Update version number for poetry (#3637)

### What problem does this PR solve?



### Type of change

- [x] Refactoring

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -15,16 +15,16 @@ body:
     value: "Please provide the following information to help us understand the issue."
 - type: input
   attributes:
-    label: Branch name
-    description: Enter the name of the branch where you encountered the issue.
-    placeholder: e.g., main
+    label: RAGFlow workspace code commit ID
+    description: Enter the commit ID associated with the issue.
+    placeholder: e.g., 26d3480e
   validations:
     required: true
 - type: input
   attributes:
-    label: Commit ID
-    description: Enter the commit ID associated with the issue.
-    placeholder: e.g., c3b2a1
+    label: RAGFlow image version
+    description: Enter the image version(shown in RAGFlow UI, `System` page) associated with the issue.
+    placeholder: e.g., 26d3480e(v0.13.0~174)
   validations:
     required: true
 - type: textarea

.github/workflows/tests.yml

@@ -0,0 +1,130 @@
+name: tests
+
+on:
+  push:
+    branches:
+      - 'main'
+      - '*.*.*'
+    paths-ignore:
+      - 'docs/**'
+      - '*.md'
+      - '*.mdx'
+  pull_request:
+    types: [ opened, synchronize, reopened, labeled ]
+    paths-ignore:
+      - 'docs/**'
+      - '*.md'
+      - '*.mdx'
+
+# https://docs.github.com/en/actions/using-jobs/using-concurrency
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  ragflow_tests:
+    name: ragflow_tests
+    # https://docs.github.com/en/actions/using-jobs/using-conditions-to-control-job-execution
+    # https://github.com/orgs/community/discussions/26261
+    if: ${{ github.event_name != 'pull_request' || contains(github.event.pull_request.labels.*.name, 'ci') }}
+    runs-on: [ "self-hosted", "debug" ]
+    steps:
+      # https://github.com/hmarr/debug-action
+      #- uses: hmarr/debug-action@v2
+
+      - name: Show PR labels
+        run: |
+          echo "Workflow triggered by ${{ github.event_name }}"
+          if [[ ${{ github.event_name }} == 'pull_request' ]]; then
+            echo "PR labels: ${{ join(github.event.pull_request.labels.*.name, ', ') }}"
+          fi
+
+      - name: Ensure workspace ownership
+        run: echo "chown -R $USER $GITHUB_WORKSPACE" && sudo chown -R $USER $GITHUB_WORKSPACE
+
+      # https://github.com/actions/checkout/issues/1781
+      - name: Check out code
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          fetch-tags: true
+
+      - name: Build ragflow:dev-slim
+        run: |
+          RUNNER_WORKSPACE_PREFIX=${RUNNER_WORKSPACE_PREFIX:-$HOME}
+          cp -r ${RUNNER_WORKSPACE_PREFIX}/huggingface.co ${RUNNER_WORKSPACE_PREFIX}/nltk_data ${RUNNER_WORKSPACE_PREFIX}/libssl*.deb  ${RUNNER_WORKSPACE_PREFIX}/tika-server*.jar* ${RUNNER_WORKSPACE_PREFIX}/chrome* ${RUNNER_WORKSPACE_PREFIX}/cl100k_base.tiktoken .
+          sudo docker pull ubuntu:22.04
+          sudo docker build --progress=plain -f Dockerfile.slim -t infiniflow/ragflow:dev-slim .
+
+      - name: Build ragflow:dev
+        run: |
+          sudo docker build --progress=plain -f Dockerfile -t infiniflow/ragflow:dev .
+
+      - name: Start ragflow:dev-slim
+        run: |
+          sudo docker compose -f docker/docker-compose.yml up -d
+
+      - name: Stop ragflow:dev-slim
+        if: always()  # always run this step even if previous steps failed
+        run: |
+          sudo docker compose -f docker/docker-compose.yml down -v
+
+      - name: Start ragflow:dev
+        run: |
+          echo "RAGFLOW_IMAGE=infiniflow/ragflow:dev" >> docker/.env
+          sudo docker compose -f docker/docker-compose.yml up -d
+
+      - 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
+            echo "Waiting for service to be available..."
+            sleep 5
+          done
+          cd sdk/python && poetry install && source .venv/bin/activate && cd test/test_sdk_api && pytest -s --tb=short get_email.py t_dataset.py t_chat.py t_session.py t_document.py t_chunk.py
+
+      - 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
+            echo "Waiting for service to be available..."
+            sleep 5
+          done
+          cd sdk/python && poetry install && source .venv/bin/activate && cd test/test_frontend_api && pytest -s --tb=short get_email.py test_dataset.py 
+          
+
+      - name: Stop ragflow:dev
+        if: always()  # always run this step even if previous steps failed
+        run: |
+          sudo docker compose -f docker/docker-compose.yml down -v
+
+      - name: Start ragflow:dev
+        run: |
+          sudo DOC_ENGINE=infinity docker compose -f docker/docker-compose.yml 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
+            echo "Waiting for service to be available..."
+            sleep 5
+          done
+          cd sdk/python && poetry install && source .venv/bin/activate && cd test/test_sdk_api && pytest -s --tb=short get_email.py t_dataset.py t_chat.py t_session.py t_document.py t_chunk.py
+
+      - 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
+            echo "Waiting for service to be available..."
+            sleep 5
+          done
+          cd sdk/python && poetry install && source .venv/bin/activate && cd test/test_frontend_api && pytest -s --tb=short get_email.py test_dataset.py
+
+      - name: Stop ragflow:dev
+        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

CONTRIBUTING.md

@@ -1,16 +1,10 @@
----
-sidebar_position: 0
-slug: /contribution_guidelines
----
-
 # Contribution guidelines
 
-Thanks for wanting to contribute to RAGFlow. This document offers guidlines and major considerations for submitting your contributions. 
+This document offers guidlines and major considerations for submitting your contributions to RAGFlow.
 
 - To report a bug, file a [GitHub issue](https://github.com/infiniflow/ragflow/issues/new/choose) with us.
 - For further questions, you can explore existing discussions or initiate a new one in [Discussions](https://github.com/orgs/infiniflow/discussions).
 
-
 ## What you can contribute
 
 The list below mentions some contributions you can make, but it is not a complete list.
@@ -27,7 +21,7 @@ The list below mentions some contributions you can make, but it is not a complet
 ### General workflow
 
 1. Fork our GitHub repository.
-2. Clone your fork to your local machine: 
+2. Clone your fork to your local machine:
 `git clone git@github.com:<yourname>/ragflow.git`
 3. Create a local branch: 
 `git checkout -b my-branch`
@@ -39,14 +33,16 @@ The list below mentions some contributions you can make, but it is not a complet
 
 ### Before filing a PR
 
-- Consider splitting a large PR into multiple smaller, standalone PRs to keep a traceable development history. 
+- Consider splitting a large PR into multiple smaller, standalone PRs to keep a traceable development history.
 - Ensure that your PR addresses just one issue, or keep any unrelated changes small.
 - Add test cases when contributing new features. They demonstrate that your code functions correctly and protect against potential issues from future changes.
-### Describing your PR 
+
+### Describing your PR
 
 - Ensure that your PR title is concise and clear, providing all the required information.
-- Refer to a corresponding GitHub issue in your PR description if applicable. 
+- Refer to a corresponding GitHub issue in your PR description if applicable.
 - Include sufficient design details for *breaking changes* or *API changes* in your description.
 
 ### Reviewing & merging a PR
-- Ensure that your PR passes all Continuous Integration (CI) tests before merging it.
+
+Ensure that your PR passes all Continuous Integration (CI) tests before merging it.

Dockerfile

@@ -1,23 +1,170 @@
-FROM infiniflow/ragflow-base:v2.0
-USER  root
+# base stage
+FROM ubuntu:22.04 AS base
+USER root
+SHELL ["/bin/bash", "-c"]
+
+ENV LIGHTEN=0
 
 WORKDIR /ragflow
 
-ADD ./web ./web
-RUN cd ./web && npm i --force && npm run build
+RUN rm -f /etc/apt/apt.conf.d/docker-clean \
+    && echo 'Binary::apt::APT::Keep-Downloaded-Packages "true";' > /etc/apt/apt.conf.d/keep-cache
 
-ADD ./api ./api
-ADD ./conf ./conf
-ADD ./deepdoc ./deepdoc
-ADD ./rag ./rag
-ADD ./agent ./agent
-ADD ./graphrag ./graphrag
+RUN --mount=type=cache,id=ragflow_base_apt,target=/var/cache/apt,sharing=locked \
+    apt update && apt-get --no-install-recommends install -y ca-certificates
+
+# Setup apt mirror site
+RUN sed -i 's|http://archive.ubuntu.com|https://mirrors.tuna.tsinghua.edu.cn|g' /etc/apt/sources.list
+
+RUN --mount=type=cache,id=ragflow_base_apt,target=/var/cache/apt,sharing=locked \
+    apt update && DEBIAN_FRONTEND=noninteractive apt install -y curl libpython3-dev nginx libglib2.0-0 libglx-mesa0 pkg-config libicu-dev libgdiplus default-jdk python3-pip pipx \
+    libatk-bridge2.0-0 libgtk-4-1 libnss3 xdg-utils unzip libgbm-dev wget git \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN pip3 config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple && pip3 config set global.trusted-host "pypi.tuna.tsinghua.edu.cn mirrors.pku.edu.cn" && pip3 config set global.extra-index-url "https://mirrors.pku.edu.cn/pypi/web/simple" \
+    && pipx install poetry \
+    && /root/.local/bin/poetry self add poetry-plugin-pypi-mirror
+
+# https://forum.aspose.com/t/aspose-slides-for-net-no-usable-version-of-libssl-found-with-linux-server/271344/13
+# aspose-slides on linux/arm64 is unavailable
+RUN --mount=type=bind,source=libssl1.1_1.1.1f-1ubuntu2_amd64.deb,target=/root/libssl1.1_1.1.1f-1ubuntu2_amd64.deb \
+    --mount=type=bind,source=libssl1.1_1.1.1f-1ubuntu2_arm64.deb,target=/root/libssl1.1_1.1.1f-1ubuntu2_arm64.deb \
+    if [ "$(uname -m)" = "x86_64" ]; then \
+        dpkg -i /root/libssl1.1_1.1.1f-1ubuntu2_amd64.deb; \
+    elif [ "$(uname -m)" = "aarch64" ]; then \
+        dpkg -i /root/libssl1.1_1.1.1f-1ubuntu2_arm64.deb; \
+    fi
+
+ENV PYTHONDONTWRITEBYTECODE=1 DOTNET_SYSTEM_GLOBALIZATION_INVARIANT=1
+ENV PATH=/root/.local/bin:$PATH
+# Configure Poetry
+ENV POETRY_NO_INTERACTION=1
+ENV POETRY_VIRTUALENVS_IN_PROJECT=true
+ENV POETRY_VIRTUALENVS_CREATE=true
+ENV POETRY_REQUESTS_TIMEOUT=15
+ENV POETRY_PYPI_MIRROR_URL=https://pypi.tuna.tsinghua.edu.cn/simple/
+
+# nodejs 12.22 on Ubuntu 22.04 is too old
+RUN --mount=type=cache,id=ragflow_base_apt,target=/var/cache/apt,sharing=locked \
+    curl -fsSL https://deb.nodesource.com/setup_20.x | bash - && \
+    apt purge -y nodejs npm && \
+    apt autoremove && \
+    apt update && \
+    apt install -y nodejs cargo && \
+    rm -rf /var/lib/apt/lists/*
+
+# builder stage
+FROM base AS builder
+USER root
+
+WORKDIR /ragflow
+
+COPY .git /ragflow/.git
+
+RUN current_commit=$(git rev-parse --short HEAD); \
+    last_tag=$(git describe --tags --abbrev=0); \
+    commit_count=$(git rev-list --count "$last_tag..HEAD"); \
+    version_info=""; \
+    if [ "$commit_count" -eq 0 ]; then \
+        version_info=$last_tag; \
+    else \
+        version_info="$current_commit($last_tag~$commit_count)"; \
+    fi; \
+    if [ "$LIGHTEN" == "1" ]; then \
+        version_info="$version_info slim"; \
+    else \
+        version_info="$version_info full"; \
+    fi; \
+    echo "RAGFlow version: $version_info"; \
+    echo $version_info > /ragflow/VERSION
+
+COPY web web
+COPY docs docs
+RUN --mount=type=cache,id=ragflow_builder_npm,target=/root/.npm,sharing=locked \
+    cd web && npm install --force && npm run build
+
+# install dependencies from poetry.lock file
+COPY pyproject.toml poetry.toml poetry.lock ./
+
+RUN --mount=type=cache,id=ragflow_builder_poetry,target=/root/.cache/pypoetry,sharing=locked \
+    if [ "$LIGHTEN" == "1" ]; then \
+        poetry install --no-root; \
+    else \
+        poetry install --no-root --with=full; \
+    fi
+
+# production stage
+FROM base AS production
+USER root
+
+WORKDIR /ragflow
+
+COPY --from=builder /ragflow/VERSION /ragflow/VERSION
+
+# Install python packages' dependencies
+# cv2 requires libGL.so.1
+RUN --mount=type=cache,id=ragflow_production_apt,target=/var/cache/apt,sharing=locked \
+    apt update && apt install -y --no-install-recommends nginx libgl1 vim less && \
+    rm -rf /var/lib/apt/lists/*
+
+COPY web web
+COPY api api
+COPY conf conf
+COPY deepdoc deepdoc
+COPY rag rag
+COPY agent agent
+COPY graphrag graphrag
+COPY pyproject.toml poetry.toml poetry.lock ./
+
+# Copy models downloaded via download_deps.py
+RUN mkdir -p /ragflow/rag/res/deepdoc /root/.ragflow
+RUN --mount=type=bind,source=huggingface.co,target=/huggingface.co \
+    tar --exclude='.*' -cf - \
+        /huggingface.co/InfiniFlow/text_concat_xgb_v1.0 \
+        /huggingface.co/InfiniFlow/deepdoc \
+        | tar -xf - --strip-components=3 -C /ragflow/rag/res/deepdoc
+RUN --mount=type=bind,source=huggingface.co,target=/huggingface.co \
+    tar -cf - \
+        /huggingface.co/BAAI/bge-large-zh-v1.5 \
+        /huggingface.co/BAAI/bge-reranker-v2-m3 \
+        /huggingface.co/maidalun1020/bce-embedding-base_v1 \
+        /huggingface.co/maidalun1020/bce-reranker-base_v1 \
+        | tar -xf - --strip-components=2 -C /root/.ragflow
+
+# Copy nltk data downloaded via download_deps.py
+COPY nltk_data /root/nltk_data
+
+# 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.
+COPY tika-server-standard-3.0.0.jar /ragflow/tika-server-standard.jar
+COPY tika-server-standard-3.0.0.jar.md5 /ragflow/tika-server-standard.jar.md5
+ENV TIKA_SERVER_JAR="file:///ragflow/tika-server-standard.jar"
+
+# Copy cl100k_base
+COPY cl100k_base.tiktoken /ragflow/9b5ad71b2ce5302211f9c61530b329a4922fc6a4
+
+# Add dependencies of selenium
+RUN --mount=type=bind,source=chrome-linux64-121-0-6167-85,target=/chrome-linux64.zip \
+    unzip /chrome-linux64.zip && \
+    mv chrome-linux64 /opt/chrome && \
+    ln -s /opt/chrome/chrome /usr/local/bin/
+RUN --mount=type=bind,source=chromedriver-linux64-121-0-6167-85,target=/chromedriver-linux64.zip \
+    unzip -j /chromedriver-linux64.zip chromedriver-linux64/chromedriver && \
+    mv chromedriver /usr/local/bin/ && \
+    rm -f /usr/bin/google-chrome
+
+# Copy compiled web pages
+COPY --from=builder /ragflow/web/dist /ragflow/web/dist
+
+# Copy Python environment and packages
+ENV VIRTUAL_ENV=/ragflow/.venv
+COPY --from=builder ${VIRTUAL_ENV} ${VIRTUAL_ENV}
+ENV PATH="${VIRTUAL_ENV}/bin:${PATH}"
 
 ENV PYTHONPATH=/ragflow/
-ENV HF_ENDPOINT=https://hf-mirror.com
 
-ADD docker/entrypoint.sh ./entrypoint.sh
-ADD docker/.env ./
+COPY docker/service_conf.yaml.template ./conf/service_conf.yaml.template
+COPY docker/entrypoint.sh ./entrypoint.sh
 RUN chmod +x ./entrypoint.sh
 
-ENTRYPOINT ["./entrypoint.sh"]
+ENTRYPOINT ["./entrypoint.sh"]

Dockerfile.arm

@@ -1,34 +0,0 @@
-FROM python:3.11
-USER  root
-
-WORKDIR /ragflow
-
-COPY requirements_arm.txt /ragflow/requirements.txt
-RUN pip install -i https://mirrors.aliyun.com/pypi/simple/ --default-timeout=1000 -r requirements.txt &&\
-python -c "import nltk;nltk.download('punkt');nltk.download('wordnet')"
-
-RUN apt-get update && \
-    apt-get install -y curl gnupg && \
-    rm -rf /var/lib/apt/lists/*
-
-RUN curl -sL https://deb.nodesource.com/setup_20.x | bash - && \
-    apt-get install -y --fix-missing nodejs nginx ffmpeg libsm6 libxext6 libgl1
-
-ADD ./web ./web
-RUN cd ./web && npm i --force && npm run build
-
-ADD ./api ./api
-ADD ./conf ./conf
-ADD ./deepdoc ./deepdoc
-ADD ./rag ./rag
-ADD ./agent ./agent
-ADD ./graphrag ./graphrag
-
-ENV PYTHONPATH=/ragflow/
-ENV HF_ENDPOINT=https://hf-mirror.com
-
-ADD docker/entrypoint.sh ./entrypoint.sh
-ADD docker/.env ./
-RUN chmod +x ./entrypoint.sh
-
-ENTRYPOINT ["./entrypoint.sh"]

Dockerfile.cuda

@@ -1,27 +0,0 @@
-FROM infiniflow/ragflow-base:v2.0
-USER  root
-
-WORKDIR /ragflow
-
-## for cuda > 12.0
-RUN pip uninstall -y onnxruntime-gpu
-RUN pip install onnxruntime-gpu --extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
-
-
-ADD ./web ./web
-RUN cd ./web && npm i --force && npm run build
-
-ADD ./api ./api
-ADD ./conf ./conf
-ADD ./deepdoc ./deepdoc
-ADD ./rag ./rag
-ADD ./agent ./agent
-ADD ./graphrag ./graphrag
-
-ENV PYTHONPATH=/ragflow/
-ENV HF_ENDPOINT=https://hf-mirror.com
-
-ADD docker/entrypoint.sh ./entrypoint.sh
-RUN chmod +x ./entrypoint.sh
-
-ENTRYPOINT ["./entrypoint.sh"]

Dockerfile.scratch

@@ -1,56 +0,0 @@
-FROM ubuntu:22.04
-USER root
-
-WORKDIR /ragflow
-
-RUN apt-get update && apt-get install -y wget curl build-essential libopenmpi-dev
-
-RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda.sh && \
-    bash ~/miniconda.sh -b -p /root/miniconda3 && \
-    rm ~/miniconda.sh && ln -s /root/miniconda3/etc/profile.d/conda.sh /etc/profile.d/conda.sh && \
-    echo ". /root/miniconda3/etc/profile.d/conda.sh" >> ~/.bashrc && \
-    echo "conda activate base" >> ~/.bashrc
-
-ENV PATH /root/miniconda3/bin:$PATH
-
-RUN conda create -y --name py11 python=3.11
-
-ENV CONDA_DEFAULT_ENV py11
-ENV CONDA_PREFIX /root/miniconda3/envs/py11
-ENV PATH $CONDA_PREFIX/bin:$PATH
-
-RUN curl -sL https://deb.nodesource.com/setup_14.x | bash -
-RUN apt-get install -y nodejs
-
-RUN apt-get install -y nginx
-
-ADD ./web ./web
-ADD ./api ./api
-ADD ./conf ./conf
-ADD ./deepdoc ./deepdoc
-ADD ./rag ./rag
-ADD ./requirements.txt ./requirements.txt
-ADD ./agent ./agent
-ADD ./graphrag ./graphrag
-
-RUN apt install openmpi-bin openmpi-common libopenmpi-dev
-ENV LD_LIBRARY_PATH /usr/lib/x86_64-linux-gnu/openmpi/lib:$LD_LIBRARY_PATH
-RUN rm /root/miniconda3/envs/py11/compiler_compat/ld
-RUN cd ./web && npm i --force && npm run build
-RUN conda run -n py11 pip install -i https://mirrors.aliyun.com/pypi/simple/ -r ./requirements.txt
-
-RUN apt-get update && \
-    apt-get install -y libglib2.0-0 libgl1-mesa-glx && \
-    rm -rf /var/lib/apt/lists/*
-
-RUN conda run -n py11 pip install -i https://mirrors.aliyun.com/pypi/simple/ ollama
-RUN conda run -n py11 python -m nltk.downloader punkt
-RUN conda run -n py11 python -m nltk.downloader wordnet
-
-ENV PYTHONPATH=/ragflow/
-ENV HF_ENDPOINT=https://hf-mirror.com
-
-ADD docker/entrypoint.sh ./entrypoint.sh
-RUN chmod +x ./entrypoint.sh
-
-ENTRYPOINT ["./entrypoint.sh"]

Dockerfile.scratch.oc9

@@ -26,6 +26,7 @@ RUN dnf install -y nginx
 
 ADD ./web ./web
 ADD ./api ./api
+ADD ./docs ./docs
 ADD ./conf ./conf
 ADD ./deepdoc ./deepdoc
 ADD ./rag ./rag
@@ -37,7 +38,7 @@ RUN dnf install -y openmpi openmpi-devel python3-openmpi
 ENV C_INCLUDE_PATH /usr/include/openmpi-x86_64:$C_INCLUDE_PATH
 ENV LD_LIBRARY_PATH /usr/lib64/openmpi/lib:$LD_LIBRARY_PATH
 RUN rm /root/miniconda3/envs/py11/compiler_compat/ld
-RUN cd ./web && npm i --force && npm run build
+RUN cd ./web && npm i && npm run build
 RUN conda run -n py11 pip install $(grep -ivE "mpi4py" ./requirements.txt) # without mpi4py==3.1.5
 RUN conda run -n py11 pip install redis
 
@@ -52,6 +53,7 @@ RUN conda run -n py11 python -m nltk.downloader wordnet
 ENV PYTHONPATH=/ragflow/
 ENV HF_ENDPOINT=https://hf-mirror.com
 
+COPY docker/service_conf.yaml.template ./conf/service_conf.yaml.template
 ADD docker/entrypoint.sh ./entrypoint.sh
 RUN chmod +x ./entrypoint.sh
 

Dockerfile.slim

@@ -0,0 +1,163 @@
+# base stage
+FROM ubuntu:22.04 AS base
+USER root
+SHELL ["/bin/bash", "-c"]
+
+ENV LIGHTEN=1
+
+WORKDIR /ragflow
+
+RUN rm -f /etc/apt/apt.conf.d/docker-clean \
+    && echo 'Binary::apt::APT::Keep-Downloaded-Packages "true";' > /etc/apt/apt.conf.d/keep-cache
+
+RUN --mount=type=cache,id=ragflow_base_apt,target=/var/cache/apt,sharing=locked \
+    apt update && apt-get --no-install-recommends install -y ca-certificates
+
+# Setup apt mirror site
+RUN sed -i 's|http://archive.ubuntu.com|https://mirrors.tuna.tsinghua.edu.cn|g' /etc/apt/sources.list
+
+RUN --mount=type=cache,id=ragflow_base_apt,target=/var/cache/apt,sharing=locked \
+    apt update && DEBIAN_FRONTEND=noninteractive apt install -y curl libpython3-dev nginx libglib2.0-0 libglx-mesa0 pkg-config libicu-dev libgdiplus default-jdk python3-pip pipx \
+    libatk-bridge2.0-0 libgtk-4-1 libnss3 xdg-utils unzip libgbm-dev wget git \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN pip3 config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple && pip3 config set global.trusted-host "pypi.tuna.tsinghua.edu.cn mirrors.pku.edu.cn" && pip3 config set global.extra-index-url "https://mirrors.pku.edu.cn/pypi/web/simple" \
+    && pipx install poetry \
+    && /root/.local/bin/poetry self add poetry-plugin-pypi-mirror
+
+# https://forum.aspose.com/t/aspose-slides-for-net-no-usable-version-of-libssl-found-with-linux-server/271344/13
+# aspose-slides on linux/arm64 is unavailable
+RUN --mount=type=bind,source=libssl1.1_1.1.1f-1ubuntu2_amd64.deb,target=/root/libssl1.1_1.1.1f-1ubuntu2_amd64.deb \
+    --mount=type=bind,source=libssl1.1_1.1.1f-1ubuntu2_arm64.deb,target=/root/libssl1.1_1.1.1f-1ubuntu2_arm64.deb \
+    if [ "$(uname -m)" = "x86_64" ]; then \
+        dpkg -i /root/libssl1.1_1.1.1f-1ubuntu2_amd64.deb; \
+    elif [ "$(uname -m)" = "aarch64" ]; then \
+        dpkg -i /root/libssl1.1_1.1.1f-1ubuntu2_arm64.deb; \
+    fi
+
+ENV PYTHONDONTWRITEBYTECODE=1 DOTNET_SYSTEM_GLOBALIZATION_INVARIANT=1
+ENV PATH=/root/.local/bin:$PATH
+# Configure Poetry
+ENV POETRY_NO_INTERACTION=1
+ENV POETRY_VIRTUALENVS_IN_PROJECT=true
+ENV POETRY_VIRTUALENVS_CREATE=true
+ENV POETRY_REQUESTS_TIMEOUT=15
+ENV POETRY_PYPI_MIRROR_URL=https://pypi.tuna.tsinghua.edu.cn/simple/
+
+# nodejs 12.22 on Ubuntu 22.04 is too old
+RUN --mount=type=cache,id=ragflow_base_apt,target=/var/cache/apt,sharing=locked \
+    curl -fsSL https://deb.nodesource.com/setup_20.x | bash - && \
+    apt purge -y nodejs npm && \
+    apt autoremove && \
+    apt update && \
+    apt install -y nodejs cargo && \
+    rm -rf /var/lib/apt/lists/*
+
+# builder stage
+FROM base AS builder
+USER root
+
+WORKDIR /ragflow
+
+COPY .git /ragflow/.git
+
+RUN current_commit=$(git rev-parse --short HEAD); \
+    last_tag=$(git describe --tags --abbrev=0); \
+    commit_count=$(git rev-list --count "$last_tag..HEAD"); \
+    version_info=""; \
+    if [ "$commit_count" -eq 0 ]; then \
+        version_info=$last_tag; \
+    else \
+        version_info="$current_commit($last_tag~$commit_count)"; \
+    fi; \
+    if [ "$LIGHTEN" == "1" ]; then \
+        version_info="$version_info slim"; \
+    else \
+        version_info="$version_info full"; \
+    fi; \
+    echo "RAGFlow version: $version_info"; \
+    echo $version_info > /ragflow/VERSION
+
+COPY web web
+COPY docs docs
+RUN --mount=type=cache,id=ragflow_builder_npm,target=/root/.npm,sharing=locked \
+    cd web && npm install --force && npm run build
+
+# install dependencies from poetry.lock file
+COPY pyproject.toml poetry.toml poetry.lock ./
+
+RUN --mount=type=cache,id=ragflow_builder_poetry,target=/root/.cache/pypoetry,sharing=locked \
+    if [ "$LIGHTEN" == "1" ]; then \
+        poetry install --no-root; \
+    else \
+        poetry install --no-root --with=full; \
+    fi
+
+# production stage
+FROM base AS production
+USER root
+
+WORKDIR /ragflow
+
+COPY --from=builder /ragflow/VERSION /ragflow/VERSION
+
+# Install python packages' dependencies
+# cv2 requires libGL.so.1
+RUN --mount=type=cache,id=ragflow_production_apt,target=/var/cache/apt,sharing=locked \
+    apt update && apt install -y --no-install-recommends nginx libgl1 vim less && \
+    rm -rf /var/lib/apt/lists/*
+
+COPY web web
+COPY api api
+COPY conf conf
+COPY deepdoc deepdoc
+COPY rag rag
+COPY agent agent
+COPY graphrag graphrag
+COPY pyproject.toml poetry.toml poetry.lock ./
+
+# Copy models downloaded via download_deps.py
+RUN mkdir -p /ragflow/rag/res/deepdoc /root/.ragflow
+RUN --mount=type=bind,source=huggingface.co,target=/huggingface.co \
+    tar --exclude='.*' -cf - \
+        /huggingface.co/InfiniFlow/text_concat_xgb_v1.0 \
+        /huggingface.co/InfiniFlow/deepdoc \
+        | tar -xf - --strip-components=3 -C /ragflow/rag/res/deepdoc
+
+# Copy nltk data downloaded via download_deps.py
+COPY nltk_data /root/nltk_data
+
+# 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.
+COPY tika-server-standard-3.0.0.jar /ragflow/tika-server-standard.jar
+COPY tika-server-standard-3.0.0.jar.md5 /ragflow/tika-server-standard.jar.md5
+ENV TIKA_SERVER_JAR="file:///ragflow/tika-server-standard.jar"
+
+# Copy cl100k_base
+COPY cl100k_base.tiktoken /ragflow/9b5ad71b2ce5302211f9c61530b329a4922fc6a4
+
+# Add dependencies of selenium
+RUN --mount=type=bind,source=chrome-linux64-121-0-6167-85,target=/chrome-linux64.zip \
+    unzip /chrome-linux64.zip && \
+    mv chrome-linux64 /opt/chrome && \
+    ln -s /opt/chrome/chrome /usr/local/bin/
+RUN --mount=type=bind,source=chromedriver-linux64-121-0-6167-85,target=/chromedriver-linux64.zip \
+    unzip -j /chromedriver-linux64.zip chromedriver-linux64/chromedriver && \
+    mv chromedriver /usr/local/bin/ && \
+    rm -f /usr/bin/google-chrome
+
+# Copy compiled web pages
+COPY --from=builder /ragflow/web/dist /ragflow/web/dist
+
+# Copy Python environment and packages
+ENV VIRTUAL_ENV=/ragflow/.venv
+COPY --from=builder ${VIRTUAL_ENV} ${VIRTUAL_ENV}
+ENV PATH="${VIRTUAL_ENV}/bin:${PATH}"
+
+ENV PYTHONPATH=/ragflow/
+
+COPY docker/service_conf.yaml.template ./conf/service_conf.yaml.template
+COPY docker/entrypoint.sh ./entrypoint.sh
+RUN chmod +x ./entrypoint.sh
+
+ENTRYPOINT ["./entrypoint.sh"]

README.md

@@ -8,20 +8,26 @@
   <a href="./README.md">English</a> |
   <a href="./README_zh.md">简体中文</a> |
   <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a>
+  <a href="./README_ko.md">한국어</a> |
+  <a href="./README_id.md">Bahasa Indonesia</a>
 </p>
 
 <p align="center">
+    <a href="https://x.com/intent/follow?screen_name=infiniflowai" target="_blank">
+        <img src="https://img.shields.io/twitter/follow/infiniflow?logo=X&color=%20%23f5f5f5" alt="follow on X(Twitter)">
+    </a>
+    <a href="https://demo.ragflow.io" target="_blank">
+        <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/badge/docker_pull-ragflow:v0.14.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.14.0">
+    </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">
     </a>
-    <a href="https://demo.ragflow.io" target="_blank">
-        <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/badge/docker_pull-ragflow:v0.10.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.10.0"></a>
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
-    <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
-  </a>
+        <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
+    </a>
 </p>
 
 <h4 align="center">
@@ -34,7 +40,7 @@
 
 <details open>
 <summary></b>📕 Table of Contents</b></summary>
-  
+
 - 💡 [What is RAGFlow?](#-what-is-ragflow)
 - 🎮 [Demo](#-demo)
 - 📌 [Latest Updates](#-latest-updates)
@@ -42,8 +48,9 @@
 - 🔎 [System Architecture](#-system-architecture)
 - 🎬 [Get Started](#-get-started)
 - 🔧 [Configurations](#-configurations)
-- 🛠️ [Build from source](#-build-from-source)
-- 🛠️ [Launch service from source](#-launch-service-from-source)
+- 🔧 [Build a docker image without embedding models](#-build-a-docker-image-without-embedding-models)
+- 🔧 [Build a docker image including embedding models](#-build-a-docker-image-including-embedding-models)
+- 🔨 [Launch service from source for development](#-launch-service-from-source-for-development)
 - 📚 [Documentation](#-documentation)
 - 📜 [Roadmap](#-roadmap)
 - 🏄 [Community](#-community)
@@ -53,43 +60,41 @@
 
 ## 💡 What is RAGFlow?
 
-[RAGFlow](https://ragflow.io/) is an open-source RAG (Retrieval-Augmented Generation) engine based on deep document understanding. It offers a streamlined RAG workflow for businesses of any scale, combining LLM (Large Language Models) to provide truthful question-answering capabilities, backed by well-founded citations from various complex formatted data.
+[RAGFlow](https://ragflow.io/) is an open-source RAG (Retrieval-Augmented Generation) engine based on deep document
+understanding. It offers a streamlined RAG workflow for businesses of any scale, combining LLM (Large Language Models)
+to provide truthful question-answering capabilities, backed by well-founded citations from various complex formatted
+data.
 
 ## 🎮 Demo
 
 Try our demo at [https://demo.ragflow.io](https://demo.ragflow.io).
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
 <img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/b083d173-dadc-4ea9-bdeb-180d7df514eb" width="1200"/>
+<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
 </div>
 
-
 ## 🔥 Latest Updates
 
+- 2024-11-22 Adds more variables to Agent.
+- 2024-11-01 Adds keyword extraction and related question generation to the parsed chunks to improve the accuracy of retrieval.
+- 2024-09-13 Adds search mode for knowledge base Q&A.
 - 2024-08-22 Support text to SQL statements through RAG.
-  
 - 2024-08-02 Supports GraphRAG inspired by [graphrag](https://github.com/microsoft/graphrag) and mind map.
-  
-- 2024-07-23 Supports audio file parsing.
 
-- 2024-07-21 Supports more LLMs (LocalAI, OpenRouter, StepFun, and Nvidia).
+## 🎉 Stay Tuned
 
-- 2024-07-18 Adds more components (Wikipedia, PubMed, Baidu, and Duckduckgo) to the graph.
-
-- 2024-07-08 Supports workflow based on [Graph](./graph/README.md).
-- 2024-06-27 Supports Markdown and Docx in the Q&A parsing method. 
-- 2024-06-27 Supports extracting images from Docx files. 
-- 2024-06-27 Supports extracting tables from Markdown files.
-- 2024-06-06 Supports [Self-RAG](https://huggingface.co/papers/2310.11511), which is enabled by default in dialog settings.
-- 2024-05-30 Integrates [BCE](https://github.com/netease-youdao/BCEmbedding) and [BGE](https://github.com/FlagOpen/FlagEmbedding) reranker models.
-- 2024-05-23 Supports [RAPTOR](https://arxiv.org/html/2401.18059v1) for better text retrieval.
-- 2024-05-15 Integrates OpenAI GPT-4o.
+⭐️ Star our repository to stay up-to-date with exciting new features and improvements! Get instant notifications for new
+releases! 🌟
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://github.com/user-attachments/assets/18c9707e-b8aa-4caf-a154-037089c105ba" width="1200"/>
+</div>
 
 ## 🌟 Key Features
 
 ### 🍭 **"Quality in, quality out"**
 
-- [Deep document understanding](./deepdoc/README.md)-based knowledge extraction from unstructured data with complicated formats.
+- [Deep document understanding](./deepdoc/README.md)-based knowledge extraction from unstructured data with complicated
+  formats.
 - Finds "needle in a data haystack" of literally unlimited tokens.
 
 ### 🍱 **Template-based chunking**
@@ -127,7 +132,8 @@ Try our demo at [https://demo.ragflow.io](https://demo.ragflow.io).
 - RAM >= 16 GB
 - Disk >= 50 GB
 - Docker >= 24.0.0 & Docker Compose >= v2.26.1
-  > If you have not installed Docker on your local machine (Windows, Mac, or Linux), see [Install Docker Engine](https://docs.docker.com/engine/install/).
+  > If you have not installed Docker on your local machine (Windows, Mac, or Linux),
+  see [Install Docker Engine](https://docs.docker.com/engine/install/).
 
 ### 🚀 Start up the server
 
@@ -146,7 +152,8 @@ Try our demo at [https://demo.ragflow.io](https://demo.ragflow.io).
    > $ sudo sysctl -w vm.max_map_count=262144
    > ```
    >
-   > This change will be reset after a system reboot. To ensure your change remains permanent, add or update the `vm.max_map_count` value in **/etc/sysctl.conf** accordingly:
+   > This change will be reset after a system reboot. To ensure your change remains permanent, add or update the
+   `vm.max_map_count` value in **/etc/sysctl.conf** accordingly:
    >
    > ```bash
    > vm.max_map_count=262144
@@ -160,16 +167,27 @@ Try our demo at [https://demo.ragflow.io](https://demo.ragflow.io).
 
 3. Build the pre-built Docker images and start up the server:
 
-   > Running the following commands automatically downloads the *dev* version RAGFlow Docker image. To download and run a specified Docker version, update `RAGFLOW_VERSION` in **docker/.env** to the intended version, for example `RAGFLOW_VERSION=v0.10.0`, before running the following commands.
+   > The command below downloads the dev version Docker image for RAGFlow slim (`dev-slim`). Note that RAGFlow slim
+   Docker images do not include embedding models or Python libraries and hence are approximately 1GB in size.
 
    ```bash
    $ cd ragflow/docker
-   $ chmod +x ./entrypoint.sh
-   $ docker compose up -d
+   $ docker compose -f docker-compose.yml up -d
    ```
-   
 
-   > The core image is about 9 GB in size and may take a while to load.
+   > - To download a RAGFlow slim Docker image of a specific version, update the `RAGFLOW_IMAGE` variable in *
+       *docker/.env** to your desired version. For example, `RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0-slim`. After
+       making this change, rerun the command above to initiate the download.
+   > - To download the dev version of RAGFlow Docker image *including* embedding models and Python libraries, update the
+       `RAGFLOW_IMAGE` variable in **docker/.env** to `RAGFLOW_IMAGE=infiniflow/ragflow:dev`. After making this change,
+       rerun the command above to initiate the download.
+   > - To download a specific version of RAGFlow Docker image *including* embedding models and Python libraries, update
+       the `RAGFLOW_IMAGE` variable in **docker/.env** to your desired version. For example,
+       `RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0`. After making this change, rerun the command above to initiate the
+       download.
+
+   > **NOTE:** A RAGFlow Docker image that includes embedding models and Python libraries is approximately 9GB in size
+   and may take significantly longer time to load.
 
 4. Check the server status after having the server up and running:
 
@@ -180,159 +198,155 @@ Try our demo at [https://demo.ragflow.io](https://demo.ragflow.io).
    _The following output confirms a successful launch of the system:_
 
    ```bash
-       ____                 ______ __
-      / __ \ ____ _ ____ _ / ____// /____  _      __
-     / /_/ // __ `// __ `// /_   / // __ \| | /| / /
-    / _, _// /_/ // /_/ // __/  / // /_/ /| |/ |/ /
-   /_/ |_| \__,_/ \__, //_/    /_/ \____/ |__/|__/
-                 /____/
+
+         ____   ___    ______ ______ __               
+        / __ \ /   |  / ____// ____// /____  _      __
+       / /_/ // /| | / / __ / /_   / // __ \| | /| / /
+      / _, _// ___ |/ /_/ // __/  / // /_/ /| |/ |/ / 
+     /_/ |_|/_/  |_|\____//_/    /_/ \____/ |__/|__/ 
 
     * Running on all addresses (0.0.0.0)
     * Running on http://127.0.0.1:9380
     * Running on http://x.x.x.x:9380
     INFO:werkzeug:Press CTRL+C to quit
    ```
-   > If you skip this confirmation step and directly log in to RAGFlow, your browser may prompt a `network anomaly` error because, at that moment, your RAGFlow may not be fully initialized.  
+   > 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](./docker/service_conf.yaml), select the desired LLM factory in `user_default_llm` and update the `API_KEY` field with the corresponding API key.
+   > 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 now on!_
+   _The show is on!_
 
 ## 🔧 Configurations
 
 When it comes to system configurations, you will need to manage the following files:
 
-- [.env](./docker/.env): Keeps the fundamental setups for the system, such as `SVR_HTTP_PORT`, `MYSQL_PASSWORD`, and `MINIO_PASSWORD`.
-- [service_conf.yaml](./docker/service_conf.yaml): Configures the back-end services.
+- [.env](./docker/.env): Keeps the fundamental setups for the system, such as `SVR_HTTP_PORT`, `MYSQL_PASSWORD`, and
+  `MINIO_PASSWORD`.
+- [service_conf.yaml.template](./docker/service_conf.yaml.template): Configures the back-end services. The environment variables in this file will be automatically populated when the Docker container starts. Any environment variables set within the Docker container will be available for use, allowing you to customize service behavior based on the deployment environment.
 - [docker-compose.yml](./docker/docker-compose.yml): The system relies on [docker-compose.yml](./docker/docker-compose.yml) to start up.
 
-You must ensure that changes to the [.env](./docker/.env) file are in line with what are in the [service_conf.yaml](./docker/service_conf.yaml) file.
+> The [./docker/README](./docker/README.md) file provides a detailed description of the environment settings and service
+> configurations which can be used as `${ENV_VARS}` in the [service_conf.yaml.template](./docker/service_conf.yaml.template) file.
 
