34,8 @@ from api.db.services.file_service import FileService
 from api.db.services.llm_service import get_init_tenant_llm
 from api.db.services.tenant_llm_service import TenantLLMService
 from api.db.services.user_service import TenantService, UserService, UserTenantService
-from api.utils import (
-    current_timestamp,
-    datetime_format,
-    download_img,
-    get_format_time,
-    get_uuid,
-)
+from common.time_utils import current_timestamp, datetime_format, get_format_time
+from api.utils import download_img, get_uuid
 from api.utils.api_utils import (
     construct_response,
     get_data_error_result,
@@ -46,6 +44,19 @@ from api.utils.api_utils import (
     validate_request,
 )
 from api.utils.crypt import decrypt
+from rag.utils.redis_conn import REDIS_CONN
+from api.apps import smtp_mail_server
+from api.utils.web_utils import (
+    send_email_html,
+    OTP_LENGTH,
+    OTP_TTL_SECONDS,
+    ATTEMPT_LIMIT,
+    ATTEMPT_LOCK_SECONDS,
+    RESEND_COOLDOWN_SECONDS,
+    otp_keys,
+    hash_code,
+    captcha_key,
+)
 
 
 @manager.route("/login", methods=["POST", "GET"])  # noqa: F821
@@ -825,3 +836,172 @@ def set_tenant_info():
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)
+        
+
+@manager.route("/forget/captcha", methods=["GET"])  # noqa: F821
+def forget_get_captcha():
+    """
+    GET /forget/captcha?email=<email>
+    - Generate an image captcha and cache it in Redis under key captcha:{email} with TTL = OTP_TTL_SECONDS.
+    - Returns the captcha as a PNG image.
+    """
+    email = (request.args.get("email") or "")
+    if not email:
+        return get_json_result(data=False, code=settings.RetCode.ARGUMENT_ERROR, message="email is required")
+
+    users = UserService.query(email=email)
+    if not users:
+        return get_json_result(data=False, code=settings.RetCode.DATA_ERROR, message="invalid email")
+
+    # Generate captcha text
+    allowed = string.ascii_uppercase + string.digits
+    captcha_text = "".join(secrets.choice(allowed) for _ in range(OTP_LENGTH))
+    REDIS_CONN.set(captcha_key(email), captcha_text, 60) # Valid for 60 seconds
+
+    from captcha.image import ImageCaptcha
+    image = ImageCaptcha(width=300, height=120, font_sizes=[50, 60, 70])
+    img_bytes = image.generate(captcha_text).read()
+    response = make_response(img_bytes)
+    response.headers.set("Content-Type", "image/JPEG")
+    return response
+
+
+@manager.route("/forget/otp", methods=["POST"])  # noqa: F821
+def forget_send_otp():
+    """
+    POST /forget/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 = request.get_json()
+    email = req.get("email") or ""
+    captcha = (req.get("captcha") or "").strip()
+
+    if not email or not captcha:
+        return get_json_result(data=False, code=settings.RetCode.ARGUMENT_ERROR, message="email and captcha required")
+
+    users = UserService.query(email=email)
+    if not users:
+        return get_json_result(data=False, code=settings.RetCode.DATA_ERROR, message="invalid email")
+
+    stored_captcha = REDIS_CONN.get(captcha_key(email))
+    if not stored_captcha:
+        return get_json_result(data=False, code=settings.RetCode.NOT_EFFECTIVE, message="invalid or expired captcha")
+    if (stored_captcha or "").strip().lower() != captcha.lower():
+        return get_json_result(data=False, code=settings.RetCode.AUTHENTICATION_ERROR, message="invalid or expired captcha")
+
+    # Delete captcha to prevent reuse
+    REDIS_CONN.delete(captcha_key(email))
+
+    k_code, k_attempts, k_last, k_lock = otp_keys(email)
+    now = int(time.time())
+    last_ts = REDIS_CONN.get(k_last)
+    if last_ts:
+        try:
+            elapsed = now - int(last_ts)
+        except Exception:
+            elapsed = RESEND_COOLDOWN_SECONDS
+        remaining = RESEND_COOLDOWN_SECONDS - elapsed
+        if remaining > 0:
+            return get_json_result(data=False, code=settings.RetCode.NOT_EFFECTIVE, message=f"you still have to wait {remaining} seconds")
+
+    # Generate OTP (uppercase letters only) and store hashed
+    otp = "".join(secrets.choice(string.ascii_uppercase) for _ in range(OTP_LENGTH))
+    salt = os.urandom(16)
+    code_hash = hash_code(otp, salt)
+    REDIS_CONN.set(k_code, f"{code_hash}:{salt.hex()}", OTP_TTL_SECONDS)
+    REDIS_CONN.set(k_attempts, 0, OTP_TTL_SECONDS)
+    REDIS_CONN.set(k_last, now, OTP_TTL_SECONDS)
+    REDIS_CONN.delete(k_lock)
+
+    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=settings.RetCode.SERVER_ERROR, message="failed to send email")
+        
+    return get_json_result(data=True, code=settings.RetCode.SUCCESS, message="verification passed, email sent")
+
+
+@manager.route("/forget", methods=["POST"])  # noqa: F821
+def forget():
+    """
+    POST: Verify email + OTP and reset password, then log the user in.
+    Request JSON: { email, otp, new_password, confirm_new_password }
+    """
+    req = request.get_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=settings.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=settings.RetCode.ARGUMENT_ERROR, message="passwords do not match")
+
+    users = UserService.query(email=email)
+    if not users:
+        return get_json_result(data=False, code=settings.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):
+        return get_json_result(data=False, code=settings.RetCode.NOT_EFFECTIVE, message="too many attempts, try later")
+
+    stored = REDIS_CONN.get(k_code)
+    if not stored:
+        return get_json_result(data=False, code=settings.RetCode.NOT_EFFECTIVE, message="expired otp")
+
+    try:
+        stored_hash, salt_hex = str(stored).split(":", 1)
+        salt = bytes.fromhex(salt_hex)
+    except Exception:
+        return get_json_result(data=False, code=settings.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
+        try:
+            attempts = int(REDIS_CONN.get(k_attempts) or 0) + 1
+        except Exception:
+            attempts = 1
+        REDIS_CONN.set(k_attempts, attempts, OTP_TTL_SECONDS)
+        if attempts >= ATTEMPT_LIMIT:
+            REDIS_CONN.set(k_lock, int(time.time()), ATTEMPT_LOCK_SECONDS)
+        return get_json_result(data=False, code=settings.RetCode.AUTHENTICATION_ERROR, message="expired otp")
+
+    # Success: consume OTP and reset password
+    REDIS_CONN.delete(k_code)
+    REDIS_CONN.delete(k_attempts)
+    REDIS_CONN.delete(k_last)
+    REDIS_CONN.delete(k_lock)
+
+    try:
+        UserService.update_user_password(user.id, new_pwd)
+    except Exception as e:
+        logging.exception(e)
+        return get_json_result(data=False, code=settings.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()
+    msg = "Password reset successful. Logged in."
+    return construct_response(data=user.to_json(), auth=user.get_id(), message=msg)

api/common/exceptions.py

@@ -36,3 +36,8 @@ class UserAlreadyExistsError(AdminException):
 class CannotDeleteAdminError(AdminException):
     def __init__(self):
         super().__init__("Cannot delete admin account", 403)
+
+
+class NotAdminError(AdminException):
+    def __init__(self, username):
+        super().__init__(f"User '{username}' is not admin", 403)

api/db/db_models.py

@@ -32,9 +32,11 @@ from playhouse.pool import PooledMySQLDatabase, PooledPostgresqlDatabase
 
 from api import settings, utils
 from api.db import ParserType, SerializedType
-from api.utils.json import json_dumps, json_loads
+from api.utils.json_encode import json_dumps, json_loads
 from api.utils.configs import deserialize_b64, serialize_b64
 
+from common.time_utils import current_timestamp, timestamp_to_date, date_string_to_timestamp
+
 
 def singleton(cls, *args, **kw):
     instances = {}
@@ -189,7 +191,7 @@ class BaseModel(Model):
                         for i, v in enumerate(f_v):
                             if isinstance(v, str) and f_n in auto_date_timestamp_field():
                                 # time type: %Y-%m-%d %H:%M:%S
-                                f_v[i] = utils.date_string_to_timestamp(v)
+                                f_v[i] = date_string_to_timestamp(v)
                         lt_value = f_v[0]
                         gt_value = f_v[1]
                         if lt_value is not None and gt_value is not None:
@@ -218,9 +220,9 @@ class BaseModel(Model):
     @classmethod
     def insert(cls, __data=None, **insert):
         if isinstance(__data, dict) and __data:
-            __data[cls._meta.combined["create_time"]] = utils.current_timestamp()
+            __data[cls._meta.combined["create_time"]] = current_timestamp()
         if insert:
-            insert["create_time"] = utils.current_timestamp()
+            insert["create_time"] = current_timestamp()
 
         return super().insert(__data, **insert)
 
@@ -231,11 +233,11 @@ class BaseModel(Model):
         if not normalized:
             return {}
 
-        normalized[cls._meta.combined["update_time"]] = utils.current_timestamp()
+        normalized[cls._meta.combined["update_time"]] = current_timestamp()
 
         for f_n in AUTO_DATE_TIMESTAMP_FIELD_PREFIX:
             if {f"{f_n}_time", f"{f_n}_date"}.issubset(cls._meta.combined.keys()) and cls._meta.combined[f"{f_n}_time"] in normalized and normalized[cls._meta.combined[f"{f_n}_time"]] is not None:
-                normalized[cls._meta.combined[f"{f_n}_date"]] = utils.timestamp_to_date(normalized[cls._meta.combined[f"{f_n}_time"]])
+                normalized[cls._meta.combined[f"{f_n}_date"]] = timestamp_to_date(normalized[cls._meta.combined[f"{f_n}_time"]])
 
         return normalized
 
@@ -313,9 +315,75 @@ class RetryingPooledMySQLDatabase(PooledMySQLDatabase):
                     raise
 
 
+class RetryingPooledPostgresqlDatabase(PooledPostgresqlDatabase):
+    def __init__(self, *args, **kwargs):
+        self.max_retries = kwargs.pop("max_retries", 5)
+        self.retry_delay = kwargs.pop("retry_delay", 1)
+        super().__init__(*args, **kwargs)
+
+    def execute_sql(self, sql, params=None, commit=True):
+        for attempt in range(self.max_retries + 1):
+            try:
+                return super().execute_sql(sql, params, commit)
+            except (OperationalError, InterfaceError) as e:
+                # PostgreSQL specific error codes
+                # 57P01: admin_shutdown
+                # 57P02: crash_shutdown
+                # 57P03: cannot_connect_now
+                # 08006: connection_failure
+                # 08003: connection_does_not_exist
+                # 08000: connection_exception
+                error_messages = ['connection', 'server closed', 'connection refused',
+                                'no connection to the server', 'terminating connection']
+
+                should_retry = any(msg in str(e).lower() for msg in error_messages)
+
+                if should_retry and attempt < self.max_retries:
+                    logging.warning(
+                        f"PostgreSQL connection issue (attempt {attempt+1}/{self.max_retries}): {e}"
+                    )
+                    self._handle_connection_loss()
+                    time.sleep(self.retry_delay * (2 ** attempt))
+                else:
+                    logging.error(f"PostgreSQL execution failure: {e}")
+                    raise
+        return None
+
+    def _handle_connection_loss(self):
+        try:
+            self.close()
+        except Exception:
+            pass
+        try:
+            self.connect()
+        except Exception as e:
+            logging.error(f"Failed to reconnect to PostgreSQL: {e}")
+            time.sleep(0.1)
+            self.connect()
+
+    def begin(self):
+        for attempt in range(self.max_retries + 1):
+            try:
+                return super().begin()
+            except (OperationalError, InterfaceError) as e:
+                error_messages = ['connection', 'server closed', 'connection refused',
+                                'no connection to the server', 'terminating connection']
+
+                should_retry = any(msg in str(e).lower() for msg in error_messages)
+
+                if should_retry and attempt < self.max_retries:
+                    logging.warning(
+                        f"PostgreSQL connection lost during transaction (attempt {attempt+1}/{self.max_retries})"
+                    )
+                    self._handle_connection_loss()
+                    time.sleep(self.retry_delay * (2 ** attempt))
+                else:
+                    raise
+
+
 class PooledDatabase(Enum):
     MYSQL = RetryingPooledMySQLDatabase
-    POSTGRES = PooledPostgresqlDatabase
+    POSTGRES = RetryingPooledPostgresqlDatabase
 
 
 class DatabaseMigrator(Enum):
@@ -328,7 +396,7 @@ class BaseDataBase:
     def __init__(self):
         database_config = settings.DATABASE.copy()
         db_name = database_config.pop("name")
-        
+
         pool_config = {
             'max_retries': 5,
             'retry_delay': 1,

api/db/db_utils.py

@@ -18,7 +18,7 @@ from functools import reduce
 
 from playhouse.pool import PooledMySQLDatabase
 
-from api.utils import current_timestamp, timestamp_to_date
+from common.time_utils import current_timestamp, timestamp_to_date
 
 from api.db.db_models import DB, DataBaseModel
 

api/db/services/api_service.py

@@ -19,7 +19,7 @@ import peewee
 
 from api.db.db_models import DB, API4Conversation, APIToken, Dialog
 from api.db.services.common_service import CommonService
-from api.utils import current_timestamp, datetime_format
+from common.time_utils import current_timestamp, datetime_format
 
 
 class APITokenService(CommonService):

api/db/services/common_service.py

@@ -19,7 +19,8 @@ import peewee
 from peewee import InterfaceError, OperationalError
 
 from api.db.db_models import DB
-from api.utils import current_timestamp, datetime_format, get_uuid
+from api.utils import get_uuid
+from common.time_utils import current_timestamp, datetime_format
 
 def retry_db_operation(func):
     @retry(

api/db/services/dialog_service.py

@@ -34,15 +34,16 @@ from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.langfuse_service import TenantLangfuseService
 from api.db.services.llm_service import LLMBundle
 from api.db.services.tenant_llm_service import TenantLLMService
-from api.utils import current_timestamp, datetime_format
+from common.time_utils import current_timestamp, datetime_format
 from graphrag.general.mind_map_extractor import MindMapExtractor
 from rag.app.resume import forbidden_select_fields4resume
 from rag.app.tag import label_question
 from rag.nlp.search import index_name
 from rag.prompts.generator import chunks_format, citation_prompt, cross_languages, full_question, kb_prompt, keyword_extraction, message_fit_in, \
     gen_meta_filter, PROMPT_JINJA_ENV, ASK_SUMMARY
-from rag.utils import num_tokens_from_string, rmSpace
+from rag.utils import num_tokens_from_string
 from rag.utils.tavily_conn import Tavily
+from common.string_utils import remove_redundant_spaces
 
 
 class DialogService(CommonService):
@@ -706,7 +707,7 @@ Please write the SQL, only SQL, without any other explanations or text.
 
     line = "|" + "|".join(["------" for _ in range(len(column_idx))]) + ("|------|" if docid_idx and docid_idx else "")
 