-> The [./docker/README](./docker/README.md) file provides a detailed description of the environment settings and service configurations, and you are REQUIRED to ensure that all environment settings listed in the [./docker/README](./docker/README.md) file are aligned with the corresponding configurations in the [service_conf.yaml](./docker/service_conf.yaml) file.
+To update the default HTTP serving port (80), go to [docker-compose.yml](./docker/docker-compose.yml) and change `80:80`
+to `<YOUR_SERVING_PORT>:80`.
 
-To update the default HTTP serving port (80), go to [docker-compose.yml](./docker/docker-compose.yml) and change `80:80` to `<YOUR_SERVING_PORT>:80`.
+Updates to the above configurations require a reboot of all containers to take effect:
 
-> Updates to all system configurations require a system reboot to take effect:
->
 > ```bash
-> $ docker-compose up -d
+> $ docker compose -f docker/docker-compose.yml up -d
 > ```
 
-## 🛠️ Build from source
+### Switch doc engine from Elasticsearch to Infinity
 
-To build the Docker images from source:
+RAGFlow uses Elasticsearch by default for storing full text and vectors. To switch to [Infinity](https://github.com/infiniflow/infinity/), follow these steps:
+
+1. Stop all running containers:
+
+   ```bash
+   $ docker compose -f docker/docker-compose.yml down -v
+   ```
+
+2. Set `DOC_ENGINE` in **docker/.env** to `infinity`.
+
+3. Start the containers:
+
+   ```bash
+   $ docker compose -f docker/docker-compose.yml up -d
+   ```
+
+> [!WARNING] 
+> Switching to Infinity on a Linux/arm64 machine is not yet officially supported.
+
+## 🔧 Build a Docker image without embedding models
+
+This image is approximately 1 GB in size and relies on external LLM and embedding services.
 
 ```bash
-$ git clone https://github.com/infiniflow/ragflow.git
-$ cd ragflow/
-$ docker build -t infiniflow/ragflow:dev .
-$ cd ragflow/docker
-$ chmod +x ./entrypoint.sh
-$ docker compose up -d
+git clone https://github.com/infiniflow/ragflow.git
+cd ragflow/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile.slim -t infiniflow/ragflow:dev-slim .
 ```
 
-## 🛠️ Launch service from source
+## 🔧 Build a Docker image including embedding models
 
-To launch the service from source:
+This image is approximately 9 GB in size. As it includes embedding models, it relies on external LLM services only.
 
-1. Clone the repository: 
+```bash
+git clone https://github.com/infiniflow/ragflow.git
+cd ragflow/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile -t infiniflow/ragflow:dev .
+```
 
+## 🔨 Launch service from source for development
+
+1. Install Poetry, or skip this step if it is already installed:
    ```bash
-   $ git clone https://github.com/infiniflow/ragflow.git
-   $ cd ragflow/
+   curl -sSL https://install.python-poetry.org | python3 -
    ```
 
-2. Create a virtual environment, ensuring that Anaconda or Miniconda is installed:
-
+2. Clone the source code and install Python dependencies:
    ```bash
-   $ conda create -n ragflow python=3.11.0
-   $ conda activate ragflow
-   $ pip install -r requirements.txt
-   ```
-   
-   ```bash
-   # If your CUDA version is higher than 12.0, run the following additional commands:
-   $ pip uninstall -y onnxruntime-gpu
-   $ pip install onnxruntime-gpu --extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
+   git clone https://github.com/infiniflow/ragflow.git
+   cd ragflow/
+   export POETRY_VIRTUALENVS_CREATE=true POETRY_VIRTUALENVS_IN_PROJECT=true
+   ~/.local/bin/poetry install --sync --no-root --with=full # install RAGFlow dependent python modules
    ```
 
-3. Copy the entry script and configure environment variables:
-
+3. Launch the dependent services (MinIO, Elasticsearch, Redis, and MySQL) using Docker Compose:
    ```bash
-   # Get the Python path:
-   $ which python
-   # Get the ragflow project path:
-   $ pwd
-   ```
-   
-   ```bash
-   $ cp docker/entrypoint.sh .
-   $ vi entrypoint.sh
+   docker compose -f docker/docker-compose-base.yml up -d
    ```
 
+   Add the following line to `/etc/hosts` to resolve all hosts specified in **docker/.env** to `127.0.0.1`:
+   ```
+   127.0.0.1       es01 infinity mysql minio redis
+   ```  
+   In **docker/service_conf.yaml.template**, update mysql port to `5455` and es port to `1200`, as specified in **docker/.env**.
+
+4. If you cannot access HuggingFace, set the `HF_ENDPOINT` environment variable to use a mirror site:
+
    ```bash
-   # Adjust configurations according to your actual situation (the following two export commands are newly added):
-   # - Assign the result of `which python` to `PY`.
-   # - Assign the result of `pwd` to `PYTHONPATH`.
-   # - Comment out `LD_LIBRARY_PATH`, if it is configured.
-   # - Optional: Add Hugging Face mirror.
-   PY=${PY}
-   export PYTHONPATH=${PYTHONPATH}
    export HF_ENDPOINT=https://hf-mirror.com
    ```
 
-4. Launch the third-party services (MinIO, Elasticsearch, Redis, and MySQL):
-
+5. Launch backend service:
    ```bash
-   $ cd docker
-   $ docker compose -f docker-compose-base.yml up -d 
+   source .venv/bin/activate
+   export PYTHONPATH=$(pwd)
+   bash docker/launch_backend_service.sh
    ```
 
-5. Check the configuration files, ensuring that:
-
-   - The settings in **docker/.env** match those in **conf/service_conf.yaml**. 
-   - The IP addresses and ports for related services in **service_conf.yaml** match the local machine IP and ports exposed by the container.
-
-6. Launch the RAGFlow backend service:
-
+6. Install frontend dependencies:
    ```bash
-   $ chmod +x ./entrypoint.sh
-   $ bash ./entrypoint.sh
-   ```
-
-7. Launch the frontend service:
-
+   cd web
+   npm install --force
+   ```  
+7. Configure frontend to update `proxy.target` in **.umirc.ts** to `http://127.0.0.1:9380`:
+8. Launch frontend service:
    ```bash
-   $ cd web
-   $ npm install --registry=https://registry.npmmirror.com --force
-   $ vim .umirc.ts
-   # Update proxy.target to http://127.0.0.1:9380
-   $ npm run dev 
-   ```
+   npm run dev 
+   ```  
 
-8. Deploy the frontend service:
+   _The following output confirms a successful launch of the system:_
 
-   ```bash
-   $ cd web
-   $ npm install --registry=https://registry.npmmirror.com --force
-   $ umi build
-   $ mkdir -p /ragflow/web
-   $ cp -r dist /ragflow/web
-   $ apt install nginx -y
-   $ cp ../docker/nginx/proxy.conf /etc/nginx
-   $ cp ../docker/nginx/nginx.conf /etc/nginx
-   $ cp ../docker/nginx/ragflow.conf /etc/nginx/conf.d
-   $ systemctl start nginx
-   ```
+   ![](https://github.com/user-attachments/assets/0daf462c-a24d-4496-a66f-92533534e187)
 
 ## 📚 Documentation
 
 - [Quickstart](https://ragflow.io/docs/dev/)
-- [User guide](https://ragflow.io/docs/dev/category/user-guides)
+- [User guide](https://ragflow.io/docs/dev/category/guides)
 - [References](https://ragflow.io/docs/dev/category/references)
 - [FAQ](https://ragflow.io/docs/dev/faq)
 
@@ -348,4 +362,5 @@ See the [RAGFlow Roadmap 2024](https://github.com/infiniflow/ragflow/issues/162)
 
 ## 🙌 Contributing
 
-RAGFlow flourishes via open-source collaboration. In this spirit, we embrace diverse contributions from the community. If you would like to be a part, review our [Contribution Guidelines](./docs/references/CONTRIBUTING.md) first.
+RAGFlow flourishes via open-source collaboration. In this spirit, we embrace diverse contributions from the community.
+If you would like to be a part, review our [Contribution Guidelines](./CONTRIBUTING.md) first.

README_id.md

@@ -0,0 +1,341 @@
+<div align="center">
+<a href="https://demo.ragflow.io/">
+<img src="web/src/assets/logo-with-text.png" width="520" alt="Logo ragflow">
+</a>
+</div>
+
+<p align="center">
+  <a href="./README.md">English</a> |
+  <a href="./README_zh.md">简体中文</a> |
+  <a href="./README_ja.md">日本語</a> |
+  <a href="./README_ko.md">한국어</a> |
+  <a href="./README_id.md">Bahasa Indonesia</a>
+</p>
+
+<p align="center">
+    <a href="https://x.com/intent/follow?screen_name=infiniflowai" target="_blank">
+        <img src="https://img.shields.io/twitter/follow/infiniflow?logo=X&color=%20%23f5f5f5" alt="Ikuti di X (Twitter)">
+    </a>
+    <a href="https://demo.ragflow.io" target="_blank">
+        <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/badge/docker_pull-ragflow:v0.14.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.14.0">
+    </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">
+    </a>
+    <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
+        <img height="21" src="https://img.shields.io/badge/Lisensi-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="Lisensi">
+    </a>
+</p>
+
+<h4 align="center">
+  <a href="https://ragflow.io/docs/dev/">Dokumentasi</a> |
+  <a href="https://github.com/infiniflow/ragflow/issues/162">Peta Jalan</a> |
+  <a href="https://twitter.com/infiniflowai">Twitter</a> |
+  <a href="https://discord.gg/4XxujFgUN7">Discord</a> |
+  <a href="https://demo.ragflow.io">Demo</a>
+</h4>
+
+<details open>
+<summary></b>📕 Daftar Isi</b></summary>
+
+- 💡 [Apa Itu RAGFlow?](#-apa-itu-ragflow)
+- 🎮 [Demo](#-demo)
+- 📌 [Pembaruan Terbaru](#-pembaruan-terbaru)
+- 🌟 [Fitur Utama](#-fitur-utama)
+- 🔎 [Arsitektur Sistem](#-arsitektur-sistem)
+- 🎬 [Mulai](#-mulai)
+- 🔧 [Konfigurasi](#-konfigurasi)
+- 🔧 [Membangun Image Docker tanpa Model Embedding](#-membangun-image-docker-tanpa-model-embedding)
+- 🔧 [Membangun Image Docker dengan Model Embedding](#-membangun-image-docker-dengan-model-embedding)
+- 🔨 [Meluncurkan aplikasi dari Sumber untuk Pengembangan](#-meluncurkan-aplikasi-dari-sumber-untuk-pengembangan)
+- 📚 [Dokumentasi](#-dokumentasi)
+- 📜 [Peta Jalan](#-peta-jalan)
+- 🏄 [Komunitas](#-komunitas)
+- 🙌 [Kontribusi](#-kontribusi)
+
+</details>
+
+## 💡 Apa Itu RAGFlow?
+
+[RAGFlow](https://ragflow.io/) adalah mesin RAG (Retrieval-Augmented Generation) open-source berbasis pemahaman dokumen yang mendalam. Platform ini menyediakan alur kerja RAG yang efisien untuk bisnis dengan berbagai skala, menggabungkan LLM (Large Language Models) untuk menyediakan kemampuan tanya-jawab yang benar dan didukung oleh referensi dari data terstruktur kompleks.
+
+## 🎮 Demo
+
+Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
+<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
+</div>
+
+## 🔥 Pembaruan Terbaru
+
+- 22-11-2024 Peningkatan definisi dan penggunaan variabel di Agen.
+- 2024-11-01: Penambahan ekstraksi kata kunci dan pembuatan pertanyaan terkait untuk meningkatkan akurasi pengambilan.
+- 2024-09-13: Penambahan mode pencarian untuk Q&A basis pengetahuan.
+- 2024-08-22: Dukungan untuk teks ke pernyataan SQL melalui RAG.
+- 2024-08-02: Dukungan GraphRAG yang terinspirasi oleh [graphrag](https://github.com/microsoft/graphrag) dan mind map.
+
+## 🎉 Tetap Terkini
+
+⭐️ Star repositori kami untuk tetap mendapat informasi tentang fitur baru dan peningkatan menarik! 🌟
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://github.com/user-attachments/assets/18c9707e-b8aa-4caf-a154-037089c105ba" width="1200"/>
+</div>
+
+## 🌟 Fitur Utama
+
+### 🍭 **"Kualitas Masuk, Kualitas Keluar"**
+
+- Ekstraksi pengetahuan berbasis pemahaman dokumen mendalam dari data tidak terstruktur dengan format yang rumit.
+- Menemukan "jarum di tumpukan data" dengan token yang hampir tidak terbatas.
+
+### 🍱 **Pemotongan Berbasis Template**
+
+- Cerdas dan dapat dijelaskan.
+- Banyak pilihan template yang tersedia.
+
+### 🌱 **Referensi yang Didasarkan pada Data untuk Mengurangi Hallusinasi**
+
+- Visualisasi pemotongan teks memungkinkan intervensi manusia.
+- Tampilan cepat referensi kunci dan referensi yang dapat dilacak untuk mendukung jawaban yang didasarkan pada fakta.
+
+### 🍔 **Kompatibilitas dengan Sumber Data Heterogen**
+
+- Mendukung Word, slide, excel, txt, gambar, salinan hasil scan, data terstruktur, halaman web, dan banyak lagi.
+
+### 🛀 **Alur Kerja RAG yang Otomatis dan Mudah**
+
+- Orkestrasi RAG yang ramping untuk bisnis kecil dan besar.
+- LLM yang dapat dikonfigurasi serta model embedding.
+- Peringkat ulang berpasangan dengan beberapa pengambilan ulang.
+- API intuitif untuk integrasi yang mudah dengan bisnis.
+
+## 🔎 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"/>
+</div>
+
+## 🎬 Mulai
+
+### 📝 Prasyarat
+
+- CPU >= 4 inti
+- RAM >= 16 GB
+- Disk >= 50 GB
+- Docker >= 24.0.0 & Docker Compose >= v2.26.1
+
+### 🚀 Menjalankan Server
+
+1. Pastikan `vm.max_map_count` >= 262144:
+
+   > Untuk memeriksa nilai `vm.max_map_count`:
+   >
+   > ```bash
+   > $ sysctl vm.max_map_count
+   > ```
+   >
+   > Jika nilainya kurang dari 262144, setel ulang `vm.max_map_count` ke setidaknya 262144:
+   >
+   > ```bash
+   > # Dalam contoh ini, kita atur menjadi 262144:
+   > $ sudo sysctl -w vm.max_map_count=262144
+   > ```
+   >
+   > Perubahan ini akan hilang setelah sistem direboot. Untuk membuat perubahan ini permanen, tambahkan atau perbarui nilai
+   `vm.max_map_count` di **/etc/sysctl.conf**:
+   >
+   > ```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:
+
+   > Perintah di bawah ini akan mengunduh versi dev dari Docker image RAGFlow slim (`dev-slim`). Image RAGFlow slim
+   tidak termasuk model embedding atau library Python dan berukuran sekitar 1GB.
+
+   ```bash
+   $ cd ragflow/docker
+   $ docker compose -f docker-compose.yml up -d
+   ```
+
+   > - Untuk mengunduh versi tertentu dari image Docker RAGFlow slim, perbarui variabel `RAGFlow_IMAGE` di *
+       *docker/.env** sesuai dengan versi yang diinginkan. Misalnya, `RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0-slim`.
+       Setelah mengubah ini, jalankan ulang perintah di atas untuk memulai unduhan.
+   > - Untuk mengunduh versi dev dari image Docker RAGFlow *termasuk* model embedding dan library Python, perbarui
+       variabel `RAGFlow_IMAGE` di **docker/.env** menjadi `RAGFLOW_IMAGE=infiniflow/ragflow:dev`. Setelah mengubah ini,
+       jalankan ulang perintah di atas untuk memulai unduhan.
+   > - Untuk mengunduh versi tertentu dari image Docker RAGFlow *termasuk* model embedding dan library Python, perbarui
+       variabel `RAGFlow_IMAGE` di **docker/.env** sesuai dengan versi yang diinginkan. Misalnya,
+       `RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0`. Setelah mengubah ini, jalankan ulang perintah di atas untuk memulai unduhan.
+
+   > **CATATAN:** Image Docker RAGFlow yang mencakup model embedding dan library Python berukuran sekitar 9GB
+   dan mungkin memerlukan waktu lebih lama untuk dimuat.
+
+4. Periksa status server setelah server aktif dan berjalan:
+
+   ```bash
+   $ docker logs -f ragflow-server
+   ```
+
+   _Output berikut menandakan bahwa sistem berhasil diluncurkan:_
+
+   ```bash
+
+         ____   ___    ______ ______ __               
+        / __ \ /   |  / ____// ____// /____  _      __
+       / /_/ // /| | / / __ / /_   / // __ \| | /| / /
+      / _, _// ___ |/ /_/ // __/  / // /_/ /| |/ |/ / 
+     /_/ |_|/_/  |_|\____//_/    /_/ \____/ |__/|__/ 
+
+    * Running on all addresses (0.0.0.0)
+    * Running on http://127.0.0.1:9380
+    * Running on http://x.x.x.x:9380
+    INFO:werkzeug:Press CTRL+C to quit
+   ```
+   > Jika Anda melewatkan langkah ini dan langsung login ke RAGFlow, browser Anda mungkin menampilkan error `network anormal`
+   karena RAGFlow mungkin belum sepenuhnya siap.
+
+5. 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.
+6. Dalam [service_conf.yaml](./docker/service_conf.yaml), 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!_
+
+## 🔧 Konfigurasi
+
+Untuk konfigurasi sistem, Anda perlu mengelola file-file berikut:
+
+- [.env](./docker/.env): Menyimpan pengaturan dasar sistem, seperti `SVR_HTTP_PORT`, `MYSQL_PASSWORD`, dan
+  `MINIO_PASSWORD`.
+- [service_conf.yaml](./docker/service_conf.yaml): Mengonfigurasi aplikasi backend.
+- [docker-compose.yml](./docker/docker-compose.yml): Sistem ini bergantung pada [docker-compose.yml](./docker/docker-compose.yml) untuk memulai.
+
+Anda harus memastikan bahwa perubahan pada file [.env](./docker/.env) sesuai dengan yang ada di file [service_conf.yaml](./docker/service_conf.yaml).
+
+> File [./docker/README](./docker/README.md) menyediakan penjelasan detail tentang pengaturan lingkungan dan konfigurasi aplikasi,
+> dan Anda DIWAJIBKAN memastikan bahwa semua pengaturan lingkungan yang tercantum di
+> [./docker/README](./docker/README.md) selaras dengan konfigurasi yang sesuai di
+> [service_conf.yaml](./docker/service_conf.yaml).
+
+Untuk memperbarui port HTTP default (80), buka [docker-compose.yml](./docker/docker-compose.yml) dan ubah `80:80`
+menjadi `<YOUR_SERVING_PORT>:80`.
+
+Pembaruan konfigurasi ini memerlukan reboot semua kontainer agar efektif:
+
+> ```bash
+> $ docker compose -f docker/docker-compose.yml up -d
+> ```
+
+## 🔧 Membangun Docker Image tanpa Model Embedding
+
+Image ini berukuran sekitar 1 GB dan bergantung pada aplikasi LLM eksternal dan embedding.
+
+```bash
+git clone https://github.com/infiniflow/ragflow.git
+cd ragflow/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile.slim -t infiniflow/ragflow:dev-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/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile -t infiniflow/ragflow:dev .
+```
+
+## 🔨 Menjalankan Aplikasi dari untuk Pengembangan
+
+1. Instal Poetry, atau lewati langkah ini jika sudah terinstal:
+   ```bash
+   curl -sSL https://install.python-poetry.org | python3 -
+   ```
+
+2. Clone kode sumber dan instal dependensi Python:
+   ```bash
+   git clone https://github.com/infiniflow/ragflow.git
+   cd ragflow/
+   export POETRY_VIRTUALENVS_CREATE=true POETRY_VIRTUALENVS_IN_PROJECT=true
+   ~/.local/bin/poetry install --sync --no-root # install modul python RAGFlow
+   ```
+
+3. Jalankan aplikasi yang diperlukan (MinIO, Elasticsearch, Redis, dan MySQL) menggunakan Docker Compose:
+   ```bash
+   docker compose -f docker/docker-compose-base.yml up -d
+   ```
+
+   Tambahkan baris berikut ke `/etc/hosts` untuk memetakan semua host yang ditentukan di **docker/service_conf.yaml** ke `127.0.0.1`:
+   ```
+   127.0.0.1       es01 infinity mysql minio redis
+   ```  
+   Di **docker/service_conf.yaml**, perbarui port mysql ke `5455` dan es ke `1200`, sesuai dengan yang ditentukan di **docker/.env**.
+
+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. Jalankan aplikasi backend:
+   ```bash
+   source .venv/bin/activate
+   export PYTHONPATH=$(pwd)
+   bash docker/launch_backend_service.sh
+   ```
+
+6. Instal dependensi frontend:
+   ```bash
+   cd web
+   npm install --force
+   ```  
+7. Konfigurasikan frontend untuk memperbarui `proxy.target` di **.umirc.ts** menjadi `http://127.0.0.1:9380`:
+8. Jalankan aplikasi frontend:
+   ```bash
+   npm run dev 
+   ```  
+
+   _Output berikut menandakan bahwa sistem berhasil diluncurkan:_
+
+   ![](https://github.com/user-attachments/assets/0daf462c-a24d-4496-a66f-92533534e187)
+
+## 📚 Dokumentasi
+
+- [Quickstart](https://ragflow.io/docs/dev/)
+- [Panduan Pengguna](https://ragflow.io/docs/dev/category/guides)
+- [Referensi](https://ragflow.io/docs/dev/category/references)
+- [FAQ](https://ragflow.io/docs/dev/faq)
+
+## 📜 Roadmap
+
+Lihat [Roadmap RAGFlow 2024](https://github.com/infiniflow/ragflow/issues/162)
+
+## 🏄 Komunitas
+
+- [Discord](https://discord.gg/4XxujFgUN7)
+- [Twitter](https://twitter.com/infiniflowai)
+- [GitHub Discussions](https://github.com/orgs/infiniflow/discussions)
+
+## 🙌 Kontribusi
+
+RAGFlow berkembang melalui kolaborasi open-source. Dalam semangat ini, kami menerima kontribusi dari komunitas.
+Jika Anda ingin berpartisipasi, tinjau terlebih dahulu [Panduan Kontribusi](./CONTRIBUTING.md).

README_ja.md

@@ -8,23 +8,29 @@
   <a href="./README.md">English</a> |
   <a href="./README_zh.md">简体中文</a> |
   <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a>
+  <a href="./README_ko.md">한국어</a> |
+  <a href="./README_id.md">Bahasa Indonesia</a>
 </p>
 
 <p align="center">
+    <a href="https://x.com/intent/follow?screen_name=infiniflowai" target="_blank">
+        <img src="https://img.shields.io/twitter/follow/infiniflow?logo=X&color=%20%23f5f5f5" alt="follow on X(Twitter)">
+    </a>
+    <a href="https://demo.ragflow.io" target="_blank">
+        <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/badge/docker_pull-ragflow:v0.14.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.14.0">
+    </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">
     </a>
-    <a href="https://demo.ragflow.io" target="_blank">
-        <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/badge/docker_pull-ragflow:v0.10.0-brightgreen"
-            alt="docker pull infiniflow/ragflow:v0.10.0"></a>
-      <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
-    <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
-  </a>
+    <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
+        <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
+    </a>
 </p>
 
+
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Document</a> |
   <a href="https://github.com/infiniflow/ragflow/issues/162">Roadmap</a> |
@@ -42,25 +48,23 @@
 デモをお試しください：[https://demo.ragflow.io](https://demo.ragflow.io)。
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
 <img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/b083d173-dadc-4ea9-bdeb-180d7df514eb" width="1200"/>
+<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
 </div>
 
 
 ## 🔥 最新情報
 
+- 2024-11-22 エージェントでの変数の定義と使用法を改善しました。
+- 2024-11-01 再現の精度を向上させるために、解析されたチャンクにキーワード抽出と関連質問の生成を追加しました。
+- 2024-09-13 ナレッジベース Q&A の検索モードを追加しました。
 - 2024-08-22 RAG を介して SQL ステートメントへのテキストをサポートします。
 - 2024-08-02 [graphrag](https://github.com/microsoft/graphrag) からインスピレーションを得た GraphRAG とマインド マップをサポートします。
-- 2024-07-23 音声ファイルの解析をサポートしました。
-- 2024-07-21 より多くの LLM サプライヤー (LocalAI/OpenRouter/StepFun/Nvidia) をサポートします。
-- 2024-07-18 グラフにコンポーネント(Wikipedia/PubMed/Baidu/Duckduckgo)を追加しました。
-- 2024-07-08 [Graph](./graph/README.md) ベースのワークフローをサポート
-- 2024-06-27 Q&A解析方式はMarkdownファイルとDocxファイルをサポートしています。
-- 2024-06-27 Docxファイルからの画像の抽出をサポートします。
-- 2024-06-27 Markdownファイルからテーブルを抽出することをサポートします。
-- 2024-06-06 会話設定でデフォルトでチェックされている [Self-RAG](https://huggingface.co/papers/2310.11511) をサポートします。
-- 2024-05-30 [BCE](https://github.com/netease-youdao/BCEmbedding) 、[BGE](https://github.com/FlagOpen/FlagEmbedding) reranker を統合。
-- 2024-05-23 より良いテキスト検索のために [RAPTOR](https://arxiv.org/html/2401.18059v1) をサポート。
-- 2024-05-15 OpenAI GPT-4oを統合しました。
+
+## 🎉 続きを楽しみに
+⭐️ リポジトリをスター登録して、エキサイティングな新機能やアップデートを最新の状態に保ちましょう！すべての新しいリリースに関する即時通知を受け取れます！ 🌟
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://github.com/user-attachments/assets/18c9707e-b8aa-4caf-a154-037089c105ba" width="1200"/>
+</div>
 
 ## 🌟 主な特徴
 
@@ -137,15 +141,18 @@
 
 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:
 
+   > 以下のコマンドは、RAGFlow slim（`dev-slim`）の開発版Dockerイメージをダウンロードします。RAGFlow slimのDockerイメージには、埋め込みモデルやPythonライブラリが含まれていないため、サイズは約1GBです。
+
    ```bash
    $ cd ragflow/docker
-   $ chmod +x ./entrypoint.sh
-   $ docker compose up -d
+   $ docker compose -f docker-compose.yml up -d
    ```
 
-   > 上記のコマンドを実行すると、RAGFlowの開発版dockerイメージが自動的にダウンロードされます。 特定のバージョンのDockerイメージをダウンロードして実行したい場合は、docker/.envファイルのRAGFLOW_VERSION変数を見つけて、対応するバージョンに変更してください。 例えば、RAGFLOW_VERSION=v0.10.0として、上記のコマンドを実行してください。
-
-   > コアイメージのサイズは約 9 GB で、ロードに時間がかかる場合があります。
+   > - 特定のバージョンのRAGFlow slim Dockerイメージをダウンロードするには、**docker/.env**内の`RAGFlow_IMAGE`変数を希望のバージョンに更新します。例えば、`RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0`とします。この変更を行った後、上記のコマンドを再実行してダウンロードを開始してください。
+   > - RAGFlowの埋め込みモデルとPythonライブラリを含む開発版Dockerイメージをダウンロードするには、**docker/.env**内の`RAGFlow_IMAGE`変数を`RAGFLOW_IMAGE=infiniflow/ragflow:dev`に更新します。この変更を行った後、上記のコマンドを再実行してダウンロードを開始してください。
+   > - 特定のバージョンのRAGFlow Dockerイメージ（埋め込みモデルとPythonライブラリを含む）をダウンロードするには、**docker/.env**内の`RAGFlow_IMAGE`変数を希望のバージョンに更新します。例えば、`RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0`とします。この変更を行った後、上記のコマンドを再実行してダウンロードを開始してください。  
+   
+   > **NOTE:** 埋め込みモデルとPythonライブラリを含むRAGFlow Dockerイメージのサイズは約9GBであり、読み込みにかなりの時間がかかる場合があります。
 
 4. サーバーを立ち上げた後、サーバーの状態を確認する:
 
@@ -156,12 +163,11 @@
    _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:_
 
    ```bash
-       ____                 ______ __
-      / __ \ ____ _ ____ _ / ____// /____  _      __
-     / /_/ // __ `// __ `// /_   / // __ \| | /| / /
-    / _, _// /_/ // /_/ // __/  / // /_/ /| |/ |/ /
-   /_/ |_| \__,_/ \__, //_/    /_/ \____/ |__/|__/
-                 /____/
+        ____   ___    ______ ______ __               
+       / __ \ /   |  / ____// ____// /____  _      __
+      / /_/ // /| | / / __ / /_   / // __ \| | /| / /
+     / _, _// ___ |/ /_/ // __/  / // /_/ /| |/ |/ / 
+    /_/ |_|/_/  |_|\____//_/    /_/ \____/ |__/|__/  
 
     * Running on all addresses (0.0.0.0)
     * Running on http://127.0.0.1:9380
@@ -195,86 +201,108 @@
 > すべてのシステム設定のアップデートを有効にするには、システムの再起動が必要です:
 >
 > ```bash
-> $ docker-compose up -d
+> $ docker compose -f docker/docker-compose.yml up -d
 > ```
 
-## 🛠️ ソースからビルドする
+### Elasticsearch から Infinity にドキュメントエンジンを切り替えます
 
-ソースからDockerイメージをビルドするには:
+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
+   ```
+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/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 -t infiniflow/ragflow:v0.10.0 .
-$ cd ragflow/docker
-$ chmod +x ./entrypoint.sh
-$ docker compose up -d
+git clone https://github.com/infiniflow/ragflow.git
+cd ragflow/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile.slim -t infiniflow/ragflow:dev-slim .
 ```
 
-## 🛠️ ソースコードからサービスを起動する方法
+## 🔧 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）
 
-ソースコードからサービスを起動する場合は、以下の手順に従ってください:
-
-1. リポジトリをクローンします
-```bash
-$ git clone https://github.com/infiniflow/ragflow.git
-$ cd ragflow/
-```
-
-2. 仮想環境を作成します（AnacondaまたはMinicondaがインストールされていることを確認してください）
-```bash
-$ conda create -n ragflow python=3.11.0
-$ conda activate ragflow
-$ pip install -r requirements.txt
-```
-CUDAのバージョンが12.0以上の場合、以下の追加コマンドを実行してください：
-```bash
-$ pip uninstall -y onnxruntime-gpu
-$ pip install onnxruntime-gpu --extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
-```
-
-3. エントリースクリプトをコピーし、環境変数を設定します
-```bash
-$ cp docker/entrypoint.sh .
-$ vi entrypoint.sh
-```
-以下のコマンドで Python のパスとragflowプロジェクトのパスを取得します：
-```bash
-$ which python
-$ pwd
-```
-
-`which python` の出力を `PY` の値として、`pwd` の出力を `PYTHONPATH` の値として設定します。
-
-`LD_LIBRARY_PATH` が既に設定されている場合は、コメントアウトできます。
+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
-# 実際の状況に応じて設定を調整してください。以下の二つの export は新たに追加された設定です
-PY=${PY}
-export PYTHONPATH=${PYTHONPATH}
-# オプション：Hugging Face ミラーを追加
-export HF_ENDPOINT=https://hf-mirror.com
+git clone https://github.com/infiniflow/ragflow.git
+cd ragflow/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile -t infiniflow/ragflow:dev .
 ```
 
-4. 基本サービスを起動します
-```bash
-$ cd docker
-$ docker compose -f docker-compose-base.yml up -d 
-```
+## 🔨 ソースコードからサービスを起動する方法
 
-5. 設定ファイルを確認します
-**docker/.env** 内の設定が**conf/service_conf.yaml**内の設定と一致していることを確認してください。**service_conf.yaml**内の関連サービスのIPアドレスとポートは、ローカルマシンのIPアドレスとコンテナが公開するポートに変更する必要があります。
+1. Poetry をインストールする。すでにインストールされている場合は、このステップをスキップしてください:
+   ```bash
+   curl -sSL https://install.python-poetry.org | python3 -
+   ```
 
-6. サービスを起動します
-```bash
-$ chmod +x ./entrypoint.sh
-$ bash ./entrypoint.sh
-```
+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/
+   export POETRY_VIRTUALENVS_CREATE=true POETRY_VIRTUALENVS_IN_PROJECT=true
+   ~/.local/bin/poetry install --sync --no-root # install RAGFlow dependent python modules
+   ```
+
+3. Docker Compose を使用して依存サービス（MinIO、Elasticsearch、Redis、MySQL）を起動する:
+   ```bash
+   docker compose -f docker/docker-compose-base.yml up -d
+   ```
+
+   `/etc/hosts` に以下の行を追加して、**docker/service_conf.yaml** に指定されたすべてのホストを `127.0.0.1` に解決します:  
+   ```
+   127.0.0.1       es01 infinity mysql minio redis
+   ```  
+   **docker/service_conf.yaml** で mysql のポートを `5455` に、es のポートを `1200` に更新します（**docker/.env** に指定された通り）.
+
+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. バックエンドサービスを起動する:
+   ```bash
+   source .venv/bin/activate
+   export PYTHONPATH=$(pwd)
+   bash docker/launch_backend_service.sh
+   ```
+
+6. 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 --force
+   ```  
+7. フロントエンドを設定し、**.umirc.ts** の `proxy.target` を `http://127.0.0.1:9380` に更新します:
+8. フロントエンドサービスを起動する:  
+   ```bash
+   npm run dev 
+   ```
+
+   _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)
 
 ## 📚 ドキュメンテーション
 
 - [Quickstart](https://ragflow.io/docs/dev/)
-- [User guide](https://ragflow.io/docs/dev/category/user-guides)
+- [User guide](https://ragflow.io/docs/dev/category/guides)
 - [References](https://ragflow.io/docs/dev/category/references)
 - [FAQ](https://ragflow.io/docs/dev/faq)
 
@@ -290,4 +318,4 @@ $ bash ./entrypoint.sh
 
 ## 🙌 コントリビュート
 
-RAGFlow はオープンソースのコラボレーションによって発展してきました。この精神に基づき、私たちはコミュニティからの多様なコントリビュートを受け入れています。 参加を希望される方は、まず [コントリビューションガイド](./docs/references/CONTRIBUTING.md)をご覧ください。
+RAGFlow はオープンソースのコラボレーションによって発展してきました。この精神に基づき、私たちはコミュニティからの多様なコントリビュートを受け入れています。 参加を希望される方は、まず [コントリビューションガイド](./CONTRIBUTING.md)をご覧ください。

README_ko.md

@@ -9,21 +9,28 @@
   <a href="./README_zh.md">简体中文</a> |
   <a href="./README_ja.md">日本語</a> |
   <a href="./README_ko.md">한국어</a> |
+  <a href="./README_id.md">Bahasa Indonesia</a>
 </p>
 
 <p align="center">
+    <a href="https://x.com/intent/follow?screen_name=infiniflowai" target="_blank">
+        <img src="https://img.shields.io/twitter/follow/infiniflow?logo=X&color=%20%23f5f5f5" alt="follow on X(Twitter)">
+    </a>
+    <a href="https://demo.ragflow.io" target="_blank">
+        <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/badge/docker_pull-ragflow:v0.14.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.14.0">
+    </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">
     </a>
-    <a href="https://demo.ragflow.io" target="_blank">
-        <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/badge/docker_pull-ragflow:v0.10.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.10.0"></a>
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
-    <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
-  </a>
+        <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
+    </a>
 </p>
 
+
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Document</a> |
   <a href="https://github.com/infiniflow/ragflow/issues/162">Roadmap</a> |
@@ -43,37 +50,28 @@
 데모를 [https://demo.ragflow.io](https://demo.ragflow.io)에서 실행해 보세요.
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
 <img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/b083d173-dadc-4ea9-bdeb-180d7df514eb" width="1200"/>
+<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
 </div>
 
 
 ## 🔥 업데이트
 
+- 2024-11-22 에이전트의 변수 정의 및 사용을 개선했습니다.
+  
+- 2024-11-01 파싱된 청크에 키워드 추출 및 관련 질문 생성을 추가하여 재현율을 향상시킵니다.
+  
+- 2024-09-13 지식베이스 Q&A 검색 모드를 추가합니다.
+  
 - 2024-08-22 RAG를 통해 SQL 문에 텍스트를 지원합니다.
   
 - 2024-08-02: [graphrag](https://github.com/microsoft/graphrag)와 마인드맵에서 영감을 받은 GraphRAG를 지원합니다.
 
-- 2024-07-23: 오디오 파일 분석을 지원합니다.
 
-- 2024-07-21: 더 많은 LLMs(LocalAI, OpenRouter, StepFun, Nvidia)를 지원합니다.
-
-- 2024-07-18: 그래프에 더 많은 구성요소(Wikipedia, PubMed, Baidu, Duckduckgo)를 추가합니다.
-
-- 2024-07-08: [Graph](./graph/README.md)를 기반으로 한 워크플로우를 지원합니다.
-
-- 2024-06-27: Q&A 분석 방법에서 Markdown과 Docx를 지원합니다.
-
-- 2024-06-27: Docx 파일에서 이미지 추출을 지원합니다.
-
-- 2024-06-27: Markdown 파일에서 표 추출을 지원합니다.
-
-- 2024-06-06: 대화 설정에서 기본으로 [Self-RAG](https://huggingface.co/papers/2310.11511)를 지원합니다.
-
-- 2024-05-30: [BCE](https://github.com/netease-youdao/BCEmbedding) 및 [BGE](https://github.com/FlagOpen/FlagEmbedding) reranker 모델을 통합합니다.
-
-- 2024-05-23: 더 나은 텍스트 검색을 위해 [RAPTOR](https://arxiv.org/html/2401.18059v1)를 지원합니다.
-
-- 2024-05-15: OpenAI GPT-4o를 통합합니다.
+## 🎉 계속 지켜봐 주세요
+⭐️우리의 저장소를 즐겨찾기에 등록하여 흥미로운 새로운 기능과 업데이트를 최신 상태로 유지하세요! 모든 새로운 릴리스에 대한 즉시 알림을 받으세요! 🌟
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://github.com/user-attachments/assets/18c9707e-b8aa-4caf-a154-037089c105ba" width="1200"/>
+</div>
 
 
 ## 🌟 주요 기능
@@ -147,14 +145,18 @@
 
 3. 미리 빌드된 Docker 이미지를 생성하고 서버를 시작하세요:
 
-   > 다음 명령어를 실행하면 *dev* 버전의 RAGFlow Docker 이미지가 자동으로 다운로드됩니다. 특정 Docker 버전을 다운로드하고 실행하려면, **docker/.env** 파일에서 `RAGFLOW_VERSION`을 원하는 버전으로 업데이트한 후, 예를 들어 `RAGFLOW_VERSION=v0.10.0`로 업데이트 한 뒤, 다음 명령어를 실행하세요.
+   > 아래의 명령은 RAGFlow slim(dev-slim)의 개발 버전 Docker 이미지를 다운로드합니다. RAGFlow slim Docker 이미지에는 임베딩 모델이나 Python 라이브러리가 포함되어 있지 않으므로 크기는 약 1GB입니다.
+
    ```bash
    $ cd ragflow/docker
-   $ chmod +x ./entrypoint.sh
-   $ docker compose up -d
+   $ docker compose -f docker-compose.yml up -d
    ```
    
-   > 기본 이미지는 약 9GB 크기이며 로드하는 데 시간이 걸릴 수 있습니다.
+   > - 특정 버전의 RAGFlow slim Docker 이미지를 다운로드하려면, **docker/.env**에서 `RAGFlow_IMAGE` 변수를 원하는 버전으로 업데이트하세요. 예를 들어, `RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0-slim`으로 설정합니다. 이 변경을 완료한 후, 위의 명령을 다시 실행하여 다운로드를 시작하세요.
+   > - RAGFlow의 임베딩 모델과 Python 라이브러리를 포함한 개발 버전 Docker 이미지를 다운로드하려면, **docker/.env**에서 `RAGFlow_IMAGE` 변수를 `RAGFLOW_IMAGE=infiniflow/ragflow:dev`로 업데이트하세요. 이 변경을 완료한 후, 위의 명령을 다시 실행하여 다운로드를 시작하세요.
+   > - 특정 버전의 RAGFlow Docker 이미지를 임베딩 모델과 Python 라이브러리를 포함하여 다운로드하려면, **docker/.env**에서 `RAGFlow_IMAGE` 변수를 원하는 버전으로 업데이트하세요. 예를 들어, `RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0` 로 설정합니다. 이 변경을 완료한 후, 위의 명령을 다시 실행하여 다운로드를 시작하세요.  
+   
+   > **NOTE:** 임베딩 모델과 Python 라이브러리를 포함한 RAGFlow Docker 이미지의 크기는 약 9GB이며, 로드하는 데 상당히 오랜 시간이 걸릴 수 있습니다.
 
 
 4. 서버가 시작된 후 서버 상태를 확인하세요:
@@ -166,19 +168,18 @@
    _다음 출력 결과로 시스템이 성공적으로 시작되었음을 확인합니다:_
 
    ```bash
-       ____                 ______ __
-      / __ \ ____ _ ____ _ / ____// /____  _      __
-     / /_/ // __ `// __ `// /_   / // __ \| | /| / /
-    / _, _// /_/ // /_/ // __/  / // /_/ /| |/ |/ /
-   /_/ |_| \__,_/ \__, //_/    /_/ \____/ |__/|__/
-                 /____/
+        ____   ___    ______ ______ __               
+       / __ \ /   |  / ____// ____// /____  _      __
+      / /_/ // /| | / / __ / /_   / // __ \| | /| / /
+     / _, _// ___ |/ /_/ // __/  / // /_/ /| |/ |/ / 
+    /_/ |_|/_/  |_|\____//_/    /_/ \____/ |__/|__/  
 
     * Running on all addresses (0.0.0.0)
     * Running on http://127.0.0.1:9380
     * Running on http://x.x.x.x:9380
     INFO:werkzeug:Press CTRL+C to quit
    ```
-   > 만약 확인 단계를 건너뛰고 바로 RAGFlow에 로그인하면, RAGFlow가 완전히 초기화되지 않았기 때문에 브라우저에서 `network anomaly` 오류가 발생할 수 있습니다.
+   > 만약 확인 단계를 건너뛰고 바로 RAGFlow에 로그인하면, RAGFlow가 완전히 초기화되지 않았기 때문에 브라우저에서 `network anormal` 오류가 발생할 수 있습니다.
 