-    rows = ["|" + "|".join([rmSpace(str(r[i])) for i in column_idx]).replace("None", " ") + "|" for r in tbl["rows"]]
+    rows = ["|" + "|".join([remove_redundant_spaces(str(r[i])) for i in column_idx]).replace("None", " ") + "|" for r in tbl["rows"]]
     rows = [r for r in rows if re.sub(r"[ |]+", "", r)]
     if quota:
         rows = "\n".join([r + f" ##{ii}$$ |" for ii, r in enumerate(rows)])

api/db/services/document_service.py

@@ -34,7 +34,8 @@ from api.db.db_models import DB, Document, Knowledgebase, Task, Tenant, UserTena
 from api.db.db_utils import bulk_insert_into_db
 from api.db.services.common_service import CommonService
 from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.utils import current_timestamp, get_format_time, get_uuid
+from api.utils import get_uuid
+from common.time_utils import current_timestamp, get_format_time
 from rag.nlp import rag_tokenizer, search
 from rag.settings import get_svr_queue_name, SVR_CONSUMER_GROUP_NAME
 from rag.utils.redis_conn import REDIS_CONN
@@ -79,7 +80,7 @@ class DocumentService(CommonService):
     @classmethod
     @DB.connection_context()
     def get_list(cls, kb_id, page_number, items_per_page,
-                 orderby, desc, keywords, id, name):
+                 orderby, desc, keywords, id, name, suffix=None, run = None):
         fields = cls.get_cls_model_fields()
         docs = cls.model.select(*[*fields, UserCanvas.title]).join(File2Document, on = (File2Document.document_id == cls.model.id))\
             .join(File, on = (File.id == File2Document.file_id))\
@@ -96,6 +97,10 @@ class DocumentService(CommonService):
             docs = docs.where(
                 fn.LOWER(cls.model.name).contains(keywords.lower())
             )
+        if suffix:
+            docs = docs.where(cls.model.suffix.in_(suffix))
+        if run:
+            docs = docs.where(cls.model.run.in_(run))
         if desc:
             docs = docs.order_by(cls.model.getter_by(orderby).desc())
         else:
@@ -667,9 +672,11 @@ class DocumentService(CommonService):
     @classmethod
     @DB.connection_context()
     def _sync_progress(cls, docs:list[dict]):
+        from api.db.services.task_service import TaskService
+
         for d in docs:
             try:
-                tsks = Task.query(doc_id=d["id"], order_by=Task.create_time)
+                tsks = TaskService.query(doc_id=d["id"], order_by=Task.create_time)
                 if not tsks:
                     continue
                 msg = []
@@ -787,21 +794,23 @@ class DocumentService(CommonService):
             "cancelled": int(cancelled),
         }
 
-def queue_raptor_o_graphrag_tasks(doc, ty, priority, fake_doc_id="", doc_ids=[]):
+def queue_raptor_o_graphrag_tasks(sample_doc_id, ty, priority, fake_doc_id="", doc_ids=[]):
     """
     You can provide a fake_doc_id to bypass the restriction of tasks at the knowledgebase level.
     Optionally, specify a list of doc_ids to determine which documents participate in the task.
     """
-    chunking_config = DocumentService.get_chunking_config(doc["id"])
+    assert ty in ["graphrag", "raptor", "mindmap"], "type should be graphrag, raptor or mindmap"
+
+    chunking_config = DocumentService.get_chunking_config(sample_doc_id["id"])
     hasher = xxhash.xxh64()
     for field in sorted(chunking_config.keys()):
         hasher.update(str(chunking_config[field]).encode("utf-8"))
 
     def new_task():
-        nonlocal doc
+        nonlocal sample_doc_id
         return {
             "id": get_uuid(),
-            "doc_id": fake_doc_id if fake_doc_id else doc["id"],
+            "doc_id": sample_doc_id["id"],
             "from_page": 100000000,
             "to_page": 100000000,
             "task_type": ty,
@@ -816,9 +825,9 @@ def queue_raptor_o_graphrag_tasks(doc, ty, priority, fake_doc_id="", doc_ids=[])
     task["digest"] = hasher.hexdigest()
     bulk_insert_into_db(Task, [task], True)
 
-    if ty in ["graphrag", "raptor", "mindmap"]:
-        task["doc_ids"] = doc_ids
-        DocumentService.begin2parse(doc["id"])
+    task["doc_id"] = fake_doc_id
+    task["doc_ids"] = doc_ids
+    DocumentService.begin2parse(sample_doc_id["id"])
     assert REDIS_CONN.queue_product(get_svr_queue_name(priority), message=task), "Can't access Redis. Please check the Redis' status."
     return task["id"]
 

api/db/services/file2document_service.py

@@ -20,7 +20,7 @@ from api.db.db_models import DB
 from api.db.db_models import File, File2Document
 from api.db.services.common_service import CommonService
 from api.db.services.document_service import DocumentService
-from api.utils import current_timestamp, datetime_format
+from common.time_utils import current_timestamp, datetime_format
 
 
 class File2DocumentService(CommonService):

api/db/services/file_service.py

@@ -476,6 +476,16 @@ class FileService(CommonService):
 
         return err, files
 
+    @classmethod
+    @DB.connection_context()
+    def list_all_files_by_parent_id(cls, parent_id):
+        try:
+            files = cls.model.select().where((cls.model.parent_id == parent_id) & (cls.model.id != parent_id))
+            return list(files)
+        except Exception:
+            logging.exception("list_by_parent_id failed")
+            raise RuntimeError("Database error (list_by_parent_id)!")
+
     @staticmethod
     def parse_docs(file_objs, user_id):
         exe = ThreadPoolExecutor(max_workers=12)

api/db/services/knowledgebase_service.py

@@ -20,7 +20,7 @@ from peewee import fn, JOIN
 from api.db import StatusEnum, TenantPermission
 from api.db.db_models import DB, Document, Knowledgebase, User, UserTenant, UserCanvas
 from api.db.services.common_service import CommonService
-from api.utils import current_timestamp, datetime_format
+from common.time_utils import current_timestamp, datetime_format
 
 
 class KnowledgebaseService(CommonService):

api/db/services/langfuse_service.py

@@ -20,7 +20,7 @@ import peewee
 
 from api.db.db_models import DB, TenantLangfuse
 from api.db.services.common_service import CommonService
-from api.utils import current_timestamp, datetime_format
+from common.time_utils import current_timestamp, datetime_format
 
 
 class TenantLangfuseService(CommonService):

api/db/services/llm_service.py

@@ -16,6 +16,7 @@
 import inspect
 import logging
 import re
+from rag.utils import num_tokens_from_string
 from functools import partial
 from typing import Generator
 from api.db.db_models import LLM
@@ -59,21 +60,6 @@ def get_init_tenant_llm(user_id):
                 }
             )
 
-    if settings.LIGHTEN != 1:
-        for buildin_embedding_model in settings.BUILTIN_EMBEDDING_MODELS:
-            mdlnm, fid = TenantLLMService.split_model_name_and_factory(buildin_embedding_model)
-            tenant_llm.append(
-                {
-                    "tenant_id": user_id,
-                    "llm_factory": fid,
-                    "llm_name": mdlnm,
-                    "model_type": "embedding",
-                    "api_key": "",
-                    "api_base": "",
-                    "max_tokens": 1024 if buildin_embedding_model == "BAAI/bge-large-zh-v1.5@BAAI" else 512,
-                }
-            )
-
     unique = {}
     for item in tenant_llm:
         key = (item["tenant_id"], item["llm_factory"], item["llm_name"])
@@ -94,9 +80,19 @@ class LLMBundle(LLM4Tenant):
 
     def encode(self, texts: list):
         if self.langfuse:
-            generation = self.langfuse.start_generation(trace_context=self.trace_context, name="encode", model=self.llm_name, input={"texts": texts})
+            generation = self.langfuse.start_generation(trace_context=self.trace_context, name="encode", model=self.llm_name, input={"texts": texts})        
+    
+        safe_texts = []
+        for text in texts:
+            token_size = num_tokens_from_string(text)
+            if token_size > self.max_length:
+                target_len = int(self.max_length * 0.95)
+                safe_texts.append(text[:target_len])
+            else:
+                safe_texts.append(text)
+                
+        embeddings, used_tokens = self.mdl.encode(safe_texts)
 
-        embeddings, used_tokens = self.mdl.encode(texts)
         llm_name = getattr(self, "llm_name", None)
         if not TenantLLMService.increase_usage(self.tenant_id, self.llm_type, used_tokens, llm_name):
             logging.error("LLMBundle.encode can't update token usage for {}/EMBEDDING used_tokens: {}".format(self.tenant_id, used_tokens))
@@ -205,32 +201,31 @@ class LLMBundle(LLM4Tenant):
             return txt
 
         return txt[last_think_end + len("</think>") :]
-    
+
     @staticmethod
     def _clean_param(chat_partial, **kwargs):
         func = chat_partial.func
         sig = inspect.signature(func)
-        keyword_args = []
         support_var_args = False
-        for param in sig.parameters.values():
-            if param.kind == inspect.Parameter.VAR_KEYWORD or param.kind == inspect.Parameter.VAR_POSITIONAL:
-                support_var_args = True
-            elif param.kind == inspect.Parameter.KEYWORD_ONLY:
-                keyword_args.append(param.name)
+        allowed_params = set()
 
-        use_kwargs = kwargs
-        if not support_var_args:
-            use_kwargs = {k: v for k, v in kwargs.items() if k in keyword_args}
-        return use_kwargs
-        
+        for param in sig.parameters.values():
+            if param.kind == inspect.Parameter.VAR_KEYWORD:
+                support_var_args = True
+            elif param.kind in (inspect.Parameter.POSITIONAL_OR_KEYWORD, inspect.Parameter.KEYWORD_ONLY):
+                allowed_params.add(param.name)
+        if support_var_args:
+            return kwargs
+        else:
+            return {k: v for k, v in kwargs.items() if k in allowed_params}
     def chat(self, system: str, history: list, gen_conf: dict = {}, **kwargs) -> str:
         if self.langfuse:
             generation = self.langfuse.start_generation(trace_context=self.trace_context, name="chat", model=self.llm_name, input={"system": system, "history": history})
 
-        chat_partial = partial(self.mdl.chat, system, history, gen_conf)
+        chat_partial = partial(self.mdl.chat, system, history, gen_conf, **kwargs)
         if self.is_tools and self.mdl.is_tools:
-            chat_partial = partial(self.mdl.chat_with_tools, system, history, gen_conf)
-            
+            chat_partial = partial(self.mdl.chat_with_tools, system, history, gen_conf, **kwargs)
+
         use_kwargs = self._clean_param(chat_partial, **kwargs)
         txt, used_tokens = chat_partial(**use_kwargs)
         txt = self._remove_reasoning_content(txt)
@@ -266,7 +261,7 @@ class LLMBundle(LLM4Tenant):
                 break
 
             if txt.endswith("</think>"):
-                ans = ans.rstrip("</think>")
+                ans = ans[: -len("</think>")]
 
             if not self.verbose_tool_use:
                 txt = re.sub(r"<tool_call>.*?</tool_call>", "", txt, flags=re.DOTALL)

api/db/services/pipeline_operation_log_service.py

@@ -27,7 +27,8 @@ from api.db.services.common_service import CommonService
 from api.db.services.document_service import DocumentService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.task_service import GRAPH_RAPTOR_FAKE_DOC_ID
-from api.utils import current_timestamp, datetime_format, get_uuid
+from api.utils import get_uuid
+from common.time_utils import current_timestamp, datetime_format
 
 
 class PipelineOperationLogService(CommonService):

api/db/services/search_service.py

@@ -20,7 +20,7 @@ from peewee import fn
 from api.db import StatusEnum
 from api.db.db_models import DB, Search, User
 from api.db.services.common_service import CommonService
-from api.utils import current_timestamp, datetime_format
+from common.time_utils import current_timestamp, datetime_format
 
 
 class SearchService(CommonService):

api/db/services/task_service.py

@@ -27,7 +27,8 @@ from api.db import StatusEnum, FileType, TaskStatus
 from api.db.db_models import Task, Document, Knowledgebase, Tenant
 from api.db.services.common_service import CommonService
 from api.db.services.document_service import DocumentService
-from api.utils import current_timestamp, get_uuid
+from api.utils import get_uuid
+from common.time_utils import current_timestamp
 from deepdoc.parser.excel_parser import RAGFlowExcelParser
 from rag.settings import get_svr_queue_name
 from rag.utils.storage_factory import STORAGE_IMPL
@@ -351,7 +352,7 @@ def queue_tasks(doc: dict, bucket: str, name: str, priority: int):
             "progress": 0.0,
             "from_page": 0,
             "to_page": 100000000,
-            "begin_at": datetime.now(),
+            "begin_at": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
         }
 
     parse_task_array = []
@@ -503,7 +504,7 @@ def queue_dataflow(tenant_id:str, flow_id:str, task_id:str, doc_id:str=CANVAS_DE
         to_page=100000000,
         task_type="dataflow" if not rerun else "dataflow_rerun",
         priority=priority,
-        begin_at=datetime.now(),
+        begin_at= datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
     )
     if doc_id not in [CANVAS_DEBUG_DOC_ID, GRAPH_RAPTOR_FAKE_DOC_ID]:
         TaskService.model.delete().where(TaskService.model.doc_id == doc_id).execute()

api/db/services/tenant_llm_service.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import os
 import logging
 from langfuse import Langfuse
 from api import settings
@@ -112,24 +113,17 @@ class TenantLLMService(CommonService):
             model_config = cls.get_api_key(tenant_id, mdlnm)
         if model_config:
             model_config = model_config.to_dict()
-            llm = LLMService.query(llm_name=mdlnm) if not fid else LLMService.query(llm_name=mdlnm, fid=fid)
-            if not llm and fid:  # for some cases seems fid mismatch
-                llm = LLMService.query(llm_name=mdlnm)
-            if llm:
-                model_config["is_tools"] = llm[0].is_tools
-        if not model_config:
-            if llm_type in [LLMType.EMBEDDING, LLMType.RERANK]:
-                llm = LLMService.query(llm_name=mdlnm) if not fid else LLMService.query(llm_name=mdlnm, fid=fid)
-                if llm and llm[0].fid in ["Youdao", "FastEmbed", "BAAI"]:
-                    model_config = {"llm_factory": llm[0].fid, "api_key": "", "llm_name": mdlnm, "api_base": ""}
-            if not model_config:
-                if mdlnm == "flag-embedding":
-                    model_config = {"llm_factory": "Tongyi-Qianwen", "api_key": "", "llm_name": llm_name,
-                                    "api_base": ""}
-                else:
-                    if not mdlnm:
-                        raise LookupError(f"Type of {llm_type} model is not set.")
-                    raise LookupError("Model({}) not authorized".format(mdlnm))
+        elif llm_type == LLMType.EMBEDDING and fid == 'Builtin' and "tei-" in os.getenv("COMPOSE_PROFILES", "") and mdlnm == os.getenv('TEI_MODEL', ''):
+            embedding_cfg = settings.EMBEDDING_CFG
+            model_config = {"llm_factory": 'Builtin', "api_key": embedding_cfg["api_key"], "llm_name": mdlnm, "api_base": embedding_cfg["base_url"]}
+        else:
+            raise LookupError(f"Model({mdlnm}@{fid}) not authorized")
+
+        llm = LLMService.query(llm_name=mdlnm) if not fid else LLMService.query(llm_name=mdlnm, fid=fid)
+        if not llm and fid:  # for some cases seems fid mismatch
+            llm = LLMService.query(llm_name=mdlnm)
+        if llm:
+            model_config["is_tools"] = llm[0].is_tools
         return model_config
 
     @classmethod

api/db/services/user_service.py

@@ -24,7 +24,8 @@ from api.db import UserTenantRole
 from api.db.db_models import DB, UserTenant
 from api.db.db_models import User, Tenant
 from api.db.services.common_service import CommonService
-from api.utils import get_uuid, current_timestamp, datetime_format
+from api.utils import get_uuid
+from common.time_utils import current_timestamp, datetime_format
 from api.db import StatusEnum
 from rag.settings import MINIO
 

api/settings.py

@@ -28,8 +28,6 @@ from api.utils.configs import decrypt_database_config, get_base_config
 from api.utils.file_utils import get_project_base_directory
 from rag.nlp import search
 
-LIGHTEN = int(os.environ.get("LIGHTEN", "0"))
-
 LLM = None
 LLM_FACTORY = None
 LLM_BASE_URL = None
@@ -77,8 +75,6 @@ SANDBOX_ENABLED = 0
 SANDBOX_HOST = None
 STRONG_TEST_COUNT = int(os.environ.get("STRONG_TEST_COUNT", "8"))
 
-BUILTIN_EMBEDDING_MODELS = ["BAAI/bge-large-zh-v1.5@BAAI", "maidalun1020/bce-embedding-base_v1@Youdao"]
-
 SMTP_CONF = None
 MAIL_SERVER = ""
 MAIL_PORT = 000
@@ -109,8 +105,7 @@ def get_or_create_secret_key():
 
 
 def init_settings():
-    global LLM, LLM_FACTORY, LLM_BASE_URL, LIGHTEN, DATABASE_TYPE, DATABASE, FACTORY_LLM_INFOS, REGISTER_ENABLED
-    LIGHTEN = int(os.environ.get("LIGHTEN", "0"))
+    global LLM, LLM_FACTORY, LLM_BASE_URL, DATABASE_TYPE, DATABASE, FACTORY_LLM_INFOS, REGISTER_ENABLED
     DATABASE_TYPE = os.getenv("DB_TYPE", "mysql")
     DATABASE = decrypt_database_config(name=DATABASE_TYPE)
     LLM = get_base_config("user_default_llm", {}) or {}
@@ -130,8 +125,6 @@ def init_settings():
 
     global CHAT_MDL, EMBEDDING_MDL, RERANK_MDL, ASR_MDL, IMAGE2TEXT_MDL
     global CHAT_CFG, EMBEDDING_CFG, RERANK_CFG, ASR_CFG, IMAGE2TEXT_CFG
-    if not LIGHTEN:
-        EMBEDDING_MDL = BUILTIN_EMBEDDING_MODELS[0]
 
     global API_KEY, PARSERS, HOST_IP, HOST_PORT, SECRET_KEY
     API_KEY = LLM.get("api_key")
@@ -152,7 +145,7 @@ def init_settings():
     IMAGE2TEXT_CFG = _resolve_per_model_config(image2text_entry, LLM_FACTORY, API_KEY, LLM_BASE_URL)
 
     CHAT_MDL = CHAT_CFG.get("model", "") or ""
-    EMBEDDING_MDL = EMBEDDING_CFG.get("model", "") or ""
+    EMBEDDING_MDL = os.getenv("TEI_MODEL", "BAAI/bge-small-en-v1.5") if "tei-" in os.getenv("COMPOSE_PROFILES", "") else ""
     RERANK_MDL = RERANK_CFG.get("model", "") or ""
     ASR_MDL = ASR_CFG.get("model", "") or ""
     IMAGE2TEXT_MDL = IMAGE2TEXT_CFG.get("model", "") or ""

api/utils/__init__.py

@@ -14,11 +14,9 @@
 #  limitations under the License.
 #
 import base64
-import datetime
 import hashlib
 import os
 import socket
-import time
 import uuid
 import requests
 
@@ -26,26 +24,6 @@ import importlib
 
 from .common import string_to_bytes
 
-
-def current_timestamp():
-    return int(time.time() * 1000)
-
-
-def timestamp_to_date(timestamp, format_string="%Y-%m-%d %H:%M:%S"):
-    if not timestamp:
-        timestamp = time.time()
-    timestamp = int(timestamp) / 1000
-    time_array = time.localtime(timestamp)
-    str_date = time.strftime(format_string, time_array)
-    return str_date
-
-
-def date_string_to_timestamp(time_str, format_string="%Y-%m-%d %H:%M:%S"):
-    time_array = time.strptime(time_str, format_string)
-    time_stamp = int(time.mktime(time_array) * 1000)
-    return time_stamp
-
-
 def get_lan_ip():
     if os.name != "nt":
         import fcntl
@@ -94,26 +72,6 @@ def get_uuid():
     return uuid.uuid1().hex
 
 
-def datetime_format(date_time: datetime.datetime) -> datetime.datetime:
-    return datetime.datetime(date_time.year, date_time.month, date_time.day,
-                             date_time.hour, date_time.minute, date_time.second)
-
-
-def get_format_time() -> datetime.datetime:
-    return datetime_format(datetime.datetime.now())
-
-
-def str2date(date_time: str):
-    return datetime.datetime.strptime(date_time, '%Y-%m-%d')
-
-
-def elapsed2time(elapsed):
-    seconds = elapsed / 1000
-    minuter, second = divmod(seconds, 60)
-    hour, minuter = divmod(minuter, 60)
-    return '%02d:%02d:%02d' % (hour, minuter, second)
-
-
 def download_img(url):
     if not url:
         return ""
@@ -123,10 +81,5 @@ def download_img(url):
         "base64," + base64.b64encode(response.content).decode("utf-8")
 
 
-def delta_seconds(date_string: str):
-    dt = datetime.datetime.strptime(date_string, "%Y-%m-%d %H:%M:%S")
-    return (datetime.datetime.now() - dt).total_seconds()
-
-
 def hash_str2int(line: str, mod: int = 10 ** 8) -> int:
     return int(hashlib.sha1(line.encode("utf-8")).hexdigest(), 16) % mod

api/utils/api_utils.py

@@ -43,7 +43,6 @@ from flask_login import current_user
 from flask import (
     request as flask_request,
 )
-from itsdangerous import URLSafeTimedSerializer
 from peewee import OperationalError
 from werkzeug.http import HTTP_STATUS_CODES
 
@@ -51,8 +50,7 @@ from api import settings
 from api.constants import REQUEST_MAX_WAIT_SEC, REQUEST_WAIT_SEC
 from api.db import ActiveEnum
 from api.db.db_models import APIToken
-from api.utils.json import CustomJSONEncoder, json_dumps
-from api.utils import get_uuid
+from api.utils.json_encode import CustomJSONEncoder, json_dumps
 from rag.utils.mcp_tool_call_conn import MCPToolCallSession, close_multiple_mcp_toolcall_sessions
 
 requests.models.complexjson.dumps = functools.partial(json.dumps, cls=CustomJSONEncoder)
@@ -410,9 +408,9 @@ def get_error_operating_result(message="Operating error"):
     return get_result(code=settings.RetCode.OPERATING_ERROR, message=message)
 
 
-def generate_confirmation_token(tenant_id):
-    serializer = URLSafeTimedSerializer(tenant_id)
-    return "ragflow-" + serializer.dumps(get_uuid(), salt=tenant_id)[2:34]
+def generate_confirmation_token():
+    import secrets
+    return "ragflow-" + secrets.token_urlsafe(32)
 
 
 def get_parser_config(chunk_method, parser_config):
@@ -588,7 +586,7 @@ def verify_embedding_availability(embd_id: str, tenant_id: str) -> tuple[bool, R
             llm["llm_name"] == llm_name and llm["llm_factory"] == llm_factory and llm["model_type"] == "embedding" for
             llm in tenant_llms)
 
-        is_builtin_model = embd_id in settings.BUILTIN_EMBEDDING_MODELS
+        is_builtin_model = llm_factory=='Builtin'
         if not (is_builtin_model or is_tenant_model or in_llm_service):
             return False, get_error_argument_result(f"Unsupported model: <{embd_id}>")
 

api/utils/common.py

@@ -14,6 +14,12 @@
 #  limitations under the License.
 #
 
+import threading
+import subprocess
+import sys
+import os
+import logging
+
 def string_to_bytes(string):
     return string if isinstance(
         string, bytes) else string.encode(encoding="utf-8")
@@ -44,3 +50,48 @@ def convert_bytes(size_in_bytes: int) -> str:
         return f"{size:.1f} {units[i]}"
     else:
         return f"{size:.2f} {units[i]}"
+
+
+def once(func):
+    """
+    A thread-safe decorator that ensures the decorated function runs exactly once,
+    caching and returning its result for all subsequent calls. This prevents
+    race conditions in multi-threaded environments by using a lock to protect
+    the execution state.
+
+    Args:
+        func (callable): The function to be executed only once.
+
+    Returns:
+        callable: A wrapper function that executes `func` on the first call
+                  and returns the cached result thereafter.
+
+    Example:
+        @once
+        def compute_expensive_value():
+            print("Computing...")
+            return 42
+
+        # First call: executes and prints
+        # Subsequent calls: return 42 without executing
+    """
+    executed = False
+    result = None
+    lock = threading.Lock()
+    def wrapper(*args, **kwargs):
+        nonlocal executed, result
+        with lock:
+            if not executed:
+                executed = True
+                result = func(*args, **kwargs)
+        return result
+    return wrapper
+
+@once
+def pip_install_torch():
+    device = os.getenv("DEVICE", "cpu")
+    if device=="cpu":
+        return
+    logging.info("Installing pytorch")
+    pkg_names = ["torch>=2.5.0,<3.0.0"]
+    subprocess.check_call([sys.executable, "-m", "pip", "install", *pkg_names])

api/utils/email_templates.py

@@ -0,0 +1,25 @@
+"""
+Reusable HTML email templates and registry.
+"""
+
+# Invitation email template
+INVITE_EMAIL_TMPL = """
+<p>Hi {{email}},</p>
+<p>{{inviter}} has invited you to join their team (ID: {{tenant_id}}).</p>
+<p>Click the link below to complete your registration:<br>
+<a href="{{invite_url}}">{{invite_url}}</a></p>
+<p>If you did not request this, please ignore this email.</p>
+"""
+
+# Password reset code template
+RESET_CODE_EMAIL_TMPL = """
+<p>Hello,</p>
+<p>Your password reset code is: <b>{{ code }}</b></p>
+<p>This code will expire in {{ ttl_min }} minutes.</p>
+"""
+
+# Template registry
+EMAIL_TEMPLATES = {
+    "invite": INVITE_EMAIL_TMPL,
+    "reset_code": RESET_CODE_EMAIL_TMPL,
+}

api/utils/file_utils.py

@@ -13,7 +13,12 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+
+
+# Standard library imports
 import base64
+import hashlib
+import io
 import json
 import os
 import re
@@ -22,13 +27,20 @@ import subprocess
 import sys
 import tempfile
 import threading
+import zipfile
 from io import BytesIO
 
+# Typing
+from typing import List, Union, Tuple
+
+# Third-party imports
+import olefile
 import pdfplumber
 from cachetools import LRUCache, cached
 from PIL import Image
 from ruamel.yaml import YAML
 
+# Local imports
 from api.constants import IMG_BASE64_PREFIX
 from api.db import FileType
 
@@ -161,7 +173,7 @@ def filename_type(filename):
     if re.match(r".*\.(wav|flac|ape|alac|wavpack|wv|mp3|aac|ogg|vorbis|opus)$", filename):
         return FileType.AURAL.value
 
-    if re.match(r".*\.(jpg|jpeg|png|tif|gif|pcx|tga|exif|fpx|svg|psd|cdr|pcd|dxf|ufo|eps|ai|raw|WMF|webp|avif|apng|icon|ico|mpg|mpeg|avi|rm|rmvb|mov|wmv|asf|dat|asx|wvx|mpe|mpa|mp4)$", filename):
+    if re.match(r".*\.(jpg|jpeg|png|tif|gif|pcx|tga|exif|fpx|svg|psd|cdr|pcd|dxf|ufo|eps|ai|raw|WMF|webp|avif|apng|icon|ico|mpg|mpeg|avi|rm|rmvb|mov|wmv|asf|dat|asx|wvx|mpe|mpa|mp4|avi|mkv)$", filename):
         return FileType.VISUAL.value
 
     return FileType.OTHER.value
@@ -284,3 +296,125 @@ def read_potential_broken_pdf(blob):
         return repaired
 
     return blob
+
+
+
+def _is_zip(h: bytes) -> bool:
+    return h.startswith(b"PK\x03\x04") or h.startswith(b"PK\x05\x06") or h.startswith(b"PK\x07\x08")
+
+def _is_pdf(h: bytes) -> bool:
+    return h.startswith(b"%PDF-")
+
+def _is_ole(h: bytes) -> bool:
+    return h.startswith(b"\xD0\xCF\x11\xE0\xA1\xB1\x1A\xE1")
+
+def _sha10(b: bytes) -> str:
+    return hashlib.sha256(b).hexdigest()[:10]
+
+def _guess_ext(b: bytes) -> str:
+    h = b[:8]
+    if _is_zip(h):
+        try:
+            with zipfile.ZipFile(io.BytesIO(b), "r") as z:
+                names = [n.lower() for n in z.namelist()]
+                if any(n.startswith("word/") for n in names):
+                    return ".docx"
+                if any(n.startswith("ppt/") for n in names):
+                    return ".pptx"
+                if any(n.startswith("xl/") for n in names):
+                    return ".xlsx"
+        except Exception:
+            pass
+        return ".zip"
+    if _is_pdf(h):
+        return ".pdf"
+    if _is_ole(h):
+        return ".doc"
+    return ".bin"
+
+# Try to extract the real embedded payload from OLE's Ole10Native
+def _extract_ole10native_payload(data: bytes) -> bytes:
+    try:
+        pos = 0
+        if len(data) < 4:
+            return data
+        _ = int.from_bytes(data[pos:pos+4], "little")
+        pos += 4
+        # filename/src/tmp (NUL-terminated ANSI)
+        for _ in range(3):
+            z = data.index(b"\x00", pos)
+            pos = z + 1
+        # skip unknown 4 bytes
+        pos += 4
+        if pos + 4 > len(data):
+            return data
+        size = int.from_bytes(data[pos:pos+4], "little")
+        pos += 4
+        if pos + size <= len(data):
+            return data[pos:pos+size]
+    except Exception:
+        pass
+    return data
+
+def extract_embed_file(target: Union[bytes, bytearray]) -> List[Tuple[str, bytes]]:
+    """
+    Only extract the 'first layer' of embedding, returning raw (filename, bytes).
+    """
+    top = bytes(target)
+    head = top[:8]
+    out: List[Tuple[str, bytes]] = []
+    seen = set()
+
+    def push(b: bytes, name_hint: str = ""):
+        h10 = _sha10(b)
+        if h10 in seen:
+            return
+        seen.add(h10)
+        ext = _guess_ext(b)
+        # If name_hint has an extension use its basename; else fallback to guessed ext
+        if "." in name_hint:
+            fname = name_hint.split("/")[-1]
+        else:
+            fname = f"{h10}{ext}"
+        out.append((fname, b))
+
+    # OOXML/ZIP container (docx/xlsx/pptx)
+    if _is_zip(head):
+        try:
+            with zipfile.ZipFile(io.BytesIO(top), "r") as z:
+                embed_dirs = (
+                    "word/embeddings/", "word/objects/", "word/activex/",
+                    "xl/embeddings/", "ppt/embeddings/"
+                )
+                for name in z.namelist():
+                    low = name.lower()
+                    if any(low.startswith(d) for d in embed_dirs):
+                        try:
+                            b = z.read(name)
+                            push(b, name)
+                        except Exception:
+                            pass
+        except Exception:
+            pass
+        return out
+
+    # OLE container (doc/ppt/xls)
+    if _is_ole(head):
+        try:
+            with olefile.OleFileIO(io.BytesIO(top)) as ole:
+                for entry in ole.listdir():
+                    p = "/".join(entry)
+                    try:
+                        data = ole.openstream(entry).read()
+                    except Exception:
+                        continue
+                    if not data:
+                        continue
+                    if "Ole10Native" in p or "ole10native" in p.lower():
+                        data = _extract_ole10native_payload(data)
+                    push(data, p)
+        except Exception:
+            pass
+        return out
+
+    return out

api/utils/health_utils.py

@@ -74,12 +74,12 @@ def get_es_cluster_stats() -> dict:
         raise Exception("Elasticsearch is not in use.")
     try:
         return {
-            "alive": True,
+            "status": "alive",
             "message": ESConnection().get_cluster_stats()
         }
     except Exception as e:
         return {
-            "alive": False,
+            "status": "timeout",
             "message": f"error: {str(e)}",
         }
 
@@ -90,12 +90,12 @@ def get_infinity_status():
         raise Exception("Infinity is not in use.")
     try:
         return {
-            "alive": True,
+            "status": "alive",
             "message": InfinityConnection().health()
         }
     except Exception as e:
         return {
-            "alive": False,
+            "status": "timeout",
             "message": f"error: {str(e)}",
         }
 
@@ -107,12 +107,12 @@ def get_mysql_status():
         headers = ['id', 'user', 'host', 'db', 'command', 'time', 'state', 'info']
         cursor.close()
         return {
-            "alive": True,
+            "status": "alive",
             "message": [dict(zip(headers, r)) for r in res_rows]
         }
     except Exception as e:
         return {
-            "alive": False,
+            "status": "timeout",
             "message": f"error: {str(e)}",
         }
 
@@ -122,12 +122,12 @@ def check_minio_alive():
     try:
         response = requests.get(f'http://{rag_settings.MINIO["host"]}/minio/health/live')
         if response.status_code == 200:
-            return {'alive': True, "message": f"Confirm elapsed: {(timer() - start_time) * 1000.0:.1f} ms."}
+            return {"status": "alive", "message": f"Confirm elapsed: {(timer() - start_time) * 1000.0:.1f} ms."}
         else:
-            return {'alive': False, "message": f"Confirm elapsed: {(timer() - start_time) * 1000.0:.1f} ms."}
+            return {"status": "timeout", "message": f"Confirm elapsed: {(timer() - start_time) * 1000.0:.1f} ms."}
     except Exception as e:
         return {
-            "alive": False,
+            "status": "timeout",
             "message": f"error: {str(e)}",
         }
 