 5. 웹 브라우저에 서버의 IP 주소를 입력하고 RAGFlow에 로그인하세요.
    > 기본 설정을 사용할 경우, `http://IP_OF_YOUR_MACHINE`만 입력하면 됩니다 (포트 번호는 제외). 기본 HTTP 서비스 포트 `80`은 기본 구성으로 사용할 때 생략할 수 있습니다.
@@ -204,118 +205,106 @@
 > 모든 시스템 구성 업데이트는 적용되기 위해 시스템 재부팅이 필요합니다.
 >
 > ```bash
-> $ docker-compose up -d
+> $ docker compose -f docker/docker-compose.yml up -d
 > ```
 
-## 🛠️ 소스에서 빌드하기
+### Elasticsearch 에서 Infinity 로 문서 엔진 전환
 
-Docker 이미지를 소스에서 빌드하려면:
+RAGFlow 는 기본적으로 Elasticsearch 를 사용하여 전체 텍스트 및 벡터를 저장합니다. [Infinity]로 전환(https://github.com/infiniflow/infinity/), 다음 절차를 따르십시오.
+1. 실행 중인 모든 컨테이너를 중지합니다.
+   ```bash
+   $docker compose-f docker/docker-compose.yml down-v
+   ```
+2. **docker/.env**의 "DOC_ENGINE" 을 "infinity" 로 설정합니다.
+3. 컨테이너 부팅:
+   ```bash
+   $docker compose-f docker/docker-compose.yml up-d
+   ``` 
+> [!WARNING]
+> Linux/arm64 시스템에서 Infinity로 전환하는 것은 공식적으로 지원되지 않습니다.
+ 
+## 🔧 소스 코드로 Docker 이미지를 컴파일합니다(임베딩 모델 포함하지 않음)
+
+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 -t infiniflow/ragflow:dev .
-$ cd ragflow/docker
-$ chmod +x ./entrypoint.sh
-$ docker compose up -d
+git clone https://github.com/infiniflow/ragflow.git
+cd ragflow/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile.slim -t infiniflow/ragflow:dev-slim .
 ```
 
+## 🔧 소스 코드로 Docker 이미지를 컴파일합니다(임베딩 모델 포함)
 
-## 🛠️ 소스에서 서비스 시작하기
+이 Docker의 크기는 약 9GB이며, 이미 임베딩 모델을 포함하고 있으므로 외부 대형 모델 서비스에만 의존하면 됩니다.
 
-서비스를 소스에서 시작하려면:
+```bash
+git clone https://github.com/infiniflow/ragflow.git
+cd ragflow/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile -t infiniflow/ragflow:dev .
+```
 
-1. 레포지토리를 클론하세요: 
+## 🔨 소스 코드로 서비스를 시작합니다.
 
+1. Poetry를 설치하거나 이미 설치된 경우 이 단계를 건너뜁니다:
    ```bash
-   $ git clone https://github.com/infiniflow/ragflow.git
-   $ cd ragflow/
+   curl -sSL https://install.python-poetry.org | python3 -
    ```
 
-2. 가상 환경을 생성하고, Anaconda 또는 Miniconda가 설치되어 있는지 확인하세요:
+2. 소스 코드를 클론하고 Python 의존성을 설치합니다:
    ```bash
-   $ conda create -n ragflow python=3.11.0
-   $ conda activate ragflow
-   $ pip install -r requirements.txt
-   ```
-   
-   ```bash
-   # CUDA 버전이 12.0보다 높은 경우, 다음 명령어를 추가로 실행하세요:
-   $ pip uninstall -y onnxruntime-gpu
-   $ pip install onnxruntime-gpu --extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
+   git clone https://github.com/infiniflow/ragflow.git
+   cd ragflow/
+   export POETRY_VIRTUALENVS_CREATE=true POETRY_VIRTUALENVS_IN_PROJECT=true
+   ~/.local/bin/poetry install --sync --no-root # install RAGFlow dependent python modules
    ```
 
-3. 진입 스크립트를 복사하고 환경 변수를 설정하세요:
+3. Docker Compose를 사용하여 의존 서비스(MinIO, Elasticsearch, Redis 및 MySQL)를 시작합니다:
    ```bash
-   # 파이썬 경로를 받아옵니다:
-   $ which python
-   # RAGFlow 프로젝트 경로를 받아옵니다:
-   $ pwd
-   ```
-   
-   ```bash
-   $ cp docker/entrypoint.sh .
-   $ vi entrypoint.sh
+   docker compose -f docker/docker-compose-base.yml up -d
    ```
 
+   `/etc/hosts` 에 다음 줄을 추가하여 **docker/service_conf.yaml** 에 지정된 모든 호스트를 `127.0.0.1` 로 해결합니다:  
+   ```
+   127.0.0.1       es01 infinity mysql minio redis
+   ```  
+   **docker/service_conf.yaml** 에서 mysql 포트를 `5455` 로, es 포트를 `1200` 으로 업데이트합니다( **docker/.env** 에 지정된 대로).
+
+4. HuggingFace에 접근할 수 없는 경우, `HF_ENDPOINT` 환경 변수를 설정하여 미러 사이트를 사용하세요:
+ 
    ```bash
-   # 실제 상황에 맞게 설정 조정하기 (다음 두 개의 export 명령어는 새로 추가되었습니다):
-   # - `which python`의 결과를 `PY`에 할당합니다.
-   # - `pwd`의 결과를 `PYTHONPATH`에 할당합니다.
-   # - `LD_LIBRARY_PATH`가 설정되어 있는 경우 주석 처리합니다.
-   # - 선택 사항: Hugging Face 미러 추가.
-   PY=${PY}
-   export PYTHONPATH=${PYTHONPATH}
    export HF_ENDPOINT=https://hf-mirror.com
    ```
 
-4. 다른 서비스(MinIO, Elasticsearch, Redis, MySQL)를 시작하세요:
+5. 백엔드 서비스를 시작합니다:
    ```bash
-   $ cd docker
-   $ docker compose -f docker-compose-base.yml up -d 
+   source .venv/bin/activate
+   export PYTHONPATH=$(pwd)
+   bash docker/launch_backend_service.sh
    ```
 
-5. 설정 파일을 확인하여 다음 사항을 확인하세요:
-  - **docker/.env**의 설정이 **conf/service_conf.yaml**의 설정과 일치하는지 확인합니다.
-  - **service_conf.yaml**의 관련 서비스에 대한 IP 주소와 포트가 로컬 머신의 IP 주소와 컨테이너에서 노출된 포트와 일치하는지 확인합니다.
-
-
-6. RAGFlow 백엔드 서비스를 시작합니다:
-
+6. 프론트엔드 의존성을 설치합니다:  
    ```bash
-   $ chmod +x ./entrypoint.sh
-   $ bash ./entrypoint.sh
+   cd web
+   npm install --force
+   ```  
+7. **.umirc.ts** 에서 `proxy.target` 을 `http://127.0.0.1:9380` 으로 업데이트합니다:
+8. 프론트엔드 서비스를 시작합니다:  
+   ```bash
+   npm run dev 
    ```
 
-7. 프론트엔드 서비스를 시작합니다:
+   _다음 인터페이스는 시스템이 성공적으로 시작되었음을 나타냅니다:_  
 
-   ```bash
-   $ cd web
-   $ npm install --registry=https://registry.npmmirror.com --force
-   $ vim .umirc.ts
-   # proxy.target을 http://127.0.0.1:9380로 업데이트합니다.
-   $ npm run dev 
-   ```
-
-8. 프론트엔드 서비스를 배포합니다:
-
-   ```bash
-   $ cd web
-   $ npm install --registry=https://registry.npmmirror.com --force
-   $ umi build
-   $ mkdir -p /ragflow/web
-   $ cp -r dist /ragflow/web
-   $ apt install nginx -y
-   $ cp ../docker/nginx/proxy.conf /etc/nginx
-   $ cp ../docker/nginx/nginx.conf /etc/nginx
-   $ cp ../docker/nginx/ragflow.conf /etc/nginx/conf.d
-   $ systemctl start nginx
-   ```
+   ![](https://github.com/user-attachments/assets/0daf462c-a24d-4496-a66f-92533534e187)
 
 ## 📚 문서
 
 - [Quickstart](https://ragflow.io/docs/dev/)
-- [User guide](https://ragflow.io/docs/dev/category/user-guides)
+- [User guide](https://ragflow.io/docs/dev/category/guides)
 - [References](https://ragflow.io/docs/dev/category/references)
 - [FAQ](https://ragflow.io/docs/dev/faq)
 
@@ -331,4 +320,4 @@ $ docker compose up -d
 
 ## 🙌 컨트리뷰션
 
-RAGFlow는 오픈소스 협업을 통해 발전합니다. 이러한 정신을 바탕으로, 우리는 커뮤니티의 다양한 기여를 환영합니다. 참여하고 싶으시다면, 먼저 [가이드라인](./docs/references/CONTRIBUTING.md)을 검토해 주세요.
+RAGFlow는 오픈소스 협업을 통해 발전합니다. 이러한 정신을 바탕으로, 우리는 커뮤니티의 다양한 기여를 환영합니다. 참여하고 싶으시다면, 먼저 [가이드라인](./CONTRIBUTING.md)을 검토해 주세요.

README_zh.md

@@ -8,22 +8,29 @@
   <a href="./README.md">English</a> |
   <a href="./README_zh.md">简体中文</a> |
   <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a>
+  <a href="./README_ko.md">한국어</a> |
+  <a href="./README_id.md">Bahasa Indonesia</a>
 </p>
 
 <p align="center">
+    <a href="https://x.com/intent/follow?screen_name=infiniflowai" target="_blank">
+        <img src="https://img.shields.io/twitter/follow/infiniflow?logo=X&color=%20%23f5f5f5" alt="follow on X(Twitter)">
+    </a>
+    <a href="https://demo.ragflow.io" target="_blank">
+        <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/badge/docker_pull-ragflow:v0.14.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.14.0">
+    </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">
     </a>
-    <a href="https://demo.ragflow.io" target="_blank">
-        <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/badge/docker_pull-ragflow:v0.10.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.10.0"></a>
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
-    <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
-  </a>
+        <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
+    </a>
 </p>
 
+
 <h4 align="center">
   <a href="https://ragflow.io/docs/dev/">Document</a> |
   <a href="https://github.com/infiniflow/ragflow/issues/162">Roadmap</a> |
@@ -41,25 +48,24 @@
 请登录网址 [https://demo.ragflow.io](https://demo.ragflow.io) 试用 demo。
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
 <img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/b083d173-dadc-4ea9-bdeb-180d7df514eb" width="1200"/>
+<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
 </div>
 
 
 ## 🔥 近期更新
 
-- 2024-08-22 支持用RAG技术实现从自然语言到SQL语句的转换。
+- 2024-11-22 完善了 Agent 中的变量定义和使用。
+- 2024-11-01 对解析后的 chunk 加入关键词抽取和相关问题生成以提高召回的准确度。
+- 2024-09-13 增加知识库问答搜索模式。
+- 2024-08-22 支持用 RAG 技术实现从自然语言到 SQL 语句的转换。
 - 2024-08-02 支持 GraphRAG 启发于 [graphrag](https://github.com/microsoft/graphrag) 和思维导图。
-- 2024-07-23 支持解析音频文件。
-- 2024-07-21 支持更多的大模型供应商(LocalAI/OpenRouter/StepFun/Nvidia)。
-- 2024-07-18 在Graph中支持算子：Wikipedia、PubMed、Baidu和Duckduckgo。
-- 2024-07-08 支持 Agentic RAG: 基于 [Graph](./graph/README.md) 的工作流。
-- 2024-06-27 Q&A 解析方式支持 Markdown 文件和 Docx 文件。
-- 2024-06-27 支持提取出 Docx 文件中的图片。
-- 2024-06-27 支持提取出 Markdown 文件中的表格。
-- 2024-06-06 支持 [Self-RAG](https://huggingface.co/papers/2310.11511) ，在对话设置里面默认勾选。
-- 2024-05-30 集成 [BCE](https://github.com/netease-youdao/BCEmbedding) 和 [BGE](https://github.com/FlagOpen/FlagEmbedding) 重排序模型。
-- 2024-05-23 实现 [RAPTOR](https://arxiv.org/html/2401.18059v1) 提供更好的文本检索。
-- 2024-05-15 集成大模型 OpenAI GPT-4o。
+
+## 🎉 关注项目
+⭐️点击右上角的 Star 关注RAGFlow，可以获取最新发布的实时通知 !🌟
+<div align="center" style="margin-top:20px;margin-bottom:20px;">
+<img src="https://github.com/user-attachments/assets/18c9707e-b8aa-4caf-a154-037089c105ba" width="1200"/>
+</div>
+
 
 ## 🌟 主要功能
 
@@ -136,16 +142,18 @@
 
 3. 进入 **docker** 文件夹，利用提前编译好的 Docker 镜像启动服务器：
 
+   > 运行以下命令会自动下载 dev 版的 RAGFlow slim Docker 镜像（`dev-slim`），该镜像并不包含 embedding 模型以及一些 Python 库，因此镜像大小约 1GB。
+
    ```bash
    $ cd ragflow/docker
-   $ chmod +x ./entrypoint.sh
-   $ docker compose -f docker-compose-CN.yml up -d
+   $ docker compose -f docker-compose.yml up -d
    ```
 
-   > 请注意，运行上述命令会自动下载 RAGFlow 的开发版本 docker 镜像。如果你想下载并运行特定版本的 docker 镜像，请在 docker/.env 文件中找到 RAGFLOW_VERSION 变量，将其改为对应版本。例如 RAGFLOW_VERSION=v0.10.0，然后运行上述命令。
-
-   > 核心镜像文件大约 9 GB，可能需要一定时间拉取。请耐心等待。
-
+   > - 如果你想下载并运行特定版本的 RAGFlow slim Docker 镜像，请在 **docker/.env** 文件中找到 `RAGFLOW_IMAGE` 变量，将其改为对应版本。例如 `RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0-slim`，然后再运行上述命令。
+   > - 如果您想安装内置 embedding 模型和 Python 库的 dev 版本的 Docker 镜像，需要将 **docker/.env** 文件中的 `RAGFLOW_IMAGE` 变量修改为： `RAGFLOW_IMAGE=infiniflow/ragflow:dev`。
+   > - 如果您想安装内置 embedding 模型和 Python 库的指定版本的 RAGFlow Docker 镜像，需要将 **docker/.env** 文件中的 `RAGFLOW_IMAGE` 变量修改为： `RAGFLOW_IMAGE=infiniflow/ragflow:v0.14.0`。修改后，再运行上面的命令。
+   > **注意：** 安装内置 embedding 模型和 Python 库的指定版本的 RAGFlow Docker 镜像大小约 9 GB，可能需要更长时间下载，请耐心等待。
+   
 4. 服务器启动成功后再次确认服务器状态：
 
    ```bash
@@ -155,19 +163,18 @@
    _出现以下界面提示说明服务器启动成功：_
 
    ```bash
-       ____                 ______ __
-      / __ \ ____ _ ____ _ / ____// /____  _      __
-     / /_/ // __ `// __ `// /_   / // __ \| | /| / /
-    / _, _// /_/ // /_/ // __/  / // /_/ /| |/ |/ /
-   /_/ |_| \__,_/ \__, //_/    /_/ \____/ |__/|__/
-                 /____/
+        ____   ___    ______ ______ __               
+       / __ \ /   |  / ____// ____// /____  _      __
+      / /_/ // /| | / / __ / /_   / // __ \| | /| / /
+     / _, _// ___ |/ /_/ // __/  / // /_/ /| |/ |/ / 
+    /_/ |_|/_/  |_|\____//_/    /_/ \____/ |__/|__/  
 
     * Running on all addresses (0.0.0.0)
     * Running on http://127.0.0.1:9380
     * Running on http://x.x.x.x:9380
     INFO:werkzeug:Press CTRL+C to quit
    ```
-   > 如果您跳过这一步系统确认步骤就登录 RAGFlow，你的浏览器有可能会提示 `network anomaly` 或 `网络异常`，因为 RAGFlow 可能并未完全启动成功。
+   > 如果您跳过这一步系统确认步骤就登录 RAGFlow，你的浏览器有可能会提示 `network anormal` 或 `网络异常`，因为 RAGFlow 可能并未完全启动成功。
 
 5. 在你的浏览器中输入你的服务器对应的 IP 地址并登录 RAGFlow。
    > 上面这个例子中，您只需输入 http://IP_OF_YOUR_MACHINE 即可：未改动过配置则无需输入端口（默认的 HTTP 服务端口 80）。
@@ -183,126 +190,126 @@
 
 - [.env](./docker/.env)：存放一些基本的系统环境变量，比如 `SVR_HTTP_PORT`、`MYSQL_PASSWORD`、`MINIO_PASSWORD` 等。
 - [service_conf.yaml](./docker/service_conf.yaml)：配置各类后台服务。
-- [docker-compose-CN.yml](./docker/docker-compose-CN.yml): 系统依赖该文件完成启动。
+- [docker-compose.yml](./docker/docker-compose.yml): 系统依赖该文件完成启动。
 
 请务必确保 [.env](./docker/.env) 文件中的变量设置与 [service_conf.yaml](./docker/service_conf.yaml) 文件中的配置保持一致！
 
+如果不能访问镜像站点hub.docker.com或者模型站点huggingface.co，请按照[.env](./docker/.env)注释修改`RAGFLOW_IMAGE`和`HF_ENDPOINT`。
+
 > [./docker/README](./docker/README.md) 文件提供了环境变量设置和服务配置的详细信息。请**一定要**确保 [./docker/README](./docker/README.md) 文件当中列出来的环境变量的值与 [service_conf.yaml](./docker/service_conf.yaml) 文件当中的系统配置保持一致。
 
-如需更新默认的 HTTP 服务端口(80), 可以在 [docker-compose-CN.yml](./docker/docker-compose-CN.yml) 文件中将配置 `80:80` 改为 `<YOUR_SERVING_PORT>:80`。
+如需更新默认的 HTTP 服务端口(80), 可以在 [docker-compose.yml](./docker/docker-compose.yml) 文件中将配置 `80:80` 改为 `<YOUR_SERVING_PORT>:80`。
 
 > 所有系统配置都需要通过系统重启生效：
 >
 > ```bash
-> $ docker compose -f docker-compose-CN.yml up -d
+> $ docker compose -f docker-compose.yml up -d
 > ```
 
-## 🛠️ 源码编译、安装 Docker 镜像
+### 把文档引擎从 Elasticsearch 切换成为 Infinity
 
-如需从源码安装 Docker 镜像：
+RAGFlow 默认使用 Elasticsearch 存储文本和向量数据. 如果要切换为 [Infinity](https://github.com/infiniflow/infinity/), 可以按照下面步骤进行:
+
+1. 停止所有容器运行:
+
+   ```bash
+   $ docker compose -f docker/docker-compose.yml down -v
+   ```
+
+2. 设置 **docker/.env** 目录中的 `DOC_ENGINE` 为 `infinity`.
+
+3. 启动容器:
+
+   ```bash
+   $ docker compose -f docker/docker-compose.yml up -d
+   ```
+
+> [!WARNING] 
+> Infinity 目前官方并未正式支持在 Linux/arm64 架构下的机器上运行.
+
+
+## 🔧 源码编译 Docker 镜像（不含 embedding 模型）
+
+本 Docker 镜像大小约 1 GB 左右并且依赖外部的大模型和 embedding 服务。
 
 ```bash
-$ git clone https://github.com/infiniflow/ragflow.git
-$ cd ragflow/
-$ docker build -t infiniflow/ragflow:v0.10.0 .
-$ cd ragflow/docker
-$ chmod +x ./entrypoint.sh
-$ docker compose up -d
+git clone https://github.com/infiniflow/ragflow.git
+cd ragflow/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile.slim -t infiniflow/ragflow:dev-slim .
 ```
 
-## 🛠️ 源码启动服务
+## 🔧 源码编译 Docker 镜像（包含 embedding 模型）
 
-如需从源码启动服务，请参考以下步骤：
-
-1. 克隆仓库
+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/
+git clone https://github.com/infiniflow/ragflow.git
+cd ragflow/
+pip3 install huggingface-hub nltk
+python3 download_deps.py
+docker build -f Dockerfile -t infiniflow/ragflow:dev .
 ```
 
-2. 创建虚拟环境（确保已安装 Anaconda 或 Miniconda）
+## 🔨 以源代码启动服务
 
-```bash
-$ conda create -n ragflow python=3.11.0
-$ conda activate ragflow
-$ pip install -r requirements.txt
-```
-如果 cuda > 12.0，需额外执行以下命令：
-```bash
-$ pip uninstall -y onnxruntime-gpu
-$ pip install onnxruntime-gpu --extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
-```
+1. 安装 Poetry。如已经安装，可跳过本步骤：  
+   ```bash
+   curl -sSL https://install.python-poetry.org | python3 -
+   ```
 
-3. 拷贝入口脚本并配置环境变量
+2. 下载源代码并安装 Python 依赖：  
+   ```bash
+   git clone https://github.com/infiniflow/ragflow.git
+   cd ragflow/
+   export POETRY_VIRTUALENVS_CREATE=true POETRY_VIRTUALENVS_IN_PROJECT=true
+   ~/.local/bin/poetry install --sync --no-root # install RAGFlow dependent python modules
+   ```
 
-```bash
-$ cp docker/entrypoint.sh .
-$ vi entrypoint.sh
-```
-使用以下命令获取python路径及ragflow项目路径：
-```bash
-$ which python
-$ pwd
-```
+3. 通过 Docker Compose 启动依赖的服务（MinIO, Elasticsearch, Redis, and MySQL）：  
+   ```bash
+   docker compose -f docker/docker-compose-base.yml up -d
+   ```
 
-将上述 `which python` 的输出作为 `PY` 的值，将 `pwd` 的输出作为 `PYTHONPATH` 的值。
+   在 `/etc/hosts` 中添加以下代码，将 **docker/service_conf.yaml** 文件中的所有 host 地址都解析为 `127.0.0.1`：  
+   ```
+   127.0.0.1       es01 infinity mysql minio redis
+   ```  
+   在文件 **docker/service_conf.yaml** 中，对照 **docker/.env** 的配置将 mysql 端口更新为 `5455`，es 端口更新为 `1200`。
 
-`LD_LIBRARY_PATH` 如果环境已经配置好，可以注释掉。
+4. 如果无法访问 HuggingFace，可以把环境变量 `HF_ENDPOINT` 设成相应的镜像站点：  
+ 
+   ```bash
+   export HF_ENDPOINT=https://hf-mirror.com
+   ```
 
-```bash
-# 此处配置需要按照实际情况调整，两个 export 为新增配置
-PY=${PY}
-export PYTHONPATH=${PYTHONPATH}
-# 可选：添加 Hugging Face 镜像
-export HF_ENDPOINT=https://hf-mirror.com
-```
+5. 启动后端服务：  
+   ```bash
+   source .venv/bin/activate
+   export PYTHONPATH=$(pwd)
+   bash docker/launch_backend_service.sh
+   ```
 
-4. 启动基础服务
+6. 安装前端依赖：  
+   ```bash
+   cd web
+   npm install --force
+   ```  
+7. 配置前端，将 **.umirc.ts** 的 `proxy.target` 更新为 `http://127.0.0.1:9380`：  
+8. 启动前端服务：  
+   ```bash
+   npm run dev 
+   ```  
 
-```bash
-$ cd docker
-$ docker compose -f docker-compose-base.yml up -d 
-```
+   _以下界面说明系统已经成功启动：_  
 
-5. 检查配置文件
-确保**docker/.env**中的配置与**conf/service_conf.yaml**中配置一致， **service_conf.yaml**中相关服务的IP地址与端口应该改成本机IP地址及容器映射出来的端口。
+   ![](https://github.com/user-attachments/assets/0daf462c-a24d-4496-a66f-92533534e187)
 
-6. 启动服务
-
-```bash
-$ chmod +x ./entrypoint.sh
-$ bash ./entrypoint.sh
-```
-
-7. 启动WebUI服务
-
-```bash
-$ cd web
-$ npm install --registry=https://registry.npmmirror.com --force
-$ vim .umirc.ts
-# 修改proxy.target为http://127.0.0.1:9380
-$ npm run dev 
-```
-
-8. 部署WebUI服务
-
-```bash
-$ cd web
-$ npm install --registry=https://registry.npmmirror.com --force
-$ umi build
-$ mkdir -p /ragflow/web
-$ cp -r dist /ragflow/web
-$ apt install nginx -y
-$ cp ../docker/nginx/proxy.conf /etc/nginx
-$ cp ../docker/nginx/nginx.conf /etc/nginx
-$ cp ../docker/nginx/ragflow.conf /etc/nginx/conf.d
-$ systemctl start nginx
-```
 ## 📚 技术文档
 
 - [Quickstart](https://ragflow.io/docs/dev/)
-- [User guide](https://ragflow.io/docs/dev/category/user-guides)
+- [User guide](https://ragflow.io/docs/dev/category/guides)
 - [References](https://ragflow.io/docs/dev/category/references)
 - [FAQ](https://ragflow.io/docs/dev/faq)
 
@@ -318,7 +325,7 @@ $ systemctl start nginx
 
 ## 🙌 贡献指南
 
-RAGFlow 只有通过开源协作才能蓬勃发展。秉持这一精神,我们欢迎来自社区的各种贡献。如果您有意参与其中,请查阅我们的 [贡献者指南](./docs/references/CONTRIBUTING.md) 。
+RAGFlow 只有通过开源协作才能蓬勃发展。秉持这一精神,我们欢迎来自社区的各种贡献。如果您有意参与其中,请查阅我们的 [贡献者指南](./CONTRIBUTING.md) 。
 
 ## 🤝 商务合作
 

SECURITY.md

@@ -18,7 +18,7 @@ main
 ### Actual behavior
 
 The restricted_loads function at [api/utils/__init__.py#L215](https://github.com/infiniflow/ragflow/blob/main/api/utils/__init__.py#L215) is still vulnerable leading via code execution.
-The main reson is that numpy module has a numpy.f2py.diagnose.run_command function directly execute commands, but the restricted_loads function allows users import functions in module numpy.
+The main reason is that numpy module has a numpy.f2py.diagnose.run_command function directly execute commands, but the restricted_loads function allows users import functions in module numpy.
 
 
 ### Steps to reproduce

agent/canvas.py

@@ -13,19 +13,13 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-import importlib
+import logging
 import json
-import traceback
 from abc import ABC
 from copy import deepcopy
 from functools import partial
-
-import pandas as pd
-
 from agent.component import component_class
 from agent.component.base import ComponentBase
-from agent.settings import flow_logger, DEBUG
-
 
 class Canvas(ABC):
     """
@@ -162,8 +156,12 @@ class Canvas(ABC):
             self.components[k]["obj"].reset()
         self._embed_id = ""
 
+    def get_compnent_name(self, cid):
+        for n in self.dsl["graph"]["nodes"]:
+            if cid == n["id"]: return n["data"]["name"]
+        return ""
+
     def run(self, **kwargs):
-        ans = ""
         if self.answer:
             cpn_id = self.answer[0]
             self.answer.pop(0)
@@ -173,10 +171,10 @@ class Canvas(ABC):
                 ans = ComponentBase.be_output(str(e))
             self.path[-1].append(cpn_id)
             if kwargs.get("stream"):
-                assert isinstance(ans, partial)
-                return ans
-            self.history.append(("assistant", ans.to_dict("records")))
-            return ans
+                for an in ans():
+                    yield an
+            else: yield ans
+            return
 
         if not self.path:
             self.components["begin"]["obj"].run(self.history, **kwargs)
@@ -184,6 +182,8 @@ class Canvas(ABC):
 
         self.path.append([])
         ran = -1
+        waiting = []
+        without_dependent_checking = []
 
         def prepare2run(cpns):
             nonlocal ran, ans
@@ -193,18 +193,22 @@ class Canvas(ABC):
                 if cpn.component_name == "Answer":
                     self.answer.append(c)
                 else:
-                    if DEBUG: print("RUN: ", c)
-                    if cpn.component_name == "Generate":
+                    logging.debug(f"Canvas.prepare2run: {c}")
+                    if c not in without_dependent_checking:
                         cpids = cpn.get_dependent_components()
-                        if any([c not in self.path[-1] for c in cpids]):
+                        if any([cc not in self.path[-1] for cc in cpids]):
+                            if c not in waiting: waiting.append(c)
                             continue
+                    yield "*'{}'* is running...🕞".format(self.get_compnent_name(c))
                     ans = cpn.run(self.history, **kwargs)
                     self.path[-1].append(c)
             ran += 1
 
-        prepare2run(self.components[self.path[-2][-1]]["downstream"])
+        for m in prepare2run(self.components[self.path[-2][-1]]["downstream"]):
+            yield {"content": m, "running_status": True}
+
         while 0 <= ran < len(self.path[-1]):
-            if DEBUG: print(ran, self.path)
+            logging.debug(f"Canvas.run: {ran} {self.path}")
             cpn_id = self.path[-1][ran]
             cpn = self.get_component(cpn_id)
             if not cpn["downstream"]: break
@@ -217,27 +221,26 @@ class Canvas(ABC):
                 assert switch_out in self.components, \
                     "{}'s output: {} not valid.".format(cpn_id, switch_out)
                 try:
-                    prepare2run([switch_out])
+                    for m in prepare2run([switch_out]):
+                        yield {"content": m, "running_status": True}
                 except Exception as e:
-                    for p in [c for p in self.path for c in p][::-1]:
-                        if p.lower().find("answer") >= 0:
-                            self.get_component(p)["obj"].set_exception(e)
-                            prepare2run([p])
-                            break
-                    traceback.print_exc()
-                    break
+                    yield {"content": "*Exception*: {}".format(e), "running_status": True}
+                    logging.exception("Canvas.run got exception")
                 continue
 
             try:
-                prepare2run(cpn["downstream"])
+                for m in prepare2run(cpn["downstream"]):
+                    yield {"content": m, "running_status": True}
             except Exception as e:
-                for p in [c for p in self.path for c in p][::-1]:
-                    if p.lower().find("answer") >= 0:
-                        self.get_component(p)["obj"].set_exception(e)
-                        prepare2run([p])
-                        break
-                traceback.print_exc()
-                break
+                yield {"content": "*Exception*: {}".format(e), "running_status": True}
+                logging.exception("Canvas.run got exception")
+
+            if ran >= len(self.path[-1]) and waiting:
+                without_dependent_checking = waiting
+                waiting = []
+                for m in prepare2run(without_dependent_checking):
+                    yield {"content": m, "running_status": True}
+                ran -= 1
 
         if self.answer:
             cpn_id = self.answer[0]
@@ -246,11 +249,13 @@ class Canvas(ABC):
             self.path[-1].append(cpn_id)
             if kwargs.get("stream"):
                 assert isinstance(ans, partial)
-                return ans
+                for an in ans():
+                    yield an
+            else:
+                yield ans
 
-            self.history.append(("assistant", ans.to_dict("records")))
-
-        return ans
+        else:
+            raise Exception("The dialog flow has no way to interact with you. Please add an 'Interact' component to the end of the flow.")
 
     def get_component(self, cpn_id):
         return self.components[cpn_id]
@@ -260,9 +265,11 @@ class Canvas(ABC):
 
     def get_history(self, window_size):
         convs = []
-        for role, obj in self.history[window_size * -2:]:
-            convs.append({"role": role, "content": (obj if role == "user" else
-                                                    '\n'.join(pd.DataFrame(obj)['content']))})
+        for role, obj in self.history[window_size * -1:]:
+            if isinstance(obj, list) and obj and all([isinstance(o, dict) for o in obj]):
+                convs.append({"role": role, "content": '\n'.join([str(s.get("content", "")) for s in obj])})
+            else:
+                convs.append({"role": role, "content": str(obj)})
         return convs
 
     def add_user_input(self, question):
@@ -300,3 +307,6 @@ class Canvas(ABC):
                 return pat + " => " + pat
 
         return False
+
+    def get_prologue(self):
+        return self.components["begin"]["obj"]._param.prologue

agent/component/__init__.py

@@ -9,6 +9,7 @@ from .relevant import Relevant, RelevantParam
 from .message import Message, MessageParam
 from .rewrite import RewriteQuestion, RewriteQuestionParam
 from .keyword import KeywordExtract, KeywordExtractParam
+from .concentrator import Concentrator, ConcentratorParam
 from .baidu import Baidu, BaiduParam
 from .duckduckgo import DuckDuckGo, DuckDuckGoParam
 from .wikipedia import Wikipedia, WikipediaParam
@@ -22,6 +23,15 @@ from .github import GitHub, GitHubParam
 from .baidufanyi import BaiduFanyi, BaiduFanyiParam
 from .qweather import QWeather, QWeatherParam
 from .exesql import ExeSQL, ExeSQLParam
+from .yahoofinance import YahooFinance, YahooFinanceParam
+from .wencai import WenCai, WenCaiParam
+from .jin10 import Jin10, Jin10Param
+from .tushare import TuShare, TuShareParam
+from .akshare import AkShare, AkShareParam
+from .crawler import Crawler, CrawlerParam
+from .invoke import Invoke, InvokeParam
+from .template import Template, TemplateParam
+
 
 def component_class(class_name):
     m = importlib.import_module("agent.component")

agent/component/akshare.py

@@ -0,0 +1,56 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+from abc import ABC
+import pandas as pd
+import akshare as ak
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class AkShareParam(ComponentParamBase):
+    """
+    Define the AkShare component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.top_n = 10
+
+    def check(self):
+        self.check_positive_integer(self.top_n, "Top N")
+
+
+class AkShare(ComponentBase, ABC):
+    component_name = "AkShare"
+
+    def _run(self, history, **kwargs):
+        ans = self.get_input()
+        ans = ",".join(ans["content"]) if "content" in ans else ""
+        if not ans:
+            return AkShare.be_output("")
+
+        try:
+            ak_res = []
+            stock_news_em_df = ak.stock_news_em(symbol=ans)
+            stock_news_em_df = stock_news_em_df.head(self._param.top_n)
+            ak_res = [{"content": '<a href="' + i["新闻链接"] + '">' + i["新闻标题"] + '</a>\n 新闻内容: ' + i[
+                "新闻内容"] + " \n发布时间:" + i["发布时间"] + " \n文章来源: " + i["文章来源"]} for index, i in stock_news_em_df.iterrows()]
+        except Exception as e:
+            return AkShare.be_output("**ERROR**: " + str(e))
+
+        if not ak_res:
+            return AkShare.be_output("")
+
+        return pd.DataFrame(ak_res)

agent/component/arxiv.py

@@ -13,13 +13,12 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 import arxiv
 import pandas as pd
-from agent.settings import DEBUG
 from agent.component.base import ComponentBase, ComponentParamBase
 
-
 class ArXivParam(ComponentParamBase):
     """
     Define the ArXiv component parameters.
@@ -65,5 +64,5 @@ class ArXiv(ComponentBase, ABC):
             return ArXiv.be_output("")
 
         df = pd.DataFrame(arxiv_res)
-        if DEBUG: print(df, ":::::::::::::::::::::::::::::::::")
+        logging.debug(f"df: {str(df)}")
         return df

agent/component/baidu.py

@@ -13,13 +13,11 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-import random
+import logging
 from abc import ABC
-from functools import partial
 import pandas as pd
 import requests
 import re
-from agent.settings import DEBUG
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -64,6 +62,6 @@ class Baidu(ComponentBase, ABC):
             return Baidu.be_output("")
 
         df = pd.DataFrame(baidu_res)
-        if DEBUG: print(df, ":::::::::::::::::::::::::::::::::")
+        logging.debug(f"df: {str(df)}")
         return df
 

agent/component/baidufanyi.py

@@ -36,7 +36,6 @@ class BaiduFanyiParam(ComponentParamBase):
         self.domain = 'finance'
 
     def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
         self.check_empty(self.appid, "BaiduFanyi APPID")
         self.check_empty(self.secret_key, "BaiduFanyi Secret Key")
         self.check_valid_value(self.trans_type, "Translate type", ['translate', 'fieldtranslate'])

agent/component/base.py

@@ -17,14 +17,13 @@ from abc import ABC
 import builtins
 import json
 import os
-from copy import deepcopy
+import logging
 from functools import partial
-from typing import List, Dict, Tuple, Union
+from typing import Tuple, Union
 
 import pandas as pd
 
 from agent import settings
-from agent.settings import flow_logger, DEBUG
 
 _FEEDED_DEPRECATED_PARAMS = "_feeded_deprecated_params"
 _DEPRECATED_PARAMS = "_deprecated_params"
@@ -36,6 +35,8 @@ class ComponentParamBase(ABC):
     def __init__(self):
         self.output_var_name = "output"
         self.message_history_window_size = 22
+        self.query = []
+        self.inputs = []
 
     def set_name(self, name: str):
         self._name = name
@@ -81,7 +82,6 @@ class ComponentParamBase(ABC):
         return {name: True for name in self.get_feeded_deprecated_params()}
 
     def __str__(self):
-
         return json.dumps(self.as_dict(), ensure_ascii=False)
 
     def as_dict(self):
@@ -359,13 +359,13 @@ class ComponentParamBase(ABC):
 
     def _warn_deprecated_param(self, param_name, descr):
         if self._deprecated_params_set.get(param_name):
-            flow_logger.warning(
+            logging.warning(
                 f"{descr} {param_name} is deprecated and ignored in this version."
             )
 
     def _warn_to_deprecate_param(self, param_name, descr, new_param):
         if self._deprecated_params_set.get(param_name):
-            flow_logger.warning(
+            logging.warning(
                 f"{descr} {param_name} will be deprecated in future release; "
                 f"please use {new_param} instead."
             )
@@ -385,10 +385,14 @@ class ComponentBase(ABC):
         """
         return """{{
             "component_name": "{}",
-            "params": {}
+            "params": {},
+            "output": {},
+            "inputs": {}
         }}""".format(self.component_name,
-                     self._param
-                     )
+                     self._param,
+                     json.dumps(json.loads(str(self._param)).get("output", {}), ensure_ascii=False),
+                     json.dumps(json.loads(str(self._param)).get("inputs", []), ensure_ascii=False)
+        )
 
     def __init__(self, canvas, id, param: ComponentParamBase):
         self._canvas = canvas
@@ -396,8 +400,15 @@ class ComponentBase(ABC):
         self._param = param
         self._param.check()
 
+    def get_dependent_components(self):
+        cpnts = set([para["component_id"].split("@")[0] for para in self._param.query \
+                     if para.get("component_id") \
+                     and para["component_id"].lower().find("answer") < 0 \
+                     and para["component_id"].lower().find("begin") < 0])
+        return list(cpnts)
+
     def run(self, history, **kwargs):
-        flow_logger.info("{}, history: {}, kwargs: {}".format(self, json.dumps(history, ensure_ascii=False),
+        logging.debug("{}, history: {}, kwargs: {}".format(self, json.dumps(history, ensure_ascii=False),
                                                               json.dumps(kwargs, ensure_ascii=False)))
         try:
             res = self._run(history, **kwargs)
@@ -431,48 +442,85 @@ class ComponentBase(ABC):
 
     def reset(self):
         setattr(self._param, self._param.output_var_name, None)
+        self._param.inputs = []
 
-    def set_output(self, v: pd.DataFrame):
+    def set_output(self, v: partial | pd.DataFrame):
         setattr(self._param, self._param.output_var_name, v)
 
     def get_input(self):
-        upstream_outs = []
         reversed_cpnts = []
         if len(self._canvas.path) > 1:
             reversed_cpnts.extend(self._canvas.path[-2])
         reversed_cpnts.extend(self._canvas.path[-1])
 
-        if DEBUG: print(self.component_name, reversed_cpnts[::-1])
+        if self._param.query:
+            self._param.inputs = []
+            outs = []
+            for q in self._param.query:
+                if q["component_id"]:
+                    if q["component_id"].split("@")[0].lower().find("begin") >= 0:
+                        cpn_id, key = q["component_id"].split("@")
+                        for p in self._canvas.get_component(cpn_id)["obj"]._param.query:
+                            if p["key"] == key:
+                                outs.append(pd.DataFrame([{"content": p.get("value", "")}]))
+                                self._param.inputs.append({"component_id": q["component_id"],
+                                                           "content": p.get("value", "")})
+                                break
+                        else:
+                            assert False, f"Can't find parameter '{key}' for {cpn_id}"
+                        continue
+
+                    outs.append(self._canvas.get_component(q["component_id"])["obj"].output(allow_partial=False)[1])
+                    self._param.inputs.append({"component_id": q["component_id"],
+                                               "content": "\n".join(
+                                                   [str(d["content"]) for d in outs[-1].to_dict('records')])})
+                elif q["value"]:
+                    self._param.inputs.append({"component_id": None, "content": q["value"]})
+                    outs.append(pd.DataFrame([{"content": q["value"]}]))
+            if outs:
+                df = pd.concat(outs, ignore_index=True)
+                if "content" in df: df = df.drop_duplicates(subset=['content']).reset_index(drop=True)
+                return df
+
+        upstream_outs = []
+
         for u in reversed_cpnts[::-1]:
-            if self.get_component_name(u) in ["switch"]: continue
+            if self.get_component_name(u) in ["switch", "concentrator"]: continue
             if self.component_name.lower() == "generate" and self.get_component_name(u) == "retrieval":
                 o = self._canvas.get_component(u)["obj"].output(allow_partial=False)[1]
                 if o is not None:
+                    o["component_id"] = u
                     upstream_outs.append(o)
                     continue
-            if u not in self._canvas.get_component(self._id)["upstream"]: continue
+            #if self.component_name.lower()!="answer" and u not in self._canvas.get_component(self._id)["upstream"]: continue
             if self.component_name.lower().find("switch") < 0 \
                     and self.get_component_name(u) in ["relevant", "categorize"]:
                 continue
             if u.lower().find("answer") >= 0:
                 for r, c in self._canvas.history[::-1]:
                     if r == "user":
-                        upstream_outs.append(pd.DataFrame([{"content": c}]))
+                        upstream_outs.append(pd.DataFrame([{"content": c, "component_id": u}]))
                         break
                 break
             if self.component_name.lower().find("answer") >= 0 and self.get_component_name(u) in ["relevant"]:
                 continue
             o = self._canvas.get_component(u)["obj"].output(allow_partial=False)[1]
             if o is not None:
+                o["component_id"] = u
                 upstream_outs.append(o)
             break
 