@@ -135,12 +135,12 @@ def check_minio_alive():
 def get_redis_info():
     try:
         return {
-            "alive": True,
+            "status": "alive",
             "message": REDIS_CONN.info()
         }
     except Exception as e:
         return {
-            "alive": False,
+            "status": "timeout",
             "message": f"error: {str(e)}",
         }
 
@@ -150,12 +150,12 @@ def check_ragflow_server_alive():
     try:
         response = requests.get(f'http://{settings.HOST_IP}:{settings.HOST_PORT}/v1/system/ping')
         if response.status_code == 200:
-            return {'alive': True, "message": f"Confirm elapsed: {(timer() - start_time) * 1000.0:.1f} ms."}
+            return {"status": "alive", "message": f"Confirm elapsed: {(timer() - start_time) * 1000.0:.1f} ms."}
         else:
-            return {'alive': False, "message": f"Confirm elapsed: {(timer() - start_time) * 1000.0:.1f} ms."}
+            return {"status": "timeout", "message": f"Confirm elapsed: {(timer() - start_time) * 1000.0:.1f} ms."}
     except Exception as e:
         return {
-            "alive": False,
+            "status": "timeout",
             "message": f"error: {str(e)}",
         }
 
@@ -192,9 +192,7 @@ def run_health_checks() -> tuple[dict, bool]:
     except Exception:
         result["storage"] = "nok"
 
-
-    all_ok = (result.get("db") == "ok") and (result.get("redis") == "ok") and (result.get("doc_engine") == "ok") and (result.get("storage") == "ok")
+    all_ok = (result.get("db") == "ok") and (result.get("redis") == "ok") and (result.get("doc_engine") == "ok") and (
+                result.get("storage") == "ok")
     result["status"] = "ok" if all_ok else "nok"
     return result, all_ok
-
-

api/utils/web_utils.py

@@ -24,6 +24,7 @@ from urllib.parse import urlparse
 from api.apps import smtp_mail_server
 from flask_mail import Message
 from flask import render_template_string
+from api.utils.email_templates import EMAIL_TEMPLATES
 from selenium import webdriver
 from selenium.common.exceptions import TimeoutException
 from selenium.webdriver.chrome.options import Options
@@ -34,6 +35,12 @@ from selenium.webdriver.support.ui import WebDriverWait
 from webdriver_manager.chrome import ChromeDriverManager
 
 
+OTP_LENGTH = 8
+OTP_TTL_SECONDS = 5 * 60
+ATTEMPT_LIMIT = 5
+ATTEMPT_LOCK_SECONDS = 30 * 60
+RESEND_COOLDOWN_SECONDS = 60
+
 
 CONTENT_TYPE_MAP = {
     # Office
@@ -178,24 +185,49 @@ def get_float(req: dict, key: str, default: float | int = 10.0) -> float:
         return default
 
 
-INVITE_EMAIL_TMPL = """
-<p>Hi {{email}},</p>
-<p>{{inviter}} has invited you to join their team (ID: {{tenant_id}}).</p>
-<p>Click the link below to complete your registration:<br>
-<a href="{{invite_url}}">{{invite_url}}</a></p>
-<p>If you did not request this, please ignore this email.</p>
-"""
+def send_email_html(subject: str, to_email: str, template_key: str, **context):
+    """Generic HTML email sender using shared templates.
+    template_key must exist in EMAIL_TEMPLATES.
+    """
+    from api.apps import app
+    tmpl = EMAIL_TEMPLATES.get(template_key)
+    if not tmpl:
+        raise ValueError(f"Unknown email template: {template_key}")
+    with app.app_context():
+        msg = Message(subject=subject, recipients=[to_email])
+        msg.html = render_template_string(tmpl, **context)
+        smtp_mail_server.send(msg)
+
 
 def send_invite_email(to_email, invite_url, tenant_id, inviter):
-    from api.apps import  app
-    with app.app_context():
-        msg = Message(subject="RAGFlow Invitation",
-                      recipients=[to_email])
-        msg.html = render_template_string(
-            INVITE_EMAIL_TMPL,
-            email=to_email,
-            invite_url=invite_url,
-            tenant_id=tenant_id,
-            inviter=inviter,
-        )
-        smtp_mail_server.send(msg)
+    # Reuse the generic HTML sender with 'invite' template
+    send_email_html(
+        subject="RAGFlow Invitation",
+        to_email=to_email,
+        template_key="invite",
+        email=to_email,
+        invite_url=invite_url,
+        tenant_id=tenant_id,
+        inviter=inviter,
+    )
+
+
+def otp_keys(email: str):
+    email = (email or "").strip().lower()
+    return (
+        f"otp:{email}",
+        f"otp_attempts:{email}",
+        f"otp_last_sent:{email}",
+        f"otp_lock:{email}",
+    )
+
+
+def hash_code(code: str, salt: bytes) -> str:
+    import hashlib
+    import hmac 
+    return hmac.new(salt, (code or "").encode("utf-8"), hashlib.sha256).hexdigest()
+    
+
+def captcha_key(email: str) -> str:
+    return f"captcha:{email}"
+    

api/versions.py

@@ -34,8 +34,6 @@ def get_ragflow_version() -> str:
             RAGFLOW_VERSION_INFO = f.read().strip()
     else:
         RAGFLOW_VERSION_INFO = get_closest_tag_and_count()
-        LIGHTEN = int(os.environ.get("LIGHTEN", "0"))
-        RAGFLOW_VERSION_INFO += " slim" if LIGHTEN == 1 else " full"
     return RAGFLOW_VERSION_INFO
 
 

common/__init__.py

@@ -0,0 +1,15 @@
+#
+#  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.
+#

common/float_utils.py

@@ -0,0 +1,46 @@
+#
+#  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.
+#
+
+def get_float(v):
+    """
+    Convert a value to float, handling None and exceptions gracefully.
+
+    Attempts to convert the input value to a float. If the value is None or
+    cannot be converted to float, returns negative infinity as a default value.
+
+    Args:
+        v: The value to convert to float. Can be any type that float() accepts,
+           or None.
+
+    Returns:
+        float: The converted float value if successful, otherwise float('-inf').
+
+    Examples:
+        >>> get_float("3.14")
+        3.14
+        >>> get_float(None)
+        -inf
+        >>> get_float("invalid")
+        -inf
+        >>> get_float(42)
+        42.0
+    """
+    if v is None:
+        return float('-inf')
+    try:
+        return float(v)
+    except Exception:
+        return float('-inf')

common/string_utils.py

@@ -0,0 +1,73 @@
+#
+#  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 re
+
+
+def remove_redundant_spaces(txt: str):
+    """
+    Remove redundant spaces around punctuation marks while preserving meaningful spaces.
+
+    This function performs two main operations:
+    1. Remove spaces after left-boundary characters (opening brackets, etc.)
+    2. Remove spaces before right-boundary characters (closing brackets, punctuation, etc.)
+
+    Args:
+        txt (str): Input text to process
+
+    Returns:
+        str: Text with redundant spaces removed
+    """
+    # First pass: Remove spaces after left-boundary characters
+    # Matches: [non-alphanumeric-and-specific-right-punctuation] + [non-space]
+    # Removes spaces after characters like '(', '<', and other non-alphanumeric chars
+    # Examples:
+    #   "( test" → "(test"
+    txt = re.sub(r"([^a-z0-9.,\)>]) +([^ ])", r"\1\2", txt, flags=re.IGNORECASE)
+
+    # Second pass: Remove spaces before right-boundary characters
+    # Matches: [non-space] + [non-alphanumeric-and-specific-left-punctuation]
+    # Removes spaces before characters like non-')', non-',', non-'.', and non-alphanumeric chars
+    # Examples:
+    #   "world !" → "world!"
+    return re.sub(r"([^ ]) +([^a-z0-9.,\(<])", r"\1\2", txt, flags=re.IGNORECASE)
+
+
+def clean_markdown_block(text):
+    """
+    Remove Markdown code block syntax from the beginning and end of text.
+
+    This function cleans Markdown code blocks by removing:
+    - Opening ```Markdown tags (with optional whitespace and newlines)
+    - Closing ``` tags (with optional whitespace and newlines)
+
+    Args:
+        text (str): Input text that may be wrapped in Markdown code blocks
+
+    Returns:
+        str: Cleaned text with Markdown code block syntax removed, and stripped of surrounding whitespace
+
+    """
+    # Remove opening ```markdown tag with optional whitespace and newlines
+    # Matches: optional whitespace + ```markdown + optional whitespace + optional newline
+    text = re.sub(r'^\s*```markdown\s*\n?', '', text)
+
+    # Remove closing ``` tag with optional whitespace and newlines
+    # Matches: optional newline + optional whitespace + ``` + optional whitespace at end
+    text = re.sub(r'\n?\s*```\s*$', '', text)
+
+    # Return text with surrounding whitespace removed
+    return text.strip()

common/time_utils.py

@@ -0,0 +1,126 @@
+#
+#  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 datetime
+import time
+
+def current_timestamp():
+    """
+    Get the current timestamp in milliseconds.
+
+    Returns:
+        int: Current Unix timestamp in milliseconds (13 digits)
+
+    Example:
+        >>> current_timestamp()
+        1704067200000
+    """
+    return int(time.time() * 1000)
+
+
+def timestamp_to_date(timestamp, format_string="%Y-%m-%d %H:%M:%S"):
+    """
+    Convert a timestamp to formatted date string.
+
+    Args:
+        timestamp: Unix timestamp in milliseconds. If None or empty, uses current time.
+        format_string: Format string for the output date (default: "%Y-%m-%d %H:%M:%S")
+
+    Returns:
+        str: Formatted date string
+
+    Example:
+        >>> timestamp_to_date(1704067200000)
+        '2024-01-01 08:00:00'
+    """
+    if not timestamp:
+        timestamp = time.time()
+    timestamp = int(timestamp) / 1000
+    time_array = time.localtime(timestamp)
+    str_date = time.strftime(format_string, time_array)
+    return str_date
+
+
+def date_string_to_timestamp(time_str, format_string="%Y-%m-%d %H:%M:%S"):
+    """
+    Convert a date string to timestamp in milliseconds.
+
+    Args:
+        time_str: Date string to convert
+        format_string: Format of the input date string (default: "%Y-%m-%d %H:%M:%S")
+
+    Returns:
+        int: Unix timestamp in milliseconds
+
+    Example:
+        >>> date_string_to_timestamp("2024-01-01 00:00:00")
+        1704067200000
+    """
+    time_array = time.strptime(time_str, format_string)
+    time_stamp = int(time.mktime(time_array) * 1000)
+    return time_stamp
+
+def datetime_format(date_time: datetime.datetime) -> datetime.datetime:
+    """
+    Normalize a datetime object by removing microsecond component.
+
+    Creates a new datetime object with only year, month, day, hour, minute, second.
+    Microseconds are set to 0.
+
+    Args:
+        date_time: datetime object to normalize
+
+    Returns:
+        datetime.datetime: New datetime object without microseconds
+
+    Example:
+        >>> dt = datetime.datetime(2024, 1, 1, 12, 30, 45, 123456)
+        >>> datetime_format(dt)
+        datetime.datetime(2024, 1, 1, 12, 30, 45)
+    """
+    return datetime.datetime(date_time.year, date_time.month, date_time.day,
+                             date_time.hour, date_time.minute, date_time.second)
+
+
+def get_format_time() -> datetime.datetime:
+    """
+    Get current datetime normalized without microseconds.
+
+    Returns:
+        datetime.datetime: Current datetime with microseconds set to 0
+
+    Example:
+        >>> get_format_time()
+        datetime.datetime(2024, 1, 1, 12, 30, 45)
+    """
+    return datetime_format(datetime.datetime.now())
+
+
+def delta_seconds(date_string: str):
+    """
+    Calculate seconds elapsed from a given date string to now.
+
+    Args:
+        date_string: Date string in "YYYY-MM-DD HH:MM:SS" format
+
+    Returns:
+        float: Number of seconds between the given date and current time
+
+    Example:
+        >>> delta_seconds("2024-01-01 12:00:00")
+        3600.0  # If current time is 2024-01-01 13:00:00
+    """
+    dt = datetime.datetime.strptime(date_string, "%Y-%m-%d %H:%M:%S")
+    return (datetime.datetime.now() - dt).total_seconds()

conf/infinity_mapping.json

@@ -31,7 +31,6 @@
 	"entities_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
 	"pagerank_fea": {"type": "integer", "default":  0},
 	"tag_feas": {"type": "varchar", "default": "", "analyzer": "rankfeatures"},
-
 	"from_entity_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
 	"to_entity_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
 	"entity_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
@@ -39,6 +38,6 @@
 	"source_id": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
 	"n_hop_with_weight": {"type": "varchar", "default": ""},
 	"removed_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
-
-	"doc_type_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"}
+	"doc_type_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
+	"toc_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"}
 }