-        if upstream_outs:
-            df = pd.concat(upstream_outs, ignore_index=True)
-            if "content" in df:
-                df = df.drop_duplicates(subset=['content']).reset_index(drop=True)
-            return df
-        return pd.DataFrame()
+        assert upstream_outs, "Can't inference the where the component input is. Please identify whose output is this component's input."
+
+        df = pd.concat(upstream_outs, ignore_index=True)
+        if "content" in df:
+            df = df.drop_duplicates(subset=['content']).reset_index(drop=True)
+
+        self._param.inputs = []
+        for _, r in df.iterrows():
+            self._param.inputs.append({"component_id": r["component_id"], "content": r["content"]})
+
+        return df
 
     def get_stream_input(self):
         reversed_cpnts = []

agent/component/bing.py

@@ -13,13 +13,12 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 import requests
 import pandas as pd
-from agent.settings import DEBUG
 from agent.component.base import ComponentBase, ComponentParamBase
 
-
 class BingParam(ComponentParamBase):
     """
     Define the Bing component parameters.
@@ -81,5 +80,5 @@ class Bing(ComponentBase, ABC):
             return Bing.be_output("")
 
         df = pd.DataFrame(bing_res)
-        if DEBUG: print(df, ":::::::::::::::::::::::::::::::::")
+        logging.debug(f"df: {str(df)}")
         return df

agent/component/categorize.py

@@ -13,11 +13,11 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 from api.db import LLMType
 from api.db.services.llm_service import LLMBundle
 from agent.component import GenerateParam, Generate
-from agent.settings import DEBUG
 
 
 class CategorizeParam(GenerateParam):
@@ -34,7 +34,7 @@ class CategorizeParam(GenerateParam):
         super().check()
         self.check_empty(self.category_description, "[Categorize] Category examples")
         for k, v in self.category_description.items():
-            if not k: raise ValueError(f"[Categorize] Category name can not be empty!")
+            if not k: raise ValueError("[Categorize] Category name can not be empty!")
             if not v.get("to"): raise ValueError(f"[Categorize] 'To' of category {k} can not be empty!")
 
     def get_prompt(self):
@@ -73,15 +73,15 @@ class Categorize(Generate, ABC):
 
     def _run(self, history, **kwargs):
         input = self.get_input()
-        input = "Question: " + ("; ".join(input["content"]) if "content" in input else "") + "Category: "
+        input = "Question: " + (list(input["content"])[-1] if "content" in input else "") + "\tCategory: "
         chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
         ans = chat_mdl.chat(self._param.get_prompt(), [{"role": "user", "content": input}],
                             self._param.gen_conf())
-        if DEBUG: print(ans, ":::::::::::::::::::::::::::::::::", input)
+        logging.debug(f"input: {input}, answer: {str(ans)}")
         for c in self._param.category_description.keys():
             if ans.lower().find(c.lower()) >= 0:
                 return Categorize.be_output(self._param.category_description[c]["to"])
 
-        return Categorize.be_output(self._param.category_description.items()[-1][1]["to"])
+        return Categorize.be_output(list(self._param.category_description.items())[-1][1]["to"])
 
 

agent/component/cite.py

@@ -1,75 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-from abc import ABC
-
-import pandas as pd
-
-from api.db import LLMType
-from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.db.services.llm_service import LLMBundle
-from api.settings import retrievaler
-from agent.component.base import ComponentBase, ComponentParamBase
-
-
-class CiteParam(ComponentParamBase):
-
-    """
-    Define the Retrieval component parameters.
-    """
-    def __init__(self):
-        super().__init__()
-        self.cite_sources = []
-
-    def check(self):
-        self.check_empty(self.cite_source, "Please specify where you want to cite from.")
-
-
-class Cite(ComponentBase, ABC):
-    component_name = "Cite"
-
-    def _run(self, history, **kwargs):
-        input = "\n- ".join(self.get_input()["content"])
-        sources = [self._canvas.get_component(cpn_id).output()[1] for cpn_id in self._param.cite_source]
-        query = []
-        for role, cnt in history[::-1][:self._param.message_history_window_size]:
-            if role != "user":continue
-            query.append(cnt)
-        query = "\n".join(query)
-
-        kbs = KnowledgebaseService.get_by_ids(self._param.kb_ids)
-        if not kbs:
-            raise ValueError("Can't find knowledgebases by {}".format(self._param.kb_ids))
-        embd_nms = list(set([kb.embd_id for kb in kbs]))
-        assert len(embd_nms) == 1, "Knowledge bases use different embedding models."
-
-        embd_mdl = LLMBundle(kbs[0].tenant_id, LLMType.EMBEDDING, embd_nms[0])
-
-        rerank_mdl = None
-        if self._param.rerank_id:
-            rerank_mdl = LLMBundle(kbs[0].tenant_id, LLMType.RERANK, self._param.rerank_id)
-
-        kbinfos = retrievaler.retrieval(query, embd_mdl, kbs[0].tenant_id, self._param.kb_ids,
-                                        1, self._param.top_n,
-                                        self._param.similarity_threshold, 1 - self._param.keywords_similarity_weight,
-                                        aggs=False, rerank_mdl=rerank_mdl)
-
-        if not kbinfos["chunks"]: return pd.DataFrame()
-        df = pd.DataFrame(kbinfos["chunks"])
-        df["content"] = df["content_with_weight"]
-        del df["content_with_weight"]
-        return df
-
-

agent/component/concentrator.py

@@ -0,0 +1,36 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+from abc import ABC
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class ConcentratorParam(ComponentParamBase):
+    """
+    Define the Concentrator component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+
+    def check(self):
+        return True
+
+
+class Concentrator(ComponentBase, ABC):
+    component_name = "Concentrator"
+
+    def _run(self, history, **kwargs):
+        return Concentrator.be_output("")

agent/component/crawler.py

@@ -0,0 +1,70 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+from abc import ABC
+import asyncio
+from crawl4ai import AsyncWebCrawler
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class CrawlerParam(ComponentParamBase):
+    """
+    Define the Crawler component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.proxy = None
+        self.extract_type = "markdown"
+    
+    def check(self):
+        self.check_valid_value(self.extract_type, "Type of content from the crawler", ['html', 'markdown', 'content'])
+
+
+class Crawler(ComponentBase, ABC):
+    component_name = "Crawler"
+
+    def _run(self, history, **kwargs):
+        ans = self.get_input()
+        ans = " - ".join(ans["content"]) if "content" in ans else ""
+        if not ans:
+            return Crawler.be_output("")
+        try:
+            result = asyncio.run(self.get_web(ans))
+
+            return Crawler.be_output(result)
+            
+        except Exception as e:
+            return Crawler.be_output(f"An unexpected error occurred: {str(e)}")
+
+    async def get_web(self, url):
+        proxy = self._param.proxy if self._param.proxy else None
+        async with AsyncWebCrawler(verbose=True, proxy=proxy) as crawler:
+            result = await crawler.arun(
+                url=url,
+                bypass_cache=True
+            )
+            
+            if self._param.extract_type == 'html':
+                return result.cleaned_html
+            elif self._param.extract_type == 'markdown':
+                return result.markdown
+            elif self._param.extract_type == 'content':
+                result.extracted_content
+            return result.markdown
+            
+
+
+    

agent/component/duckduckgo.py

@@ -13,10 +13,10 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 from duckduckgo_search import DDGS
 import pandas as pd
-from agent.settings import DEBUG
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -62,5 +62,5 @@ class DuckDuckGo(ComponentBase, ABC):
             return DuckDuckGo.be_output("")
 
         df = pd.DataFrame(duck_res)
-        if DEBUG: print(df, ":::::::::::::::::::::::::::::::::")
+        logging.debug("df: {df}")
         return df

agent/component/exesql.py

@@ -16,7 +16,8 @@
 from abc import ABC
 import re
 import pandas as pd
-from peewee import MySQLDatabase, PostgresqlDatabase
+import pymysql
+import psycopg2
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -44,6 +45,9 @@ class ExeSQLParam(ComponentParamBase):
         self.check_positive_integer(self.port, "IP Port")
         self.check_empty(self.password, "Database password")
         self.check_positive_integer(self.top_n, "Number of records")
+        if self.database == "rag_flow":
+            if self.host == "ragflow-mysql": raise ValueError("The host is not accessible.")
+            if self.password == "infini_rag_flow": raise ValueError("The host is not accessible.")
 
 
 class ExeSQL(ComponentBase, ABC):
@@ -54,7 +58,7 @@ class ExeSQL(ComponentBase, ABC):
             setattr(self, "_loop", 0)
         if self._loop >= self._param.loop:
             self._loop = 0
-            raise Exception("Maximum loop time exceeds. Can't query the correct data via sql statement.")
+            raise Exception("Maximum loop time exceeds. Can't query the correct data via SQL statement.")
         self._loop += 1
 
         ans = self.get_input()
@@ -63,34 +67,37 @@ class ExeSQL(ComponentBase, ABC):
         ans = re.sub(r';.*?SELECT ', '; SELECT ', ans, flags=re.IGNORECASE)
         ans = re.sub(r';[^;]*$', r';', ans)
         if not ans:
-            return ExeSQL.be_output("SQL statement not found!")
+            raise Exception("SQL statement not found!")
 
         if self._param.db_type in ["mysql", "mariadb"]:
-            db = MySQLDatabase(self._param.database, user=self._param.username, host=self._param.host,
-                               port=self._param.port, password=self._param.password)
+            db = pymysql.connect(db=self._param.database, user=self._param.username, host=self._param.host,
+                                 port=self._param.port, password=self._param.password)
         elif self._param.db_type == 'postgresql':
-            db = PostgresqlDatabase(self._param.database, user=self._param.username, host=self._param.host,
-                                    port=self._param.port, password=self._param.password)
+            db = psycopg2.connect(dbname=self._param.database, user=self._param.username, host=self._param.host,
+                                  port=self._param.port, password=self._param.password)
 
         try:
-            db.connect()
+            cursor = db.cursor()
         except Exception as e:
-            return ExeSQL.be_output("**Error**: \nDatabase Connection Failed! \n" + str(e))
+            raise Exception("Database Connection Failed! \n" + str(e))
         sql_res = []
-        for single_sql in re.split(r';', ans):
+        for single_sql in re.split(r';', ans.replace(r"\n", " ")):
             if not single_sql:
                 continue
             try:
-                query = db.execute_sql(single_sql)
-                single_res = pd.DataFrame([i for i in query.fetchmany(size=self._param.top_n)])
-                single_res.columns = [i[0] for i in query.description]
-                sql_res.append({"content": "\nTotal: " + str(query.rowcount) + "\n" + single_res.to_markdown()})
+                cursor.execute(single_sql)
+                if cursor.rowcount == 0:
+                    sql_res.append({"content": "\nTotal: 0\n No record in the database!"})
+                    continue
+                single_res = pd.DataFrame([i for i in cursor.fetchmany(size=self._param.top_n)])
+                single_res.columns = [i[0] for i in cursor.description]
+                sql_res.append({"content": "\nTotal: " + str(cursor.rowcount) + "\n" + single_res.to_markdown()})
             except Exception as e:
                 sql_res.append({"content": "**Error**:" + str(e) + "\nError SQL Statement:" + single_sql})
                 pass
         db.close()
 
         if not sql_res:
-            return ExeSQL.be_output("No record in the database!")
+            return ExeSQL.be_output("")
 
         return pd.DataFrame(sql_res)

agent/component/generate.py

@@ -17,8 +17,9 @@ import re
 from functools import partial
 import pandas as pd
 from api.db import LLMType
+from api.db.services.dialog_service import message_fit_in
 from api.db.services.llm_service import LLMBundle
-from api.settings import retrievaler
+from api import settings
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -62,18 +63,22 @@ class Generate(ComponentBase):
     component_name = "Generate"
 
     def get_dependent_components(self):
-        cpnts = [para["component_id"] for para in self._param.parameters]
-        return cpnts
+        cpnts = set([para["component_id"].split("@")[0] for para in self._param.parameters \
+                     if para.get("component_id") \
+                     and para["component_id"].lower().find("answer") < 0 \
+                     and para["component_id"].lower().find("begin") < 0])
+        return list(cpnts)
 
     def set_cite(self, retrieval_res, answer):
         retrieval_res = retrieval_res.dropna(subset=["vector", "content_ltks"]).reset_index(drop=True)
         if "empty_response" in retrieval_res.columns:
             retrieval_res["empty_response"].fillna("", inplace=True)
-        answer, idx = retrievaler.insert_citations(answer, [ck["content_ltks"] for _, ck in retrieval_res.iterrows()],
-                                                   [ck["vector"] for _, ck in retrieval_res.iterrows()],
-                                                   LLMBundle(self._canvas.get_tenant_id(), LLMType.EMBEDDING,
-                                                             self._canvas.get_embedding_model()), tkweight=0.7,
-                                                   vtweight=0.3)
+        answer, idx = settings.retrievaler.insert_citations(answer,
+                                                            [ck["content_ltks"] for _, ck in retrieval_res.iterrows()],
+                                                            [ck["vector"] for _, ck in retrieval_res.iterrows()],
+                                                            LLMBundle(self._canvas.get_tenant_id(), LLMType.EMBEDDING,
+                                                                      self._canvas.get_embedding_model()), tkweight=0.7,
+                                                            vtweight=0.3)
         doc_ids = set([])
         recall_docs = []
         for i in idx:
@@ -100,20 +105,53 @@ class Generate(ComponentBase):
         chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
         prompt = self._param.prompt
 
-        retrieval_res = self.get_input()
-        input = ("  - " + "\n  - ".join(retrieval_res["content"])) if "content" in retrieval_res else ""
+        retrieval_res = []
+        self._param.inputs = []
         for para in self._param.parameters:
-            cpn = self._canvas.get_component(para["component_id"])["obj"]
+            if not para.get("component_id"): continue
+            component_id = para["component_id"].split("@")[0]
+            if para["component_id"].lower().find("@") >= 0:
+                cpn_id, key = para["component_id"].split("@")
+                for p in self._canvas.get_component(cpn_id)["obj"]._param.query:
+                    if p["key"] == key:
+                        kwargs[para["key"]] = p.get("value", "")
+                        self._param.inputs.append(
+                            {"component_id": para["component_id"], "content": kwargs[para["key"]]})
+                        break
+                else:
+                    assert False, f"Can't find parameter '{key}' for {cpn_id}"
+                continue
+
+            cpn = self._canvas.get_component(component_id)["obj"]
+            if cpn.component_name.lower() == "answer":
+                hist = self._canvas.get_history(1)
+                if hist:
+                    hist = hist[0]["content"]
+                else:
+                    hist = ""
+                kwargs[para["key"]] = hist
+                continue
             _, out = cpn.output(allow_partial=False)
             if "content" not in out.columns:
-                kwargs[para["key"]] = "Nothing"
+                kwargs[para["key"]] = ""
             else:
-                kwargs[para["key"]] = "  - " + "\n  - ".join(out["content"])
+                if cpn.component_name.lower() == "retrieval":
+                    retrieval_res.append(out)
+                kwargs[para["key"]] = "  - "+"\n - ".join([o if isinstance(o, str) else str(o) for o in out["content"]])
+            self._param.inputs.append({"component_id": para["component_id"], "content": kwargs[para["key"]]})
+
+        if retrieval_res:
+            retrieval_res = pd.concat(retrieval_res, ignore_index=True)
+        else: retrieval_res = pd.DataFrame([])
 
-        kwargs["input"] = input
         for n, v in kwargs.items():
-            # prompt = re.sub(r"\{%s\}"%n, re.escape(str(v)), prompt)
-            prompt = re.sub(r"\{%s\}" % n, str(v), prompt)
+            prompt = re.sub(r"\{%s\}" % re.escape(n), str(v).replace("\\", " "), prompt)
+
+        if not self._param.inputs and prompt.find("{input}") >= 0:
+            retrieval_res = self.get_input()
+            input = ("  - " + "\n  - ".join(
+                [c for c in retrieval_res["content"] if isinstance(c, str)])) if "content" in retrieval_res else ""
+            prompt = re.sub(r"\{input\}", re.escape(input), prompt)
 
         downstreams = self._canvas.get_component(self._id)["downstream"]
         if kwargs.get("stream") and len(downstreams) == 1 and self._canvas.get_component(downstreams[0])[
@@ -123,13 +161,17 @@ class Generate(ComponentBase):
         if "empty_response" in retrieval_res.columns and not "".join(retrieval_res["content"]):
             res = {"content": "\n- ".join(retrieval_res["empty_response"]) if "\n- ".join(
                 retrieval_res["empty_response"]) else "Nothing found in knowledgebase!", "reference": []}
-            return Generate.be_output(res)
+            return pd.DataFrame([res])
+
+        msg = self._canvas.get_history(self._param.message_history_window_size)
+        if len(msg) < 1: msg.append({"role": "user", "content": ""})
+        _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(chat_mdl.max_length * 0.97))
+        if len(msg) < 2: msg.append({"role": "user", "content": ""})
+        ans = chat_mdl.chat(msg[0]["content"], msg[1:], self._param.gen_conf())
 
-        ans = chat_mdl.chat(prompt, self._canvas.get_history(self._param.message_history_window_size),
-                            self._param.gen_conf())
         if self._param.cite and "content_ltks" in retrieval_res.columns and "vector" in retrieval_res.columns:
-            df = self.set_cite(retrieval_res, ans)
-            return pd.DataFrame(df)
+            res = self.set_cite(retrieval_res, ans)
+            return pd.DataFrame([res])
 
         return Generate.be_output(ans)
 
@@ -142,9 +184,12 @@ class Generate(ComponentBase):
             self.set_output(res)
             return
 
+        msg = self._canvas.get_history(self._param.message_history_window_size)
+        if len(msg) < 1: msg.append({"role": "user", "content": ""})
+        _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(chat_mdl.max_length * 0.97))
+        if len(msg) < 2: msg.append({"role": "user", "content": ""})
         answer = ""
-        for ans in chat_mdl.chat_streamly(prompt, self._canvas.get_history(self._param.message_history_window_size),
-                                          self._param.gen_conf()):
+        for ans in chat_mdl.chat_streamly(msg[0]["content"], msg[1:], self._param.gen_conf()):
             res = {"content": ans, "reference": []}
             answer = ans
             yield res

agent/component/github.py

@@ -13,10 +13,10 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 import pandas as pd
 import requests
-from agent.settings import DEBUG
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -57,5 +57,5 @@ class GitHub(ComponentBase, ABC):
             return GitHub.be_output("")
 
         df = pd.DataFrame(github_res)
-        if DEBUG: print(df, ":::::::::::::::::::::::::::::::::")
+        logging.debug(f"df: {df}")
         return df

agent/component/google.py

@@ -13,10 +13,10 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 from serpapi import GoogleSearch
 import pandas as pd
-from agent.settings import DEBUG
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -85,12 +85,12 @@ class Google(ComponentBase, ABC):
                  "hl": self._param.language, "num": self._param.top_n})
             google_res = [{"content": '<a href="' + i["link"] + '">' + i["title"] + '</a>    ' + i["snippet"]} for i in
                           client.get_dict()["organic_results"]]
-        except Exception as e:
+        except Exception:
             return Google.be_output("**ERROR**: Existing Unavailable Parameters!")
 
         if not google_res:
             return Google.be_output("")
 
         df = pd.DataFrame(google_res)
-        if DEBUG: print(df, ":::::::::::::::::::::::::::::::::")
+        logging.debug(f"df: {df}")
         return df

agent/component/googlescholar.py

@@ -13,9 +13,9 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 import pandas as pd
-from agent.settings import DEBUG
 from agent.component.base import ComponentBase, ComponentParamBase
 from scholarly import scholarly
 
@@ -58,13 +58,13 @@ class GoogleScholar(ComponentBase, ABC):
                     'pub_url'] + '"></a> ' + "\n author: " + ",".join(pub['bib']['author']) + '\n Abstract: ' + pub[
                                                    'bib'].get('abstract', 'no abstract')})
 
-            except StopIteration or Exception as e:
-                print("**ERROR** " + str(e))
+            except StopIteration or Exception:
+                logging.exception("GoogleScholar")
                 break
 
         if not scholar_res:
             return GoogleScholar.be_output("")
 
         df = pd.DataFrame(scholar_res)
-        if DEBUG: print(df, ":::::::::::::::::::::::::::::::::")
+        logging.debug(f"df: {df}")
         return df

agent/component/invoke.py

@@ -0,0 +1,106 @@
+#
+#  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 json
+import re
+from abc import ABC
+import requests
+from deepdoc.parser import HtmlParser
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class InvokeParam(ComponentParamBase):
+    """
+    Define the Crawler component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.proxy = None
+        self.headers = ""
+        self.method = "get"
+        self.variables = []
+        self.url = ""
+        self.timeout = 60
+        self.clean_html = False
+
+    def check(self):
+        self.check_valid_value(self.method.lower(), "Type of content from the crawler", ['get', 'post', 'put'])
+        self.check_empty(self.url, "End point URL")
+        self.check_positive_integer(self.timeout, "Timeout time in second")
+        self.check_boolean(self.clean_html, "Clean HTML")
+
+
+class Invoke(ComponentBase, ABC):
+    component_name = "Invoke"
+
+    def _run(self, history, **kwargs):
+        args = {}
+        for para in self._param.variables:
+            if para.get("component_id"):
+                cpn = self._canvas.get_component(para["component_id"])["obj"]
+                if cpn.component_name.lower() == "answer":
+                    args[para["key"]] = self._canvas.get_history(1)[0]["content"]
+                    continue
+                _, out = cpn.output(allow_partial=False)
+                args[para["key"]] = "\n".join(out["content"])
+            else:
+                args[para["key"]] = "\n".join(para["value"])
+
+        url = self._param.url.strip()
+        if url.find("http") != 0:
+            url = "http://" + url
+
+        method = self._param.method.lower()
+        headers = {}
+        if self._param.headers:
+            headers = json.loads(self._param.headers)
+        proxies = None
+        if re.sub(r"https?:?/?/?", "", self._param.proxy):
+            proxies = {"http": self._param.proxy, "https": self._param.proxy}
+
+        if method == 'get':
+            response = requests.get(url=url,
+                                    params=args,
+                                    headers=headers,
+                                    proxies=proxies,
+                                    timeout=self._param.timeout)
+            if self._param.clean_html:
+                sections = HtmlParser()(None, response.content)
+                return Invoke.be_output("\n".join(sections))
+
+            return Invoke.be_output(response.text)
+
+        if method == 'put':
+            response = requests.put(url=url,
+                                    data=args,
+                                    headers=headers,
+                                    proxies=proxies,
+                                    timeout=self._param.timeout)
+            if self._param.clean_html:
+                sections = HtmlParser()(None, response.content)
+                return Invoke.be_output("\n".join(sections))
+            return Invoke.be_output(response.text)
+
+        if method == 'post':
+            response = requests.post(url=url,
+                                    json=args,
+                                    headers=headers,
+                                    proxies=proxies,
+                                    timeout=self._param.timeout)
+            if self._param.clean_html:
+                sections = HtmlParser()(None, response.content)
+                return Invoke.be_output("\n".join(sections))
+            return Invoke.be_output(response.text)

agent/component/jin10.py

@@ -0,0 +1,130 @@
+#
+#  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 json
+from abc import ABC
+import pandas as pd
+import requests
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class Jin10Param(ComponentParamBase):
+    """
+    Define the Jin10 component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.type = "flash"
+        self.secret_key = "xxx"
+        self.flash_type = '1'
+        self.calendar_type = 'cj'
+        self.calendar_datatype = 'data'
+        self.symbols_type = 'GOODS'
+        self.symbols_datatype = 'symbols'
+        self.contain = ""
+        self.filter = ""
+
+    def check(self):
+        self.check_valid_value(self.type, "Type", ['flash', 'calendar', 'symbols', 'news'])
+        self.check_valid_value(self.flash_type, "Flash Type", ['1', '2', '3', '4', '5'])
+        self.check_valid_value(self.calendar_type, "Calendar Type", ['cj', 'qh', 'hk', 'us'])
+        self.check_valid_value(self.calendar_datatype, "Calendar DataType", ['data', 'event', 'holiday'])
+        self.check_valid_value(self.symbols_type, "Symbols Type", ['GOODS', 'FOREX', 'FUTURE', 'CRYPTO'])
+        self.check_valid_value(self.symbols_datatype, 'Symbols DataType', ['symbols', 'quotes'])
+
+
+class Jin10(ComponentBase, ABC):
+    component_name = "Jin10"
+
+    def _run(self, history, **kwargs):
+        ans = self.get_input()
+        ans = " - ".join(ans["content"]) if "content" in ans else ""
+        if not ans:
+            return Jin10.be_output("")
+
+        jin10_res = []
+        headers = {'secret-key': self._param.secret_key}
+        try:
+            if self._param.type == "flash":
+                params = {
+                    'category': self._param.flash_type,
+                    'contain': self._param.contain,
+                    'filter': self._param.filter
+                }
+                response = requests.get(
+                    url='https://open-data-api.jin10.com/data-api/flash?category=' + self._param.flash_type,
+                    headers=headers, data=json.dumps(params))
+                response = response.json()
+                for i in response['data']:
+                    jin10_res.append({"content": i['data']['content']})
+            if self._param.type == "calendar":
+                params = {
+                    'category': self._param.calendar_type
+                }
+                response = requests.get(
+                    url='https://open-data-api.jin10.com/data-api/calendar/' + self._param.calendar_datatype + '?category=' + self._param.calendar_type,
+                    headers=headers, data=json.dumps(params))
+
+                response = response.json()
+                jin10_res.append({"content": pd.DataFrame(response['data']).to_markdown()})
+            if self._param.type == "symbols":
+                params = {
+                    'type': self._param.symbols_type
+                }
+                if self._param.symbols_datatype == "quotes":
+                    params['codes'] = 'BTCUSD'
+                response = requests.get(
+                    url='https://open-data-api.jin10.com/data-api/' + self._param.symbols_datatype + '?type=' + self._param.symbols_type,
+                    headers=headers, data=json.dumps(params))
+                response = response.json()
+                if self._param.symbols_datatype == "symbols":
+                    for i in response['data']:
+                        i['Commodity Code'] = i['c']
+                        i['Stock Exchange'] = i['e']
+                        i['Commodity Name'] = i['n']
+                        i['Commodity Type'] = i['t']
+                        del i['c'], i['e'], i['n'], i['t']
+                if self._param.symbols_datatype == "quotes":
+                    for i in response['data']:
+                        i['Selling Price'] = i['a']
+                        i['Buying Price'] = i['b']
+                        i['Commodity Code'] = i['c']
+                        i['Stock Exchange'] = i['e']
+                        i['Highest Price'] = i['h']
+                        i['Yesterday’s Closing Price'] = i['hc']
+                        i['Lowest Price'] = i['l']
+                        i['Opening Price'] = i['o']
+                        i['Latest Price'] = i['p']
+                        i['Market Quote Time'] = i['t']
+                        del i['a'], i['b'], i['c'], i['e'], i['h'], i['hc'], i['l'], i['o'], i['p'], i['t']
+                jin10_res.append({"content": pd.DataFrame(response['data']).to_markdown()})
+            if self._param.type == "news":
+                params = {
+                    'contain': self._param.contain,
+                    'filter': self._param.filter
+                }
+                response = requests.get(
+                    url='https://open-data-api.jin10.com/data-api/news',
+                    headers=headers, data=json.dumps(params))
+                response = response.json()
+                jin10_res.append({"content": pd.DataFrame(response['data']).to_markdown()})
+        except Exception as e:
+            return Jin10.be_output("**ERROR**: " + str(e))
+
+        if not jin10_res:
+            return Jin10.be_output("")
+
+        return pd.DataFrame(jin10_res)

agent/component/keyword.py

@@ -13,12 +13,12 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 import re
 from abc import ABC
 from api.db import LLMType
 from api.db.services.llm_service import LLMBundle
 from agent.component import GenerateParam, Generate
-from agent.settings import DEBUG
 
 
 class KeywordExtractParam(GenerateParam):
@@ -50,16 +50,13 @@ class KeywordExtract(Generate, ABC):
     component_name = "KeywordExtract"
 
     def _run(self, history, **kwargs):
-        q = ""
-        for r, c in self._canvas.history[::-1]:
-            if r == "user":
-                q += c
-                break
+        query = self.get_input()
+        query = str(query["content"][0]) if "content" in query else ""
 
         chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
-        ans = chat_mdl.chat(self._param.get_prompt(), [{"role": "user", "content": q}],
+        ans = chat_mdl.chat(self._param.get_prompt(), [{"role": "user", "content": query}],
                             self._param.gen_conf())
 
         ans = re.sub(r".*keyword:", "", ans).strip()
-        if DEBUG: print(ans, ":::::::::::::::::::::::::::::::::")
+        logging.debug(f"ans: {ans}")
         return KeywordExtract.be_output(ans)

agent/component/pubmed.py

@@ -13,11 +13,12 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 from Bio import Entrez
+import re
 import pandas as pd
 import xml.etree.ElementTree as ET
-from agent.settings import DEBUG
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -47,12 +48,15 @@ class PubMed(ComponentBase, ABC):
         try:
             Entrez.email = self._param.email
             pubmedids = Entrez.read(Entrez.esearch(db='pubmed', retmax=self._param.top_n, term=ans))['IdList']
-            pubmedcnt = ET.fromstring(
-                Entrez.efetch(db='pubmed', id=",".join(pubmedids), retmode="xml").read().decode("utf-8"))
+            pubmedcnt = ET.fromstring(re.sub(r'<(/?)b>|<(/?)i>', '', Entrez.efetch(db='pubmed', id=",".join(pubmedids),
+                                                                                   retmode="xml").read().decode(
+                "utf-8")))
             pubmed_res = [{"content": 'Title:' + child.find("MedlineCitation").find("Article").find(
                 "ArticleTitle").text + '\nUrl:<a href=" https://pubmed.ncbi.nlm.nih.gov/' + child.find(
-                "MedlineCitation").find("PMID").text + '">' + '</a>\n' + 'Abstract:' + child.find(
-                "MedlineCitation").find("Article").find("Abstract").find("AbstractText").text} for child in
+                "MedlineCitation").find("PMID").text + '">' + '</a>\n' + 'Abstract:' + (
+                                          child.find("MedlineCitation").find("Article").find("Abstract").find(
+                                              "AbstractText").text if child.find("MedlineCitation").find(
+                                              "Article").find("Abstract") else "No abstract available")} for child in
                           pubmedcnt.findall("PubmedArticle")]
         except Exception as e:
             return PubMed.be_output("**ERROR**: " + str(e))
@@ -61,5 +65,5 @@ class PubMed(ComponentBase, ABC):
             return PubMed.be_output("")
 
         df = pd.DataFrame(pubmed_res)
-        if DEBUG: print(df, ":::::::::::::::::::::::::::::::::")
+        logging.debug(f"df: {df}")
         return df

agent/component/qweather.py

@@ -51,7 +51,7 @@ class QWeatherParam(ComponentParamBase):
                                ['zh', 'zh-hant', 'en', 'de', 'es', 'fr', 'it', 'ja', 'ko', 'ru', 'hi', 'th', 'ar', 'pt',
                                 'bn', 'ms', 'nl', 'el', 'la', 'sv', 'id', 'pl', 'tr', 'cs', 'et', 'vi', 'fil', 'fi',
                                 'he', 'is', 'nb'])
-        self.check_vaild_value(self.time_period, "Time period", ['now', '3d', '7d', '10d', '15d', '30d'])
+        self.check_valid_value(self.time_period, "Time period", ['now', '3d', '7d', '10d', '15d', '30d'])
 
 
 class QWeather(ComponentBase, ABC):

agent/component/relevant.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 from api.db import LLMType
 from api.db.services.llm_service import LLMBundle
@@ -70,7 +71,7 @@ class Relevant(Generate, ABC):
         ans = chat_mdl.chat(self._param.get_prompt(), [{"role": "user", "content": ans}],
                             self._param.gen_conf())
 
-        print(ans, ":::::::::::::::::::::::::::::::::")
+        logging.debug(ans)
         if ans.lower().find("yes") >= 0:
             return Relevant.be_output(self._param.yes)
         if ans.lower().find("no") >= 0:

agent/component/retrieval.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 
 import pandas as pd
@@ -20,7 +21,7 @@ import pandas as pd
 from api.db import LLMType
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMBundle
-from api.settings import retrievaler
+from api import settings
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -43,22 +44,19 @@ class RetrievalParam(ComponentParamBase):
         self.check_decimal_float(self.similarity_threshold, "[Retrieval] Similarity threshold")
         self.check_decimal_float(self.keywords_similarity_weight, "[Retrieval] Keywords similarity weight")
         self.check_positive_number(self.top_n, "[Retrieval] Top N")
-        self.check_empty(self.kb_ids, "[Retrieval] Knowledge bases")
 
 
 class Retrieval(ComponentBase, ABC):
     component_name = "Retrieval"
 
     def _run(self, history, **kwargs):
-        query = []
-        for role, cnt in history[::-1][:self._param.message_history_window_size]:
-            if role != "user":continue
-            query.append(cnt)
-        query = "\n".join(query)
+        query = self.get_input()
+        query = str(query["content"][0]) if "content" in query else ""
 
         kbs = KnowledgebaseService.get_by_ids(self._param.kb_ids)
         if not kbs:
-            raise ValueError("Can't find knowledgebases by {}".format(self._param.kb_ids))
+            return Retrieval.be_output("")
+
         embd_nms = list(set([kb.embd_id for kb in kbs]))
         assert len(embd_nms) == 1, "Knowledge bases use different embedding models."
 
@@ -69,20 +67,21 @@ class Retrieval(ComponentBase, ABC):
         if self._param.rerank_id:
             rerank_mdl = LLMBundle(kbs[0].tenant_id, LLMType.RERANK, self._param.rerank_id)
 
-        kbinfos = retrievaler.retrieval(query, embd_mdl, kbs[0].tenant_id, self._param.kb_ids,
+        kbinfos = settings.retrievaler.retrieval(query, embd_mdl, kbs[0].tenant_id, self._param.kb_ids,
                                         1, self._param.top_n,
                                         self._param.similarity_threshold, 1 - self._param.keywords_similarity_weight,
                                         aggs=False, rerank_mdl=rerank_mdl)
 
         if not kbinfos["chunks"]:
             df = Retrieval.be_output("")
-            df["empty_response"] = self._param.empty_response
+            if self._param.empty_response and self._param.empty_response.strip():
+                df["empty_response"] = self._param.empty_response
             return df
 
         df = pd.DataFrame(kbinfos["chunks"])
         df["content"] = df["content_with_weight"]
         del df["content_with_weight"]
-        print(">>>>>>>>>>>>>>>>>>>>>>>>>>\n", query, df)
+        logging.debug("{} {}".format(query, df))
         return df
 
 

agent/component/rewrite.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from abc import ABC
 from api.db import LLMType
 from api.db.services.llm_service import LLMBundle
@@ -33,7 +34,7 @@ class RewriteQuestionParam(GenerateParam):
     def check(self):
         super().check()
 
-    def get_prompt(self):
+    def get_prompt(self, conv):
         self.prompt = """
         You are an expert at query expansion to generate a paraphrasing of a question.
         I can't retrieval relevant information from the knowledge base by using user's question directly.     
@@ -43,6 +44,40 @@ class RewriteQuestionParam(GenerateParam):
         And return 5 versions of question and one is from translation.
         Just list the question. No other words are needed.
         """
+        return f"""
+Role: A helpful assistant
+Task: Generate a full user question that would follow the conversation.
+Requirements & Restrictions:
+  - Text generated MUST be in the same language of the original user's question.
+  - If the user's latest question is completely, don't do anything, just return the original question.
+  - DON'T generate anything except a refined question.
+
+######################
+-Examples-
+######################
+# Example 1
+## Conversation
+USER: What is the name of Donald Trump's father?
+ASSISTANT:  Fred Trump.
+USER: And his mother?
+###############
+Output: What's the name of Donald Trump's mother?
+------------
+# Example 2
+## Conversation
+USER: What is the name of Donald Trump's father?
+ASSISTANT:  Fred Trump.
+USER: And his mother?
+ASSISTANT:  Mary Trump.
+User: What's her full name?
+###############
+Output: What's the full name of Donald Trump's mother Mary Trump?
+######################
+# Real Data
+## Conversation
+{conv}
+###############
+    """
         return self.prompt
 
 
@@ -54,19 +89,23 @@ class RewriteQuestion(Generate, ABC):
             setattr(self, "_loop", 0)
         if self._loop >= self._param.loop:
             self._loop = 0
-            raise Exception("Maximum loop time exceeds. Can't find relevant information.")
+            raise Exception("Sorry! Nothing relevant found.")
         self._loop += 1
-        q = "Question: "
-        for r, c in self._canvas.history[::-1]:
-            if r == "user":
-                q += c
-                break
+
+        hist = self._canvas.get_history(4)
+        conv = []
+        for m in hist:
+            if m["role"] not in ["user", "assistant"]: continue
+            conv.append("{}: {}".format(m["role"].upper(), m["content"]))
+        conv = "\n".join(conv)
 
         chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
-        ans = chat_mdl.chat(self._param.get_prompt(), [{"role": "user", "content": q}],
+        ans = chat_mdl.chat(self._param.get_prompt(conv), [{"role": "user", "content": "Output: "}],
                             self._param.gen_conf())
+        self._canvas.history.pop()
+        self._canvas.history.append(("user", ans))
 
-        print(ans, ":::::::::::::::::::::::::::::::::")
+        logging.debug(ans)
         return RewriteQuestion.be_output(ans)
 
 

agent/component/switch.py

@@ -42,43 +42,44 @@ class SwitchParam(ComponentParamBase):
         self.check_empty(self.conditions, "[Switch] conditions")
         for cond in self.conditions:
             if not cond["to"]: raise ValueError(f"[Switch] 'To' can not be empty!")
-            if cond["logical_operator"] not in ["and", "or"] and len(cond["items"]) > 1:
-                raise ValueError(f"[Switch] Please set logical_operator correctly!")
 
 
 class Switch(ComponentBase, ABC):
     component_name = "Switch"
 
+    def get_dependent_components(self):
+        res = []
+        for cond in self._param.conditions:
+            for item in cond["items"]:
+                if not item["cpn_id"]: continue
+                if item["cpn_id"].find("begin") >= 0:
+                    continue
+                cid = item["cpn_id"].split("@")[0]
+                res.append(cid)
+
+        return list(set(res))
+
     def _run(self, history, **kwargs):
         for cond in self._param.conditions:
-
-            if len(cond["items"]) == 1:
-                out = self._canvas.get_component(cond["items"][0]["cpn_id"])["obj"].output()[1]
-                cpn_input = "" if "content" not in out.columns else " ".join(out["content"])
-                if self.process_operator(cpn_input, cond["items"][0]["operator"], cond["items"][0]["value"]):
-                    return Switch.be_output(cond["to"])
-                continue
-
-            if cond["logical_operator"] == "and":
-                res = True
-                for item in cond["items"]:
-                    out = self._canvas.get_component(item["cpn_id"])["obj"].output()[1]
-                    cpn_input = "" if "content" not in out.columns else " ".join(out["content"])
-                    if not self.process_operator(cpn_input, item["operator"], item["value"]):
-                        res = False
-                        break
-                if res:
-                    return Switch.be_output(cond["to"])
-                continue
-
-            res = False
+            res = []
             for item in cond["items"]:
-                out = self._canvas.get_component(item["cpn_id"])["obj"].output()[1]
-                cpn_input = "" if "content" not in out.columns else " ".join(out["content"])
-                if self.process_operator(cpn_input, item["operator"], item["value"]):
-                    res = True
-                    break
-            if res:
+                if not item["cpn_id"]:continue
+                cid = item["cpn_id"].split("@")[0]
+                if item["cpn_id"].find("@") > 0:
+                    cpn_id, key = item["cpn_id"].split("@")
+                    for p in self._canvas.get_component(cid)["obj"]._param.query:
+                        if p["key"] == key:
+                            res.append(self.process_operator(p.get("value",""), item["operator"], item.get("value", "")))
+                            break
+                else:
+                    out = self._canvas.get_component(cid)["obj"].output()[1]
+                    cpn_input = "" if "content" not in out.columns else " ".join([str(s) for s in out["content"]])
+                    res.append(self.process_operator(cpn_input, item["operator"], item.get("value", "")))
+
+                if cond["logical_operator"] != "and" and any(res):
+                    return Switch.be_output(cond["to"])
+
+            if all(res):
                 return Switch.be_output(cond["to"])
 
         return Switch.be_output(self._param.end_cpn_id)
@@ -124,4 +125,4 @@ class Switch(ComponentBase, ABC):
             except Exception as e:
                 return True if input <= value else False
 