conf/llm_factories.json

@@ -227,8 +227,8 @@
             "llm": [
                 {
                     "llm_name": "qwen3-8b",
-                    "tags": "LLM,CHAT,131k",
-                    "max_tokens": 131000,
+                    "tags": "LLM,CHAT,128k",
+                    "max_tokens": 128000,
                     "model_type": "chat",
                     "is_tools": true
                 },
@@ -241,15 +241,15 @@
                 },
                 {
                     "llm_name": "qwen3-32b",
-                    "tags": "LLM,CHAT,131k",
-                    "max_tokens": 131000,
+                    "tags": "LLM,CHAT,128k",
+                    "max_tokens": 128000,
                     "model_type": "chat",
                     "is_tools": true
                 },
                 {
-                    "llm_name": "kimi-k2-instruct",
-                    "tags": "LLM,CHAT,128K",
-                    "max_tokens": 128000,
+                    "llm_name": "kimi-k2-instruct-0905",
+                    "tags": "LLM,CHAT,256K",
+                    "max_tokens": 256000,
                     "model_type": "chat",
                     "is_tools": true
                 },
@@ -280,6 +280,48 @@
                     "max_tokens": 128000,
                     "model_type": "chat",
                     "is_tools": true
+                },
+                {
+                    "llm_name": "hunyuan-a13b-instruct",
+                    "tags": "LLM,CHAT,256k",
+                    "max_tokens": 256000,
+                    "model_type": "chat",
+                    "is_tools": true
+                },
+                {
+                    "llm_name": "qwen3-next-80b-a3b-instruct",
+                    "tags": "LLM,CHAT,1024k",
+                    "max_tokens": 1024000,
+                    "model_type": "chat",
+                    "is_tools": true
+                },
+                {
+                    "llm_name": "deepseek-v3.2-exp",
+                    "tags": "LLM,CHAT,128k",
+                    "max_tokens": 128000,
+                    "model_type": "chat",
+                    "is_tools": true
+                },
+                {
+                    "llm_name": "deepseek-v3.1-terminus",
+                    "tags": "LLM,CHAT,128k",
+                    "max_tokens": 128000,
+                    "model_type": "chat",
+                    "is_tools": true
+                },
+                {
+                    "llm_name": "qwen3-vl-235b-a22b-instruct",
+                    "tags": "LLM,CHAT,262k",
+                    "max_tokens": 262000,
+                    "model_type": "chat",
+                    "is_tools": true
+                },
+                {
+                    "llm_name": "qwen3-vl-30b-a3b-instruct",
+                    "tags": "LLM,CHAT,262k",
+                    "max_tokens": 262000,
+                    "model_type": "chat",
+                    "is_tools": true
                 }
             ]
         },
@@ -932,20 +974,6 @@
             "status": "1",
             "llm": []
         },
-        {
-            "name": "Youdao",
-            "logo": "",
-            "tags": "TEXT EMBEDDING",
-            "status": "1",
-            "llm": [
-                {
-                    "llm_name": "maidalun1020/bce-embedding-base_v1",
-                    "tags": "TEXT EMBEDDING,",
-                    "max_tokens": 512,
-                    "model_type": "embedding"
-                }
-            ]
-        },
         {
             "name": "DeepSeek",
             "logo": "",
@@ -971,31 +999,9 @@
         {
             "name": "VolcEngine",
             "logo": "",
-            "tags": "LLM, TEXT EMBEDDING",
+            "tags": "LLM, TEXT EMBEDDING, IMAGE2TEXT",
             "status": "1",
-            "llm": [
-                {
-                    "llm_name": "Doubao-pro-128k",
-                    "tags": "LLM,CHAT,128k",
-                    "max_tokens": 131072,
-                    "model_type": "chat",
-                    "is_tools": true
-                },
-                {
-                    "llm_name": "Doubao-pro-32k",
-                    "tags": "LLM,CHAT,32k",
-                    "max_tokens": 32768,
-                    "model_type": "chat",
-                    "is_tools": true
-                },
-                {
-                    "llm_name": "Doubao-pro-4k",
-                    "tags": "LLM,CHAT,4k",
-                    "max_tokens": 4096,
-                    "model_type": "chat",
-                    "is_tools": true
-                }
-            ]
+            "llm": []
         },
         {
             "name": "BaiChuan",
@@ -1121,15 +1127,27 @@
             ]
         },
         {
-            "name": "BAAI",
+            "name": "Builtin",
             "logo": "",
             "tags": "TEXT EMBEDDING",
             "status": "1",
             "llm": [
                 {
-                    "llm_name": "BAAI/bge-large-zh-v1.5",
-                    "tags": "TEXT EMBEDDING,",
-                    "max_tokens": 1024,
+                    "llm_name": "BAAI/bge-small-en-v1.5",
+                    "tags": "TEXT EMBEDDING,512",
+                    "max_tokens": 512,
+                    "model_type": "embedding"
+                },
+                {
+                    "llm_name": "BAAI/bge-m3",
+                    "tags": "TEXT EMBEDDING,8k",
+                    "max_tokens": 8192,
+                    "model_type": "embedding"
+                },
+                {
+                    "llm_name": "Qwen/Qwen3-Embedding-0.6B",
+                    "tags": "TEXT EMBEDDING,32k",
+                    "max_tokens": 32768,
                     "model_type": "embedding"
                 }
             ]
@@ -1367,35 +1385,35 @@
                     "llm_name": "gemini-2.5-flash",
                     "tags": "LLM,CHAT,1024K,IMAGE2TEXT",
                     "max_tokens": 1048576,
-                    "model_type": "chat",
+                    "model_type": "image2text",
                     "is_tools": true
                 },
                 {
                     "llm_name": "gemini-2.5-pro",
                     "tags": "LLM,CHAT,IMAGE2TEXT,1024K",
                     "max_tokens": 1048576,
-                    "model_type": "chat",
+                    "model_type": "image2text",
                     "is_tools": true
                 },
                 {
                     "llm_name": "gemini-2.5-flash-lite",
                     "tags": "LLM,CHAT,1024K,IMAGE2TEXT",
                     "max_tokens": 1048576,
-                    "model_type": "chat",
+                    "model_type": "image2text",
                     "is_tools": true
                 },
                 {
                     "llm_name": "gemini-2.0-flash",
                     "tags": "LLM,CHAT,1024K",
                     "max_tokens": 1048576,
-                    "model_type": "chat",
+                    "model_type": "image2text",
                     "is_tools": true
                 },
                 {
                     "llm_name": "gemini-2.0-flash-lite",
                     "tags": "LLM,CHAT,1024K",
                     "max_tokens": 1048576,
-                    "model_type": "chat",
+                    "model_type": "image2text",
                     "is_tools": true
                 },
                 {
@@ -3009,7 +3027,7 @@
                     "tags": "LLM,CHAT,IMAGE2TEXT,32k",
                     "max_tokens": 32000,
                     "model_type": "image2text",
-                    "is_tools": true
+                    "is_tools": false
                 },
                 {
                     "llm_name": "THUDM/GLM-Z1-32B-0414",

conf/service_conf.yaml

@@ -32,6 +32,11 @@ redis:
   db: 1
   password: 'infini_rag_flow'
   host: 'localhost:6379'
+user_default_llm:
+  default_models:
+    embedding_model:
+      api_key: 'xxx'
+      base_url: 'http://localhost:6380'
 # postgres:
 #   name: 'rag_flow'
 #   user: 'rag_flow'
@@ -77,7 +82,8 @@ redis:
 #       api_key: 'xxxx'
 #       base_url: 'https://api.xx.com'
 #     embedding_model:
-#       name: 'bge-m3'
+#       api_key: 'xxx'
+#       base_url: 'http://localhost:6380'
 #     rerank_model: 'bge-reranker-v2'
 #     asr_model:
 #       model: 'whisper-large-v3' # alias of name
@@ -127,3 +133,9 @@ redis:
 #     - "RAGFlow" # display name
 #     - "" # sender email address
 #   mail_frontend_url: "https://your-frontend.example.com"
+# tcadp_config:
+#  secret_id: 'tencent_secret_id'
+#  secret_key: 'tencent_secret_key'
+#  region: 'tencent_region'
+#  table_result_type: '1'
+#  markdown_image_response_type: '1'

deepdoc/parser/docling_parser.py

@@ -0,0 +1,344 @@
+#
+#  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 __future__ import annotations
+
+import logging
+import re
+from dataclasses import dataclass
+from enum import Enum
+from io import BytesIO
+from os import PathLike
+from pathlib import Path
+from typing import Any, Callable, Iterable, Optional
+
+import pdfplumber
+from PIL import Image
+
+try:
+    from docling.document_converter import DocumentConverter
+except Exception:
+    DocumentConverter = None  
+
+try:
+    from deepdoc.parser.pdf_parser import RAGFlowPdfParser
+except Exception:
+    class RAGFlowPdfParser:  
+        pass
+
+
+class DoclingContentType(str, Enum):
+    IMAGE = "image"
+    TABLE = "table"
+    TEXT = "text"
+    EQUATION = "equation"
+
+
+@dataclass
+class _BBox:
+    page_no: int  
+    x0: float
+    y0: float
+    x1: float
+    y1: float
+
+
+class DoclingParser(RAGFlowPdfParser):
+    def __init__(self):
+        self.logger = logging.getLogger(self.__class__.__name__)
+        self.page_images: list[Image.Image] = []
+        self.page_from = 0
+        self.page_to = 10_000
+
+    def check_installation(self) -> bool:
+        if DocumentConverter is None:
+            self.logger.warning("[Docling] 'docling' is not importable, please: pip install docling")
+            return False
+        try:
+            _ = DocumentConverter()
+            return True
+        except Exception as e:
+            self.logger.error(f"[Docling] init DocumentConverter failed: {e}")
+            return False
+
+    def __images__(self, fnm, zoomin: int = 1, page_from=0, page_to=600, callback=None):
+        self.page_from = page_from
+        self.page_to = page_to
+        try:
+            opener = pdfplumber.open(fnm) if isinstance(fnm, (str, PathLike)) else pdfplumber.open(BytesIO(fnm))
+            with opener as pdf:
+                pages = pdf.pages[page_from:page_to]
+                self.page_images = [p.to_image(resolution=72 * zoomin, antialias=True).original for p in pages]
+        except Exception as e:
+            self.page_images = []
+            self.logger.exception(e)
+
+    def _make_line_tag(self,bbox: _BBox) -> str:
+        if bbox is None:
+            return ""
+        x0,x1, top, bott = bbox.x0, bbox.x1, bbox.y0, bbox.y1
+        if hasattr(self, "page_images") and self.page_images and len(self.page_images) >= bbox.page_no:
+            _, page_height = self.page_images[bbox.page_no-1].size
+            top, bott = page_height-top ,page_height-bott
+        return "@@{}\t{:.1f}\t{:.1f}\t{:.1f}\t{:.1f}##".format(
+            bbox.page_no, x0,x1, top, bott
+        )
+
+    @staticmethod
+    def extract_positions(txt: str) -> list[tuple[list[int], float, float, float, float]]:
+        poss = []
+        for tag in re.findall(r"@@[0-9-]+\t[0-9.\t]+##", txt):
+            pn, left, right, top, bottom = tag.strip("#").strip("@").split("\t")
+            left, right, top, bottom = float(left), float(right), float(top), float(bottom)
+            poss.append(([int(p) - 1 for p in pn.split("-")], left, right, top, bottom))
+        return poss
+
+    def crop(self, text: str, ZM: int = 1, need_position: bool = False):
+        imgs = []
+        poss = self.extract_positions(text)
+        if not poss:
+            return (None, None) if need_position else None
+
+        GAP = 6
+        pos = poss[0]
+        poss.insert(0, ([pos[0][0]], pos[1], pos[2], max(0, pos[3] - 120), max(pos[3] - GAP, 0)))
+        pos = poss[-1]
+        poss.append(([pos[0][-1]], pos[1], pos[2], min(self.page_images[pos[0][-1]].size[1], pos[4] + GAP), min(self.page_images[pos[0][-1]].size[1], pos[4] + 120)))
+        positions = []
+        for ii, (pns, left, right, top, bottom) in enumerate(poss):
+            if bottom <= top:
+                bottom = top + 4
+            img0 = self.page_images[pns[0]]
+            x0, y0, x1, y1 = int(left), int(top), int(right), int(min(bottom, img0.size[1]))
+            
+            crop0 = img0.crop((x0, y0, x1, y1))
+            imgs.append(crop0)
+            if 0 < ii < len(poss)-1:
+                positions.append((pns[0] + self.page_from, x0, x1, y0, y1))
+            remain_bottom = bottom - img0.size[1]
+            for pn in pns[1:]:
+                if remain_bottom <= 0:
+                    break
+                page = self.page_images[pn]
+                x0, y0, x1, y1 = int(left), 0, int(right), int(min(remain_bottom, page.size[1]))
+                cimgp = page.crop((x0, y0, x1, y1))
+                imgs.append(cimgp)
+                if 0 < ii < len(poss) - 1:
+                    positions.append((pn + self.page_from, x0, x1, y0, y1))
+                remain_bottom -= page.size[1]
+
+        if not imgs:
+            return (None, None) if need_position else None
+
+        height = sum(i.size[1] + GAP for i in imgs)
+        width = max(i.size[0] for i in imgs)
+        pic = Image.new("RGB", (width, int(height)), (245, 245, 245))
+        h = 0
+        for ii, img in enumerate(imgs):
+            if ii == 0 or ii + 1 == len(imgs):
+                img = img.convert("RGBA")
+                overlay = Image.new("RGBA", img.size, (0, 0, 0, 0))
+                overlay.putalpha(128)
+                img = Image.alpha_composite(img, overlay).convert("RGB")
+            pic.paste(img, (0, int(h)))
+            h += img.size[1] + GAP
+
+        return (pic, positions) if need_position else pic
+
+    def _iter_doc_items(self, doc) -> Iterable[tuple[str, Any, Optional[_BBox]]]:
+        for t in getattr(doc, "texts", []):
+            parent=getattr(t, "parent", "")
+            ref=getattr(parent,"cref","")
+            label=getattr(t, "label", "")
+            if (label in ("section_header","text",) and ref in ("#/body",)) or label in ("list_item",):
+                text = getattr(t, "text", "") or ""
+                bbox = None
+                if getattr(t, "prov", None):
+                    pn = getattr(t.prov[0], "page_no", None)
+                    bb = getattr(t.prov[0], "bbox", None)
+                    bb = [getattr(bb, "l", None),getattr(bb, "t", None),getattr(bb, "r", None),getattr(bb, "b", None)]
+                    if pn and bb and len(bb) == 4:
+                        bbox = _BBox(page_no=int(pn), x0=bb[0], y0=bb[1], x1=bb[2], y1=bb[3])
+                yield (DoclingContentType.TEXT.value, text, bbox)
+
+        for item in getattr(doc, "texts", []):
+            if getattr(item, "label", "") in ("FORMULA",):
+                text = getattr(item, "text", "") or ""
+                bbox = None
+                if getattr(item, "prov", None):
+                    pn = getattr(item.prov, "page_no", None)
+                    bb = getattr(item.prov, "bbox", None)
+                    bb = [getattr(bb, "l", None),getattr(bb, "t", None),getattr(bb, "r", None),getattr(bb, "b", None)]
+                    if pn and bb and len(bb) == 4:
+                        bbox = _BBox(int(pn), bb[0], bb[1], bb[2], bb[3])
+                yield (DoclingContentType.EQUATION.value, text, bbox)
+
+    def _transfer_to_sections(self, doc) -> list[tuple[str, str]]:
+        """
+        和 MinerUParser 保持一致：返回 [(section_text, line_tag), ...]
+        """
+        sections: list[tuple[str, str]] = []
+        for typ, payload, bbox in self._iter_doc_items(doc):
+            if typ == DoclingContentType.TEXT.value:
+                section = payload.strip()
+                if not section:
+                    continue
+            elif typ == DoclingContentType.EQUATION.value:
+                section = payload.strip()
+            else:
+                continue
+            
+            tag = self._make_line_tag(bbox) if isinstance(bbox,_BBox) else ""
+            sections.append((section, tag))
+        return sections
+
+    def cropout_docling_table(self, page_no: int, bbox: tuple[float, float, float, float], zoomin: int = 1):
+        if not getattr(self, "page_images", None):
+            return None, ""
+
+        idx = (page_no - 1) - getattr(self, "page_from", 0)
+        if idx < 0 or idx >= len(self.page_images):
+            return None, ""
+
+        page_img = self.page_images[idx]
+        W, H = page_img.size
+        left, top, right, bott = bbox
+
+        x0 = float(left)
+        y0 = float(H-top)
+        x1 = float(right)
+        y1 = float(H-bott)
+
+        x0, y0 = max(0.0, min(x0, W - 1)), max(0.0, min(y0, H - 1))
+        x1, y1 = max(x0 + 1.0, min(x1, W)), max(y0 + 1.0, min(y1, H))
+
+        try:
+            crop = page_img.crop((int(x0), int(y0), int(x1), int(y1))).convert("RGB")
+        except Exception:
+            return None, ""
+
+        pos = (page_no-1 if page_no>0 else 0, x0, x1, y0, y1)
+        return crop, [pos]
+
+    def _transfer_to_tables(self, doc):
+        tables = []
+        for tab in getattr(doc, "tables", []):
+            img = None
+            positions = ""
+            if getattr(tab, "prov", None):
+                pn = getattr(tab.prov[0], "page_no", None)
+                bb = getattr(tab.prov[0], "bbox", None)
+                if pn is not None and bb is not None:
+                    left = getattr(bb, "l", None)
+                    top = getattr(bb, "t", None)
+                    right = getattr(bb, "r", None)
+                    bott = getattr(bb, "b", None)
+                    if None not in (left, top, right, bott):
+                        img, positions = self.cropout_docling_table(int(pn), (float(left), float(top), float(right), float(bott)))
+            html = ""
+            try:
+                html = tab.export_to_html(doc=doc)
+            except Exception:
+                pass
+            tables.append(((img, html), positions if positions else ""))
+        for pic in getattr(doc, "pictures", []):
+            img = None
+            positions = ""
+            if getattr(pic, "prov", None):
+                pn = getattr(pic.prov[0], "page_no", None)
+                bb = getattr(pic.prov[0], "bbox", None)
+                if pn is not None and bb is not None:
+                    left = getattr(bb, "l", None)
+                    top = getattr(bb, "t", None)
+                    right = getattr(bb, "r", None)
+                    bott = getattr(bb, "b", None)
+                    if None not in (left, top, right, bott):
+                        img, positions = self.cropout_docling_table(int(pn), (float(left), float(top), float(right), float(bott)))
+            captions = ""
+            try:
+                captions = pic.caption_text(doc=doc)
+            except Exception:
+                pass
+            tables.append(((img, [captions]), positions if positions else ""))
+        return tables
+
+    def parse_pdf(
+        self,
+        filepath: str | PathLike[str],
+        binary: BytesIO | bytes | None = None,
+        callback: Optional[Callable] = None,
+        *,
+        output_dir: Optional[str] = None, 
+        lang: Optional[str] = None,        
+        method: str = "auto",             
+        delete_output: bool = True,       
+    ):
+
+        if not self.check_installation():
+            raise RuntimeError("Docling not available, please install `docling`")
+
+        if binary is not None:
+            tmpdir = Path(output_dir) if output_dir else Path.cwd() / ".docling_tmp"
+            tmpdir.mkdir(parents=True, exist_ok=True)
+            name = Path(filepath).name or "input.pdf"
+            tmp_pdf = tmpdir / name
+            with open(tmp_pdf, "wb") as f:
+                if isinstance(binary, (bytes, bytearray)):
+                    f.write(binary)
+                else:
+                    f.write(binary.getbuffer())
+            src_path = tmp_pdf
+        else:
+            src_path = Path(filepath)
+            if not src_path.exists():
+                raise FileNotFoundError(f"PDF not found: {src_path}")
+
+        if callback:
+            callback(0.1, f"[Docling] Converting: {src_path}")
+
+        try:
+            self.__images__(str(src_path), zoomin=1)
+        except Exception as e:
+            self.logger.warning(f"[Docling] render pages failed: {e}")
+
+        conv = DocumentConverter()  
+        conv_res = conv.convert(str(src_path))
+        doc = conv_res.document
+        if callback:
+            callback(0.7, f"[Docling] Parsed doc: {getattr(doc, 'num_pages', 'n/a')} pages")
+
+        sections = self._transfer_to_sections(doc)
+        tables = self._transfer_to_tables(doc)
+
+        if callback:
+            callback(0.95, f"[Docling] Sections: {len(sections)}, Tables: {len(tables)}")
+
+        if binary is not None and delete_output:
+            try:
+                Path(src_path).unlink(missing_ok=True)
+            except Exception:
+                pass
+
+        if callback:
+            callback(1.0, "[Docling] Done.")
+        return sections, tables
+
+
+if __name__ == "__main__":
+    logging.basicConfig(level=logging.INFO)
+    parser = DoclingParser()
+    print("Docling available:", parser.check_installation())
+    sections, tables = parser.parse_pdf(filepath="test_docling/toc.pdf", binary=None)
+    print(len(sections), len(tables))

deepdoc/parser/excel_parser.py

@@ -54,8 +54,8 @@ class RAGFlowExcelParser:
             try:
                 file_like_object.seek(0)
                 try:
-                    df = pd.read_excel(file_like_object)
-                    return RAGFlowExcelParser._dataframe_to_workbook(df)
+                    dfs = pd.read_excel(file_like_object, sheet_name=None)
+                    return RAGFlowExcelParser._dataframe_to_workbook(dfs)
                 except Exception as ex:
                     logging.info(f"pandas with default engine load error: {ex}, try calamine instead")
                     file_like_object.seek(0)
@@ -75,6 +75,10 @@ class RAGFlowExcelParser:
 