-        raise ValueError('Not supported operator' + operator)
+        raise ValueError('Not supported operator' + operator)

agent/component/template.py

@@ -0,0 +1,85 @@
+#
+#  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 re
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class TemplateParam(ComponentParamBase):
+    """
+    Define the Generate component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.content = ""
+        self.parameters = []
+
+    def check(self):
+        self.check_empty(self.content, "[Template] Content")
+        return True
+
+
+class Template(ComponentBase):
+    component_name = "Template"
+
+    def get_dependent_components(self):
+        cpnts = set([para["component_id"].split("@")[0] for para in self._param.parameters \
+                     if para.get("component_id") \
+                     and para["component_id"].lower().find("answer") < 0 \
+                     and para["component_id"].lower().find("begin") < 0])
+        return list(cpnts)
+
+    def _run(self, history, **kwargs):
+        content = self._param.content
+
+        self._param.inputs = []
+        for para in self._param.parameters:
+            if not para.get("component_id"): continue
+            component_id = para["component_id"].split("@")[0]
+            if para["component_id"].lower().find("@") >= 0:
+                cpn_id, key = para["component_id"].split("@")
+                for p in self._canvas.get_component(cpn_id)["obj"]._param.query:
+                    if p["key"] == key:
+                        kwargs[para["key"]] = p.get("value", "")
+                        self._param.inputs.append(
+                            {"component_id": para["component_id"], "content": kwargs[para["key"]]})
+                        break
+                else:
+                    assert False, f"Can't find parameter '{key}' for {cpn_id}"
+                continue
+
+            cpn = self._canvas.get_component(component_id)["obj"]
+            if cpn.component_name.lower() == "answer":
+                hist = self._canvas.get_history(1)
+                if hist:
+                    hist = hist[0]["content"]
+                else:
+                    hist = ""
+                kwargs[para["key"]] = hist
+                continue
+
+            _, out = cpn.output(allow_partial=False)
+            if "content" not in out.columns:
+                kwargs[para["key"]] = ""
+            else:
+                kwargs[para["key"]] = "  - "+"\n - ".join([o if isinstance(o, str) else str(o) for o in out["content"]])
+            self._param.inputs.append({"component_id": para["component_id"], "content": kwargs[para["key"]]})
+
+        for n, v in kwargs.items():
+            content = re.sub(r"\{%s\}" % re.escape(n), str(v).replace("\\", " "), content)
+
+        return Template.be_output(content)
+

agent/component/tushare.py

@@ -0,0 +1,72 @@
+#
+#  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 json
+from abc import ABC
+import pandas as pd
+import time
+import requests
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class TuShareParam(ComponentParamBase):
+    """
+    Define the TuShare component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.token = "xxx"
+        self.src = "eastmoney"
+        self.start_date = "2024-01-01 09:00:00"
+        self.end_date = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
+        self.keyword = ""
+
+    def check(self):
+        self.check_valid_value(self.src, "Quick News Source",
+                               ["sina", "wallstreetcn", "10jqka", "eastmoney", "yuncaijing", "fenghuang", "jinrongjie"])
+
+
+class TuShare(ComponentBase, ABC):
+    component_name = "TuShare"
+
+    def _run(self, history, **kwargs):
+        ans = self.get_input()
+        ans = ",".join(ans["content"]) if "content" in ans else ""
+        if not ans:
+            return TuShare.be_output("")
+
+        try:
+            tus_res = []
+            params = {
+                "api_name": "news",
+                "token": self._param.token,
+                "params": {"src": self._param.src, "start_date": self._param.start_date,
+                           "end_date": self._param.end_date}
+            }
+            response = requests.post(url="http://api.tushare.pro", data=json.dumps(params).encode('utf-8'))
+            response = response.json()
+            if response['code'] != 0:
+                return TuShare.be_output(response['msg'])
+            df = pd.DataFrame(response['data']['items'])
+            df.columns = response['data']['fields']
+            tus_res.append({"content": (df[df['content'].str.contains(self._param.keyword, case=False)]).to_markdown()})
+        except Exception as e:
+            return TuShare.be_output("**ERROR**: " + str(e))
+
+        if not tus_res:
+            return TuShare.be_output("")
+
+        return pd.DataFrame(tus_res)

agent/component/wencai.py

@@ -0,0 +1,80 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+from abc import ABC
+import pandas as pd
+import pywencai
+from agent.component.base import ComponentBase, ComponentParamBase
+
+
+class WenCaiParam(ComponentParamBase):
+    """
+    Define the WenCai component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.top_n = 10
+        self.query_type = "stock"
+
+    def check(self):
+        self.check_positive_integer(self.top_n, "Top N")
+        self.check_valid_value(self.query_type, "Query type",
+                               ['stock', 'zhishu', 'fund', 'hkstock', 'usstock', 'threeboard', 'conbond', 'insurance',
+                                'futures', 'lccp',
+                                'foreign_exchange'])
+
+
+class WenCai(ComponentBase, ABC):
+    component_name = "WenCai"
+
+    def _run(self, history, **kwargs):
+        ans = self.get_input()
+        ans = ",".join(ans["content"]) if "content" in ans else ""
+        if not ans:
+            return WenCai.be_output("")
+
+        try:
+            wencai_res = []
+            res = pywencai.get(query=ans, query_type=self._param.query_type, perpage=self._param.top_n)
+            if isinstance(res, pd.DataFrame):
+                wencai_res.append({"content": res.to_markdown()})
+            if isinstance(res, dict):
+                for item in res.items():
+                    if isinstance(item[1], list):
+                        wencai_res.append({"content": item[0] + "\n" + pd.DataFrame(item[1]).to_markdown()})
+                        continue
+                    if isinstance(item[1], str):
+                        wencai_res.append({"content": item[0] + "\n" + item[1]})
+                        continue
+                    if isinstance(item[1], dict):
+                        if "meta" in item[1].keys():
+                            continue
+                        wencai_res.append({"content": pd.DataFrame.from_dict(item[1], orient='index').to_markdown()})
+                        continue
+                    if isinstance(item[1], pd.DataFrame):
+                        if "image_url" in item[1].columns:
+                            continue
+                        wencai_res.append({"content": item[1].to_markdown()})
+                        continue
+                        
+                    wencai_res.append({"content": item[0] + "\n" + str(item[1])})
+        except Exception as e:
+            return WenCai.be_output("**ERROR**: " + str(e))
+
+        if not wencai_res:
+            return WenCai.be_output("")
+
+        return pd.DataFrame(wencai_res)

agent/component/wikipedia.py

@@ -13,12 +13,10 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-import random
+import logging
 from abc import ABC
-from functools import partial
 import wikipedia
 import pandas as pd
-from agent.settings import DEBUG
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -65,5 +63,5 @@ class Wikipedia(ComponentBase, ABC):
             return Wikipedia.be_output("")
 
         df = pd.DataFrame(wiki_res)
-        if DEBUG: print(df, ":::::::::::::::::::::::::::::::::")
+        logging.debug(f"df: {df}")
         return df

agent/component/yahoofinance.py

@@ -0,0 +1,84 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import logging
+from abc import ABC
+import pandas as pd
+from agent.component.base import ComponentBase, ComponentParamBase
+import yfinance as yf
+
+
+class YahooFinanceParam(ComponentParamBase):
+    """
+    Define the YahooFinance component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.info = True
+        self.history = False
+        self.count = False
+        self.financials = False
+        self.income_stmt = False
+        self.balance_sheet = False
+        self.cash_flow_statement = False
+        self.news = True
+
+    def check(self):
+        self.check_boolean(self.info, "get all stock info")
+        self.check_boolean(self.history, "get historical market data")
+        self.check_boolean(self.count, "show share count")
+        self.check_boolean(self.financials, "show financials")
+        self.check_boolean(self.income_stmt, "income statement")
+        self.check_boolean(self.balance_sheet, "balance sheet")
+        self.check_boolean(self.cash_flow_statement, "cash flow statement")
+        self.check_boolean(self.news, "show news")
+
+
+class YahooFinance(ComponentBase, ABC):
+    component_name = "YahooFinance"
+
+    def _run(self, history, **kwargs):
+        ans = self.get_input()
+        ans = "".join(ans["content"]) if "content" in ans else ""
+        if not ans:
+            return YahooFinance.be_output("")
+
+        yohoo_res = []
+        try:
+            msft = yf.Ticker(ans)
+            if self._param.info:
+                yohoo_res.append({"content": "info:\n" + pd.Series(msft.info).to_markdown() + "\n"})
+            if self._param.history:
+                yohoo_res.append({"content": "history:\n" + msft.history().to_markdown() + "\n"})
+            if self._param.financials:
+                yohoo_res.append({"content": "calendar:\n" + pd.DataFrame(msft.calendar).to_markdown() + "\n"})
+            if self._param.balance_sheet:
+                yohoo_res.append({"content": "balance sheet:\n" + msft.balance_sheet.to_markdown() + "\n"})
+                yohoo_res.append(
+                    {"content": "quarterly balance sheet:\n" + msft.quarterly_balance_sheet.to_markdown() + "\n"})
+            if self._param.cash_flow_statement:
+                yohoo_res.append({"content": "cash flow statement:\n" + msft.cashflow.to_markdown() + "\n"})
+                yohoo_res.append(
+                    {"content": "quarterly cash flow statement:\n" + msft.quarterly_cashflow.to_markdown() + "\n"})
+            if self._param.news:
+                yohoo_res.append({"content": "news:\n" + pd.DataFrame(msft.news).to_markdown() + "\n"})
+        except Exception:
+            logging.exception("YahooFinance got exception")
+
+        if not yohoo_res:
+            return YahooFinance.be_output("")
+
+        return pd.DataFrame(yohoo_res)

agent/settings.py

@@ -13,22 +13,6 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-# Logger
-import os
 
-from api.utils.file_utils import get_project_base_directory
-from api.utils.log_utils import LoggerFactory, getLogger
-
-DEBUG = 0
-LoggerFactory.set_directory(
-    os.path.join(
-        get_project_base_directory(),
-        "logs",
-        "flow"))
-# {CRITICAL: 50, FATAL:50, ERROR:40, WARNING:30, WARN:30, INFO:20, DEBUG:10, NOTSET:0}
-LoggerFactory.LEVEL = 30
-
-flow_logger = getLogger("flow")
-database_logger = getLogger("database")
 FLOAT_ZERO = 1e-8
 PARAM_MAXDEPTH = 5

agent/test/dsl_examples/categorize.json

@@ -26,20 +26,48 @@
                       "category_description": {
                         "product_related": {
                           "description": "The question is about the product usage, appearance and how it works.",
-                          "examples": "Why it always beaming?\nHow to install it onto the wall?\nIt leaks, what to do?"
+                          "examples": "Why it always beaming?\nHow to install it onto the wall?\nIt leaks, what to do?",
+                          "to": "message:0"
                         },
                         "others": {
                           "description": "The question is not about the product usage, appearance and how it works.",
-                          "examples": "How are you doing?\nWhat is your name?\nAre you a robot?\nWhat's the weather?\nWill it rain?"
+                          "examples": "How are you doing?\nWhat is your name?\nAre you a robot?\nWhat's the weather?\nWill it rain?",
+                          "to": "message:1"
                         }
                       }
                     }
                 },
-                "downstream": [],
+                "downstream": ["message:0","message:1"],
                 "upstream": ["answer:0"]
+            },
+            "message:0": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "messages": [
+                        "Message 0!!!!!!!"
+                      ]
+                    }
+                },
+                "downstream": ["answer:0"],
+                "upstream": ["categorize:0"]
+            },
+            "message:1": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "messages": [
+                        "Message 1!!!!!!!"
+                      ]
+                    }
+                },
+                "downstream": ["answer:0"],
+                "upstream": ["categorize:0"]
             }
   },
   "history": [],
+  "messages": [],
   "path": [],
+  "reference": [],
   "answer": []
-}
+}

agent/test/dsl_examples/concentrator_message.json

@@ -0,0 +1,113 @@
+{
+  "components": {
+            "begin": {
+                "obj":{
+                    "component_name": "Begin",
+                    "params": {
+                      "prologue": "Hi there!"
+                    }
+                },
+                "downstream": ["answer:0"],
+                "upstream": []
+            },
+            "answer:0": {
+                "obj": {
+                    "component_name": "Answer",
+                    "params": {}
+                },
+                "downstream": ["categorize:0"],
+                "upstream": ["begin"]
+            },
+            "categorize:0": {
+                "obj": {
+                    "component_name": "Categorize",
+                    "params": {
+                      "llm_id": "deepseek-chat",
+                      "category_description": {
+                        "product_related": {
+                          "description": "The question is about the product usage, appearance and how it works.",
+                          "examples": "Why it always beaming?\nHow to install it onto the wall?\nIt leaks, what to do?",
+                          "to": "concentrator:0"
+                        },
+                        "others": {
+                          "description": "The question is not about the product usage, appearance and how it works.",
+                          "examples": "How are you doing?\nWhat is your name?\nAre you a robot?\nWhat's the weather?\nWill it rain?",
+                          "to": "concentrator:1"
+                        }
+                      }
+                    }
+                },
+                "downstream": ["concentrator:0","concentrator:1"],
+                "upstream": ["answer:0"]
+            },
+            "concentrator:0": {
+                "obj": {
+                    "component_name": "Concentrator",
+                    "params": {}
+                },
+                "downstream": ["message:0"],
+                "upstream": ["categorize:0"]
+            },
+            "concentrator:1": {
+                "obj": {
+                    "component_name": "Concentrator",
+                    "params": {}
+                },
+                "downstream": ["message:1_0","message:1_1","message:1_2"],
+                "upstream": ["categorize:0"]
+            },
+            "message:0": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "messages": [
+                        "Message 0_0!!!!!!!"
+                      ]
+                    }
+                },
+                "downstream": ["answer:0"],
+                "upstream": ["concentrator:0"]
+            },
+            "message:1_0": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "messages": [
+                        "Message 1_0!!!!!!!"
+                      ]
+                    }
+                },
+                "downstream": ["answer:0"],
+                "upstream": ["concentrator:1"]
+            },
+            "message:1_1": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "messages": [
+                        "Message 1_1!!!!!!!"
+                      ]
+                    }
+                },
+                "downstream": ["answer:0"],
+                "upstream": ["concentrator:1"]
+            },
+            "message:1_2": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "messages": [
+                        "Message 1_2!!!!!!!"
+                      ]
+                    }
+                },
+                "downstream": ["answer:0"],
+                "upstream": ["concentrator:1"]
+            }
+  },
+  "history": [],
+  "messages": [],
+  "path": [],
+  "reference": [],
+  "answer": []
+}

api/apps/__init__.py

@@ -13,14 +13,16 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-import logging
 import os
 import sys
+import logging
 from importlib.util import module_from_spec, spec_from_file_location
 from pathlib import Path
 from flask import Blueprint, Flask
 from werkzeug.wrappers.request import Request
 from flask_cors import CORS
+from flasgger import Swagger
+from itsdangerous.url_safe import URLSafeTimedSerializer as Serializer
 
 from api.db import StatusEnum
 from api.db.db_models import close_connection
@@ -29,32 +31,60 @@ from api.utils import CustomJSONEncoder, commands
 
 from flask_session import Session
 from flask_login import LoginManager
-from api.settings import SECRET_KEY, stat_logger
-from api.settings import API_VERSION, access_logger
+from api import settings
 from api.utils.api_utils import server_error_response
-from itsdangerous.url_safe import URLSafeTimedSerializer as Serializer
+from api.constants import API_VERSION
 
-__all__ = ['app']
-
-
-logger = logging.getLogger('flask.app')
-for h in access_logger.handlers:
-    logger.addHandler(h)
+__all__ = ["app"]
 
 Request.json = property(lambda self: self.get_json(force=True, silent=True))
 
 app = Flask(__name__)
-CORS(app, supports_credentials=True,max_age=2592000)
+
+# Add this at the beginning of your file to configure Swagger UI
+swagger_config = {
+    "headers": [],
+    "specs": [
+        {
+            "endpoint": "apispec",
+            "route": "/apispec.json",
+            "rule_filter": lambda rule: True,  # Include all endpoints
+            "model_filter": lambda tag: True,  # Include all models
+        }
+    ],
+    "static_url_path": "/flasgger_static",
+    "swagger_ui": True,
+    "specs_route": "/apidocs/",
+}
+
+swagger = Swagger(
+    app,
+    config=swagger_config,
+    template={
+        "swagger": "2.0",
+        "info": {
+            "title": "RAGFlow API",
+            "description": "",
+            "version": "1.0.0",
+        },
+        "securityDefinitions": {
+            "ApiKeyAuth": {"type": "apiKey", "name": "Authorization", "in": "header"}
+        },
+    },
+)
+
+CORS(app, supports_credentials=True, max_age=2592000)
 app.url_map.strict_slashes = False
 app.json_encoder = CustomJSONEncoder
 app.errorhandler(Exception)(server_error_response)
 
-
 ## convince for dev and debug
-#app.config["LOGIN_DISABLED"] = True
+# app.config["LOGIN_DISABLED"] = True
 app.config["SESSION_PERMANENT"] = False
 app.config["SESSION_TYPE"] = "filesystem"
-app.config['MAX_CONTENT_LENGTH'] = int(os.environ.get("MAX_CONTENT_LENGTH", 128 * 1024 * 1024))
+app.config["MAX_CONTENT_LENGTH"] = int(
+    os.environ.get("MAX_CONTENT_LENGTH", 128 * 1024 * 1024)
+)
 
 Session(app)
 login_manager = LoginManager()
@@ -64,17 +94,23 @@ commands.register_commands(app)
 
 
 def search_pages_path(pages_dir):
-    app_path_list = [path for path in pages_dir.glob('*_app.py') if not path.name.startswith('.')]
-    api_path_list = [path for path in pages_dir.glob('*_api.py') if not path.name.startswith('.')]
+    app_path_list = [
+        path for path in pages_dir.glob("*_app.py") if not path.name.startswith(".")
+    ]
+    api_path_list = [
+        path for path in pages_dir.glob("*sdk/*.py") if not path.name.startswith(".")
+    ]
     app_path_list.extend(api_path_list)
     return app_path_list
 
 
 def register_page(page_path):
-    path = f'{page_path}'
+    path = f"{page_path}"
 
-    page_name = page_path.stem.rstrip('_api') if "_api" in path else page_path.stem.rstrip('_app')
-    module_name = '.'.join(page_path.parts[page_path.parts.index('api'):-1] + (page_name,))
+    page_name = page_path.stem.rstrip("_app")
+    module_name = ".".join(
+        page_path.parts[page_path.parts.index("api"): -1] + (page_name,)
+    )
 
     spec = spec_from_file_location(module_name, page_path)
     page = module_from_spec(spec)
@@ -82,8 +118,10 @@ def register_page(page_path):
     page.manager = Blueprint(page_name, module_name)
     sys.modules[module_name] = page
     spec.loader.exec_module(page)
-    page_name = getattr(page, 'page_name', page_name)
-    url_prefix = f'/api/{API_VERSION}/{page_name}' if "_api" in path else f'/{API_VERSION}/{page_name}'
+    page_name = getattr(page, "page_name", page_name)
+    url_prefix = (
+        f"/api/{API_VERSION}" if "/sdk/" in path else f"/{API_VERSION}/{page_name}"
+    )
 
     app.register_blueprint(page.manager, url_prefix=url_prefix)
     return url_prefix
@@ -91,30 +129,31 @@ def register_page(page_path):
 
 pages_dir = [
     Path(__file__).parent,
-    Path(__file__).parent.parent / 'api' / 'apps', # FIXME: ragflow/api/api/apps, can be remove?
+    Path(__file__).parent.parent / "api" / "apps",
+    Path(__file__).parent.parent / "api" / "apps" / "sdk",
 ]
 
 client_urls_prefix = [
-    register_page(path)
-    for dir in pages_dir
-    for path in search_pages_path(dir)
+    register_page(path) for dir in pages_dir for path in search_pages_path(dir)
 ]
 
 
 @login_manager.request_loader
 def load_user(web_request):
-    jwt = Serializer(secret_key=SECRET_KEY)
+    jwt = Serializer(secret_key=settings.SECRET_KEY)
     authorization = web_request.headers.get("Authorization")
     if authorization:
         try:
             access_token = str(jwt.loads(authorization))
-            user = UserService.query(access_token=access_token, status=StatusEnum.VALID.value)
+            user = UserService.query(
+                access_token=access_token, status=StatusEnum.VALID.value
+            )
             if user:
                 return user[0]
             else:
                 return None
-        except Exception as e:
-            stat_logger.exception(e)
+        except Exception:
+            logging.exception("load_user got exception")
             return None
     else:
         return None
@@ -122,4 +161,4 @@ def load_user(web_request):
 
 @app.teardown_request
 def _db_close(exc):
-    close_connection()
+    close_connection()

api/apps/api_app.py

@@ -22,35 +22,29 @@ from api.db.services.llm_service import TenantLLMService
 from flask_login import login_required, current_user
 
 from api.db import FileType, LLMType, ParserType, FileSource
-from api.db.db_models import APIToken, API4Conversation, Task, File
+from api.db.db_models import APIToken, Task, File
 from api.db.services import duplicate_name
 from api.db.services.api_service import APITokenService, API4ConversationService
-from api.db.services.dialog_service import DialogService, chat
+from api.db.services.dialog_service import DialogService, chat, keyword_extraction
 from api.db.services.document_service import DocumentService, doc_upload_and_parse
 from api.db.services.file2document_service import File2DocumentService
 from api.db.services.file_service import FileService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.task_service import queue_tasks, TaskService
 from api.db.services.user_service import UserTenantService
-from api.settings import RetCode, retrievaler
+from api import settings
 from api.utils import get_uuid, current_timestamp, datetime_format
-from api.utils.api_utils import server_error_response, get_data_error_result, get_json_result, validate_request
-from itsdangerous import URLSafeTimedSerializer
+from api.utils.api_utils import server_error_response, get_data_error_result, get_json_result, validate_request, \
+    generate_confirmation_token
 
 from api.utils.file_utils import filename_type, thumbnail
-from rag.nlp import keyword_extraction
-from rag.utils.minio_conn import MINIO
+from rag.utils.storage_factory import STORAGE_IMPL
 
-from api.db.services.canvas_service import CanvasTemplateService, UserCanvasService
+from api.db.services.canvas_service import UserCanvasService
 from agent.canvas import Canvas
 from functools import partial
 
 
-def generate_confirmation_token(tenent_id):
-    serializer = URLSafeTimedSerializer(tenent_id)
-    return "ragflow-" + serializer.dumps(get_uuid(), salt=tenent_id)[2:34]
-
-
 @manager.route('/new_token', methods=['POST'])
 @login_required
 def new_token():
@@ -58,7 +52,7 @@ def new_token():
     try:
         tenants = UserTenantService.query(user_id=current_user.id)
         if not tenants:
-            return get_data_error_result(retmsg="Tenant not found!")
+            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),
@@ -74,7 +68,7 @@ def new_token():
             obj["dialog_id"] = req["dialog_id"]
 
         if not APITokenService.save(**obj):
-            return get_data_error_result(retmsg="Fail to new a dialog!")
+            return get_data_error_result(message="Fail to new a dialog!")
 
         return get_json_result(data=obj)
     except Exception as e:
@@ -87,7 +81,7 @@ def token_list():
     try:
         tenants = UserTenantService.query(user_id=current_user.id)
         if not tenants:
-            return get_data_error_result(retmsg="Tenant not found!")
+            return get_data_error_result(message="Tenant not found!")
 
         id = request.args["dialog_id"] if "dialog_id" in request.args else request.args["canvas_id"]
         objs = APITokenService.query(tenant_id=tenants[0].tenant_id, dialog_id=id)
@@ -116,7 +110,7 @@ def stats():
     try:
         tenants = UserTenantService.query(user_id=current_user.id)
         if not tenants:
-            return get_data_error_result(retmsg="Tenant not found!")
+            return get_data_error_result(message="Tenant not found!")
         objs = API4ConversationService.stats(
             tenants[0].tenant_id,
             request.args.get(
@@ -147,18 +141,21 @@ def set_conversation():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
     req = request.json
     try:
         if objs[0].source == "agent":
-            e, c = UserCanvasService.get_by_id(objs[0].dialog_id)
+            e, cvs = UserCanvasService.get_by_id(objs[0].dialog_id)
             if not e:
                 return server_error_response("canvas not found.")
+            if not isinstance(cvs.dsl, str):
+                cvs.dsl = json.dumps(cvs.dsl, ensure_ascii=False)
+            canvas = Canvas(cvs.dsl, objs[0].tenant_id)
             conv = {
                 "id": get_uuid(),
-                "dialog_id": c.id,
+                "dialog_id": cvs.id,
                 "user_id": request.args.get("user_id", ""),
-                "message": [{"role": "assistant", "content": "Hi there!"}],
+                "message": [{"role": "assistant", "content": canvas.get_prologue()}],
                 "source": "agent"
             }
             API4ConversationService.save(**conv)
@@ -166,7 +163,7 @@ def set_conversation():
         else:
             e, dia = DialogService.get_by_id(objs[0].dialog_id)
             if not e:
-                return get_data_error_result(retmsg="Dialog not found")
+                return get_data_error_result(message="Dialog not found")
             conv = {
                 "id": get_uuid(),
                 "dialog_id": dia.id,
@@ -186,11 +183,11 @@ def completion():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
     req = request.json
     e, conv = API4ConversationService.get_by_id(req["conversation_id"])
     if not e:
-        return get_data_error_result(retmsg="Conversation not found!")
+        return get_data_error_result(message="Conversation not found!")
     if "quote" not in req: req["quote"] = False
 
     msg = []
@@ -199,15 +196,18 @@ def completion():
             continue
         if m["role"] == "assistant" and not msg:
             continue
-        msg.append({"role": m["role"], "content": m["content"]})
+        msg.append(m)
+    if not msg[-1].get("id"): msg[-1]["id"] = get_uuid()
+    message_id = msg[-1]["id"]
 
     def fillin_conv(ans):
-        nonlocal conv
+        nonlocal conv, message_id
         if not conv.reference:
             conv.reference.append(ans["reference"])
         else:
             conv.reference[-1] = ans["reference"]
-        conv.message[-1] = {"role": "assistant", "content": ans["answer"]}
+        conv.message[-1] = {"role": "assistant", "content": ans["answer"], "id": message_id}
+        ans["id"] = message_id
 
     def rename_field(ans):
         reference = ans['reference']
@@ -233,7 +233,7 @@ def completion():
 
             if not conv.reference:
                 conv.reference = []
-            conv.message.append({"role": "assistant", "content": ""})
+            conv.message.append({"role": "assistant", "content": "", "id": message_id})
             conv.reference.append({"chunks": [], "doc_aggs": []})
 
             final_ans = {"reference": [], "content": ""}
@@ -257,19 +257,20 @@ def completion():
                             ans = {"answer": ans["content"], "reference": ans.get("reference", [])}
                             fillin_conv(ans)
                             rename_field(ans)
-                            yield "data:" + json.dumps({"retcode": 0, "retmsg": "", "data": ans},
+                            yield "data:" + json.dumps({"code": 0, "message": "", "data": ans},
                                                        ensure_ascii=False) + "\n\n"
 
-                        canvas.messages.append({"role": "assistant", "content": final_ans["content"]})
+                        canvas.messages.append({"role": "assistant", "content": final_ans["content"], "id": message_id})
+                        canvas.history.append(("assistant", final_ans["content"]))
                         if final_ans.get("reference"):
                             canvas.reference.append(final_ans["reference"])
                         cvs.dsl = json.loads(str(canvas))
                         API4ConversationService.append_message(conv.id, conv.to_dict())
                     except Exception as e:
-                        yield "data:" + json.dumps({"retcode": 500, "retmsg": str(e),
+                        yield "data:" + json.dumps({"code": 500, "message": str(e),
                                                     "data": {"answer": "**ERROR**: " + str(e), "reference": []}},
                                                    ensure_ascii=False) + "\n\n"
-                    yield "data:" + json.dumps({"retcode": 0, "retmsg": "", "data": True}, ensure_ascii=False) + "\n\n"
+                    yield "data:" + json.dumps({"code": 0, "message": "", "data": True}, ensure_ascii=False) + "\n\n"
 
                 resp = Response(sse(), mimetype="text/event-stream")
                 resp.headers.add_header("Cache-control", "no-cache")
@@ -279,7 +280,7 @@ def completion():
                 return resp
 
             final_ans["content"] = "\n".join(answer["content"]) if "content" in answer else ""
-            canvas.messages.append({"role": "assistant", "content": final_ans["content"]})
+            canvas.messages.append({"role": "assistant", "content": final_ans["content"], "id": message_id})
             if final_ans.get("reference"):
                 canvas.reference.append(final_ans["reference"])
             cvs.dsl = json.loads(str(canvas))
@@ -289,18 +290,18 @@ def completion():
             API4ConversationService.append_message(conv.id, conv.to_dict())
             rename_field(result)
             return get_json_result(data=result)
-        
-        #******************For dialog******************
+
+        # ******************For dialog******************
         conv.message.append(msg[-1])
         e, dia = DialogService.get_by_id(conv.dialog_id)
         if not e:
-            return get_data_error_result(retmsg="Dialog not found!")
+            return get_data_error_result(message="Dialog not found!")
         del req["conversation_id"]
         del req["messages"]
 
         if not conv.reference:
             conv.reference = []
-        conv.message.append({"role": "assistant", "content": ""})
+        conv.message.append({"role": "assistant", "content": "", "id": message_id})
         conv.reference.append({"chunks": [], "doc_aggs": []})
 
         def stream():
@@ -309,14 +310,14 @@ def completion():
                 for ans in chat(dia, msg, True, **req):
                     fillin_conv(ans)
                     rename_field(ans)
-                    yield "data:" + json.dumps({"retcode": 0, "retmsg": "", "data": ans},
+                    yield "data:" + json.dumps({"code": 0, "message": "", "data": ans},
                                                ensure_ascii=False) + "\n\n"
                 API4ConversationService.append_message(conv.id, conv.to_dict())
             except Exception as e:
-                yield "data:" + json.dumps({"retcode": 500, "retmsg": str(e),
+                yield "data:" + json.dumps({"code": 500, "message": str(e),
                                             "data": {"answer": "**ERROR**: " + str(e), "reference": []}},
                                            ensure_ascii=False) + "\n\n"
-            yield "data:" + json.dumps({"retcode": 0, "retmsg": "", "data": True}, ensure_ascii=False) + "\n\n"
+            yield "data:" + json.dumps({"code": 0, "message": "", "data": True}, ensure_ascii=False) + "\n\n"
 
         if req.get("stream", True):
             resp = Response(stream(), mimetype="text/event-stream")
@@ -325,7 +326,7 @@ def completion():
             resp.headers.add_header("X-Accel-Buffering", "no")
             resp.headers.add_header("Content-Type", "text/event-stream; charset=utf-8")
             return resp
-            
+
         answer = None
         for ans in chat(dia, msg, **req):
             answer = ans
@@ -342,12 +343,22 @@ def completion():
 @manager.route('/conversation/<conversation_id>', methods=['GET'])
 # @login_required
 def get(conversation_id):
+    token = request.headers.get('Authorization').split()[1]
+    objs = APIToken.query(token=token)
+    if not objs:
+        return get_json_result(
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
+
     try:
         e, conv = API4ConversationService.get_by_id(conversation_id)
         if not e:
-            return get_data_error_result(retmsg="Conversation not found!")
+            return get_data_error_result(message="Conversation not found!")
 
         conv = conv.to_dict()
+        if token != APIToken.query(dialog_id=conv['dialog_id'])[0].token:
+            return get_json_result(data=False, message='Token is not valid for this conversation_id!"',
+                                   code=settings.RetCode.AUTHENTICATION_ERROR)
+
         for referenct_i in conv['reference']:
             if referenct_i is None or len(referenct_i) == 0:
                 continue
@@ -367,7 +378,7 @@ def upload():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
 
     kb_name = request.form.get("kb_name").strip()
     tenant_id = objs[0].tenant_id
@@ -376,19 +387,19 @@ def upload():
         e, kb = KnowledgebaseService.get_by_name(kb_name, tenant_id)
         if not e:
             return get_data_error_result(
-                retmsg="Can't find this knowledgebase!")
+                message="Can't find this knowledgebase!")
         kb_id = kb.id
     except Exception as e:
         return server_error_response(e)
 
     if 'file' not in request.files:
         return get_json_result(
-            data=False, retmsg='No file part!', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='No file part!', code=settings.RetCode.ARGUMENT_ERROR)
 
     file = request.files['file']
     if file.filename == '':
         return get_json_result(
-            data=False, retmsg='No file selected!', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='No file selected!', code=settings.RetCode.ARGUMENT_ERROR)
 
     root_folder = FileService.get_root_folder(tenant_id)
     pf_id = root_folder["id"]
@@ -399,7 +410,7 @@ def upload():
     try:
         if DocumentService.get_doc_count(kb.tenant_id) >= int(os.environ.get('MAX_FILE_NUM_PER_USER', 8192)):
             return get_data_error_result(
-                retmsg="Exceed the maximum file number of a free user!")
+                message="Exceed the maximum file number of a free user!")
 
         filename = duplicate_name(
             DocumentService.query,
@@ -408,13 +419,13 @@ def upload():
         filetype = filename_type(filename)
         if not filetype:
             return get_data_error_result(
-                retmsg="This type of file has not been supported yet!")
+                message="This type of file has not been supported yet!")
 
         location = filename
-        while MINIO.obj_exist(kb_id, location):
+        while STORAGE_IMPL.obj_exist(kb_id, location):
             location += "_"
         blob = request.files['file'].read()
-        MINIO.put(kb_id, location, blob)
+        STORAGE_IMPL.put(kb_id, location, blob)
         doc = {
             "id": get_uuid(),
             "kb_id": kb.id,
@@ -438,6 +449,8 @@ def upload():
             doc["parser_id"] = ParserType.AUDIO.value
         if re.search(r"\.(ppt|pptx|pages)$", filename):
             doc["parser_id"] = ParserType.PRESENTATION.value
+        if re.search(r"\.(eml)$", filename):
+            doc["parser_id"] = ParserType.EMAIL.value
 
         doc_result = DocumentService.insert(doc)
         FileService.add_file_from_kb(doc, kb_folder["id"], kb.tenant_id)
@@ -455,14 +468,14 @@ def upload():
                 # if str(req["run"]) == TaskStatus.CANCEL.value:
                 tenant_id = DocumentService.get_tenant_id(doc["id"])
                 if not tenant_id:
-                    return get_data_error_result(retmsg="Tenant not found!")
+                    return get_data_error_result(message="Tenant not found!")
 
                 # e, doc = DocumentService.get_by_id(doc["id"])
                 TaskService.filter_delete([Task.doc_id == doc["id"]])
                 e, doc = DocumentService.get_by_id(doc["id"])
                 doc = doc.to_dict()
                 doc["tenant_id"] = tenant_id
-                bucket, name = File2DocumentService.get_minio_address(doc_id=doc["id"])
+                bucket, name = File2DocumentService.get_storage_address(doc_id=doc["id"])
                 queue_tasks(doc, bucket, name)
             except Exception as e:
                 return server_error_response(e)
@@ -477,17 +490,17 @@ def upload_parse():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
 
     if 'file' not in request.files:
         return get_json_result(
-            data=False, retmsg='No file part!', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='No file part!', code=settings.RetCode.ARGUMENT_ERROR)
 
     file_objs = request.files.getlist('file')
     for file_obj in file_objs:
         if file_obj.filename == '':
             return get_json_result(
-                data=False, retmsg='No file selected!', retcode=RetCode.ARGUMENT_ERROR)
+                data=False, message='No file selected!', code=settings.RetCode.ARGUMENT_ERROR)
 
     doc_ids = doc_upload_and_parse(request.form.get("conversation_id"), file_objs, objs[0].tenant_id)
     return get_json_result(data=doc_ids)
@@ -500,7 +513,7 @@ def list_chunks():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
 
     req = request.json
 
@@ -514,15 +527,16 @@ def list_chunks():
             doc_id = req['doc_id']
         else:
             return get_json_result(
-                data=False, retmsg="Can't find doc_name or doc_id"
+                data=False, message="Can't find doc_name or doc_id"
             )
+        kb_ids = KnowledgebaseService.get_kb_ids(tenant_id)
 
-        res = retrievaler.chunk_list(doc_id=doc_id, tenant_id=tenant_id)
+        res = settings.retrievaler.chunk_list(doc_id, tenant_id, kb_ids)
         res = [
             {
                 "content": res_item["content_with_weight"],
                 "doc_name": res_item["docnm_kwd"],
-                "img_id": res_item["img_id"]
+                "image_id": res_item["img_id"]
             } for res_item in res
         ]
 
@@ -539,7 +553,7 @@ def list_kb_docs():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
 
     req = request.json
     tenant_id = objs[0].tenant_id
@@ -549,7 +563,7 @@ def list_kb_docs():
         e, kb = KnowledgebaseService.get_by_name(kb_name, tenant_id)
         if not e:
             return get_data_error_result(
-                retmsg="Can't find this knowledgebase!")
+                message="Can't find this knowledgebase!")
         kb_id = kb.id
 
     except Exception as e:
@@ -571,6 +585,7 @@ def list_kb_docs():
     except Exception as e:
         return server_error_response(e)
 
+
 @manager.route('/document/infos', methods=['POST'])
 @validate_request("doc_ids")
 def docinfos():
@@ -578,7 +593,7 @@ def docinfos():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
     req = request.json
     doc_ids = req["doc_ids"]
     docs = DocumentService.get_by_ids(doc_ids)
@@ -592,7 +607,7 @@ def document_rm():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
 
     tenant_id = objs[0].tenant_id
     req = request.json
@@ -604,7 +619,7 @@ def document_rm():
 
         if not doc_ids:
             return get_json_result(
-                data=False, retmsg="Can't find doc_names or doc_ids"
+                data=False, message="Can't find doc_names or doc_ids"
             )
 
     except Exception as e:
@@ -619,27 +634,27 @@ def document_rm():
         try:
             e, doc = DocumentService.get_by_id(doc_id)
             if not e:
-                return get_data_error_result(retmsg="Document not found!")
+                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(retmsg="Tenant not found!")
+                return get_data_error_result(message="Tenant not found!")
 
-            b, n = File2DocumentService.get_minio_address(doc_id=doc_id)
+            b, n = File2DocumentService.get_storage_address(doc_id=doc_id)
 
             if not DocumentService.remove_document(doc, tenant_id):
                 return get_data_error_result(
-                    retmsg="Database error (Document removal)!")
+                    message="Database error (Document removal)!")
 
             f2d = File2DocumentService.get_by_document_id(doc_id)
             FileService.filter_delete([File.source_type == FileSource.KNOWLEDGEBASE, File.id == f2d[0].file_id])
             File2DocumentService.delete_by_document_id(doc_id)
 
-            MINIO.rm(b, n)
+            STORAGE_IMPL.rm(b, n)
         except Exception as e:
             errors += str(e)
 
     if errors:
-        return get_json_result(data=False, retmsg=errors, retcode=RetCode.SERVER_ERROR)
+        return get_json_result(data=False, message=errors, code=settings.RetCode.SERVER_ERROR)
 
     return get_json_result(data=True)
 
@@ -654,36 +669,99 @@ def completion_faq():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
 
     e, conv = API4ConversationService.get_by_id(req["conversation_id"])
     if not e:
-        return get_data_error_result(retmsg="Conversation not found!")
+        return get_data_error_result(message="Conversation not found!")
     if "quote" not in req: req["quote"] = True
 
     msg = []
     msg.append({"role": "user", "content": req["word"]})
+    if not msg[-1].get("id"): msg[-1]["id"] = get_uuid()
+    message_id = msg[-1]["id"]
+
+    def fillin_conv(ans):
+        nonlocal conv, message_id
+        if not conv.reference:
+            conv.reference.append(ans["reference"])
+        else:
+            conv.reference[-1] = ans["reference"]
+        conv.message[-1] = {"role": "assistant", "content": ans["answer"], "id": message_id}
+        ans["id"] = message_id
 
     try:
+        if conv.source == "agent":
+            conv.message.append(msg[-1])
+            e, cvs = UserCanvasService.get_by_id(conv.dialog_id)
+            if not e:
+                return server_error_response("canvas not found.")
+
+            if not isinstance(cvs.dsl, str):
+                cvs.dsl = json.dumps(cvs.dsl, ensure_ascii=False)
+
+            if not conv.reference:
+                conv.reference = []
+            conv.message.append({"role": "assistant", "content": "", "id": message_id})
+            conv.reference.append({"chunks": [], "doc_aggs": []})
+
+            final_ans = {"reference": [], "doc_aggs": []}
+            canvas = Canvas(cvs.dsl, objs[0].tenant_id)
+
+            canvas.messages.append(msg[-1])
+            canvas.add_user_input(msg[-1]["content"])
+            answer = canvas.run(stream=False)
+
+            assert answer is not None, "Nothing. Is it over?"
+
+            data_type_picture = {
+                "type": 3,
+                "url": "base64 content"
+            }
+            data = [
+                {
+                    "type": 1,
+                    "content": ""
+                }
+            ]
+            final_ans["content"] = "\n".join(answer["content"]) if "content" in answer else ""
+            canvas.messages.append({"role": "assistant", "content": final_ans["content"], "id": message_id})
+            if final_ans.get("reference"):
+                canvas.reference.append(final_ans["reference"])
+            cvs.dsl = json.loads(str(canvas))
+
+            ans = {"answer": final_ans["content"], "reference": final_ans.get("reference", [])}
+            data[0]["content"] += re.sub(r'##\d\$\$', '', ans["answer"])
+            fillin_conv(ans)
+            API4ConversationService.append_message(conv.id, conv.to_dict())
+
+            chunk_idxs = [int(match[2]) for match in re.findall(r'##\d\$\$', ans["answer"])]
+            for chunk_idx in chunk_idxs[:1]:
+                if ans["reference"]["chunks"][chunk_idx]["img_id"]:
+                    try:
+                        bkt, nm = ans["reference"]["chunks"][chunk_idx]["img_id"].split("-")
+                        response = STORAGE_IMPL.get(bkt, nm)
+                        data_type_picture["url"] = base64.b64encode(response).decode('utf-8')
+                        data.append(data_type_picture)
+                        break
+                    except Exception as e:
+                        return server_error_response(e)
+
+            response = {"code": 200, "msg": "success", "data": data}
+            return response
+
+        # ******************For dialog******************
         conv.message.append(msg[-1])
         e, dia = DialogService.get_by_id(conv.dialog_id)
         if not e:
-            return get_data_error_result(retmsg="Dialog not found!")
+            return get_data_error_result(message="Dialog not found!")
         del req["conversation_id"]
 
         if not conv.reference:
             conv.reference = []
-        conv.message.append({"role": "assistant", "content": ""})
+        conv.message.append({"role": "assistant", "content": "", "id": message_id})
         conv.reference.append({"chunks": [], "doc_aggs": []})
 
-        def fillin_conv(ans):
-            nonlocal conv
-            if not conv.reference:
-                conv.reference.append(ans["reference"])
-            else:
-                conv.reference[-1] = ans["reference"]
-            conv.message[-1] = {"role": "assistant", "content": ans["answer"]}
-
         data_type_picture = {
             "type": 3,
             "url": "base64 content"
@@ -707,7 +785,7 @@ def completion_faq():
             if ans["reference"]["chunks"][chunk_idx]["img_id"]:
                 try:
                     bkt, nm = ans["reference"]["chunks"][chunk_idx]["img_id"].split("-")
-                    response = MINIO.get(bkt, nm)
+                    response = STORAGE_IMPL.get(bkt, nm)
                     data_type_picture["url"] = base64.b64encode(response).decode('utf-8')
                     data.append(data_type_picture)
                     break
@@ -728,10 +806,10 @@ def retrieval():
     objs = APIToken.query(token=token)
     if not objs:
         return get_json_result(
-            data=False, retmsg='Token is not valid!"', retcode=RetCode.AUTHENTICATION_ERROR)
+            data=False, message='Token is not valid!"', code=settings.RetCode.AUTHENTICATION_ERROR)
 
     req = request.json
-    kb_ids = req.get("kb_id",[])
+    kb_ids = req.get("kb_id", [])
     doc_ids = req.get("doc_ids", [])
     question = req.get("question")
     page = int(req.get("page", 1))
@@ -745,26 +823,26 @@ def retrieval():
         embd_nms = list(set([kb.embd_id for kb in kbs]))
         if len(embd_nms) != 1:
             return get_json_result(
-                data=False, retmsg='Knowledge bases use different embedding models or does not exist."', retcode=RetCode.AUTHENTICATION_ERROR)
+                data=False, message='Knowledge bases use different embedding models or does not exist."',
+                code=settings.RetCode.AUTHENTICATION_ERROR)
 
         embd_mdl = TenantLLMService.model_instance(
             kbs[0].tenant_id, LLMType.EMBEDDING.value, llm_name=kbs[0].embd_id)
         rerank_mdl = None
         if req.get("rerank_id"):
             rerank_mdl = TenantLLMService.model_instance(
-            kbs[0].tenant_id, LLMType.RERANK.value, llm_name=req["rerank_id"])
+                kbs[0].tenant_id, LLMType.RERANK.value, llm_name=req["rerank_id"])
         if req.get("keyword", False):
             chat_mdl = TenantLLMService.model_instance(kbs[0].tenant_id, LLMType.CHAT)
             question += keyword_extraction(chat_mdl, question)
-        ranks = retrievaler.retrieval(question, embd_mdl, kbs[0].tenant_id, kb_ids, page, size,
-            similarity_threshold, vector_similarity_weight, top,
-            doc_ids, rerank_mdl=rerank_mdl)
+        ranks = settings.retrievaler.retrieval(question, embd_mdl, kbs[0].tenant_id, kb_ids, page, size,
+                                               similarity_threshold, vector_similarity_weight, top,
+                                               doc_ids, rerank_mdl=rerank_mdl)
         for c in ranks["chunks"]:
-            if "vector" in c:
-                del c["vector"]
+            c.pop("vector", None)
         return get_json_result(data=ranks)
     except Exception as e:
         if str(e).find("not_found") > 0:
-            return get_json_result(data=False, retmsg=f'No chunk found! Check the chunk status please!',
-                                   retcode=RetCode.DATA_ERROR)
-        return server_error_response(e)
+            return get_json_result(data=False, message='No chunk found! Check the chunk status please!',
+                                   code=settings.RetCode.DATA_ERROR)
+        return server_error_response(e)

api/apps/canvas_app.py

@@ -13,13 +13,16 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 import json
+import traceback
 from functools import partial
 from flask import request, Response
 from flask_login import login_required, current_user
 from api.db.services.canvas_service import CanvasTemplateService, UserCanvasService
+from api.settings import RetCode
 from api.utils import get_uuid
-from api.utils.api_utils import get_json_result, server_error_response, validate_request
+from api.utils.api_utils import get_json_result, server_error_response, validate_request, get_data_error_result
 from agent.canvas import Canvas
 from peewee import MySQLDatabase, PostgresqlDatabase
 
@@ -43,6 +46,10 @@ def canvas_list():
 @login_required
 def rm():
     for i in request.json["canvas_ids"]:
+        if not UserCanvasService.query(user_id=current_user.id,id=i):
+            return get_json_result(
+                data=False, message='Only owner of canvas authorized for this operation.',
+                code=RetCode.OPERATING_ERROR)
         UserCanvasService.delete_by_id(i)
     return get_json_result(data=True)
 
@@ -61,10 +68,13 @@ def save():
             return server_error_response(ValueError("Duplicated title."))
         req["id"] = get_uuid()
         if not UserCanvasService.save(**req):
-            return server_error_response("Fail to save canvas.")
+            return get_data_error_result(message="Fail to save canvas.")
     else:
+        if not UserCanvasService.query(user_id=current_user.id, id=req["id"]):
+            return get_json_result(
+                data=False, message='Only owner of canvas authorized for this operation.',
+                code=RetCode.OPERATING_ERROR)
         UserCanvasService.update_by_id(req["id"], req)
-
     return get_json_result(data=req)
 
 
@@ -73,7 +83,7 @@ def save():
 def get(canvas_id):
     e, c = UserCanvasService.get_by_id(canvas_id)
     if not e:
-        return server_error_response("canvas not found.")
+        return get_data_error_result(message="canvas not found.")
     return get_json_result(data=c.to_dict())
 
 
@@ -85,46 +95,55 @@ def run():
     stream = req.get("stream", True)
     e, cvs = UserCanvasService.get_by_id(req["id"])
     if not e:
-        return server_error_response("canvas not found.")
+        return get_data_error_result(message="canvas not found.")
+    if not UserCanvasService.query(user_id=current_user.id, id=req["id"]):
+        return get_json_result(
+            data=False, message='Only owner of canvas authorized for this operation.',
+            code=RetCode.OPERATING_ERROR)
 
     if not isinstance(cvs.dsl, str):
         cvs.dsl = json.dumps(cvs.dsl, ensure_ascii=False)
 
     final_ans = {"reference": [], "content": ""}
+    message_id = req.get("message_id", get_uuid())
     try:
         canvas = Canvas(cvs.dsl, current_user.id)
         if "message" in req:
-            canvas.messages.append({"role": "user", "content": req["message"]})
+            canvas.messages.append({"role": "user", "content": req["message"], "id": message_id})
             canvas.add_user_input(req["message"])
-        answer = canvas.run(stream=stream)
-        print(canvas)
     except Exception as e:
         return server_error_response(e)
 
-    assert answer is not None, "Nothing. Is it over?"
-
     if stream:
-        assert isinstance(answer, partial), "Nothing. Is it over?"
-
         def sse():
             nonlocal answer, cvs
             try:
-                for ans in answer():
+                for ans in canvas.run(stream=True):
+                    if ans.get("running_status"):
+                        yield "data:" + json.dumps({"code": 0, "message": "",
+                                                    "data": {"answer": ans["content"],
+                                                             "running_status": True}},
+                                                   ensure_ascii=False) + "\n\n"
+                        continue
                     for k in ans.keys():
                         final_ans[k] = ans[k]
                     ans = {"answer": ans["content"], "reference": ans.get("reference", [])}
-                    yield "data:" + json.dumps({"retcode": 0, "retmsg": "", "data": ans}, ensure_ascii=False) + "\n\n"
+                    yield "data:" + json.dumps({"code": 0, "message": "", "data": ans}, ensure_ascii=False) + "\n\n"
 
-                canvas.messages.append({"role": "assistant", "content": final_ans["content"]})
+                canvas.messages.append({"role": "assistant", "content": final_ans["content"], "id": message_id})
+                canvas.history.append(("assistant", final_ans["content"]))
                 if final_ans.get("reference"):
                     canvas.reference.append(final_ans["reference"])
                 cvs.dsl = json.loads(str(canvas))
                 UserCanvasService.update_by_id(req["id"], cvs.to_dict())
             except Exception as e:
-                yield "data:" + json.dumps({"retcode": 500, "retmsg": str(e),
+                cvs.dsl = json.loads(str(canvas))
+                UserCanvasService.update_by_id(req["id"], cvs.to_dict())
+                traceback.print_exc()
+                yield "data:" + json.dumps({"code": 500, "message": str(e),
                                             "data": {"answer": "**ERROR**: " + str(e), "reference": []}},
                                            ensure_ascii=False) + "\n\n"
-            yield "data:" + json.dumps({"retcode": 0, "retmsg": "", "data": True}, ensure_ascii=False) + "\n\n"
+            yield "data:" + json.dumps({"code": 0, "message": "", "data": True}, ensure_ascii=False) + "\n\n"
 
         resp = Response(sse(), mimetype="text/event-stream")
         resp.headers.add_header("Cache-control", "no-cache")
@@ -133,13 +152,15 @@ def run():
         resp.headers.add_header("Content-Type", "text/event-stream; charset=utf-8")
         return resp
 
-    final_ans["content"] = "\n".join(answer["content"]) if "content" in answer else ""
-    canvas.messages.append({"role": "assistant", "content": final_ans["content"]})
-    if final_ans.get("reference"):
-        canvas.reference.append(final_ans["reference"])
-    cvs.dsl = json.loads(str(canvas))
-    UserCanvasService.update_by_id(req["id"], cvs.to_dict())
-    return get_json_result(data={"answer": final_ans["content"], "reference": final_ans.get("reference", [])})
+    for answer in canvas.run(stream=False):
+        if answer.get("running_status"): continue
+        final_ans["content"] = "\n".join(answer["content"]) if "content" in answer else ""
+        canvas.messages.append({"role": "assistant", "content": final_ans["content"], "id": message_id})
+        if final_ans.get("reference"):
+            canvas.reference.append(final_ans["reference"])
+        cvs.dsl = json.loads(str(canvas))
+        UserCanvasService.update_by_id(req["id"], cvs.to_dict())
+        return get_json_result(data={"answer": final_ans["content"], "reference": final_ans.get("reference", [])})
 
 
 @manager.route('/reset', methods=['POST'])
@@ -150,7 +171,11 @@ def reset():
     try:
         e, user_canvas = UserCanvasService.get_by_id(req["id"])
         if not e:
-            return server_error_response("canvas not found.")
+            return get_data_error_result(message="canvas not found.")
+        if not UserCanvasService.query(user_id=current_user.id, id=req["id"]):
+            return get_json_result(
+                data=False, message='Only owner of canvas authorized for this operation.',
+                code=RetCode.OPERATING_ERROR)
 
         canvas = Canvas(json.dumps(user_canvas.dsl), current_user.id)
         canvas.reset()

api/apps/chunk_app.py

@@ -15,23 +15,21 @@
 #
 import datetime
 import json
-import traceback
 
 from flask import request
 from flask_login import login_required, current_user
-from elasticsearch_dsl import Q
 
+from api.db.services.dialog_service import keyword_extraction
 from rag.app.qa import rmPrefix, beAdoc
-from rag.nlp import search, rag_tokenizer, keyword_extraction
-from rag.utils.es_conn import ELASTICSEARCH
+from rag.nlp import search, rag_tokenizer
 from rag.utils import rmSpace
 from api.db import LLMType, ParserType
 from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.db.services.llm_service import TenantLLMService
+from api.db.services.llm_service import LLMBundle
 from api.db.services.user_service import UserTenantService
 from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
 from api.db.services.document_service import DocumentService
-from api.settings import RetCode, retrievaler, kg_retrievaler
+from api import settings
 from api.utils.api_utils import get_json_result
 import hashlib
 import re
@@ -49,16 +47,17 @@ def list_chunk():
     try:
         tenant_id = DocumentService.get_tenant_id(req["doc_id"])
         if not tenant_id:
-            return get_data_error_result(retmsg="Tenant not found!")
+            return get_data_error_result(message="Tenant not found!")
         e, doc = DocumentService.get_by_id(doc_id)
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
+            return get_data_error_result(message="Document not found!")
+        kb_ids = KnowledgebaseService.get_kb_ids(tenant_id)
         query = {
             "doc_ids": [doc_id], "page": page, "size": size, "question": question, "sort": True
         }
         if "available_int" in req:
             query["available_int"] = int(req["available_int"])
-        sres = retrievaler.search(query, search.index_name(tenant_id))
+        sres = settings.retrievaler.search(query, search.index_name(tenant_id), kb_ids, highlight=True)
         res = {"total": sres.total, "chunks": [], "doc": doc.to_dict()}
         for id in sres.ids:
             d = {
@@ -69,22 +68,18 @@ def list_chunk():
                 "doc_id": sres.field[id]["doc_id"],
                 "docnm_kwd": sres.field[id]["docnm_kwd"],
                 "important_kwd": sres.field[id].get("important_kwd", []),
-                "img_id": sres.field[id].get("img_id", ""),
+                "image_id": sres.field[id].get("img_id", ""),
                 "available_int": sres.field[id].get("available_int", 1),
-                "positions": sres.field[id].get("position_int", "").split("\t")
+                "positions": json.loads(sres.field[id].get("position_list", "[]")),
             }
-            if len(d["positions"]) % 5 == 0:
-                poss = []
-                for i in range(0, len(d["positions"]), 5):
-                    poss.append([float(d["positions"][i]), float(d["positions"][i + 1]), float(d["positions"][i + 2]),
-                                 float(d["positions"][i + 3]), float(d["positions"][i + 4])])
-                d["positions"] = poss
+            assert isinstance(d["positions"], list)
+            assert len(d["positions"]) == 0 or (isinstance(d["positions"][0], list) and len(d["positions"][0]) == 5)
             res["chunks"].append(d)
         return get_json_result(data=res)
     except Exception as e:
         if str(e).find("not_found") > 0:
-            return get_json_result(data=False, retmsg=f'No chunk found!',
-                                   retcode=RetCode.DATA_ERROR)
+            return get_json_result(data=False, message='No chunk found!',
+                                   code=settings.RetCode.DATA_ERROR)
         return server_error_response(e)
 
 
@@ -95,27 +90,25 @@ def get():
     try:
         tenants = UserTenantService.query(user_id=current_user.id)
         if not tenants:
-            return get_data_error_result(retmsg="Tenant not found!")
-        res = ELASTICSEARCH.get(
-            chunk_id, search.index_name(
-                tenants[0].tenant_id))
-        if not res.get("found"):
+            return get_data_error_result(message="Tenant not found!")
+        tenant_id = tenants[0].tenant_id
+
+        kb_ids = KnowledgebaseService.get_kb_ids(tenant_id)
+        chunk = settings.docStoreConn.get(chunk_id, search.index_name(tenant_id), kb_ids)
+        if chunk is None:
             return server_error_response("Chunk not found")
-        id = res["_id"]
-        res = res["_source"]
-        res["chunk_id"] = id
         k = []
-        for n in res.keys():
+        for n in chunk.keys():
             if re.search(r"(_vec$|_sm_|_tks|_ltks)", n):
                 k.append(n)
         for n in k:
-            del res[n]
+            del chunk[n]
 
-        return get_json_result(data=res)
+        return get_json_result(data=chunk)
     except Exception as e:
         if str(e).find("NotFoundError") >= 0:
-            return get_json_result(data=False, retmsg=f'Chunk not found!',
-                                   retcode=RetCode.DATA_ERROR)
+            return get_json_result(data=False, message='Chunk not found!',
+                                   code=settings.RetCode.DATA_ERROR)
         return server_error_response(e)
 
 
@@ -138,15 +131,14 @@ def set():
     try:
         tenant_id = DocumentService.get_tenant_id(req["doc_id"])
         if not tenant_id:
-            return get_data_error_result(retmsg="Tenant not found!")
+            return get_data_error_result(message="Tenant not found!")
 
         embd_id = DocumentService.get_embd_id(req["doc_id"])
-        embd_mdl = TenantLLMService.model_instance(
-            tenant_id, LLMType.EMBEDDING.value, embd_id)
+        embd_mdl = LLMBundle(tenant_id, LLMType.EMBEDDING, embd_id)
 
         e, doc = DocumentService.get_by_id(req["doc_id"])
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
+            return get_data_error_result(message="Document not found!")
 
         if doc.parser_id == ParserType.QA:
             arr = [
@@ -155,7 +147,7 @@ def set():
                     req["content_with_weight"]) if len(t) > 1]
             if len(arr) != 2:
                 return get_data_error_result(
-                    retmsg="Q&A must be separated by TAB/ENTER key.")
+                    message="Q&A must be separated by TAB/ENTER key.")
             q, a = rmPrefix(arr[0]), rmPrefix(arr[1])
             d = beAdoc(d, arr[0], arr[1], not any(
                 [rag_tokenizer.is_chinese(t) for t in q + a]))
@@ -163,7 +155,7 @@ def set():
         v, c = embd_mdl.encode([doc.name, req["content_with_weight"]])
         v = 0.1 * v[0] + 0.9 * v[1] if doc.parser_id != ParserType.QA else v[1]
         d["q_%d_vec" % len(v)] = v.tolist()
-        ELASTICSEARCH.upsert([d], search.index_name(tenant_id))
+        settings.docStoreConn.insert([d], search.index_name(tenant_id), doc.kb_id)
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)
@@ -175,12 +167,15 @@ def set():
 def switch():
     req = request.json
     try:
-        tenant_id = DocumentService.get_tenant_id(req["doc_id"])
-        if not tenant_id:
-            return get_data_error_result(retmsg="Tenant not found!")
-        if not ELASTICSEARCH.upsert([{"id": i, "available_int": int(req["available_int"])} for i in req["chunk_ids"]],
-                                    search.index_name(tenant_id)):
-            return get_data_error_result(retmsg="Index updating failure")
+        e, doc = DocumentService.get_by_id(req["doc_id"])
+        if not e:
+            return get_data_error_result(message="Document not found!")
+        for cid in req["chunk_ids"]:
+            if not settings.docStoreConn.update({"id": cid},
+                                                {"available_int": int(req["available_int"])},
+                                                search.index_name(DocumentService.get_tenant_id(req["doc_id"])),
+                                                doc.kb_id):
+                return get_data_error_result(message="Index updating failure")
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)
@@ -192,12 +187,11 @@ def switch():
 def rm():
     req = request.json
     try:
-        if not ELASTICSEARCH.deleteByQuery(
-                Q("ids", values=req["chunk_ids"]), search.index_name(current_user.id)):
-            return get_data_error_result(retmsg="Index updating failure")
         e, doc = DocumentService.get_by_id(req["doc_id"])
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
+            return get_data_error_result(message="Document not found!")
+        if not settings.docStoreConn.delete({"id": req["chunk_ids"]}, search.index_name(current_user.id), doc.kb_id):
+            return get_data_error_result(message="Index updating failure")
         deleted_chunk_ids = req["chunk_ids"]
         chunk_number = len(deleted_chunk_ids)
         DocumentService.decrement_chunk_num(doc.id, doc.kb_id, 1, chunk_number, 0)
@@ -225,23 +219,22 @@ def create():
     try:
         e, doc = DocumentService.get_by_id(req["doc_id"])
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
+            return get_data_error_result(message="Document not found!")
         d["kb_id"] = [doc.kb_id]
         d["docnm_kwd"] = doc.name
         d["doc_id"] = doc.id
 
         tenant_id = DocumentService.get_tenant_id(req["doc_id"])
         if not tenant_id:
-            return get_data_error_result(retmsg="Tenant not found!")
+            return get_data_error_result(message="Tenant not found!")
 
         embd_id = DocumentService.get_embd_id(req["doc_id"])
-        embd_mdl = TenantLLMService.model_instance(
-            tenant_id, LLMType.EMBEDDING.value, embd_id)
+        embd_mdl = LLMBundle(tenant_id, LLMType.EMBEDDING.value, embd_id)
 
         v, c = embd_mdl.encode([doc.name, req["content_with_weight"]])
         v = 0.1 * v[0] + 0.9 * v[1]
         d["q_%d_vec" % len(v)] = v.tolist()
-        ELASTICSEARCH.upsert([d], search.index_name(tenant_id))
+        settings.docStoreConn.insert([d], search.index_name(tenant_id), doc.kb_id)
 
         DocumentService.increment_chunk_num(
             doc.id, doc.kb_id, c, 1, 0)
@@ -258,41 +251,54 @@ def retrieval_test():
     page = int(req.get("page", 1))
     size = int(req.get("size", 30))
     question = req["question"]
-    kb_id = req["kb_id"]
+    kb_ids = req["kb_id"]
+    if isinstance(kb_ids, str):
+        kb_ids = [kb_ids]
     doc_ids = req.get("doc_ids", [])
-    similarity_threshold = float(req.get("similarity_threshold", 0.2))
+    similarity_threshold = float(req.get("similarity_threshold", 0.0))
     vector_similarity_weight = float(req.get("vector_similarity_weight", 0.3))
     top = int(req.get("top_k", 1024))
-    try:
-        e, kb = KnowledgebaseService.get_by_id(kb_id)
-        if not e:
-            return get_data_error_result(retmsg="Knowledgebase not found!")
+    tenant_ids = []
 
-        embd_mdl = TenantLLMService.model_instance(
-            kb.tenant_id, LLMType.EMBEDDING.value, llm_name=kb.embd_id)
+    try:
+        tenants = UserTenantService.query(user_id=current_user.id)
+        for kb_id in kb_ids:
+            for tenant in tenants:
+                if KnowledgebaseService.query(
+                        tenant_id=tenant.tenant_id, id=kb_id):
+                    tenant_ids.append(tenant.tenant_id)
+                    break
+            else:
+                return get_json_result(
+                    data=False, message='Only owner of knowledgebase authorized for this operation.',
+                    code=settings.RetCode.OPERATING_ERROR)
+
+        e, kb = KnowledgebaseService.get_by_id(kb_ids[0])
+        if not e:
+            return get_data_error_result(message="Knowledgebase not found!")
+
+        embd_mdl = LLMBundle(kb.tenant_id, LLMType.EMBEDDING.value, llm_name=kb.embd_id)
 
         rerank_mdl = None
         if req.get("rerank_id"):
-            rerank_mdl = TenantLLMService.model_instance(
-                kb.tenant_id, LLMType.RERANK.value, llm_name=req["rerank_id"])
+            rerank_mdl = LLMBundle(kb.tenant_id, LLMType.RERANK.value, llm_name=req["rerank_id"])
 
         if req.get("keyword", False):
-            chat_mdl = TenantLLMService.model_instance(kb.tenant_id, LLMType.CHAT)
+            chat_mdl = LLMBundle(kb.tenant_id, LLMType.CHAT)
             question += keyword_extraction(chat_mdl, question)
 
-        retr = retrievaler if kb.parser_id != ParserType.KG else kg_retrievaler
-        ranks = retr.retrieval(question, embd_mdl, kb.tenant_id, [kb_id], page, size,
+        retr = settings.retrievaler if kb.parser_id != ParserType.KG else settings.kg_retrievaler
+        ranks = retr.retrieval(question, embd_mdl, tenant_ids, kb_ids, page, size,
                                similarity_threshold, vector_similarity_weight, top,
-                               doc_ids, rerank_mdl=rerank_mdl)
+                               doc_ids, rerank_mdl=rerank_mdl, highlight=req.get("highlight"))
         for c in ranks["chunks"]:
-            if "vector" in c:
-                del c["vector"]
+            c.pop("vector", None)
 
         return get_json_result(data=ranks)
     except Exception as e:
         if str(e).find("not_found") > 0:
-            return get_json_result(data=False, retmsg=f'No chunk found! Check the chunk status please!',
-                                   retcode=RetCode.DATA_ERROR)
+            return get_json_result(data=False, message='No chunk found! Check the chunk status please!',
+                                   code=settings.RetCode.DATA_ERROR)
         return server_error_response(e)
 
 
@@ -300,19 +306,38 @@ def retrieval_test():
 @login_required
 def knowledge_graph():
     doc_id = request.args["doc_id"]
+    tenant_id = DocumentService.get_tenant_id(doc_id)
+    kb_ids = KnowledgebaseService.get_kb_ids(tenant_id)
     req = {
-        "doc_ids":[doc_id],
+        "doc_ids": [doc_id],
         "knowledge_graph_kwd": ["graph", "mind_map"]
     }
-    tenant_id = DocumentService.get_tenant_id(doc_id)
-    sres = retrievaler.search(req, search.index_name(tenant_id))
+    sres = settings.retrievaler.search(req, search.index_name(tenant_id), kb_ids)
     obj = {"graph": {}, "mind_map": {}}
     for id in sres.ids[:2]:
         ty = sres.field[id]["knowledge_graph_kwd"]
         try:
-            obj[ty] = json.loads(sres.field[id]["content_with_weight"])
-        except Exception as e:
-            print(traceback.format_exc(), flush=True)
+            content_json = json.loads(sres.field[id]["content_with_weight"])
+        except Exception:
+            continue
+
+        if ty == 'mind_map':
+            node_dict = {}
+
+            def repeat_deal(content_json, node_dict):
+                if 'id' in content_json:
+                    if content_json['id'] in node_dict:
+                        node_name = content_json['id']
+                        content_json['id'] += f"({node_dict[content_json['id']]})"
+                        node_dict[node_name] += 1
+                    else:
+                        node_dict[content_json['id']] = 1
+                if 'children' in content_json and content_json['children']:
+                    for item in content_json['children']:
+                        repeat_deal(item, node_dict)
+
+            repeat_deal(content_json, node_dict)
+
+        obj[ty] = content_json
 
     return get_json_result(data=obj)
-

api/apps/conversation_app.py

@@ -13,14 +13,22 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-from copy import deepcopy
-from flask import request, Response
-from flask_login import login_required
-from api.db.services.dialog_service import DialogService, ConversationService, chat
-from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
-from api.utils import get_uuid
-from api.utils.api_utils import get_json_result
 import json
+import re
+import traceback
+from copy import deepcopy
+from api.db.services.user_service import UserTenantService
+from flask import request, Response
+from flask_login import login_required, current_user
+
+from api.db import LLMType
+from api.db.services.dialog_service import DialogService, ConversationService, chat, ask
+from api.db.services.knowledgebase_service import KnowledgebaseService
+from api.db.services.llm_service import LLMBundle, TenantService, TenantLLMService
+from api import settings
+from api.utils.api_utils import get_json_result
+from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
+from graphrag.mind_map_extractor import MindMapExtractor
 
 
 @manager.route('/set', methods=['POST'])
@@ -28,15 +36,17 @@ import json
 def set_conversation():
     req = request.json
     conv_id = req.get("conversation_id")
-    if conv_id:
+    is_new = req.get("is_new")
+    del req["is_new"]
+    if not is_new:
         del req["conversation_id"]
         try:
             if not ConversationService.update_by_id(conv_id, req):
-                return get_data_error_result(retmsg="Conversation not found!")
+                return get_data_error_result(message="Conversation not found!")
             e, conv = ConversationService.get_by_id(conv_id)
             if not e:
                 return get_data_error_result(
-                    retmsg="Fail to update a conversation!")
+                    message="Fail to update a conversation!")
             conv = conv.to_dict()
             return get_json_result(data=conv)
         except Exception as e:
@@ -45,9 +55,9 @@ def set_conversation():
     try:
         e, dia = DialogService.get_by_id(req["dialog_id"])
         if not e:
-            return get_data_error_result(retmsg="Dialog not found")
+            return get_data_error_result(message="Dialog not found")
         conv = {
-            "id": get_uuid(),
+            "id": conv_id,
             "dialog_id": req["dialog_id"],
             "name": req.get("name", "New conversation"),
             "message": [{"role": "assistant", "content": dia.prompt_config["prologue"]}]
@@ -55,7 +65,7 @@ def set_conversation():
         ConversationService.save(**conv)
         e, conv = ConversationService.get_by_id(conv["id"])
         if not e:
-            return get_data_error_result(retmsg="Fail to new a conversation!")
+            return get_data_error_result(message="Fail to new a conversation!")
         conv = conv.to_dict()
         return get_json_result(data=conv)
     except Exception as e:
@@ -69,7 +79,15 @@ def get():
     try:
         e, conv = ConversationService.get_by_id(conv_id)
         if not e:
-            return get_data_error_result(retmsg="Conversation not found!")
+            return get_data_error_result(message="Conversation not found!")
+        tenants = UserTenantService.query(user_id=current_user.id)
+        for tenant in tenants:
+            if DialogService.query(tenant_id=tenant.tenant_id, id=conv.dialog_id):
+                break
+        else:
+            return get_json_result(
+                data=False, message='Only owner of conversation authorized for this operation.',
+                code=settings.RetCode.OPERATING_ERROR)
         conv = conv.to_dict()
         return get_json_result(data=conv)
     except Exception as e:
@@ -82,6 +100,17 @@ def rm():
     conv_ids = request.json["conversation_ids"]
     try:
         for cid in conv_ids:
+            exist, conv = ConversationService.get_by_id(cid)
+            if not exist:
+                return get_data_error_result(message="Conversation not found!")
+            tenants = UserTenantService.query(user_id=current_user.id)
+            for tenant in tenants:
+                if DialogService.query(tenant_id=tenant.tenant_id, id=conv.dialog_id):
+                    break
+            else:
+                return get_json_result(
+                    data=False, message='Only owner of conversation authorized for this operation.',
+                    code=settings.RetCode.OPERATING_ERROR)
             ConversationService.delete_by_id(cid)
         return get_json_result(data=True)
     except Exception as e:
@@ -93,6 +122,10 @@ def rm():
 def list_convsersation():
     dialog_id = request.args["dialog_id"]
     try:
+        if not DialogService.query(tenant_id=current_user.id, id=dialog_id):
+            return get_json_result(
+                data=False, message='Only owner of dialog authorized for this operation.',
+                code=settings.RetCode.OPERATING_ERROR)
         convs = ConversationService.query(
             dialog_id=dialog_id,
             order_by=ConversationService.model.create_time,
@@ -105,56 +138,56 @@ def list_convsersation():
 
 @manager.route('/completion', methods=['POST'])
 @login_required
-#@validate_request("conversation_id", "messages")
+@validate_request("conversation_id", "messages")
 def completion():
     req = request.json
-    #req = {"conversation_id": "9aaaca4c11d311efa461fa163e197198", "messages": [
-    #    {"role": "user", "content": "上海有吗？"}
-    #]}
     msg = []
     for m in req["messages"]:
         if m["role"] == "system":
             continue
         if m["role"] == "assistant" and not msg:
             continue
-        msg.append({"role": m["role"], "content": m["content"]})
-        if "doc_ids" in m:
-            msg[-1]["doc_ids"] = m["doc_ids"]
+        msg.append(m)
+    message_id = msg[-1].get("id")
     try:
         e, conv = ConversationService.get_by_id(req["conversation_id"])
         if not e:
-            return get_data_error_result(retmsg="Conversation not found!")
-        conv.message.append(deepcopy(msg[-1]))
+            return get_data_error_result(message="Conversation not found!")
+        conv.message = deepcopy(req["messages"])
         e, dia = DialogService.get_by_id(conv.dialog_id)
         if not e:
-            return get_data_error_result(retmsg="Dialog not found!")
+            return get_data_error_result(message="Dialog not found!")
         del req["conversation_id"]
         del req["messages"]
 
         if not conv.reference:
             conv.reference = []
-        conv.message.append({"role": "assistant", "content": ""})
+        conv.message.append({"role": "assistant", "content": "", "id": message_id})
         conv.reference.append({"chunks": [], "doc_aggs": []})
 
         def fillin_conv(ans):
-            nonlocal conv
+            nonlocal conv, message_id
             if not conv.reference:
                 conv.reference.append(ans["reference"])
-            else: conv.reference[-1] = ans["reference"]
-            conv.message[-1] = {"role": "assistant", "content": ans["answer"]}
+            else:
+                conv.reference[-1] = ans["reference"]
+            conv.message[-1] = {"role": "assistant", "content": ans["answer"],
+                                "id": message_id, "prompt": ans.get("prompt", "")}
+            ans["id"] = message_id
 
         def stream():
             nonlocal dia, msg, req, conv
             try:
                 for ans in chat(dia, msg, True, **req):
                     fillin_conv(ans)
-                    yield "data:"+json.dumps({"retcode": 0, "retmsg": "", "data": ans}, ensure_ascii=False) + "\n\n"
+                    yield "data:" + json.dumps({"code": 0, "message": "", "data": ans}, ensure_ascii=False) + "\n\n"
                 ConversationService.update_by_id(conv.id, conv.to_dict())
             except Exception as e:
-                yield "data:" + json.dumps({"retcode": 500, "retmsg": str(e),
-                                            "data": {"answer": "**ERROR**: "+str(e), "reference": []}},
+                traceback.print_exc()
+                yield "data:" + json.dumps({"code": 500, "message": str(e),
+                                            "data": {"answer": "**ERROR**: " + str(e), "reference": []}},
                                            ensure_ascii=False) + "\n\n"
-            yield "data:"+json.dumps({"retcode": 0, "retmsg": "", "data": True}, ensure_ascii=False) + "\n\n"
+            yield "data:" + json.dumps({"code": 0, "message": "", "data": True}, ensure_ascii=False) + "\n\n"
 
         if req.get("stream", True):
             resp = Response(stream(), mimetype="text/event-stream")
@@ -175,3 +208,170 @@ def completion():
     except Exception as e:
         return server_error_response(e)
 
+
+@manager.route('/tts', methods=['POST'])
+@login_required
+def tts():
+    req = request.json
+    text = req["text"]
+
+    tenants = TenantService.get_info_by(current_user.id)
+    if not tenants:
+        return get_data_error_result(message="Tenant not found!")
+
+    tts_id = tenants[0]["tts_id"]
+    if not tts_id:
+        return get_data_error_result(message="No default TTS model is set")
+
+    tts_mdl = LLMBundle(tenants[0]["tenant_id"], LLMType.TTS, tts_id)
+
+    def stream_audio():
+        try:
+            for txt in re.split(r"[，。/《》？；：！\n\r:;]+", text):
+                for chunk in tts_mdl.tts(txt):
+                    yield chunk
+        except Exception as e:
+            yield ("data:" + json.dumps({"code": 500, "message": str(e),
+                                         "data": {"answer": "**ERROR**: " + str(e)}},
+                                        ensure_ascii=False)).encode('utf-8')
+
+    resp = Response(stream_audio(), mimetype="audio/mpeg")
+    resp.headers.add_header("Cache-Control", "no-cache")
+    resp.headers.add_header("Connection", "keep-alive")
+    resp.headers.add_header("X-Accel-Buffering", "no")
+
+    return resp
+
+
+@manager.route('/delete_msg', methods=['POST'])
+@login_required
+@validate_request("conversation_id", "message_id")
+def delete_msg():
+    req = request.json
+    e, conv = ConversationService.get_by_id(req["conversation_id"])
+    if not e:
+        return get_data_error_result(message="Conversation not found!")
+
+    conv = conv.to_dict()
+    for i, msg in enumerate(conv["message"]):
+        if req["message_id"] != msg.get("id", ""):
+            continue
+        assert conv["message"][i + 1]["id"] == req["message_id"]
+        conv["message"].pop(i)
+        conv["message"].pop(i)
+        conv["reference"].pop(max(0, i // 2 - 1))
+        break
+
+    ConversationService.update_by_id(conv["id"], conv)
+    return get_json_result(data=conv)
+
+
+@manager.route('/thumbup', methods=['POST'])
+@login_required
+@validate_request("conversation_id", "message_id")
+def thumbup():
+    req = request.json
+    e, conv = ConversationService.get_by_id(req["conversation_id"])
+    if not e:
+        return get_data_error_result(message="Conversation not found!")
+    up_down = req.get("set")
+    feedback = req.get("feedback", "")
+    conv = conv.to_dict()
+    for i, msg in enumerate(conv["message"]):
+        if req["message_id"] == msg.get("id", "") and msg.get("role", "") == "assistant":
+            if up_down:
+                msg["thumbup"] = True
+                if "feedback" in msg: del msg["feedback"]
+            else:
+                msg["thumbup"] = False
+                if feedback: msg["feedback"] = feedback
+            break
+
+    ConversationService.update_by_id(conv["id"], conv)
+    return get_json_result(data=conv)
+
+
+@manager.route('/ask', methods=['POST'])
+@login_required
+@validate_request("question", "kb_ids")
+def ask_about():
+    req = request.json
+    uid = current_user.id
+
+    def stream():
+        nonlocal req, uid
+        try:
+            for ans in ask(req["question"], req["kb_ids"], uid):
+                yield "data:" + json.dumps({"code": 0, "message": "", "data": ans}, ensure_ascii=False) + "\n\n"
+        except Exception as e:
+            yield "data:" + json.dumps({"code": 500, "message": str(e),
+                                        "data": {"answer": "**ERROR**: " + str(e), "reference": []}},
+                                       ensure_ascii=False) + "\n\n"
+        yield "data:" + json.dumps({"code": 0, "message": "", "data": True}, ensure_ascii=False) + "\n\n"
+
+    resp = Response(stream(), mimetype="text/event-stream")
+    resp.headers.add_header("Cache-control", "no-cache")
+    resp.headers.add_header("Connection", "keep-alive")
+    resp.headers.add_header("X-Accel-Buffering", "no")
+    resp.headers.add_header("Content-Type", "text/event-stream; charset=utf-8")
+    return resp
+
+
+@manager.route('/mindmap', methods=['POST'])
+@login_required
+@validate_request("question", "kb_ids")
+def mindmap():
+    req = request.json
+    kb_ids = req["kb_ids"]
+    e, kb = KnowledgebaseService.get_by_id(kb_ids[0])
+    if not e:
+        return get_data_error_result(message="Knowledgebase not found!")
+
+    embd_mdl = TenantLLMService.model_instance(
+        kb.tenant_id, LLMType.EMBEDDING.value, llm_name=kb.embd_id)
+    chat_mdl = LLMBundle(current_user.id, LLMType.CHAT)
+    ranks = settings.retrievaler.retrieval(req["question"], embd_mdl, kb.tenant_id, kb_ids, 1, 12,
+                                           0.3, 0.3, aggs=False)
+    mindmap = MindMapExtractor(chat_mdl)
+    mind_map = mindmap([c["content_with_weight"] for c in ranks["chunks"]]).output
+    if "error" in mind_map:
+        return server_error_response(Exception(mind_map["error"]))
+    return get_json_result(data=mind_map)
+
+
+@manager.route('/related_questions', methods=['POST'])
+@login_required
+@validate_request("question")
+def related_questions():
+    req = request.json
+    question = req["question"]
+    chat_mdl = LLMBundle(current_user.id, LLMType.CHAT)
+    prompt = """
+Objective: To generate search terms related to the user's search keywords, helping users find more valuable information.
+Instructions:
+ - Based on the keywords provided by the user, generate 5-10 related search terms.
+ - Each search term should be directly or indirectly related to the keyword, guiding the user to find more valuable information.
+ - Use common, general terms as much as possible, avoiding obscure words or technical jargon.
+ - Keep the term length between 2-4 words, concise and clear.
+ - DO NOT translate, use the language of the original keywords.
+
+### Example:
+Keywords: Chinese football
+Related search terms:
+1. Current status of Chinese football
+2. Reform of Chinese football
+3. Youth training of Chinese football
+4. Chinese football in the Asian Cup
+5. Chinese football in the World Cup
+
+Reason:
+ - When searching, users often only use one or two keywords, making it difficult to fully express their information needs.
+ - Generating related search terms can help users dig deeper into relevant information and improve search efficiency. 
+ - At the same time, related terms can also help search engines better understand user needs and return more accurate search results.
+ 
+"""
+    ans = chat_mdl.chat(prompt, [{"role": "user", "content": f"""
+Keywords: {question}
+Related search terms:
+    """}], {"temperature": 0.9})
+    return get_json_result(data=[re.sub(r"^[0-9]\. ", "", a) for a in ans.split("\n") if re.match(r"^[0-9]\. ", a)])