     @staticmethod
     def _dataframe_to_workbook(df):
+        # if contains multiple sheets use _dataframes_to_workbook
+        if isinstance(df, dict) and len(df) > 1:
+            return RAGFlowExcelParser._dataframes_to_workbook(df)
+
         df = RAGFlowExcelParser._clean_dataframe(df)
         wb = Workbook()
         ws = wb.active
@@ -88,6 +92,22 @@ class RAGFlowExcelParser:
                 ws.cell(row=row_num, column=col_num, value=value)
 
         return wb
+    
+    @staticmethod
+    def _dataframes_to_workbook(dfs: dict):
+        wb = Workbook()
+        default_sheet = wb.active
+        wb.remove(default_sheet)
+        
+        for sheet_name, df in dfs.items():
+            df = RAGFlowExcelParser._clean_dataframe(df)
+            ws = wb.create_sheet(title=sheet_name)
+            for col_num, column_name in enumerate(df.columns, 1):
+                ws.cell(row=1, column=col_num, value=column_name)
+            for row_num, row in enumerate(df.values, 2):
+                for col_num, value in enumerate(row, 1):
+                    ws.cell(row=row_num, column=col_num, value=value)
+        return wb
 
     def html(self, fnm, chunk_rows=256):
         from html import escape
@@ -103,7 +123,12 @@ class RAGFlowExcelParser:
 
         for sheetname in wb.sheetnames:
             ws = wb[sheetname]
-            rows = list(ws.rows)
+            try:
+                rows = list(ws.rows)
+            except Exception as e:
+                logging.warning(f"Skip sheet '{sheetname}' due to rows access error: {e}")
+                continue
+
             if not rows:
                 continue
 
@@ -150,7 +175,11 @@ class RAGFlowExcelParser:
         res = []
         for sheetname in wb.sheetnames:
             ws = wb[sheetname]
-            rows = list(ws.rows)
+            try:
+                rows = list(ws.rows)
+            except Exception as e:
+                logging.warning(f"Skip sheet '{sheetname}' due to rows access error: {e}")
+                continue
             if not rows:
                 continue
             ti = list(rows[0])
@@ -173,9 +202,14 @@ class RAGFlowExcelParser:
         if fnm.split(".")[-1].lower().find("xls") >= 0:
             wb = RAGFlowExcelParser._load_excel_to_workbook(BytesIO(binary))
             total = 0
+            
             for sheetname in wb.sheetnames:
-                ws = wb[sheetname]
-                total += len(list(ws.rows))
+               try:
+                   ws = wb[sheetname]
+                   total += len(list(ws.rows))
+               except Exception as e:
+                   logging.warning(f"Skip sheet '{sheetname}' due to rows access error: {e}")
+                   continue
             return total
 
         if fnm.split(".")[-1].lower() in ["csv", "txt"]:

deepdoc/parser/figure_parser.py

@@ -17,6 +17,8 @@ from concurrent.futures import ThreadPoolExecutor, as_completed
 
 from PIL import Image
 
+from api.db import LLMType
+from api.db.services.llm_service import LLMBundle
 from api.utils.api_utils import timeout
 from rag.app.picture import vision_llm_chunk as picture_vision_llm_chunk
 from rag.prompts.generator import vision_llm_figure_describe_prompt
@@ -32,6 +34,43 @@ def vision_figure_parser_figure_data_wrapper(figures_data_without_positions):
         if isinstance(figure_data[1], Image.Image)
     ]
 
+def vision_figure_parser_docx_wrapper(sections,tbls,callback=None,**kwargs):
+    try:
+        vision_model = LLMBundle(kwargs["tenant_id"], LLMType.IMAGE2TEXT)
+        callback(0.7, "Visual model detected. Attempting to enhance figure extraction...")
+    except Exception:
+        vision_model = None
+    if vision_model:
+        figures_data = vision_figure_parser_figure_data_wrapper(sections)
+        try:
+            docx_vision_parser = VisionFigureParser(vision_model=vision_model, figures_data=figures_data, **kwargs)
+            boosted_figures = docx_vision_parser(callback=callback)
+            tbls.extend(boosted_figures)
+        except Exception as e:
+            callback(0.8, f"Visual model error: {e}. Skipping figure parsing enhancement.")
+    return tbls
+
+def vision_figure_parser_pdf_wrapper(tbls,callback=None,**kwargs):
+    try:
+        vision_model = LLMBundle(kwargs["tenant_id"], LLMType.IMAGE2TEXT)
+        callback(0.7, "Visual model detected. Attempting to enhance figure extraction...")
+    except Exception:
+        vision_model = None
+    if vision_model:
+        def is_figure_item(item):
+            return (
+                isinstance(item[0][0], Image.Image) and
+                isinstance(item[0][1], list)
+            )
+        figures_data = [item for item in tbls if is_figure_item(item)]
+        try:
+            docx_vision_parser = VisionFigureParser(vision_model=vision_model, figures_data=figures_data, **kwargs)
+            boosted_figures = docx_vision_parser(callback=callback)
+            tbls = [item for item in tbls if not is_figure_item(item)]
+            tbls.extend(boosted_figures)
+        except Exception as e:
+            callback(0.8, f"Visual model error: {e}. Skipping figure parsing enhancement.")
+    return tbls
 
 shared_executor = ThreadPoolExecutor(max_workers=10)
 

deepdoc/parser/markdown_parser.py

@@ -117,11 +117,24 @@ class MarkdownElementExtractor:
         self.markdown_content = markdown_content
         self.lines = markdown_content.split("\n")
 
-    def extract_elements(self):
+    def get_delimiters(self,delimiters):
+        toks = re.findall(r"`([^`]+)`", delimiters)
+        toks = sorted(set(toks), key=lambda x: -len(x))
+        return "|".join(re.escape(t) for t in toks if t)
+    
+    def extract_elements(self,delimiter=None):
         """Extract individual elements (headers, code blocks, lists, etc.)"""
         sections = []
 
         i = 0
+        dels=""
+        if delimiter:
+            dels = self.get_delimiters(delimiter)
+        if len(dels) > 0:
+            text = "\n".join(self.lines)
+            parts = re.split(dels, text)
+            sections = [p.strip() for p in parts if p and p.strip()]
+            return sections
         while i < len(self.lines):
             line = self.lines[i]
 