api/apps/dataset_api.py

@@ -1,878 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-import os
-import pathlib
-import re
-import warnings
-from functools import partial
-from io import BytesIO
-
-from elasticsearch_dsl import Q
-from flask import request, send_file
-from flask_login import login_required, current_user
-from httpx import HTTPError
-
-from api.contants import NAME_LENGTH_LIMIT
-from api.db import FileType, ParserType, FileSource, TaskStatus
-from api.db import StatusEnum
-from api.db.db_models import File
-from api.db.services import duplicate_name
-from api.db.services.document_service import DocumentService
-from api.db.services.file2document_service import File2DocumentService
-from api.db.services.file_service import FileService
-from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.db.services.user_service import TenantService
-from api.settings import RetCode
-from api.utils import get_uuid
-from api.utils.api_utils import construct_json_result, construct_error_response
-from api.utils.api_utils import construct_result, validate_request
-from api.utils.file_utils import filename_type, thumbnail
-from rag.app import book, laws, manual, naive, one, paper, presentation, qa, resume, table, picture, audio, email
-from rag.nlp import search
-from rag.utils.es_conn import ELASTICSEARCH
-from rag.utils.minio_conn import MINIO
-
-MAXIMUM_OF_UPLOADING_FILES = 256
-
-
-# ------------------------------ create a dataset ---------------------------------------
-
-@manager.route("/", methods=["POST"])
-@login_required  # use login
-@validate_request("name")  # check name key
-def create_dataset():
-    # Check if Authorization header is present
-    authorization_token = request.headers.get("Authorization")
-    if not authorization_token:
-        return construct_json_result(code=RetCode.AUTHENTICATION_ERROR, message="Authorization header is missing.")
-
-    # TODO: Login or API key
-    # objs = APIToken.query(token=authorization_token)
-    #
-    # # Authorization error
-    # if not objs:
-    #     return construct_json_result(code=RetCode.AUTHENTICATION_ERROR, message="Token is invalid.")
-    #
-    # tenant_id = objs[0].tenant_id
-
-    tenant_id = current_user.id
-    request_body = request.json
-
-    # In case that there's no name
-    if "name" not in request_body:
-        return construct_json_result(code=RetCode.DATA_ERROR, message="Expected 'name' field in request body")
-
-    dataset_name = request_body["name"]
-
-    # empty dataset_name
-    if not dataset_name:
-        return construct_json_result(code=RetCode.DATA_ERROR, message="Empty dataset name")
-
-    # In case that there's space in the head or the tail
-    dataset_name = dataset_name.strip()
-
-    # In case that the length of the name exceeds the limit
-    dataset_name_length = len(dataset_name)
-    if dataset_name_length > NAME_LENGTH_LIMIT:
-        return construct_json_result(
-            code=RetCode.DATA_ERROR,
-            message=f"Dataset name: {dataset_name} with length {dataset_name_length} exceeds {NAME_LENGTH_LIMIT}!")
-
-    # In case that there are other fields in the data-binary
-    if len(request_body.keys()) > 1:
-        name_list = []
-        for key_name in request_body.keys():
-            if key_name != "name":
-                name_list.append(key_name)
-        return construct_json_result(code=RetCode.DATA_ERROR,
-                                     message=f"fields: {name_list}, are not allowed in request body.")
-
-    # If there is a duplicate name, it will modify it to make it unique
-    request_body["name"] = duplicate_name(
-        KnowledgebaseService.query,
-        name=dataset_name,
-        tenant_id=tenant_id,
-        status=StatusEnum.VALID.value)
-    try:
-        request_body["id"] = get_uuid()
-        request_body["tenant_id"] = tenant_id
-        request_body["created_by"] = tenant_id
-        exist, t = TenantService.get_by_id(tenant_id)
-        if not exist:
-            return construct_result(code=RetCode.AUTHENTICATION_ERROR, message="Tenant not found.")
-        request_body["embd_id"] = t.embd_id
-        if not KnowledgebaseService.save(**request_body):
-            # failed to create new dataset
-            return construct_result()
-        return construct_json_result(code=RetCode.SUCCESS,
-                                     data={"dataset_name": request_body["name"], "dataset_id": request_body["id"]})
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# -----------------------------list datasets-------------------------------------------------------
-
-@manager.route("/", methods=["GET"])
-@login_required
-def list_datasets():
-    offset = request.args.get("offset", 0)
-    count = request.args.get("count", -1)
-    orderby = request.args.get("orderby", "create_time")
-    desc = request.args.get("desc", True)
-    try:
-        tenants = TenantService.get_joined_tenants_by_user_id(current_user.id)
-        datasets = KnowledgebaseService.get_by_tenant_ids_by_offset(
-            [m["tenant_id"] for m in tenants], current_user.id, int(offset), int(count), orderby, desc)
-        return construct_json_result(data=datasets, code=RetCode.SUCCESS, message=f"List datasets successfully!")
-    except Exception as e:
-        return construct_error_response(e)
-    except HTTPError as http_err:
-        return construct_json_result(http_err)
-
-
-# ---------------------------------delete a dataset ----------------------------
-
-@manager.route("/<dataset_id>", methods=["DELETE"])
-@login_required
-def remove_dataset(dataset_id):
-    try:
-        datasets = KnowledgebaseService.query(created_by=current_user.id, id=dataset_id)
-
-        # according to the id, searching for the dataset
-        if not datasets:
-            return construct_json_result(message=f"The dataset cannot be found for your current account.",
-                                         code=RetCode.OPERATING_ERROR)
-
-        # Iterating the documents inside the dataset
-        for doc in DocumentService.query(kb_id=dataset_id):
-            if not DocumentService.remove_document(doc, datasets[0].tenant_id):
-                # the process of deleting failed
-                return construct_json_result(code=RetCode.DATA_ERROR,
-                                             message="There was an error during the document removal process. "
-                                                     "Please check the status of the RAGFlow server and try the removal again.")
-            # delete the other files
-            f2d = File2DocumentService.get_by_document_id(doc.id)
-            FileService.filter_delete([File.source_type == FileSource.KNOWLEDGEBASE, File.id == f2d[0].file_id])
-            File2DocumentService.delete_by_document_id(doc.id)
-
-        # delete the dataset
-        if not KnowledgebaseService.delete_by_id(dataset_id):
-            return construct_json_result(code=RetCode.DATA_ERROR,
-                                         message="There was an error during the dataset removal process. "
-                                                 "Please check the status of the RAGFlow server and try the removal again.")
-        # success
-        return construct_json_result(code=RetCode.SUCCESS, message=f"Remove dataset: {dataset_id} successfully")
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# ------------------------------ get details of a dataset ----------------------------------------
-
-@manager.route("/<dataset_id>", methods=["GET"])
-@login_required
-def get_dataset(dataset_id):
-    try:
-        dataset = KnowledgebaseService.get_detail(dataset_id)
-        if not dataset:
-            return construct_json_result(code=RetCode.DATA_ERROR, message="Can't find this dataset!")
-        return construct_json_result(data=dataset, code=RetCode.SUCCESS)
-    except Exception as e:
-        return construct_json_result(e)
-
-
-# ------------------------------ update a dataset --------------------------------------------
-
-@manager.route("/<dataset_id>", methods=["PUT"])
-@login_required
-def update_dataset(dataset_id):
-    req = request.json
-    try:
-        # the request cannot be empty
-        if not req:
-            return construct_json_result(code=RetCode.DATA_ERROR, message="Please input at least one parameter that "
-                                                                          "you want to update!")
-        # check whether the dataset can be found
-        if not KnowledgebaseService.query(created_by=current_user.id, id=dataset_id):
-            return construct_json_result(message=f"Only the owner of knowledgebase is authorized for this operation!",
-                                         code=RetCode.OPERATING_ERROR)
-
-        exist, dataset = KnowledgebaseService.get_by_id(dataset_id)
-        # check whether there is this dataset
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR, message="This dataset cannot be found!")
-
-        if "name" in req:
-            name = req["name"].strip()
-            # check whether there is duplicate name
-            if name.lower() != dataset.name.lower() \
-                    and len(KnowledgebaseService.query(name=name, tenant_id=current_user.id,
-                                                       status=StatusEnum.VALID.value)) > 1:
-                return construct_json_result(code=RetCode.DATA_ERROR,
-                                             message=f"The name: {name.lower()} is already used by other "
-                                                     f"datasets. Please choose a different name.")
-
-        dataset_updating_data = {}
-        chunk_num = req.get("chunk_num")
-        # modify the value of 11 parameters
-
-        # 2 parameters: embedding id and chunk method
-        # only if chunk_num is 0, the user can update the embedding id
-        if req.get("embedding_model_id"):
-            if chunk_num == 0:
-                dataset_updating_data["embd_id"] = req["embedding_model_id"]
-            else:
-                return construct_json_result(code=RetCode.DATA_ERROR,
-                                             message="You have already parsed the document in this "
-                                                     "dataset, so you cannot change the embedding "
-                                                     "model.")
-        # only if chunk_num is 0, the user can update the chunk_method
-        if "chunk_method" in req:
-            type_value = req["chunk_method"]
-            if is_illegal_value_for_enum(type_value, ParserType):
-                return construct_json_result(message=f"Illegal value {type_value} for 'chunk_method' field.",
-                                             code=RetCode.DATA_ERROR)
-            if chunk_num != 0:
-                construct_json_result(code=RetCode.DATA_ERROR, message="You have already parsed the document "
-                                                                       "in this dataset, so you cannot "
-                                                                       "change the chunk method.")
-            dataset_updating_data["parser_id"] = req["template_type"]
-
-        # convert the photo parameter to avatar
-        if req.get("photo"):
-            dataset_updating_data["avatar"] = req["photo"]
-
-        # layout_recognize
-        if "layout_recognize" in req:
-            if "parser_config" not in dataset_updating_data:
-                dataset_updating_data['parser_config'] = {}
-            dataset_updating_data['parser_config']['layout_recognize'] = req['layout_recognize']
-
-        # TODO: updating use_raptor needs to construct a class
-
-        # 6 parameters
-        for key in ["name", "language", "description", "permission", "id", "token_num"]:
-            if key in req:
-                dataset_updating_data[key] = req.get(key)
-
-        # update
-        if not KnowledgebaseService.update_by_id(dataset.id, dataset_updating_data):
-            return construct_json_result(code=RetCode.OPERATING_ERROR, message="Failed to update! "
-                                                                               "Please check the status of RAGFlow "
-                                                                               "server and try again!")
-
-        exist, dataset = KnowledgebaseService.get_by_id(dataset.id)
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR, message="Failed to get the dataset "
-                                                                          "using the dataset ID.")
-
-        return construct_json_result(data=dataset.to_json(), code=RetCode.SUCCESS)
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# --------------------------------content management ----------------------------------------------
-
-# ----------------------------upload files-----------------------------------------------------
-@manager.route("/<dataset_id>/documents/", methods=["POST"])
-@login_required
-def upload_documents(dataset_id):
-    # no files
-    if not request.files:
-        return construct_json_result(
-            message="There is no file!", code=RetCode.ARGUMENT_ERROR)
-
-    # the number of uploading files exceeds the limit
-    file_objs = request.files.getlist("file")
-    num_file_objs = len(file_objs)
-
-    if num_file_objs > MAXIMUM_OF_UPLOADING_FILES:
-        return construct_json_result(code=RetCode.DATA_ERROR, message=f"You try to upload {num_file_objs} files, "
-                                                                      f"which exceeds the maximum number of uploading files: {MAXIMUM_OF_UPLOADING_FILES}")
-
-    # no dataset
-    exist, dataset = KnowledgebaseService.get_by_id(dataset_id)
-    if not exist:
-        return construct_json_result(message="Can't find this dataset", code=RetCode.DATA_ERROR)
-
-    for file_obj in file_objs:
-        file_name = file_obj.filename
-        # no name
-        if not file_name:
-            return construct_json_result(
-                message="There is a file without name!", code=RetCode.ARGUMENT_ERROR)
-
-        # TODO: support the remote files
-        if 'http' in file_name:
-            return construct_json_result(code=RetCode.ARGUMENT_ERROR, message="Remote files have not unsupported.")
-
-    # get the root_folder
-    root_folder = FileService.get_root_folder(current_user.id)
-    # get the id of the root_folder
-    parent_file_id = root_folder["id"]  # document id
-    # this is for the new user, create '.knowledgebase' file
-    FileService.init_knowledgebase_docs(parent_file_id, current_user.id)
-    # go inside this folder, get the kb_root_folder
-    kb_root_folder = FileService.get_kb_folder(current_user.id)
-    # link the file management to the kb_folder
-    kb_folder = FileService.new_a_file_from_kb(dataset.tenant_id, dataset.name, kb_root_folder["id"])
-
-    # grab all the errs
-    err = []
-    MAX_FILE_NUM_PER_USER = int(os.environ.get("MAX_FILE_NUM_PER_USER", 0))
-    uploaded_docs_json = []
-    for file in file_objs:
-        try:
-            # TODO: get this value from the database as some tenants have this limit while others don't
-            if MAX_FILE_NUM_PER_USER > 0 and DocumentService.get_doc_count(dataset.tenant_id) >= MAX_FILE_NUM_PER_USER:
-                return construct_json_result(code=RetCode.DATA_ERROR,
-                                             message="Exceed the maximum file number of a free user!")
-            # deal with the duplicate name
-            filename = duplicate_name(
-                DocumentService.query,
-                name=file.filename,
-                kb_id=dataset.id)
-
-            # deal with the unsupported type
-            filetype = filename_type(filename)
-            if filetype == FileType.OTHER.value:
-                return construct_json_result(code=RetCode.DATA_ERROR,
-                                             message="This type of file has not been supported yet!")
-
-            # upload to the minio
-            location = filename
-            while MINIO.obj_exist(dataset_id, location):
-                location += "_"
-
-            blob = file.read()
-
-            # the content is empty, raising a warning
-            if blob == b'':
-                warnings.warn(f"[WARNING]: The content of the file {filename} is empty.")
-
-            MINIO.put(dataset_id, location, blob)
-
-            doc = {
-                "id": get_uuid(),
-                "kb_id": dataset.id,
-                "parser_id": dataset.parser_id,
-                "parser_config": dataset.parser_config,
-                "created_by": current_user.id,
-                "type": filetype,
-                "name": filename,
-                "location": location,
-                "size": len(blob),
-                "thumbnail": thumbnail(filename, blob)
-            }
-            if doc["type"] == FileType.VISUAL:
-                doc["parser_id"] = ParserType.PICTURE.value
-            if doc["type"] == FileType.AURAL:
-                doc["parser_id"] = ParserType.AUDIO.value
-            if re.search(r"\.(ppt|pptx|pages)$", filename):
-                doc["parser_id"] = ParserType.PRESENTATION.value
-            DocumentService.insert(doc)
-
-            FileService.add_file_from_kb(doc, kb_folder["id"], dataset.tenant_id)
-            uploaded_docs_json.append(doc)
-        except Exception as e:
-            err.append(file.filename + ": " + str(e))
-
-    if err:
-        # return all the errors
-        return construct_json_result(message="\n".join(err), code=RetCode.SERVER_ERROR)
-    # success
-    return construct_json_result(data=uploaded_docs_json, code=RetCode.SUCCESS)
-
-
-# ----------------------------delete a file-----------------------------------------------------
-@manager.route("/<dataset_id>/documents/<document_id>", methods=["DELETE"])
-@login_required
-def delete_document(document_id, dataset_id):  # string
-    # get the root folder
-    root_folder = FileService.get_root_folder(current_user.id)
-    # parent file's id
-    parent_file_id = root_folder["id"]
-    # consider the new user
-    FileService.init_knowledgebase_docs(parent_file_id, current_user.id)
-    # store all the errors that may have
-    errors = ""
-    try:
-        # whether there is this document
-        exist, doc = DocumentService.get_by_id(document_id)
-        if not exist:
-            return construct_json_result(message=f"Document {document_id} not found!", code=RetCode.DATA_ERROR)
-        # whether this doc is authorized by this tenant
-        tenant_id = DocumentService.get_tenant_id(document_id)
-        if not tenant_id:
-            return construct_json_result(
-                message=f"You cannot delete this document {document_id} due to the authorization"
-                        f" reason!", code=RetCode.AUTHENTICATION_ERROR)
-
-        # get the doc's id and location
-        real_dataset_id, location = File2DocumentService.get_minio_address(doc_id=document_id)
-
-        if real_dataset_id != dataset_id:
-            return construct_json_result(message=f"The document {document_id} is not in the dataset: {dataset_id}, "
-                                                 f"but in the dataset: {real_dataset_id}.", code=RetCode.ARGUMENT_ERROR)
-
-        # there is an issue when removing
-        if not DocumentService.remove_document(doc, tenant_id):
-            return construct_json_result(
-                message="There was an error during the document removal process. Please check the status of the "
-                        "RAGFlow server and try the removal again.", code=RetCode.OPERATING_ERROR)
-
-        # fetch the File2Document record associated with the provided document ID.
-        file_to_doc = File2DocumentService.get_by_document_id(document_id)
-        # delete the associated File record.
-        FileService.filter_delete([File.source_type == FileSource.KNOWLEDGEBASE, File.id == file_to_doc[0].file_id])
-        # delete the File2Document record itself using the document ID. This removes the
-        # association between the document and the file after the File record has been deleted.
-        File2DocumentService.delete_by_document_id(document_id)
-
-        # delete it from minio
-        MINIO.rm(dataset_id, location)
-    except Exception as e:
-        errors += str(e)
-    if errors:
-        return construct_json_result(data=False, message=errors, code=RetCode.SERVER_ERROR)
-
-    return construct_json_result(data=True, code=RetCode.SUCCESS)
-
-
-# ----------------------------list files-----------------------------------------------------
-@manager.route('/<dataset_id>/documents/', methods=['GET'])
-@login_required
-def list_documents(dataset_id):
-    if not dataset_id:
-        return construct_json_result(
-            data=False, message="Lack of 'dataset_id'", code=RetCode.ARGUMENT_ERROR)
-
-    # searching keywords
-    keywords = request.args.get("keywords", "")
-
-    offset = request.args.get("offset", 0)
-    count = request.args.get("count", -1)
-    order_by = request.args.get("order_by", "create_time")
-    descend = request.args.get("descend", True)
-    try:
-        docs, total = DocumentService.list_documents_in_dataset(dataset_id, int(offset), int(count), order_by,
-                                                                descend, keywords)
-
-        return construct_json_result(data={"total": total, "docs": docs}, message=RetCode.SUCCESS)
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# ----------------------------update: enable rename-----------------------------------------------------
-@manager.route("/<dataset_id>/documents/<document_id>", methods=["PUT"])
-@login_required
-def update_document(dataset_id, document_id):
-    req = request.json
-    try:
-        legal_parameters = set()
-        legal_parameters.add("name")
-        legal_parameters.add("enable")
-        legal_parameters.add("template_type")
-
-        for key in req.keys():
-            if key not in legal_parameters:
-                return construct_json_result(code=RetCode.ARGUMENT_ERROR, message=f"{key} is an illegal parameter.")
-
-        # The request body cannot be empty
-        if not req:
-            return construct_json_result(
-                code=RetCode.DATA_ERROR,
-                message="Please input at least one parameter that you want to update!")
-
-        # Check whether there is this dataset
-        exist, dataset = KnowledgebaseService.get_by_id(dataset_id)
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR, message=f"This dataset {dataset_id} cannot be found!")
-
-        # The document does not exist
-        exist, document = DocumentService.get_by_id(document_id)
-        if not exist:
-            return construct_json_result(message=f"This document {document_id} cannot be found!",
-                                         code=RetCode.ARGUMENT_ERROR)
-
-        # Deal with the different keys
-        updating_data = {}
-        if "name" in req:
-            new_name = req["name"]
-            updating_data["name"] = new_name
-            # Check whether the new_name is suitable
-            # 1. no name value
-            if not new_name:
-                return construct_json_result(code=RetCode.DATA_ERROR, message="There is no new name.")
-
-            # 2. In case that there's space in the head or the tail
-            new_name = new_name.strip()
-
-            # 3. Check whether the new_name has the same extension of file as before
-            if pathlib.Path(new_name.lower()).suffix != pathlib.Path(
-                    document.name.lower()).suffix:
-                return construct_json_result(
-                    data=False,
-                    message="The extension of file cannot be changed",
-                    code=RetCode.ARGUMENT_ERROR)
-
-            # 4. Check whether the new name has already been occupied by other file
-            for d in DocumentService.query(name=new_name, kb_id=document.kb_id):
-                if d.name == new_name:
-                    return construct_json_result(
-                        message="Duplicated document name in the same dataset.",
-                        code=RetCode.ARGUMENT_ERROR)
-
-        if "enable" in req:
-            enable_value = req["enable"]
-            if is_illegal_value_for_enum(enable_value, StatusEnum):
-                return construct_json_result(message=f"Illegal value {enable_value} for 'enable' field.",
-                                             code=RetCode.DATA_ERROR)
-            updating_data["status"] = enable_value
-
-        # TODO: Chunk-method - update parameters inside the json object parser_config
-        if "template_type" in req:
-            type_value = req["template_type"]
-            if is_illegal_value_for_enum(type_value, ParserType):
-                return construct_json_result(message=f"Illegal value {type_value} for 'template_type' field.",
-                                             code=RetCode.DATA_ERROR)
-            updating_data["parser_id"] = req["template_type"]
-
-        # The process of updating
-        if not DocumentService.update_by_id(document_id, updating_data):
-            return construct_json_result(
-                code=RetCode.OPERATING_ERROR,
-                message="Failed to update document in the database! "
-                        "Please check the status of RAGFlow server and try again!")
-
-        # name part: file service
-        if "name" in req:
-            # Get file by document id
-            file_information = File2DocumentService.get_by_document_id(document_id)
-            if file_information:
-                exist, file = FileService.get_by_id(file_information[0].file_id)
-                FileService.update_by_id(file.id, {"name": req["name"]})
-
-        exist, document = DocumentService.get_by_id(document_id)
-
-        # Success
-        return construct_json_result(data=document.to_json(), message="Success", code=RetCode.SUCCESS)
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# Helper method to judge whether it's an illegal value
-def is_illegal_value_for_enum(value, enum_class):
-    return value not in enum_class.__members__.values()
-
-
-# ----------------------------download a file-----------------------------------------------------
-@manager.route("/<dataset_id>/documents/<document_id>", methods=["GET"])
-@login_required
-def download_document(dataset_id, document_id):
-    try:
-        # Check whether there is this dataset
-        exist, _ = KnowledgebaseService.get_by_id(dataset_id)
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR,
-                                         message=f"This dataset '{dataset_id}' cannot be found!")
-
-        # Check whether there is this document
-        exist, document = DocumentService.get_by_id(document_id)
-        if not exist:
-            return construct_json_result(message=f"This document '{document_id}' cannot be found!",
-                                         code=RetCode.ARGUMENT_ERROR)
-
-        # The process of downloading
-        doc_id, doc_location = File2DocumentService.get_minio_address(doc_id=document_id)  # minio address
-        file_stream = MINIO.get(doc_id, doc_location)
-        if not file_stream:
-            return construct_json_result(message="This file is empty.", code=RetCode.DATA_ERROR)
-
-        file = BytesIO(file_stream)
-
-        # Use send_file with a proper filename and MIME type
-        return send_file(
-            file,
-            as_attachment=True,
-            download_name=document.name,
-            mimetype='application/octet-stream'  # Set a default MIME type
-        )
-
-    # Error
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# ----------------------------start parsing a document-----------------------------------------------------
-# helper method for parsing
-# callback method
-def doc_parse_callback(doc_id, prog=None, msg=""):
-    cancel = DocumentService.do_cancel(doc_id)
-    if cancel:
-        raise Exception("The parsing process has been cancelled!")
-
-"""
-def doc_parse(binary, doc_name, parser_name, tenant_id, doc_id):
-    match parser_name:
-        case "book":
-            book.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "laws":
-            laws.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "manual":
-            manual.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "naive":
-            # It's the mode by default, which is general in the front-end
-            naive.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "one":
-            one.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "paper":
-            paper.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "picture":
-            picture.chunk(doc_name, binary=binary, tenant_id=tenant_id, lang="Chinese",
-                          callback=partial(doc_parse_callback, doc_id))
-        case "presentation":
-            presentation.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "qa":
-            qa.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "resume":
-            resume.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "table":
-            table.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "audio":
-            audio.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case "email":
-            email.chunk(doc_name, binary=binary, callback=partial(doc_parse_callback, doc_id))
-        case _:
-            return False
-
-    return True
-    """
-
-
-@manager.route("/<dataset_id>/documents/<document_id>/status", methods=["POST"])
-@login_required
-def parse_document(dataset_id, document_id):
-    try:
-        # valid dataset
-        exist, _ = KnowledgebaseService.get_by_id(dataset_id)
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR,
-                                         message=f"This dataset '{dataset_id}' cannot be found!")
-
-        return parsing_document_internal(document_id)
-
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# ----------------------------start parsing documents-----------------------------------------------------
-@manager.route("/<dataset_id>/documents/status", methods=["POST"])
-@login_required
-def parse_documents(dataset_id):
-    doc_ids = request.json["doc_ids"]
-    try:
-        exist, _ = KnowledgebaseService.get_by_id(dataset_id)
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR,
-                                         message=f"This dataset '{dataset_id}' cannot be found!")
-        # two conditions
-        if not doc_ids:
-            # documents inside the dataset
-            docs, total = DocumentService.list_documents_in_dataset(dataset_id, 0, -1, "create_time",
-                                                                    True, "")
-            doc_ids = [doc["id"] for doc in docs]
-
-        message = ""
-        # for loop
-        for id in doc_ids:
-            res = parsing_document_internal(id)
-            res_body = res.json
-            if res_body["code"] == RetCode.SUCCESS:
-                message += res_body["message"]
-            else:
-                return res
-        return construct_json_result(data=True, code=RetCode.SUCCESS, message=message)
-
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# helper method for parsing the document
-def parsing_document_internal(id):
-    message = ""
-    try:
-        # Check whether there is this document
-        exist, document = DocumentService.get_by_id(id)
-        if not exist:
-            return construct_json_result(message=f"This document '{id}' cannot be found!",
-                                         code=RetCode.ARGUMENT_ERROR)
-
-        tenant_id = DocumentService.get_tenant_id(id)
-        if not tenant_id:
-            return construct_json_result(message="Tenant not found!", code=RetCode.AUTHENTICATION_ERROR)
-
-        info = {"run": "1", "progress": 0}
-        info["progress_msg"] = ""
-        info["chunk_num"] = 0
-        info["token_num"] = 0
-
-        DocumentService.update_by_id(id, info)
-
-        ELASTICSEARCH.deleteByQuery(Q("match", doc_id=id), idxnm=search.index_name(tenant_id))
-
-        _, doc_attributes = DocumentService.get_by_id(id)
-        doc_attributes = doc_attributes.to_dict()
-        doc_id = doc_attributes["id"]
-
-        bucket, doc_name = File2DocumentService.get_minio_address(doc_id=doc_id)
-        binary = MINIO.get(bucket, doc_name)
-        parser_name = doc_attributes["parser_id"]
-        if binary:
-            res = doc_parse(binary, doc_name, parser_name, tenant_id, doc_id)
-            if res is False:
-                message += f"The parser id: {parser_name} of the document {doc_id} is not supported; "
-        else:
-            message += f"Empty data in the document: {doc_name}; "
-        # failed in parsing
-        if doc_attributes["status"] == TaskStatus.FAIL.value:
-            message += f"Failed in parsing the document: {doc_id}; "
-        return construct_json_result(code=RetCode.SUCCESS, message=message)
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# ----------------------------stop parsing a doc-----------------------------------------------------
-@manager.route("<dataset_id>/documents/<document_id>/status", methods=["DELETE"])
-@login_required
-def stop_parsing_document(dataset_id, document_id):
-    try:
-        # valid dataset
-        exist, _ = KnowledgebaseService.get_by_id(dataset_id)
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR,
-                                         message=f"This dataset '{dataset_id}' cannot be found!")
-
-        return stop_parsing_document_internal(document_id)
-
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# ----------------------------stop parsing docs-----------------------------------------------------
-@manager.route("<dataset_id>/documents/status", methods=["DELETE"])
-@login_required
-def stop_parsing_documents(dataset_id):
-    doc_ids = request.json["doc_ids"]
-    try:
-        # valid dataset?
-        exist, _ = KnowledgebaseService.get_by_id(dataset_id)
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR,
-                                         message=f"This dataset '{dataset_id}' cannot be found!")
-        if not doc_ids:
-            # documents inside the dataset
-            docs, total = DocumentService.list_documents_in_dataset(dataset_id, 0, -1, "create_time",
-                                                                        True, "")
-            doc_ids = [doc["id"] for doc in docs]
-
-        message = ""
-        # for loop
-        for id in doc_ids:
-            res = stop_parsing_document_internal(id)
-            res_body = res.json
-            if res_body["code"] == RetCode.SUCCESS:
-                message += res_body["message"]
-            else:
-                return res
-        return construct_json_result(data=True, code=RetCode.SUCCESS, message=message)
-
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# Helper method
-def stop_parsing_document_internal(document_id):
-    try:
-        # valid doc?
-        exist, doc = DocumentService.get_by_id(document_id)
-        if not exist:
-            return construct_json_result(message=f"This document '{document_id}' cannot be found!",
-                                         code=RetCode.ARGUMENT_ERROR)
-        doc_attributes = doc.to_dict()
-
-        # only when the status is parsing, we need to stop it
-        if doc_attributes["status"] == TaskStatus.RUNNING.value:
-            tenant_id = DocumentService.get_tenant_id(document_id)
-            if not tenant_id:
-                return construct_json_result(message="Tenant not found!", code=RetCode.AUTHENTICATION_ERROR)
-
-            # update successfully?
-            if not DocumentService.update_by_id(document_id, {"status": "2"}):  # cancel
-                return construct_json_result(
-                    code=RetCode.OPERATING_ERROR,
-                    message="There was an error during the stopping parsing the document process. "
-                            "Please check the status of the RAGFlow server and try the update again."
-                )
-
-            _, doc_attributes = DocumentService.get_by_id(document_id)
-            doc_attributes = doc_attributes.to_dict()
-
-            # failed in stop parsing
-            if doc_attributes["status"] == TaskStatus.RUNNING.value:
-                return construct_json_result(message=f"Failed in parsing the document: {document_id}; ", code=RetCode.SUCCESS)
-        return construct_json_result(code=RetCode.SUCCESS, message="")
-    except Exception as e:
-        return construct_error_response(e)
-
-
-# ----------------------------show the status of the file-----------------------------------------------------
-@manager.route("/<dataset_id>/documents/<document_id>/status", methods=["GET"])
-@login_required
-def show_parsing_status(dataset_id, document_id):
-    try:
-        # valid dataset
-        exist, _ = KnowledgebaseService.get_by_id(dataset_id)
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR,
-                                         message=f"This dataset: '{dataset_id}' cannot be found!")
-        # valid document
-        exist, _ = DocumentService.get_by_id(document_id)
-        if not exist:
-            return construct_json_result(code=RetCode.DATA_ERROR,
-                                         message=f"This document: '{document_id}' is not a valid document.")
-
-        _, doc = DocumentService.get_by_id(document_id)  # get doc object
-        doc_attributes = doc.to_dict()
-
-        return construct_json_result(
-            data={"progress": doc_attributes["progress"], "status": TaskStatus(doc_attributes["status"]).name},
-            code=RetCode.SUCCESS
-        )
-    except Exception as e:
-        return construct_error_response(e)
-
-# ----------------------------list the chunks of the file-----------------------------------------------------
-
-# -- --------------------------delete the chunk-----------------------------------------------------
-
-# ----------------------------edit the status of the chunk-----------------------------------------------------
-
-# ----------------------------insert a new chunk-----------------------------------------------------
-
-# ----------------------------upload a file-----------------------------------------------------
-
-# ----------------------------get a specific chunk-----------------------------------------------------
-
-# ----------------------------retrieval test-----------------------------------------------------

api/apps/dialog_app.py

@@ -19,7 +19,8 @@ from flask_login import login_required, current_user
 from api.db.services.dialog_service import DialogService
 from api.db import StatusEnum
 from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.db.services.user_service import TenantService
+from api.db.services.user_service import TenantService, UserTenantService
+from api import settings
 from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
 from api.utils import get_uuid
 from api.utils.api_utils import get_json_result
@@ -67,17 +68,17 @@ def set_dialog():
             continue
         if prompt_config["system"].find("{%s}" % p["key"]) < 0:
             return get_data_error_result(
-                retmsg="Parameter '{}' is not used".format(p["key"]))
+                message="Parameter '{}' is not used".format(p["key"]))
 
     try:
         e, tenant = TenantService.get_by_id(current_user.id)
         if not e:
-            return get_data_error_result(retmsg="Tenant not found!")
+            return get_data_error_result(message="Tenant not found!")
         llm_id = req.get("llm_id", tenant.llm_id)
         if not dialog_id:
             if not req.get("kb_ids"):
                 return get_data_error_result(
-                    retmsg="Fail! Please select knowledgebase!")
+                    message="Fail! Please select knowledgebase!")
             dia = {
                 "id": get_uuid(),
                 "tenant_id": current_user.id,
@@ -95,20 +96,20 @@ def set_dialog():
                 "icon": icon
             }
             if not DialogService.save(**dia):
-                return get_data_error_result(retmsg="Fail to new a dialog!")
+                return get_data_error_result(message="Fail to new a dialog!")
             e, dia = DialogService.get_by_id(dia["id"])
             if not e:
-                return get_data_error_result(retmsg="Fail to new a dialog!")
+                return get_data_error_result(message="Fail to new a dialog!")
             return get_json_result(data=dia.to_json())
         else:
             del req["dialog_id"]
             if "kb_names" in req:
                 del req["kb_names"]
             if not DialogService.update_by_id(dialog_id, req):
-                return get_data_error_result(retmsg="Dialog not found!")
+                return get_data_error_result(message="Dialog not found!")
             e, dia = DialogService.get_by_id(dialog_id)
             if not e:
-                return get_data_error_result(retmsg="Fail to update a dialog!")
+                return get_data_error_result(message="Fail to update a dialog!")
             dia = dia.to_dict()
             dia["kb_ids"], dia["kb_names"] = get_kb_names(dia["kb_ids"])
             return get_json_result(data=dia)
@@ -123,7 +124,7 @@ def get():
     try:
         e, dia = DialogService.get_by_id(dialog_id)
         if not e:
-            return get_data_error_result(retmsg="Dialog not found!")
+            return get_data_error_result(message="Dialog not found!")
         dia = dia.to_dict()
         dia["kb_ids"], dia["kb_names"] = get_kb_names(dia["kb_ids"])
         return get_json_result(data=dia)
@@ -164,9 +165,19 @@ def list_dialogs():
 @validate_request("dialog_ids")
 def rm():
     req = request.json
+    dialog_list=[]
+    tenants = UserTenantService.query(user_id=current_user.id)
     try:
-        DialogService.update_many_by_id(
-            [{"id": id, "status": StatusEnum.INVALID.value} for id in req["dialog_ids"]])
+        for id in req["dialog_ids"]:
+            for tenant in tenants:
+                if DialogService.query(tenant_id=tenant.tenant_id, id=id):
+                    break
+            else:
+                return get_json_result(
+                    data=False, message='Only owner of dialog authorized for this operation.',
+                    code=settings.RetCode.OPERATING_ERROR)
+            dialog_list.append({"id": id,"status":StatusEnum.INVALID.value})
+        DialogService.update_many_by_id(dialog_list)
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)

api/apps/document_app.py

@@ -13,44 +13,34 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License
 #
-import datetime
-import hashlib
 import json
-import os
+import os.path
 import pathlib
 import re
-import traceback
-from concurrent.futures import ThreadPoolExecutor
-from copy import deepcopy
-from io import BytesIO
 
 import flask
-from elasticsearch_dsl import Q
 from flask import request
 from flask_login import login_required, current_user
 
 from api.db.db_models import Task, File
-from api.db.services.dialog_service import DialogService, ConversationService
 from api.db.services.file2document_service import File2DocumentService
 from api.db.services.file_service import FileService
-from api.db.services.llm_service import LLMBundle
 from api.db.services.task_service import TaskService, queue_tasks
-from api.db.services.user_service import TenantService
-from graphrag.mind_map_extractor import MindMapExtractor
-from rag.app import naive
+from api.db.services.user_service import UserTenantService
+from deepdoc.parser.html_parser import RAGFlowHtmlParser
 from rag.nlp import search
-from rag.utils.es_conn import ELASTICSEARCH
 from api.db.services import duplicate_name
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
 from api.utils import get_uuid
-from api.db import FileType, TaskStatus, ParserType, FileSource, LLMType
+from api.db import FileType, TaskStatus, ParserType, FileSource
 from api.db.services.document_service import DocumentService, doc_upload_and_parse
-from api.settings import RetCode, stat_logger
+from api import settings
 from api.utils.api_utils import get_json_result
-from rag.utils.minio_conn import MINIO
+from rag.utils.storage_factory import STORAGE_IMPL
 from api.utils.file_utils import filename_type, thumbnail, get_project_base_directory
 from api.utils.web_utils import html2pdf, is_valid_url
+from api.constants import IMG_BASE64_PREFIX
 
 
 @manager.route('/upload', methods=['POST'])
@@ -60,16 +50,16 @@ def upload():
     kb_id = request.form.get("kb_id")
     if not kb_id:
         return get_json_result(
-            data=False, retmsg='Lack of "KB ID"', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='Lack of "KB ID"', code=settings.RetCode.ARGUMENT_ERROR)
     if 'file' not in request.files:
         return get_json_result(
-            data=False, retmsg='No file part!', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='No file part!', code=settings.RetCode.ARGUMENT_ERROR)
 
     file_objs = request.files.getlist('file')
     for file_obj in file_objs:
         if file_obj.filename == '':
             return get_json_result(
-                data=False, retmsg='No file selected!', retcode=RetCode.ARGUMENT_ERROR)
+                data=False, message='No file selected!', code=settings.RetCode.ARGUMENT_ERROR)
 
     e, kb = KnowledgebaseService.get_by_id(kb_id)
     if not e:
@@ -78,7 +68,7 @@ def upload():
     err, _ = FileService.upload_document(kb, file_objs, current_user.id)
     if err:
         return get_json_result(
-            data=False, retmsg="\n".join(err), retcode=RetCode.SERVER_ERROR)
+            data=False, message="\n".join(err), code=settings.RetCode.SERVER_ERROR)
     return get_json_result(data=True)
 
 
@@ -89,12 +79,12 @@ def web_crawl():
     kb_id = request.form.get("kb_id")
     if not kb_id:
         return get_json_result(
-            data=False, retmsg='Lack of "KB ID"', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='Lack of "KB ID"', code=settings.RetCode.ARGUMENT_ERROR)
     name = request.form.get("name")
     url = request.form.get("url")
     if not is_valid_url(url):
         return get_json_result(
-            data=False, retmsg='The URL format is invalid', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='The URL format is invalid', code=settings.RetCode.ARGUMENT_ERROR)
     e, kb = KnowledgebaseService.get_by_id(kb_id)
     if not e:
         raise LookupError("Can't find this knowledgebase!")
@@ -118,9 +108,9 @@ def web_crawl():
             raise RuntimeError("This type of file has not been supported yet!")
 
         location = filename
-        while MINIO.obj_exist(kb_id, location):
+        while STORAGE_IMPL.obj_exist(kb_id, location):
             location += "_"
-        MINIO.put(kb_id, location, blob)
+        STORAGE_IMPL.put(kb_id, location, blob)
         doc = {
             "id": get_uuid(),
             "kb_id": kb.id,
@@ -139,6 +129,8 @@ def web_crawl():
             doc["parser_id"] = ParserType.AUDIO.value
         if re.search(r"\.(ppt|pptx|pages)$", filename):
             doc["parser_id"] = ParserType.PRESENTATION.value
+        if re.search(r"\.(eml)$", filename):
+            doc["parser_id"] = ParserType.EMAIL.value
         DocumentService.insert(doc)
         FileService.add_file_from_kb(doc, kb_folder["id"], kb.tenant_id)
     except Exception as e:
@@ -154,17 +146,17 @@ def create():
     kb_id = req["kb_id"]
     if not kb_id:
         return get_json_result(
-            data=False, retmsg='Lack of "KB ID"', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='Lack of "KB ID"', code=settings.RetCode.ARGUMENT_ERROR)
 
     try:
         e, kb = KnowledgebaseService.get_by_id(kb_id)
         if not e:
             return get_data_error_result(
-                retmsg="Can't find this knowledgebase!")
+                message="Can't find this knowledgebase!")
 
         if DocumentService.query(name=req["name"], kb_id=kb_id):
             return get_data_error_result(
-                retmsg="Duplicated document name in the same knowledgebase.")
+                message="Duplicated document name in the same knowledgebase.")
 
         doc = DocumentService.insert({
             "id": get_uuid(),
@@ -188,7 +180,16 @@ def list_docs():
     kb_id = request.args.get("kb_id")
     if not kb_id:
         return get_json_result(
-            data=False, retmsg='Lack of "KB ID"', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='Lack of "KB ID"', code=settings.RetCode.ARGUMENT_ERROR)
+    tenants = UserTenantService.query(user_id=current_user.id)
+    for tenant in tenants:
+        if KnowledgebaseService.query(
+                tenant_id=tenant.tenant_id, id=kb_id):
+            break
+    else:
+        return get_json_result(
+            data=False, message='Only owner of knowledgebase authorized for this operation.',
+            code=settings.RetCode.OPERATING_ERROR)
     keywords = request.args.get("keywords", "")
 
     page_number = int(request.args.get("page", 1))
@@ -198,29 +199,47 @@ def list_docs():
     try:
         docs, tol = DocumentService.get_by_kb_id(
             kb_id, page_number, items_per_page, orderby, desc, keywords)
+
+        for doc_item in docs:
+            if doc_item['thumbnail'] and not doc_item['thumbnail'].startswith(IMG_BASE64_PREFIX):
+                doc_item['thumbnail'] = f"/v1/document/image/{kb_id}-{doc_item['thumbnail']}"
+
         return get_json_result(data={"total": tol, "docs": docs})
     except Exception as e:
         return server_error_response(e)
 
 
 @manager.route('/infos', methods=['POST'])
+@login_required
 def docinfos():
     req = request.json
     doc_ids = req["doc_ids"]
+    for doc_id in doc_ids:
+        if not DocumentService.accessible(doc_id, current_user.id):
+            return get_json_result(
+                data=False,
+                message='No authorization.',
+                code=settings.RetCode.AUTHENTICATION_ERROR
+            )
     docs = DocumentService.get_by_ids(doc_ids)
     return get_json_result(data=list(docs.dicts()))
 
 
 @manager.route('/thumbnails', methods=['GET'])
-#@login_required
+# @login_required
 def thumbnails():
     doc_ids = request.args.get("doc_ids").split(",")
     if not doc_ids:
         return get_json_result(
-            data=False, retmsg='Lack of "Document ID"', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='Lack of "Document ID"', code=settings.RetCode.ARGUMENT_ERROR)
 
     try:
         docs = DocumentService.get_thumbnails(doc_ids)
+
+        for doc_item in docs:
+            if doc_item['thumbnail'] and not doc_item['thumbnail'].startswith(IMG_BASE64_PREFIX):
+                doc_item['thumbnail'] = f"/v1/document/image/{doc_item['kb_id']}-{doc_item['thumbnail']}"
+
         return get_json_result(data={d["id"]: d["thumbnail"] for d in docs})
     except Exception as e:
         return server_error_response(e)
@@ -232,37 +251,34 @@ def thumbnails():
 def change_status():
     req = request.json
     if str(req["status"]) not in ["0", "1"]:
-        get_json_result(
+        return get_json_result(
             data=False,
-            retmsg='"Status" must be either 0 or 1!',
-            retcode=RetCode.ARGUMENT_ERROR)
+            message='"Status" must be either 0 or 1!',
+            code=settings.RetCode.ARGUMENT_ERROR)
+
+    if not DocumentService.accessible(req["doc_id"], current_user.id):
+        return get_json_result(
+            data=False,
+            message='No authorization.',
+            code=settings.RetCode.AUTHENTICATION_ERROR)
 
     try:
         e, doc = DocumentService.get_by_id(req["doc_id"])
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
+            return get_data_error_result(message="Document not found!")
         e, kb = KnowledgebaseService.get_by_id(doc.kb_id)
         if not e:
             return get_data_error_result(
-                retmsg="Can't find this knowledgebase!")
+                message="Can't find this knowledgebase!")
 
         if not DocumentService.update_by_id(
                 req["doc_id"], {"status": str(req["status"])}):
             return get_data_error_result(
-                retmsg="Database error (Document update)!")
+                message="Database error (Document update)!")
 