deepdoc/parser/mineru_parser.py

@@ -0,0 +1,499 @@
+#
+#  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 json
+import logging
+import platform
+import re
+import subprocess
+import sys
+import tempfile
+import threading
+import time
+from io import BytesIO
+from os import PathLike
+from pathlib import Path
+from queue import Empty, Queue
+from typing import Any, Callable, Optional
+import requests
+import os
+import zipfile
+
+import numpy as np
+import pdfplumber
+from PIL import Image
+from strenum import StrEnum
+
+from deepdoc.parser.pdf_parser import RAGFlowPdfParser
+
+LOCK_KEY_pdfplumber = "global_shared_lock_pdfplumber"
+if LOCK_KEY_pdfplumber not in sys.modules:
+    sys.modules[LOCK_KEY_pdfplumber] = threading.Lock()
+
+
+class MinerUContentType(StrEnum):
+    IMAGE = "image"
+    TABLE = "table"
+    TEXT = "text"
+    EQUATION = "equation"
+    CODE = "code"
+    LIST = "list"
+    DISCARDED = "discarded"
+
+
+class MinerUParser(RAGFlowPdfParser):
+    def __init__(self, mineru_path: str = "mineru", mineru_api: str = "http://host.docker.internal:9987"):
+        self.mineru_path = Path(mineru_path)
+        self.mineru_api = mineru_api.rstrip('/')
+        self.using_api = False
+        self.logger = logging.getLogger(self.__class__.__name__)
+
+    def _extract_zip_no_root(self, zip_path, extract_to, root_dir):
+        with zipfile.ZipFile(zip_path, 'r') as zip_ref:
+            if not root_dir:
+                files = zip_ref.namelist()
+                if files and files[0].endswith('/'):
+                    root_dir = files[0]
+                else:
+                    root_dir = None
+            
+            if not root_dir or not root_dir.endswith('/'):
+                self.logger.info(f"[MinerU] No root directory found, extracting all...fff{root_dir}")
+                zip_ref.extractall(extract_to)
+                return
+            
+            root_len = len(root_dir)
+            for member in zip_ref.infolist():
+                filename = member.filename
+                if filename == root_dir:
+                    self.logger.info("[MinerU] Ignore root folder...")
+                    continue
+                
+                path = filename
+                if path.startswith(root_dir):
+                    path = path[root_len:]
+                
+                full_path = os.path.join(extract_to, path)
+                if member.is_dir():
+                    os.makedirs(full_path, exist_ok=True)
+                else:
+                    os.makedirs(os.path.dirname(full_path), exist_ok=True)
+                    with open(full_path, 'wb') as f:
+                        f.write(zip_ref.read(filename))
+
+    def _is_http_endpoint_valid(self, url, timeout=5):
+        try:
+            response = requests.head(url, timeout=timeout, allow_redirects=True)
+            return response.status_code in [200, 301, 302, 307, 308]
+        except Exception:
+            return False
+
+    def check_installation(self) -> bool:
+        subprocess_kwargs = {
+            "capture_output": True,
+            "text": True,
+            "check": True,
+            "encoding": "utf-8",
+            "errors": "ignore",
+        }
+
+        if platform.system() == "Windows":
+            subprocess_kwargs["creationflags"] = getattr(subprocess, "CREATE_NO_WINDOW", 0)
+
+        try:
+            result = subprocess.run([str(self.mineru_path), "--version"], **subprocess_kwargs)
+            version_info = result.stdout.strip()
+            if version_info:
+                logging.info(f"[MinerU] Detected version: {version_info}")
+            else:
+                logging.info("[MinerU] Detected MinerU, but version info is empty.")
+            return True
+        except subprocess.CalledProcessError as e:
+            logging.warning(f"[MinerU] Execution failed (exit code {e.returncode}).")
+        except FileNotFoundError:
+            logging.warning("[MinerU] MinerU not found. Please install it via: pip install -U 'mineru[core]'")
+        except Exception as e:
+            logging.error(f"[MinerU] Unexpected error during installation check: {e}")
+
+        try:
+            if self.mineru_api:
+                # check openapi.json
+                openapi_exists = self._is_http_endpoint_valid(self.mineru_api + "/openapi.json")
+                logging.info(f"[MinerU] Detected {self.mineru_api}/openapi.json: {openapi_exists}")
+                self.using_api = openapi_exists
+                return openapi_exists
+            else:
+                logging.info("[MinerU] api not exists.")
+        except Exception as e:
+            logging.error(f"[MinerU] Unexpected error during api check: {e}")
+        return False
+
+    def _run_mineru(self, input_path: Path, output_dir: Path, method: str = "auto", backend: str = "pipeline", lang: Optional[str] = None, callback: Optional[Callable] = None):
+        if self.using_api:
+            self._run_mineru_api(input_path, output_dir, method, backend, lang, callback)
+        else:
+            self._run_mineru_executable(input_path, output_dir, method, backend, lang, callback)
+    
+    def _run_mineru_api(self, input_path: Path, output_dir: Path, method: str = "auto", backend: str = "pipeline", lang: Optional[str] = None, callback: Optional[Callable] = None):
+        OUTPUT_ZIP_PATH = os.path.join(str(output_dir), "output.zip")
+        
+        pdf_file_path = str(input_path)
+
+        if not os.path.exists(pdf_file_path):
+            raise RuntimeError(f"[MinerU] PDF file not exists: {pdf_file_path}")
+
+        pdf_file_name = Path(pdf_file_path).stem.strip()
+        output_path = os.path.join(str(output_dir), pdf_file_name, method)
+        os.makedirs(output_path, exist_ok=True)
+
+        files = {
+            "files": (pdf_file_name + ".pdf", open(pdf_file_path, "rb"), "application/pdf")
+        }
+
+        data = {
+            "output_dir": "./output",
+            "lang_list": lang,
+            "backend": backend,
+            "parse_method": method,
+            "formula_enable": True,
+            "table_enable": True,
+            "server_url": None,
+            "return_md": True,
+            "return_middle_json": True,
+            "return_model_output": True,
+            "return_content_list": True,
+            "return_images": True,
+            "response_format_zip": True,
+            "start_page_id": 0,
+            "end_page_id": 99999
+        }
+
+        headers = {
+            "Accept": "application/json"
+        }
+        try:
+            self.logger.info(f"[MinerU] invoke api: {self.mineru_api}/file_parse")
+            if callback:
+                callback(0.20, f"[MinerU] invoke api: {self.mineru_api}/file_parse")
+            response = requests.post(
+                url=f"{self.mineru_api}/file_parse",
+                files=files,
+                data=data,
+                headers=headers,
+                timeout=1800
+            )
+
+            response.raise_for_status()
+            if response.headers.get("Content-Type") == "application/zip":
+                self.logger.info(f"[MinerU] zip file returned, saving to {OUTPUT_ZIP_PATH}...")
+
+                if callback:
+                    callback(0.30, f"[MinerU] zip file returned, saving to {OUTPUT_ZIP_PATH}...")
+
+                with open(OUTPUT_ZIP_PATH, "wb") as f:
+                    f.write(response.content)
+
+                self.logger.info(f"[MinerU] Unzip to {output_path}...")
+                self._extract_zip_no_root(OUTPUT_ZIP_PATH, output_path, pdf_file_name + "/")
+
+                if callback:
+                    callback(0.40, f"[MinerU] Unzip to {output_path}...")
+            else:
+                self.logger.warning("[MinerU] not zip returned from api：%s " % response.headers.get("Content-Type"))
+        except Exception as e:
+            raise RuntimeError(f"[MinerU] api failed with exception {e}")
+        self.logger.info("[MinerU] Api completed successfully.")
+
+    def _run_mineru_executable(self, input_path: Path, output_dir: Path, method: str = "auto", backend: str = "pipeline", lang: Optional[str] = None, callback: Optional[Callable] = None):
+        cmd = [str(self.mineru_path), "-p", str(input_path), "-o", str(output_dir), "-m", method]
+        if backend:
+            cmd.extend(["-b", backend])
+        if lang:
+            cmd.extend(["-l", lang])
+
+        self.logger.info(f"[MinerU] Running command: {' '.join(cmd)}")
+
+        subprocess_kwargs = {
+            "stdout": subprocess.PIPE,
+            "stderr": subprocess.PIPE,
+            "text": True,
+            "encoding": "utf-8",
+            "errors": "ignore",
+            "bufsize": 1,
+        }
+
+        if platform.system() == "Windows":
+            subprocess_kwargs["creationflags"] = getattr(subprocess, "CREATE_NO_WINDOW", 0)
+
+        process = subprocess.Popen(cmd, **subprocess_kwargs)
+        stdout_queue, stderr_queue = Queue(), Queue()
+
+        def enqueue_output(pipe, queue, prefix):
+            for line in iter(pipe.readline, ""):
+                if line.strip():
+                    queue.put((prefix, line.strip()))
+            pipe.close()
+
+        threading.Thread(target=enqueue_output, args=(process.stdout, stdout_queue, "STDOUT"), daemon=True).start()
+        threading.Thread(target=enqueue_output, args=(process.stderr, stderr_queue, "STDERR"), daemon=True).start()
+
+        while process.poll() is None:
+            for q in (stdout_queue, stderr_queue):
+                try:
+                    while True:
+                        prefix, line = q.get_nowait()
+                        if prefix == "STDOUT":
+                            self.logger.info(f"[MinerU] {line}")
+                        else:
+                            self.logger.warning(f"[MinerU] {line}")
+                except Empty:
+                    pass
+            time.sleep(0.1)
+
+        return_code = process.wait()
+        if return_code != 0:
+            raise RuntimeError(f"[MinerU] Process failed with exit code {return_code}")
+        self.logger.info("[MinerU] Command completed successfully.")
+
+    def __images__(self, fnm, zoomin: int = 1, page_from=0, page_to=600, callback=None):
+        self.page_from = page_from
+        self.page_to = page_to
+        try:
+            with pdfplumber.open(fnm) if isinstance(fnm, (str, PathLike)) else pdfplumber.open(BytesIO(fnm)) as pdf:
+                self.pdf = pdf
+                self.page_images = [p.to_image(resolution=72 * zoomin, antialias=True).original for _, p in enumerate(self.pdf.pages[page_from:page_to])]
+        except Exception as e:
+            self.page_images = None
+            self.total_page = 0
+            logging.exception(e)
+
+    def _line_tag(self, bx):
+        pn = [bx["page_idx"] + 1]
+        positions = bx["bbox"]
+        x0, top, x1, bott = positions
+
+        if hasattr(self, "page_images") and self.page_images and len(self.page_images) > bx["page_idx"]:
+            page_width, page_height = self.page_images[bx["page_idx"]].size
+            x0 = (x0 / 1000.0) * page_width
+            x1 = (x1 / 1000.0) * page_width
+            top = (top / 1000.0) * page_height
+            bott = (bott / 1000.0) * page_height
+
+        return "@@{}\t{:.1f}\t{:.1f}\t{:.1f}\t{:.1f}##".format("-".join([str(p) for p in pn]), x0, x1, top, bott)
+
+    def crop(self, text, ZM=1, need_position=False):
+        imgs = []
+        poss = self.extract_positions(text)
+        if not poss:
+            if need_position:
+                return None, None
+            return
+
+        max_width = max(np.max([right - left for (_, left, right, _, _) in poss]), 6)
+        GAP = 6
+        pos = poss[0]
+        poss.insert(0, ([pos[0][0]], pos[1], pos[2], max(0, pos[3] - 120), max(pos[3] - GAP, 0)))
+        pos = poss[-1]
+        poss.append(([pos[0][-1]], pos[1], pos[2], min(self.page_images[pos[0][-1]].size[1], pos[4] + GAP), min(self.page_images[pos[0][-1]].size[1], pos[4] + 120)))
+
+        positions = []
+        for ii, (pns, left, right, top, bottom) in enumerate(poss):
+            right = left + max_width
+
+            if bottom <= top:
+                bottom = top + 2
+
+            for pn in pns[1:]:
+                bottom += self.page_images[pn - 1].size[1]
+
+            img0 = self.page_images[pns[0]]
+            x0, y0, x1, y1 = int(left), int(top), int(right), int(min(bottom, img0.size[1]))
+            crop0 = img0.crop((x0, y0, x1, y1))
+            imgs.append(crop0)
+            if 0 < ii < len(poss) - 1:
+                positions.append((pns[0] + self.page_from, x0, x1, y0, y1))
+
+            bottom -= img0.size[1]
+            for pn in pns[1:]:
+                page = self.page_images[pn]
+                x0, y0, x1, y1 = int(left), 0, int(right), int(min(bottom, page.size[1]))
+                cimgp = page.crop((x0, y0, x1, y1))
+                imgs.append(cimgp)
+                if 0 < ii < len(poss) - 1:
+                    positions.append((pn + self.page_from, x0, x1, y0, y1))
+                bottom -= page.size[1]
+
+        if not imgs:
+            if need_position:
+                return None, None
+            return
+
+        height = 0
+        for img in imgs:
+            height += img.size[1] + GAP
+        height = int(height)
+        width = int(np.max([i.size[0] for i in imgs]))
+        pic = Image.new("RGB", (width, height), (245, 245, 245))
+        height = 0
+        for ii, img in enumerate(imgs):
+            if ii == 0 or ii + 1 == len(imgs):
+                img = img.convert("RGBA")
+                overlay = Image.new("RGBA", img.size, (0, 0, 0, 0))
+                overlay.putalpha(128)
+                img = Image.alpha_composite(img, overlay).convert("RGB")
+            pic.paste(img, (0, int(height)))
+            height += img.size[1] + GAP
+
+        if need_position:
+            return pic, positions
+        return pic
+
+    @staticmethod
+    def extract_positions(txt: str):
+        poss = []
+        for tag in re.findall(r"@@[0-9-]+\t[0-9.\t]+##", txt):
+            pn, left, right, top, bottom = tag.strip("#").strip("@").split("\t")
+            left, right, top, bottom = float(left), float(right), float(top), float(bottom)
+            poss.append(([int(p) - 1 for p in pn.split("-")], left, right, top, bottom))
+        return poss
+
+    def _read_output(self, output_dir: Path, file_stem: str, method: str = "auto", backend: str = "pipeline") -> list[dict[str, Any]]:
+        subdir = output_dir / file_stem / method
+        if backend.startswith("vlm-"):
+            subdir = output_dir / file_stem / "vlm"
+        json_file = subdir / f"{file_stem}_content_list.json"
+
+        if not json_file.exists():
+            raise FileNotFoundError(f"[MinerU] Missing output file: {json_file}")
+
+        with open(json_file, "r", encoding="utf-8") as f:
+            data = json.load(f)
+
+        for item in data:
+            for key in ("img_path", "table_img_path", "equation_img_path"):
+                if key in item and item[key]:
+                    item[key] = str((subdir / item[key]).resolve())
+        return data
+
+    def _transfer_to_sections(self, outputs: list[dict[str, Any]]):
+        sections = []
+        for output in outputs:
+            match output["type"]:
+                case MinerUContentType.TEXT:
+                    section = output["text"]
+                case MinerUContentType.TABLE:
+                    section = output["table_body"] if "table_body" in output else "" + "\n".join(output["table_caption"]) + "\n".join(output["table_footnote"])
+                case MinerUContentType.IMAGE:
+                    section = "".join(output["image_caption"]) + "\n" + "".join(output["image_footnote"])
+                case MinerUContentType.EQUATION:
+                    section = output["text"]
+                case MinerUContentType.CODE:
+                    section = output["code_body"] + "\n".join(output.get("code_caption", []))
+                case MinerUContentType.LIST:
+                    section = "\n".join(output.get("list_items", []))
+                case MinerUContentType.DISCARDED:
+                    pass
+
+            if section:
+                sections.append((section, self._line_tag(output)))
+        return sections
+
+    def _transfer_to_tables(self, outputs: list[dict[str, Any]]):
+        return []
+
+    def parse_pdf(
+        self,
+        filepath: str | PathLike[str],
+        binary: BytesIO | bytes,
+        callback: Optional[Callable] = None,
+        *,
+        output_dir: Optional[str] = None,
+        backend: str = "pipeline",
+        lang: Optional[str] = None,
+        method: str = "auto",
+        delete_output: bool = True,
+    ) -> tuple:
+        import shutil
+
+        temp_pdf = None
+        created_tmp_dir = False
+
+        # remove spaces, or mineru crash, and _read_output fail too
+        file_path = Path(filepath)
+        pdf_file_name = file_path.stem.replace(" ", "") + ".pdf"
+        pdf_file_path_valid = os.path.join(file_path.parent, pdf_file_name)
+
+        if binary:
+            temp_dir = Path(tempfile.mkdtemp(prefix="mineru_bin_pdf_"))
+            temp_pdf = temp_dir / pdf_file_name
+            with open(temp_pdf, "wb") as f:
+                f.write(binary)
+            pdf = temp_pdf
+            self.logger.info(f"[MinerU] Received binary PDF -> {temp_pdf}")
+            if callback:
+                callback(0.15, f"[MinerU] Received binary PDF -> {temp_pdf}")
+        else:
+            if pdf_file_path_valid != filepath:
+                self.logger.info(f"[MinerU] Remove all space in file name: {pdf_file_path_valid}")
+                shutil.move(filepath, pdf_file_path_valid)
+            pdf = Path(pdf_file_path_valid)
+            if not pdf.exists():
+                if callback:
+                    callback(-1, f"[MinerU] PDF not found: {pdf}")
+                raise FileNotFoundError(f"[MinerU] PDF not found: {pdf}")
+
+        if output_dir:
+            out_dir = Path(output_dir)
+            out_dir.mkdir(parents=True, exist_ok=True)
+        else:
+            out_dir = Path(tempfile.mkdtemp(prefix="mineru_pdf_"))
+            created_tmp_dir = True
+
+        self.logger.info(f"[MinerU] Output directory: {out_dir}")
+        if callback:
+            callback(0.15, f"[MinerU] Output directory: {out_dir}")
+
+        self.__images__(pdf, zoomin=1)
+
+        try:
+            self._run_mineru(pdf, out_dir, method=method, backend=backend, lang=lang, callback=callback)
+            outputs = self._read_output(out_dir, pdf.stem, method=method, backend=backend)
+            self.logger.info(f"[MinerU] Parsed {len(outputs)} blocks from PDF.")
+            if callback:
+                callback(0.75, f"[MinerU] Parsed {len(outputs)} blocks from PDF.")
+            return self._transfer_to_sections(outputs), self._transfer_to_tables(outputs)
+        finally:
+            if temp_pdf and temp_pdf.exists():
+                try:
+                    temp_pdf.unlink()
+                    temp_pdf.parent.rmdir()
+                except Exception:
+                    pass
+            if delete_output and created_tmp_dir and out_dir.exists():
+                try:
+                    shutil.rmtree(out_dir)
+                except Exception:
+                    pass
+
+
+if __name__ == "__main__":
+    parser = MinerUParser("mineru")
+    print("MinerU available:", parser.check_installation())
+
+    filepath = ""
+    with open(filepath, "rb") as file:
+        outputs = parser.parse_pdf(filepath=filepath, binary=file.read())
+        for output in outputs:
+            print(output)

deepdoc/parser/pdf_parser.py

@@ -34,8 +34,8 @@ from huggingface_hub import snapshot_download
 from PIL import Image
 from pypdf import PdfReader as pdf2_read
 
-from api import settings
 from api.utils.file_utils import get_project_base_directory
+from api.utils.common import pip_install_torch
 from deepdoc.vision import OCR, AscendLayoutRecognizer, LayoutRecognizer, Recognizer, TableStructureRecognizer
 from rag.app.picture import vision_llm_chunk as picture_vision_llm_chunk
 from rag.nlp import rag_tokenizer
@@ -84,14 +84,13 @@ class RAGFlowPdfParser:
         self.tbl_det = TableStructureRecognizer()
 
         self.updown_cnt_mdl = xgb.Booster()
-        if not settings.LIGHTEN:
-            try:
-                import torch.cuda
-
-                if torch.cuda.is_available():
-                    self.updown_cnt_mdl.set_param({"device": "cuda"})
-            except Exception:
-                logging.exception("RAGFlowPdfParser __init__")
+        try:
+            pip_install_torch()
+            import torch.cuda
+            if torch.cuda.is_available():
+                self.updown_cnt_mdl.set_param({"device": "cuda"})
+        except Exception:
+            logging.exception("RAGFlowPdfParser __init__")
         try:
             model_dir = os.path.join(get_project_base_directory(), "rag/res/deepdoc")
             self.updown_cnt_mdl.load_model(os.path.join(model_dir, "updown_concat_xgb.model"))
@@ -1131,7 +1130,7 @@ class RAGFlowPdfParser:
             bxes = [b for bxs in self.boxes for b in bxs]
             self.is_english = re.search(r"[\na-zA-Z0-9,/¸;:'\[\]\(\)!@#$%^&*\"?<>._-]{30,}", "".join([b["text"] for b in random.choices(bxes, k=min(30, len(bxes)))]))
 
-        logging.debug("Is it English:", self.is_english)
+        logging.debug(f"Is it English: {self.is_english}")
 
         self.page_cum_height = np.cumsum(self.page_cum_height)
         assert len(self.page_cum_height) == len(self.page_images) + 1

deepdoc/parser/tcadp_parser.py

@@ -0,0 +1,504 @@
+#
+#  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 base64
+import json
+import logging
+import os
+import shutil
+import tempfile
+import time
+import traceback
+import types
+import zipfile
+from datetime import datetime
+from io import BytesIO
+from os import PathLike
+from pathlib import Path
+from typing import Any, Callable, Optional
+
+import requests
+from tencentcloud.common import credential
+from tencentcloud.common.profile.client_profile import ClientProfile
+from tencentcloud.common.profile.http_profile import HttpProfile
+from tencentcloud.common.exception.tencent_cloud_sdk_exception import TencentCloudSDKException
+from tencentcloud.lkeap.v20240522 import lkeap_client, models
+
+from api.utils.configs import get_base_config
+from deepdoc.parser.pdf_parser import RAGFlowPdfParser
+
+
+class TencentCloudAPIClient:
+    """Tencent Cloud API client using official SDK"""
+
+    def __init__(self, secret_id, secret_key, region):
+        self.secret_id = secret_id
+        self.secret_key = secret_key
+        self.region = region
+        
+        # Create credentials
+        self.cred = credential.Credential(secret_id, secret_key)
+        
+        # Instantiate an http option, optional, can be skipped if no special requirements
+        self.httpProfile = HttpProfile()
+        self.httpProfile.endpoint = "lkeap.tencentcloudapi.com"
+
+        # Instantiate a client option, optional, can be skipped if no special requirements
+        self.clientProfile = ClientProfile()
+        self.clientProfile.httpProfile = self.httpProfile
+        
+        # Instantiate the client object for the product to be requested, clientProfile is optional
+        self.client = lkeap_client.LkeapClient(self.cred, region, self.clientProfile)
+
+    def reconstruct_document_sse(self, file_type, file_url=None, file_base64=None, file_start_page=1, file_end_page=1000, config=None):
+        """Call document parsing API using official SDK"""
+        try:
+            # Instantiate a request object, each interface corresponds to a request object
+            req = models.ReconstructDocumentSSERequest()
+            
+            # Build request parameters
+            params = {
+                "FileType": file_type,
+                "FileStartPageNumber": file_start_page,
+                "FileEndPageNumber": file_end_page,
+            }
+            
+            # According to Tencent Cloud API documentation, either FileUrl or FileBase64 parameter must be provided, if both are provided only FileUrl will be used
+            if file_url:
+                params["FileUrl"] = file_url
+                logging.info(f"[TCADP] Using file URL: {file_url}")
+            elif file_base64:
+                params["FileBase64"] = file_base64
+                logging.info(f"[TCADP] Using Base64 data, length: {len(file_base64)} characters")
+            else:
+                raise ValueError("Must provide either FileUrl or FileBase64 parameter")
+
+            if config:
+                params["Config"] = config
+
+            req.from_json_string(json.dumps(params))
+
+            # The returned resp is an instance of ReconstructDocumentSSEResponse, corresponding to the request object
+            resp = self.client.ReconstructDocumentSSE(req)
+            parser_result = {}
+            
+            # Output json format string response
+            if isinstance(resp, types.GeneratorType):  # Streaming response
+                logging.info("[TCADP] Detected streaming response")
+                for event in resp:
+                    logging.info(f"[TCADP] Received event: {event}")
+                    if event.get('data'):
+                        try:
+                            data_dict = json.loads(event['data'])
+                            logging.info(f"[TCADP] Parsed data: {data_dict}")
+                            
+                            if data_dict.get('Progress') == "100":
+                                parser_result = data_dict
+                                logging.info("[TCADP] Document parsing completed!")
+                                logging.info(f"[TCADP] Task ID: {data_dict.get('TaskId')}")
+                                logging.info(f"[TCADP] Success pages: {data_dict.get('SuccessPageNum')}")
+                                logging.info(f"[TCADP] Failed pages: {data_dict.get('FailPageNum')}")
+
+                                # Print failed page information
+                                failed_pages = data_dict.get("FailedPages", [])
+                                if failed_pages:
+                                    logging.warning("[TCADP] Failed parsing pages:")
+                                    for page in failed_pages:
+                                        logging.warning(f"[TCADP]   Page number: {page.get('PageNumber')}, Error: {page.get('ErrorMsg')}")
+                                
+                                # Check if there is a download link
+                                download_url = data_dict.get("DocumentRecognizeResultUrl")
+                                if download_url:
+                                    logging.info(f"[TCADP] Got download link: {download_url}")
+                                else:
+                                    logging.warning("[TCADP] No download link obtained")
+                                
+                                break  # Found final result, exit loop
+                            else:
+                                # Print progress information
+                                progress = data_dict.get("Progress", "0")
+                                logging.info(f"[TCADP] Progress: {progress}%")
+                        except json.JSONDecodeError as e:
+                            logging.error(f"[TCADP] Failed to parse JSON data: {e}")
+                            logging.error(f"[TCADP] Raw data: {event.get('data')}")
+                            continue
+                    else:
+                        logging.info(f"[TCADP] Event without data: {event}")
+            else:  # Non-streaming response
+                logging.info("[TCADP] Detected non-streaming response")
+                if hasattr(resp, 'data') and resp.data:
+                    try:
+                        data_dict = json.loads(resp.data)
+                        parser_result = data_dict
+                        logging.info(f"[TCADP] JSON parsing successful: {parser_result}")
+                    except json.JSONDecodeError as e:
+                        logging.error(f"[TCADP] JSON parsing failed: {e}")
+                        return None
+                else:
+                    logging.error("[TCADP] No data in response")
+                    return None
+
+            return parser_result
+
+        except TencentCloudSDKException as err:
+            logging.error(f"[TCADP] Tencent Cloud SDK error: {err}")
+            return None
+        except Exception as e:
+            logging.error(f"[TCADP] Unknown error: {e}")
+            logging.error(f"[TCADP] Error stack trace: {traceback.format_exc()}")
+            return None
+
+    def download_result_file(self, download_url, output_dir):
+        """Download parsing result file"""
+        if not download_url:
+            logging.warning("[TCADP] No downloadable result file")
+            return None
+
+        try:
+            response = requests.get(download_url)
+            response.raise_for_status()
+
+            # Ensure output directory exists
+            os.makedirs(output_dir, exist_ok=True)
+
+            # Generate filename
+            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+            filename = f"tcadp_result_{timestamp}.zip"
+            file_path = os.path.join(output_dir, filename)
+
+            # Save file
+            with open(file_path, "wb") as f:
+                f.write(response.content)
+
+            logging.info(f"[TCADP] Document parsing result downloaded to: {os.path.basename(file_path)}")
+            return file_path
+
+        except requests.exceptions.RequestException as e:
+            logging.error(f"[TCADP] Failed to download file: {e}")
+            return None
+
+
+class TCADPParser(RAGFlowPdfParser):
+    def __init__(self, secret_id: str = None, secret_key: str = None, region: str = "ap-guangzhou"):
+        super().__init__()
+        
+        # First initialize logger
+        self.logger = logging.getLogger(self.__class__.__name__)
+        
+        # Priority: read configuration from RAGFlow configuration system (service_conf.yaml)
+        try:
+            tcadp_parser = get_base_config("tcadp_config", {})
+            if isinstance(tcadp_parser, dict) and tcadp_parser:
+                self.secret_id = secret_id or tcadp_parser.get("secret_id")
+                self.secret_key = secret_key or tcadp_parser.get("secret_key")
+                self.region = region or tcadp_parser.get("region", "ap-guangzhou")
+                self.table_result_type = tcadp_parser.get("table_result_type", "1")
+                self.markdown_image_response_type = tcadp_parser.get("markdown_image_response_type", "1")
+                self.logger.info("[TCADP] Configuration read from service_conf.yaml")
+            else:
+                self.logger.error("[TCADP] Please configure tcadp_config in service_conf.yaml first")
+
+        except ImportError:
+            self.logger.info("[TCADP] Configuration module import failed")
+
+        if not self.secret_id or not self.secret_key:
+            raise ValueError("[TCADP] Please set Tencent Cloud API keys, configure tcadp_config in service_conf.yaml")
+
+    def check_installation(self) -> bool:
+        """Check if Tencent Cloud API configuration is correct"""
+        try:
+            # Check necessary configuration parameters
+            if not self.secret_id or not self.secret_key:
+                self.logger.error("[TCADP] Tencent Cloud API configuration incomplete")
+                return False
+
+            # Try to create client to verify configuration
+            TencentCloudAPIClient(self.secret_id, self.secret_key, self.region)
+            self.logger.info("[TCADP] Tencent Cloud API configuration check passed")
+            return True
+        except Exception as e:
+            self.logger.error(f"[TCADP] Tencent Cloud API configuration check failed: {e}")
+            return False
+
+    def _file_to_base64(self, file_path: str, binary: bytes = None) -> str:
+        """Convert file to Base64 format"""
+        
+        if binary:
+            # If binary data is directly available, convert directly
+            return base64.b64encode(binary).decode('utf-8')
+        else:
+            # Read from file path and convert
+            with open(file_path, 'rb') as f:
+                file_data = f.read()
+                return base64.b64encode(file_data).decode('utf-8')
+
+    def _extract_content_from_zip(self, zip_path: str) -> list[dict[str, Any]]:
+        """Extract parsing results from downloaded ZIP file"""
+        results = []
+
+        try:
+            with zipfile.ZipFile(zip_path, "r") as zip_file:
+                # Find JSON result files
+                json_files = [f for f in zip_file.namelist() if f.endswith(".json")]
+
+                for json_file in json_files:
+                    with zip_file.open(json_file) as f:
+                        data = json.load(f)
+                        if isinstance(data, list):
+                            results.extend(data)
+                        else:
+                            results.append(data)
+
+                # Find Markdown files
+                md_files = [f for f in zip_file.namelist() if f.endswith(".md")]
+                for md_file in md_files:
+                    with zip_file.open(md_file) as f:
+                        content = f.read().decode("utf-8")
+                        results.append({"type": "text", "content": content, "file": md_file})
+
+        except Exception as e:
+            self.logger.error(f"[TCADP] Failed to extract ZIP file content: {e}")
+
+        return results
+
+    def _parse_content_to_sections(self, content_data: list[dict[str, Any]]) -> list[tuple[str, str]]:
+        """Convert parsing results to sections format"""
+        sections = []
+
+        for item in content_data:
+            content_type = item.get("type", "text")
+            content = item.get("content", "")
+
+            if not content:
+                continue
+
+            # Process based on content type
+            if content_type == "text" or content_type == "paragraph":
+                section_text = content
+            elif content_type == "table":
+                # Handle table content
+                table_data = item.get("table_data", {})
+                if isinstance(table_data, dict):
+                    # Convert table data to text
+                    rows = table_data.get("rows", [])
+                    section_text = "\n".join([" | ".join(row) for row in rows])
+                else:
+                    section_text = str(table_data)
+            elif content_type == "image":
+                # Handle image content
+                caption = item.get("caption", "")
+                section_text = f"[Image] {caption}" if caption else "[Image]"
+            elif content_type == "equation":
+                # Handle equation content
+                section_text = f"$${content}$$"
+            else:
+                section_text = content
+
+            if section_text.strip():
+                # Generate position tag (simplified version)
+                position_tag = "@@1\t0.0\t1000.0\t0.0\t100.0##"
+                sections.append((section_text, position_tag))
+
+        return sections
+
+    def _parse_content_to_tables(self, content_data: list[dict[str, Any]]) -> list:
+        """Convert parsing results to tables format"""
+        tables = []
+
+        for item in content_data:
+            if item.get("type") == "table":
+                table_data = item.get("table_data", {})
+                if isinstance(table_data, dict):
+                    rows = table_data.get("rows", [])
+                    if rows:
+                        # Convert to table format
+                        table_html = "<table>\n"
+                        for i, row in enumerate(rows):
+                            table_html += "  <tr>\n"
+                            for cell in row:
+                                tag = "th" if i == 0 else "td"
+                                table_html += f"    <{tag}>{cell}</{tag}>\n"
+                            table_html += "  </tr>\n"
+                        table_html += "</table>"
+                        tables.append(table_html)
+
+        return tables
+
+    def parse_pdf(
+        self,
+        filepath: str | PathLike[str],
+        binary: BytesIO | bytes,
+        callback: Optional[Callable] = None,
+        *,
+        output_dir: Optional[str] = None,
+        file_type: str = "PDF",
+        file_start_page: Optional[int] = 1,
+        file_end_page: Optional[int] = 1000,
+        delete_output: Optional[bool] = True,
+        max_retries: Optional[int] = 1,
+    ) -> tuple:
+        """Parse PDF document"""
+
+        temp_file = None
+        created_tmp_dir = False
+
+        try:
+            # Handle input file
+            if binary:
+                temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".pdf")
+                temp_file.write(binary)
+                temp_file.close()
+                file_path = temp_file.name
+                self.logger.info(f"[TCADP] Received binary PDF -> {os.path.basename(file_path)}")
+                if callback:
+                    callback(0.1, f"[TCADP] Received binary PDF -> {os.path.basename(file_path)}")
+            else:
+                file_path = str(filepath)
+                if not os.path.exists(file_path):
+                    if callback:
+                        callback(-1, f"[TCADP] PDF file does not exist: {file_path}")
+                    raise FileNotFoundError(f"[TCADP] PDF file does not exist: {file_path}")
+
+            # Convert file to Base64 format
+            if callback:
+                callback(0.2, "[TCADP] Converting file to Base64 format")
+            
+            file_base64 = self._file_to_base64(file_path, binary)
+            if callback:
+                callback(0.25, f"[TCADP] File converted to Base64, size: {len(file_base64)} characters")
+
+            # Create Tencent Cloud API client
+            client = TencentCloudAPIClient(self.secret_id, self.secret_key, self.region)
+
+            # Call document parsing API (with retry mechanism)
+            if callback:
+                callback(0.3, "[TCADP] Starting to call Tencent Cloud document parsing API")
+
+            result = None
+            for attempt in range(max_retries):
+                try:
+                    if attempt > 0:
+                        self.logger.info(f"[TCADP] Retry attempt {attempt + 1}")
+                        if callback:
+                            callback(0.3 + attempt * 0.1, f"[TCADP] Retry attempt {attempt + 1}")
+                        time.sleep(2 ** attempt)  # Exponential backoff
+
+                    config = {
+                        "TableResultType": self.table_result_type,
+                        "MarkdownImageResponseType": self.markdown_image_response_type
+                    }
+
+                    result = client.reconstruct_document_sse(
+                        file_type=file_type, 
+                        file_base64=file_base64, 
+                        file_start_page=file_start_page, 
+                        file_end_page=file_end_page, 
+                        config=config
+                    )
+                    
+                    if result:
+                        self.logger.info(f"[TCADP] Attempt {attempt + 1} successful")
+                        break
+                    else:
+                        self.logger.warning(f"[TCADP] Attempt {attempt + 1} failed, result is None")
+                        
+                except Exception as e:
+                    self.logger.error(f"[TCADP] Attempt {attempt + 1} exception: {e}")
+                    if attempt == max_retries - 1:
+                        raise
+
+            if not result:
+                error_msg = f"[TCADP] Document parsing failed, retried {max_retries} times"
+                self.logger.error(error_msg)
+                if callback:
+                    callback(-1, error_msg)
+                raise RuntimeError(error_msg)
+
+            # Get download link
+            download_url = result.get("DocumentRecognizeResultUrl")
+            if not download_url:
+                if callback:
+                    callback(-1, "[TCADP] No parsing result download link obtained")
+                raise RuntimeError("[TCADP] No parsing result download link obtained")
+
+            if callback:
+                callback(0.6, f"[TCADP] Parsing result download link: {download_url}")
+
+            # Set output directory
+            if output_dir:
+                out_dir = Path(output_dir)
+                out_dir.mkdir(parents=True, exist_ok=True)
+            else:
+                out_dir = Path(tempfile.mkdtemp(prefix="adp_pdf_"))
+                created_tmp_dir = True
+
+            # Download result file
+            zip_path = client.download_result_file(download_url, str(out_dir))
+            if not zip_path:
+                if callback:
+                    callback(-1, "[TCADP] Failed to download parsing result")
+                raise RuntimeError("[TCADP] Failed to download parsing result")
+
+            if callback:
+                # Shorten file path display, only show filename
+                zip_filename = os.path.basename(zip_path)
+                callback(0.8, f"[TCADP] Parsing result downloaded: {zip_filename}")
+
+            # Extract ZIP file content
+            content_data = self._extract_content_from_zip(zip_path)
+            self.logger.info(f"[TCADP] Extracted {len(content_data)} content blocks")
+
+            if callback:
+                callback(0.9, f"[TCADP] Extracted {len(content_data)} content blocks")
+
+            # Convert to sections and tables format
+            sections = self._parse_content_to_sections(content_data)
+            tables = self._parse_content_to_tables(content_data)
+
+            self.logger.info(f"[TCADP] Parsing completed: {len(sections)} sections, {len(tables)} tables")
+
+            if callback:
+                callback(1.0, f"[TCADP] Parsing completed: {len(sections)} sections, {len(tables)} tables")
+
+            return sections, tables
+
+        finally:
+            # Clean up temporary files
+            if temp_file and os.path.exists(temp_file.name):
+                try:
+                    os.unlink(temp_file.name)
+                except Exception:
+                    pass
+
+            if delete_output and created_tmp_dir and out_dir.exists():
+                try:
+                    shutil.rmtree(out_dir)
+                except Exception:
+                    pass
+
+
+if __name__ == "__main__":
+    # Test ADP parser
+    parser = TCADPParser()
+    print("ADP available:", parser.check_installation())
+
+    # Test parsing
+    filepath = ""
+    if filepath and os.path.exists(filepath):
+        with open(filepath, "rb") as file:
+            sections, tables = parser.parse_pdf(filepath=filepath, binary=file.read())
+            print(f"Parsing result: {len(sections)} sections, {len(tables)} tables")
+            for i, (section, tag) in enumerate(sections[:3]):  # Only print first 3
+                print(f"Section {i + 1}: {section[:100]}...")

deepdoc/vision/ocr.py

@@ -22,6 +22,7 @@ import os
 from huggingface_hub import snapshot_download
 
 from api.utils.file_utils import get_project_base_directory
+from api.utils.common import pip_install_torch
 from rag.settings import PARALLEL_DEVICES
 from .operators import *  # noqa: F403
 from . import operators
@@ -83,6 +84,7 @@ def load_model(model_dir, nm, device_id: int | None = None):
 
     def cuda_is_available():
         try:
+            pip_install_torch()
             import torch
             target_id = 0 if device_id is None else device_id
             if torch.cuda.is_available() and torch.cuda.device_count() > target_id:

deepdoc/vision/table_structure_recognizer.py

@@ -57,10 +57,10 @@ class TableStructureRecognizer(Recognizer):
             raise RuntimeError("Unsupported table structure recognizer type.")
 
         if table_structure_recognizer_type == "onnx":
-            logging.debug("Using Onnx table structure recognizer", flush=True)
+            logging.debug("Using Onnx table structure recognizer")
             tbls = super().__call__(images, thr)
         else:  # ascend
-            logging.debug("Using Ascend table structure recognizer", flush=True)
+            logging.debug("Using Ascend table structure recognizer")
             tbls = self._run_ascend_tsr(images, thr)
 
         res = []