-        if str(req["status"]) == "0":
-            ELASTICSEARCH.updateScriptByQuery(Q("term", doc_id=req["doc_id"]),
-                                              scripts="ctx._source.available_int=0;",
-                                              idxnm=search.index_name(
-                                                  kb.tenant_id)
-                                              )
-        else:
-            ELASTICSEARCH.updateScriptByQuery(Q("term", doc_id=req["doc_id"]),
-                                              scripts="ctx._source.available_int=1;",
-                                              idxnm=search.index_name(
-                                                  kb.tenant_id)
-                                              )
+        status = int(req["status"])
+        settings.docStoreConn.update({"doc_id": req["doc_id"]}, {"available_int": status},
+                                     search.index_name(kb.tenant_id), doc.kb_id)
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)
@@ -275,6 +291,15 @@ def rm():
     req = request.json
     doc_ids = req["doc_id"]
     if isinstance(doc_ids, str): doc_ids = [doc_ids]
+
+    for doc_id in doc_ids:
+        if not DocumentService.accessible4deletion(doc_id, current_user.id):
+            return get_json_result(
+                data=False,
+                message='No authorization.',
+                code=settings.RetCode.AUTHENTICATION_ERROR
+            )
+
     root_folder = FileService.get_root_folder(current_user.id)
     pf_id = root_folder["id"]
     FileService.init_knowledgebase_docs(pf_id, current_user.id)
@@ -283,27 +308,27 @@ def rm():
         try:
             e, doc = DocumentService.get_by_id(doc_id)
             if not e:
-                return get_data_error_result(retmsg="Document not found!")
+                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(retmsg="Tenant not found!")
+                return get_data_error_result(message="Tenant not found!")
 
-            b, n = File2DocumentService.get_minio_address(doc_id=doc_id)
+            b, n = File2DocumentService.get_storage_address(doc_id=doc_id)
 
             if not DocumentService.remove_document(doc, tenant_id):
                 return get_data_error_result(
-                    retmsg="Database error (Document removal)!")
+                    message="Database error (Document removal)!")
 
             f2d = File2DocumentService.get_by_document_id(doc_id)
             FileService.filter_delete([File.source_type == FileSource.KNOWLEDGEBASE, File.id == f2d[0].file_id])
             File2DocumentService.delete_by_document_id(doc_id)
 
-            MINIO.rm(b, n)
+            STORAGE_IMPL.rm(b, n)
         except Exception as e:
             errors += str(e)
 
     if errors:
-        return get_json_result(data=False, retmsg=errors, retcode=RetCode.SERVER_ERROR)
+        return get_json_result(data=False, message=errors, code=settings.RetCode.SERVER_ERROR)
 
     return get_json_result(data=True)
 
@@ -313,6 +338,13 @@ def rm():
 @validate_request("doc_ids", "run")
 def run():
     req = request.json
+    for doc_id in req["doc_ids"]:
+        if not DocumentService.accessible(doc_id, current_user.id):
+            return get_json_result(
+                data=False,
+                message='No authorization.',
+                code=settings.RetCode.AUTHENTICATION_ERROR
+            )
     try:
         for id in req["doc_ids"]:
             info = {"run": str(req["run"]), "progress": 0}
@@ -324,16 +356,19 @@ def run():
             # if str(req["run"]) == TaskStatus.CANCEL.value:
             tenant_id = DocumentService.get_tenant_id(id)
             if not tenant_id:
-                return get_data_error_result(retmsg="Tenant not found!")
-            ELASTICSEARCH.deleteByQuery(
-                Q("match", doc_id=id), idxnm=search.index_name(tenant_id))
+                return get_data_error_result(message="Tenant not found!")
+            e, doc = DocumentService.get_by_id(id)
+            if not e:
+                return get_data_error_result(message="Document not found!")
+            if settings.docStoreConn.indexExist(search.index_name(tenant_id), doc.kb_id):
+                settings.docStoreConn.delete({"doc_id": id}, search.index_name(tenant_id), doc.kb_id)
 
             if str(req["run"]) == TaskStatus.RUNNING.value:
                 TaskService.filter_delete([Task.doc_id == id])
                 e, doc = DocumentService.get_by_id(id)
                 doc = doc.to_dict()
                 doc["tenant_id"] = tenant_id
-                bucket, name = File2DocumentService.get_minio_address(doc_id=doc["id"])
+                bucket, name = File2DocumentService.get_storage_address(doc_id=doc["id"])
                 queue_tasks(doc, bucket, name)
 
         return get_json_result(data=True)
@@ -346,25 +381,31 @@ def run():
 @validate_request("doc_id", "name")
 def rename():
     req = request.json
+    if not DocumentService.accessible(req["doc_id"], current_user.id):
+        return get_json_result(
+            data=False,
+            message='No authorization.',
+            code=settings.RetCode.AUTHENTICATION_ERROR
+        )
     try:
         e, doc = DocumentService.get_by_id(req["doc_id"])
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
+            return get_data_error_result(message="Document not found!")
         if pathlib.Path(req["name"].lower()).suffix != pathlib.Path(
                 doc.name.lower()).suffix:
             return get_json_result(
                 data=False,
-                retmsg="The extension of file can't be changed",
-                retcode=RetCode.ARGUMENT_ERROR)
+                message="The extension of file can't be changed",
+                code=settings.RetCode.ARGUMENT_ERROR)
         for d in DocumentService.query(name=req["name"], kb_id=doc.kb_id):
             if d.name == req["name"]:
                 return get_data_error_result(
-                    retmsg="Duplicated document name in the same knowledgebase.")
+                    message="Duplicated document name in the same knowledgebase.")
 
         if not DocumentService.update_by_id(
                 req["doc_id"], {"name": req["name"]}):
             return get_data_error_result(
-                retmsg="Database error (Document rename)!")
+                message="Database error (Document rename)!")
 
         informs = File2DocumentService.get_by_document_id(req["doc_id"])
         if informs:
@@ -382,10 +423,10 @@ def get(doc_id):
     try:
         e, doc = DocumentService.get_by_id(doc_id)
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
+            return get_data_error_result(message="Document not found!")
 
-        b, n = File2DocumentService.get_minio_address(doc_id=doc_id)
-        response = flask.make_response(MINIO.get(b, n))
+        b, n = File2DocumentService.get_storage_address(doc_id=doc_id)
+        response = flask.make_response(STORAGE_IMPL.get(b, n))
 
         ext = re.search(r"\.([^.]+)$", doc.name)
         if ext:
@@ -406,10 +447,17 @@ def get(doc_id):
 @validate_request("doc_id", "parser_id")
 def change_parser():
     req = request.json
+
+    if not DocumentService.accessible(req["doc_id"], current_user.id):
+        return get_json_result(
+            data=False,
+            message='No authorization.',
+            code=settings.RetCode.AUTHENTICATION_ERROR
+        )
     try:
         e, doc = DocumentService.get_by_id(req["doc_id"])
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
+            return get_data_error_result(message="Document not found!")
         if doc.parser_id.lower() == req["parser_id"].lower():
             if "parser_config" in req:
                 if req["parser_config"] == doc.parser_config:
@@ -417,27 +465,28 @@ def change_parser():
             else:
                 return get_json_result(data=True)
 
-        if doc.type == FileType.VISUAL or re.search(
-                r"\.(ppt|pptx|pages)$", doc.name):
-            return get_data_error_result(retmsg="Not supported yet!")
+        if ((doc.type == FileType.VISUAL and req["parser_id"] != "picture")
+                or (re.search(
+                    r"\.(ppt|pptx|pages)$", doc.name) and req["parser_id"] != "presentation")):
+            return get_data_error_result(message="Not supported yet!")
 
         e = DocumentService.update_by_id(doc.id,
                                          {"parser_id": req["parser_id"], "progress": 0, "progress_msg": "",
                                           "run": TaskStatus.UNSTART.value})
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
+            return get_data_error_result(message="Document not found!")
         if "parser_config" in req:
             DocumentService.update_parser_config(doc.id, req["parser_config"])
         if doc.token_num > 0:
             e = DocumentService.increment_chunk_num(doc.id, doc.kb_id, doc.token_num * -1, doc.chunk_num * -1,
                                                     doc.process_duation * -1)
             if not e:
-                return get_data_error_result(retmsg="Document not found!")
+                return get_data_error_result(message="Document not found!")
             tenant_id = DocumentService.get_tenant_id(req["doc_id"])
             if not tenant_id:
-                return get_data_error_result(retmsg="Tenant not found!")
-            ELASTICSEARCH.deleteByQuery(
-                Q("match", doc_id=doc.id), idxnm=search.index_name(tenant_id))
+                return get_data_error_result(message="Tenant not found!")
+            if settings.docStoreConn.indexExist(search.index_name(tenant_id), doc.kb_id):
+                settings.docStoreConn.delete({"doc_id": doc.id}, search.index_name(tenant_id), doc.kb_id)
 
         return get_json_result(data=True)
     except Exception as e:
@@ -449,7 +498,7 @@ def change_parser():
 def get_image(image_id):
     try:
         bkt, nm = image_id.split("-")
-        response = flask.make_response(MINIO.get(bkt, nm))
+        response = flask.make_response(STORAGE_IMPL.get(bkt, nm))
         response.headers.set('Content-Type', 'image/JPEG')
         return response
     except Exception as e:
@@ -462,14 +511,74 @@ def get_image(image_id):
 def upload_and_parse():
     if 'file' not in request.files:
         return get_json_result(
-            data=False, retmsg='No file part!', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='No file part!', code=settings.RetCode.ARGUMENT_ERROR)
 
     file_objs = request.files.getlist('file')
     for file_obj in file_objs:
         if file_obj.filename == '':
             return get_json_result(
-                data=False, retmsg='No file selected!', retcode=RetCode.ARGUMENT_ERROR)
+                data=False, message='No file selected!', code=settings.RetCode.ARGUMENT_ERROR)
 
     doc_ids = doc_upload_and_parse(request.form.get("conversation_id"), file_objs, current_user.id)
 
     return get_json_result(data=doc_ids)
+
+
+@manager.route('/parse', methods=['POST'])
+@login_required
+def parse():
+    url = request.json.get("url") if request.json else ""
+    if url:
+        if not is_valid_url(url):
+            return get_json_result(
+                data=False, message='The URL format is invalid', code=settings.RetCode.ARGUMENT_ERROR)
+        download_path = os.path.join(get_project_base_directory(), "logs/downloads")
+        os.makedirs(download_path, exist_ok=True)
+        from seleniumwire.webdriver import Chrome, ChromeOptions
+        options = ChromeOptions()
+        options.add_argument('--headless')
+        options.add_argument('--disable-gpu')
+        options.add_argument('--no-sandbox')
+        options.add_argument('--disable-dev-shm-usage')
+        options.add_experimental_option('prefs', {
+            'download.default_directory': download_path,
+            'download.prompt_for_download': False,
+            'download.directory_upgrade': True,
+            'safebrowsing.enabled': True
+        })
+        driver = Chrome(options=options)
+        driver.get(url)
+        res_headers = [r.response.headers for r in driver.requests]
+        if len(res_headers) > 1:
+            sections = RAGFlowHtmlParser().parser_txt(driver.page_source)
+            driver.quit()
+            return get_json_result(data="\n".join(sections))
+
+        class File:
+            filename: str
+            filepath: str
+
+            def __init__(self, filename, filepath):
+                self.filename = filename
+                self.filepath = filepath
+
+            def read(self):
+                with open(self.filepath, "rb") as f:
+                    return f.read()
+
+        r = re.search(r"filename=\"([^\"]+)\"", str(res_headers))
+        if not r or not r.group(1):
+            return get_json_result(
+                data=False, message="Can't not identify downloaded file", code=settings.RetCode.ARGUMENT_ERROR)
+        f = File(r.group(1), os.path.join(download_path, r.group(1)))
+        txt = FileService.parse_docs([f], current_user.id)
+        return get_json_result(data=txt)
+
+    if 'file' not in request.files:
+        return get_json_result(
+            data=False, message='No file part!', code=settings.RetCode.ARGUMENT_ERROR)
+
+    file_objs = request.files.getlist('file')
+    txt = FileService.parse_docs(file_objs, current_user.id)
+
+    return get_json_result(data=txt)

api/apps/file2document_app.py

@@ -13,9 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License
 #
-from elasticsearch_dsl import Q
 
-from api.db.db_models import File2Document
 from api.db.services.file2document_service import File2DocumentService
 from api.db.services.file_service import FileService
 
@@ -26,10 +24,8 @@ from api.utils.api_utils import server_error_response, get_data_error_result, va
 from api.utils import get_uuid
 from api.db import FileType
 from api.db.services.document_service import DocumentService
-from api.settings import RetCode
+from api import settings
 from api.utils.api_utils import get_json_result
-from rag.nlp import search
-from rag.utils.es_conn import ELASTICSEARCH
 
 
 @manager.route('/convert', methods=['POST'])
@@ -54,13 +50,13 @@ def convert():
                     doc_id = inform.document_id
                     e, doc = DocumentService.get_by_id(doc_id)
                     if not e:
-                        return get_data_error_result(retmsg="Document not found!")
+                        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(retmsg="Tenant not found!")
+                        return get_data_error_result(message="Tenant not found!")
                     if not DocumentService.remove_document(doc, tenant_id):
                         return get_data_error_result(
-                            retmsg="Database error (Document removal)!")
+                            message="Database error (Document removal)!")
                 File2DocumentService.delete_by_file_id(id)
 
                 # insert
@@ -68,16 +64,16 @@ def convert():
                     e, kb = KnowledgebaseService.get_by_id(kb_id)
                     if not e:
                         return get_data_error_result(
-                            retmsg="Can't find this knowledgebase!")
+                            message="Can't find this knowledgebase!")
                     e, file = FileService.get_by_id(id)
                     if not e:
                         return get_data_error_result(
-                            retmsg="Can't find this file!")
+                            message="Can't find this file!")
 
                     doc = DocumentService.insert({
                         "id": get_uuid(),
                         "kb_id": kb.id,
-                        "parser_id": kb.parser_id,
+                        "parser_id": FileService.get_parser(file.type, file.name, kb.parser_id),
                         "parser_config": kb.parser_config,
                         "created_by": current_user.id,
                         "type": file.type,
@@ -104,26 +100,26 @@ def rm():
     file_ids = req["file_ids"]
     if not file_ids:
         return get_json_result(
-            data=False, retmsg='Lack of "Files ID"', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='Lack of "Files ID"', code=settings.RetCode.ARGUMENT_ERROR)
     try:
         for file_id in file_ids:
             informs = File2DocumentService.get_by_file_id(file_id)
             if not informs:
-                return get_data_error_result(retmsg="Inform not found!")
+                return get_data_error_result(message="Inform not found!")
             for inform in informs:
                 if not inform:
-                    return get_data_error_result(retmsg="Inform not found!")
+                    return get_data_error_result(message="Inform not found!")
                 File2DocumentService.delete_by_file_id(file_id)
                 doc_id = inform.document_id
                 e, doc = DocumentService.get_by_id(doc_id)
                 if not e:
-                    return get_data_error_result(retmsg="Document not found!")
+                    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(retmsg="Tenant not found!")
+                    return get_data_error_result(message="Tenant not found!")
                 if not DocumentService.remove_document(doc, tenant_id):
                     return get_data_error_result(
-                        retmsg="Database error (Document removal)!")
+                        message="Database error (Document removal)!")
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)

api/apps/file_app.py

@@ -18,7 +18,6 @@ import pathlib
 import re
 
 import flask
-from elasticsearch_dsl import Q
 from flask import request
 from flask_login import login_required, current_user
 
@@ -29,12 +28,10 @@ from api.utils import get_uuid
 from api.db import FileType, FileSource
 from api.db.services import duplicate_name
 from api.db.services.file_service import FileService
-from api.settings import RetCode
+from api import settings
 from api.utils.api_utils import get_json_result
 from api.utils.file_utils import filename_type
-from rag.nlp import search
-from rag.utils.es_conn import ELASTICSEARCH
-from rag.utils.minio_conn import MINIO
+from rag.utils.storage_factory import STORAGE_IMPL
 
 
 @manager.route('/upload', methods=['POST'])
@@ -49,24 +46,24 @@ def upload():
 
     if 'file' not in request.files:
         return get_json_result(
-            data=False, retmsg='No file part!', retcode=RetCode.ARGUMENT_ERROR)
+            data=False, message='No file part!', code=settings.RetCode.ARGUMENT_ERROR)
     file_objs = request.files.getlist('file')
 
     for file_obj in file_objs:
         if file_obj.filename == '':
             return get_json_result(
-                data=False, retmsg='No file selected!', retcode=RetCode.ARGUMENT_ERROR)
+                data=False, message='No file selected!', code=settings.RetCode.ARGUMENT_ERROR)
     file_res = []
     try:
         for file_obj in file_objs:
             e, file = FileService.get_by_id(pf_id)
             if not e:
                 return get_data_error_result(
-                    retmsg="Can't find this folder!")
+                    message="Can't find this folder!")
             MAX_FILE_NUM_PER_USER = int(os.environ.get('MAX_FILE_NUM_PER_USER', 0))
             if MAX_FILE_NUM_PER_USER > 0 and DocumentService.get_doc_count(current_user.id) >= MAX_FILE_NUM_PER_USER:
                 return get_data_error_result(
-                    retmsg="Exceed the maximum file number of a free user!")
+                    message="Exceed the maximum file number of a free user!")
 
             # split file name path
             if not file_obj.filename:
@@ -85,20 +82,20 @@ def upload():
             if file_len != len_id_list:
                 e, file = FileService.get_by_id(file_id_list[len_id_list - 1])
                 if not e:
-                    return get_data_error_result(retmsg="Folder not found!")
+                    return get_data_error_result(message="Folder not found!")
                 last_folder = FileService.create_folder(file, file_id_list[len_id_list - 1], file_obj_names,
                                                         len_id_list)
             else:
                 e, file = FileService.get_by_id(file_id_list[len_id_list - 2])
                 if not e:
-                    return get_data_error_result(retmsg="Folder not found!")
+                    return get_data_error_result(message="Folder not found!")
                 last_folder = FileService.create_folder(file, file_id_list[len_id_list - 2], file_obj_names,
                                                         len_id_list)
 
             # file type
             filetype = filename_type(file_obj_names[file_len - 1])
             location = file_obj_names[file_len - 1]
-            while MINIO.obj_exist(last_folder.id, location):
+            while STORAGE_IMPL.obj_exist(last_folder.id, location):
                 location += "_"
             blob = file_obj.read()
             filename = duplicate_name(
@@ -116,7 +113,7 @@ def upload():
                 "size": len(blob),
             }
             file = FileService.insert(file)
-            MINIO.put(last_folder.id, location, blob)
+            STORAGE_IMPL.put(last_folder.id, location, blob)
             file_res.append(file.to_json())
         return get_json_result(data=file_res)
     except Exception as e:
@@ -137,10 +134,10 @@ def create():
     try:
         if not FileService.is_parent_folder_exist(pf_id):
             return get_json_result(
-                data=False, retmsg="Parent Folder Doesn't Exist!", retcode=RetCode.OPERATING_ERROR)
+                data=False, message="Parent Folder Doesn't Exist!", code=settings.RetCode.OPERATING_ERROR)
         if FileService.query(name=req["name"], parent_id=pf_id):
             return get_data_error_result(
-                retmsg="Duplicated folder name in the same folder.")
+                message="Duplicated folder name in the same folder.")
 
         if input_file_type == FileType.FOLDER.value:
             file_type = FileType.FOLDER.value
@@ -181,14 +178,14 @@ def list_files():
     try:
         e, file = FileService.get_by_id(pf_id)
         if not e:
-            return get_data_error_result(retmsg="Folder not found!")
+            return get_data_error_result(message="Folder not found!")
 
         files, total = FileService.get_by_pf_id(
             current_user.id, pf_id, page_number, items_per_page, orderby, desc, keywords)
 
         parent_folder = FileService.get_parent_folder(pf_id)
         if not FileService.get_parent_folder(pf_id):
-            return get_json_result(retmsg="File not found!")
+            return get_json_result(message="File not found!")
 
         return get_json_result(data={"total": total, "files": files, "parent_folder": parent_folder.to_json()})
     except Exception as e:
@@ -212,7 +209,7 @@ def get_parent_folder():
     try:
         e, file = FileService.get_by_id(file_id)
         if not e:
-            return get_data_error_result(retmsg="Folder not found!")
+            return get_data_error_result(message="Folder not found!")
 
         parent_folder = FileService.get_parent_folder(file_id)
         return get_json_result(data={"parent_folder": parent_folder.to_json()})
@@ -227,7 +224,7 @@ def get_all_parent_folders():
     try:
         e, file = FileService.get_by_id(file_id)
         if not e:
-            return get_data_error_result(retmsg="Folder not found!")
+            return get_data_error_result(message="Folder not found!")
 
         parent_folders = FileService.get_all_parent_folders(file_id)
         parent_folders_res = []
@@ -248,9 +245,9 @@ def rm():
         for file_id in file_ids:
             e, file = FileService.get_by_id(file_id)
             if not e:
-                return get_data_error_result(retmsg="File or Folder not found!")
+                return get_data_error_result(message="File or Folder not found!")
             if not file.tenant_id:
-                return get_data_error_result(retmsg="Tenant not found!")
+                return get_data_error_result(message="Tenant not found!")
             if file.source_type == FileSource.KNOWLEDGEBASE:
                 continue
 
@@ -259,13 +256,13 @@ def rm():
                 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(retmsg="File not found!")
-                    MINIO.rm(file.parent_id, file.location)
+                        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:
                 if not FileService.delete(file):
                     return get_data_error_result(
-                        retmsg="Database error (File removal)!")
+                        message="Database error (File removal)!")
 
             # delete file2document
             informs = File2DocumentService.get_by_file_id(file_id)
@@ -273,13 +270,13 @@ def rm():
                 doc_id = inform.document_id
                 e, doc = DocumentService.get_by_id(doc_id)
                 if not e:
-                    return get_data_error_result(retmsg="Document not found!")
+                    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(retmsg="Tenant not found!")
+                    return get_data_error_result(message="Tenant not found!")
                 if not DocumentService.remove_document(doc, tenant_id):
                     return get_data_error_result(
-                        retmsg="Database error (Document removal)!")
+                        message="Database error (Document removal)!")
             File2DocumentService.delete_by_file_id(file_id)
 
         return get_json_result(data=True)
@@ -295,29 +292,30 @@ def rename():
     try:
         e, file = FileService.get_by_id(req["file_id"])
         if not e:
-            return get_data_error_result(retmsg="File not found!")
-        if pathlib.Path(req["name"].lower()).suffix != pathlib.Path(
+            return get_data_error_result(message="File not found!")
+        if file.type != FileType.FOLDER.value \
+            and pathlib.Path(req["name"].lower()).suffix != pathlib.Path(
                 file.name.lower()).suffix:
             return get_json_result(
                 data=False,
-                retmsg="The extension of file can't be changed",
-                retcode=RetCode.ARGUMENT_ERROR)
+                message="The extension of file can't be changed",
+                code=settings.RetCode.ARGUMENT_ERROR)
         for file in FileService.query(name=req["name"], pf_id=file.parent_id):
             if file.name == req["name"]:
                 return get_data_error_result(
-                    retmsg="Duplicated file name in the same folder.")
+                    message="Duplicated file name in the same folder.")
 
         if not FileService.update_by_id(
                 req["file_id"], {"name": req["name"]}):
             return get_data_error_result(
-                retmsg="Database error (File rename)!")
+                message="Database error (File rename)!")
 
         informs = File2DocumentService.get_by_file_id(req["file_id"])
         if informs:
             if not DocumentService.update_by_id(
                     informs[0].document_id, {"name": req["name"]}):
                 return get_data_error_result(
-                    retmsg="Database error (Document rename)!")
+                    message="Database error (Document rename)!")
 
         return get_json_result(data=True)
     except Exception as e:
@@ -330,9 +328,9 @@ def get(file_id):
     try:
         e, file = FileService.get_by_id(file_id)
         if not e:
-            return get_data_error_result(retmsg="Document not found!")
-        b, n = File2DocumentService.get_minio_address(file_id=file_id)
-        response = flask.make_response(MINIO.get(b, n))
+            return get_data_error_result(message="Document not found!")
+        b, n = File2DocumentService.get_storage_address(file_id=file_id)
+        response = flask.make_response(STORAGE_IMPL.get(b, n))
         ext = re.search(r"\.([^.]+)$", file.name)
         if ext:
             if file.type == FileType.VISUAL.value:
@@ -358,12 +356,12 @@ def move():
         for file_id in file_ids:
             e, file = FileService.get_by_id(file_id)
             if not e:
-                return get_data_error_result(retmsg="File or Folder not found!")
+                return get_data_error_result(message="File or Folder not found!")
             if not file.tenant_id:
-                return get_data_error_result(retmsg="Tenant not found!")
+                return get_data_error_result(message="Tenant not found!")
         fe, _ = FileService.get_by_id(parent_id)
         if not fe:
-            return get_data_error_result(retmsg="Parent Folder not found!")
+            return get_data_error_result(message="Parent Folder not found!")
         FileService.move_file(file_ids, parent_id)
         return get_json_result(data=True)
     except Exception as e:

api/apps/kb_app.py

@@ -13,7 +13,6 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-from elasticsearch_dsl import Q
 from flask import request
 from flask_login import login_required, current_user
 
@@ -23,14 +22,13 @@ from api.db.services.file2document_service import File2DocumentService
 from api.db.services.file_service import FileService
 from api.db.services.user_service import TenantService, UserTenantService
 from api.utils.api_utils import server_error_response, get_data_error_result, validate_request
-from api.utils import get_uuid, get_format_time
-from api.db import StatusEnum, UserTenantRole, FileSource
+from api.utils import get_uuid
+from api.db import StatusEnum, FileSource
 from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.db.db_models import Knowledgebase, File
-from api.settings import stat_logger, RetCode
+from api.db.db_models import File
 from api.utils.api_utils import get_json_result
+from api import settings
 from rag.nlp import search
-from rag.utils.es_conn import ELASTICSEARCH
 
 
 @manager.route('/create', methods=['post'])
@@ -50,7 +48,7 @@ def create():
         req["created_by"] = current_user.id
         e, t = TenantService.get_by_id(current_user.id)
         if not e:
-            return get_data_error_result(retmsg="Tenant not found.")
+            return get_data_error_result(message="Tenant not found.")
         req["embd_id"] = t.embd_id
         if not KnowledgebaseService.save(**req):
             return get_data_error_result()
@@ -65,21 +63,27 @@ def create():
 def update():
     req = request.json
     req["name"] = req["name"].strip()
+    if not KnowledgebaseService.accessible4deletion(req["kb_id"], current_user.id):
+        return get_json_result(
+            data=False,
+            message='No authorization.',
+            code=settings.RetCode.AUTHENTICATION_ERROR
+        )
     try:
         if not KnowledgebaseService.query(
                 created_by=current_user.id, id=req["kb_id"]):
             return get_json_result(
-                data=False, retmsg=f'Only owner of knowledgebase authorized for this operation.', retcode=RetCode.OPERATING_ERROR)
+                data=False, message='Only owner of knowledgebase authorized for this operation.', code=settings.RetCode.OPERATING_ERROR)
 
         e, kb = KnowledgebaseService.get_by_id(req["kb_id"])
         if not e:
             return get_data_error_result(
-                retmsg="Can't find this knowledgebase!")
+                message="Can't find this knowledgebase!")
 
         if req["name"].lower() != kb.name.lower() \
                 and len(KnowledgebaseService.query(name=req["name"], tenant_id=current_user.id, status=StatusEnum.VALID.value)) > 1:
             return get_data_error_result(
-                retmsg="Duplicated knowledgebase name.")
+                message="Duplicated knowledgebase name.")
 
         del req["kb_id"]
         if not KnowledgebaseService.update_by_id(kb.id, req):
@@ -88,7 +92,7 @@ def update():
         e, kb = KnowledgebaseService.get_by_id(kb.id)
         if not e:
             return get_data_error_result(
-                retmsg="Database error (Knowledgebase rename)!")
+                message="Database error (Knowledgebase rename)!")
 
         return get_json_result(data=kb.to_json())
     except Exception as e:
@@ -100,10 +104,19 @@ def update():
 def detail():
     kb_id = request.args["kb_id"]
     try:
+        tenants = UserTenantService.query(user_id=current_user.id)
+        for tenant in tenants:
+            if KnowledgebaseService.query(
+                    tenant_id=tenant.tenant_id, id=kb_id):
+                break
+        else:
+            return get_json_result(
+                data=False, message='Only owner of knowledgebase authorized for this operation.',
+                code=settings.RetCode.OPERATING_ERROR)
         kb = KnowledgebaseService.get_detail(kb_id)
         if not kb:
             return get_data_error_result(
-                retmsg="Can't find this knowledgebase!")
+                message="Can't find this knowledgebase!")
         return get_json_result(data=kb)
     except Exception as e:
         return server_error_response(e)
@@ -130,24 +143,32 @@ def list_kbs():
 @validate_request("kb_id")
 def rm():
     req = request.json
+    if not KnowledgebaseService.accessible4deletion(req["kb_id"], current_user.id):
+        return get_json_result(
+            data=False,
+            message='No authorization.',
+            code=settings.RetCode.AUTHENTICATION_ERROR
+        )
     try:
         kbs = KnowledgebaseService.query(
                 created_by=current_user.id, id=req["kb_id"])
         if not kbs:
             return get_json_result(
-                data=False, retmsg=f'Only owner of knowledgebase authorized for this operation.', retcode=RetCode.OPERATING_ERROR)
+                data=False, message='Only owner of knowledgebase authorized for this operation.', code=settings.RetCode.OPERATING_ERROR)
 
         for doc in DocumentService.query(kb_id=req["kb_id"]):
             if not DocumentService.remove_document(doc, kbs[0].tenant_id):
                 return get_data_error_result(
-                    retmsg="Database error (Document removal)!")
+                    message="Database error (Document removal)!")
             f2d = File2DocumentService.get_by_document_id(doc.id)
             FileService.filter_delete([File.source_type == FileSource.KNOWLEDGEBASE, File.id == f2d[0].file_id])
+            FileService.filter_delete([File.source_type == FileSource.KNOWLEDGEBASE, File.type == "folder", File.name == kbs[0].name])
             File2DocumentService.delete_by_document_id(doc.id)
 
         if not KnowledgebaseService.delete_by_id(req["kb_id"]):
             return get_data_error_result(
-                retmsg="Database error (Knowledgebase removal)!")
+                message="Database error (Knowledgebase removal)!")
+        settings.docStoreConn.delete({"kb_id": req["kb_id"]}, search.index_name(kbs[0].tenant_id), req["kb_id"])
         return get_json_result(data=True)
     except Exception as e:
         return server_error_response(e)

api/apps/llm_app.py

@@ -13,23 +13,39 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
+import json
+
 from flask import request
 from flask_login import login_required, current_user
 from api.db.services.llm_service import LLMFactoriesService, TenantLLMService, 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
 from api.utils.api_utils import get_json_result
-from rag.llm import EmbeddingModel, ChatModel, RerankModel,CvModel
+from rag.llm import EmbeddingModel, ChatModel, RerankModel, CvModel, TTSModel
 import requests
-import ast
+
 
 @manager.route('/factories', methods=['GET'])
 @login_required
 def factories():
     try:
         fac = LLMFactoriesService.get_all()
-        return get_json_result(data=[f.to_dict() for f in fac if f.name not in ["Youdao", "FastEmbed", "BAAI"]])
+        fac = [f.to_dict() for f in fac if f.name not in ["Youdao", "FastEmbed", "BAAI"]]
+        llms = LLMService.get_all()
+        mdl_types = {}
+        for m in llms:
+            if m.status != StatusEnum.VALID.value:
+                continue
+            if m.fid not in mdl_types:
+                mdl_types[m.fid] = set([])
+            mdl_types[m.fid].add(m.model_type)
+        for f in fac:
+            f["model_types"] = list(mdl_types.get(f["name"], [LLMType.CHAT, LLMType.EMBEDDING, LLMType.RERANK,
+                                                              LLMType.IMAGE2TEXT, LLMType.SPEECH2TEXT, LLMType.TTS]))
+        return get_json_result(data=fac)
     except Exception as e:
         return server_error_response(e)
 
@@ -58,14 +74,14 @@ def set_api_key():
             mdl = ChatModel[factory](
                 req["api_key"], llm.llm_name, base_url=req.get("base_url"))
             try:
-                m, tc = mdl.chat(None, [{"role": "user", "content": "Hello! How are you doing!"}], 
-                                 {"temperature": 0.9,'max_tokens':50})
-                if m.find("**ERROR**") >=0:
+                m, tc = mdl.chat(None, [{"role": "user", "content": "Hello! How are you doing!"}],
+                                 {"temperature": 0.9, 'max_tokens': 50})
+                if m.find("**ERROR**") >= 0:
                     raise Exception(m)
+                chat_passed = True
             except Exception as e:
                 msg += f"\nFail to access model({llm.llm_name}) using this api key." + str(
                     e)
-            chat_passed = True
         elif not rerank_passed and llm.model_type == LLMType.RERANK:
             mdl = RerankModel[factory](
                 req["api_key"], llm.llm_name, base_url=req.get("base_url"))
@@ -73,32 +89,41 @@ def set_api_key():
                 arr, tc = mdl.similarity("What's the weather?", ["Is it sunny today?"])
                 if len(arr) == 0 or tc == 0:
                     raise Exception("Fail")
+                rerank_passed = True
+                logging.debug(f'passed model rerank {llm.llm_name}')
             except Exception as e:
                 msg += f"\nFail to access model({llm.llm_name}) using this api key." + str(
                     e)
-            rerank_passed = True
+        if any([embd_passed, chat_passed, rerank_passed]):
+            msg = ''
+            break
 
     if msg:
-        return get_data_error_result(retmsg=msg)
+        return get_data_error_result(message=msg)
 
-    llm = {
+    llm_config = {
         "api_key": req["api_key"],
         "api_base": req.get("base_url", "")
     }
     for n in ["model_type", "llm_name"]:
         if n in req:
-            llm[n] = req[n]
+            llm_config[n] = req[n]
 
-    if not TenantLLMService.filter_update(
-            [TenantLLM.tenant_id == current_user.id, TenantLLM.llm_factory == factory], llm):
-        for llm in LLMService.query(fid=factory):
+    for llm in LLMService.query(fid=factory):
+        llm_config["max_tokens"]=llm.max_tokens
+        if not TenantLLMService.filter_update(
+                [TenantLLM.tenant_id == current_user.id,
+                 TenantLLM.llm_factory == factory,
+                 TenantLLM.llm_name == llm.llm_name],
+                llm_config):
             TenantLLMService.save(
                 tenant_id=current_user.id,
                 llm_factory=factory,
                 llm_name=llm.llm_name,
                 model_type=llm.model_type,
-                api_key=req["api_key"],
-                api_base=req.get("base_url", "")
+                api_key=llm_config["api_key"],
+                api_base=llm_config["api_base"],
+                max_tokens=llm_config["max_tokens"]
             )
 
     return get_json_result(data=True)
@@ -111,43 +136,67 @@ def add_llm():
     req = request.json
     factory = req["llm_factory"]
 
+    def apikey_json(keys):
+        nonlocal req
+        return json.dumps({k: req.get(k, "") for k in keys})
+
     if factory == "VolcEngine":
         # For VolcEngine, due to its special authentication method
-        # Assemble volc_ak, volc_sk, endpoint_id into api_key
-        temp = list(ast.literal_eval(req["llm_name"]).items())[0]
-        llm_name = temp[0]
-        endpoint_id = temp[1]
-        api_key = '{' + f'"volc_ak": "{req.get("volc_ak", "")}", ' \
-                        f'"volc_sk": "{req.get("volc_sk", "")}", ' \
-                        f'"ep_id": "{endpoint_id}", ' + '}'
+        # Assemble ark_api_key endpoint_id into api_key
+        llm_name = req["llm_name"]
+        api_key = apikey_json(["ark_api_key", "endpoint_id"])
+
     elif factory == "Tencent Hunyuan":
-        api_key = '{' + f'"hunyuan_sid": "{req.get("hunyuan_sid", "")}", ' \
-                        f'"hunyuan_sk": "{req.get("hunyuan_sk", "")}"' + '}'
-        req["api_key"] = api_key
+        req["api_key"] = apikey_json(["hunyuan_sid", "hunyuan_sk"])
         return set_api_key()
+
+    elif factory == "Tencent Cloud":
+        req["api_key"] = apikey_json(["tencent_cloud_sid", "tencent_cloud_sk"])
+
     elif factory == "Bedrock":
         # For Bedrock, due to its special authentication method
         # Assemble bedrock_ak, bedrock_sk, bedrock_region
         llm_name = req["llm_name"]
-        api_key = '{' + f'"bedrock_ak": "{req.get("bedrock_ak", "")}", ' \
-                        f'"bedrock_sk": "{req.get("bedrock_sk", "")}", ' \
-                        f'"bedrock_region": "{req.get("bedrock_region", "")}", ' + '}'
+        api_key = apikey_json(["bedrock_ak", "bedrock_sk", "bedrock_region"])
+
     elif factory == "LocalAI":
-        llm_name = req["llm_name"]+"___LocalAI"
+        llm_name = req["llm_name"] + "___LocalAI"
         api_key = "xxxxxxxxxxxxxxx"
+
+    elif factory == "HuggingFace":
+        llm_name = req["llm_name"] + "___HuggingFace"
+        api_key = "xxxxxxxxxxxxxxx"
+
     elif factory == "OpenAI-API-Compatible":
-        llm_name = req["llm_name"]+"___OpenAI-API"
-        api_key = req.get("api_key","xxxxxxxxxxxxxxx")
-    elif factory =="XunFei Spark":
+        llm_name = req["llm_name"] + "___OpenAI-API"
+        api_key = req.get("api_key", "xxxxxxxxxxxxxxx")
+
+    elif factory == "XunFei Spark":
         llm_name = req["llm_name"]
-        api_key = req.get("spark_api_password","xxxxxxxxxxxxxxx") 
+        if req["model_type"] == "chat":
+            api_key = req.get("spark_api_password", "xxxxxxxxxxxxxxx")
+        elif req["model_type"] == "tts":
+            api_key = apikey_json(["spark_app_id", "spark_api_secret", "spark_api_key"])
+
     elif factory == "BaiduYiyan":
         llm_name = req["llm_name"]
-        api_key = '{' + f'"yiyan_ak": "{req.get("yiyan_ak", "")}", ' \
-                f'"yiyan_sk": "{req.get("yiyan_sk", "")}"' + '}'
+        api_key = apikey_json(["yiyan_ak", "yiyan_sk"])
+
+    elif factory == "Fish Audio":
+        llm_name = req["llm_name"]
+        api_key = apikey_json(["fish_audio_ak", "fish_audio_refid"])
+
+    elif factory == "Google Cloud":
+        llm_name = req["llm_name"]
+        api_key = apikey_json(["google_project_id", "google_region", "google_service_account_key"])
+
+    elif factory == "Azure-OpenAI":
+        llm_name = req["llm_name"]
+        api_key = apikey_json(["api_key", "api_version"])
+
     else:
         llm_name = req["llm_name"]
-        api_key = req.get("api_key","xxxxxxxxxxxxxxx") 
+        api_key = req.get("api_key", "xxxxxxxxxxxxxxx")
 
     llm = {
         "tenant_id": current_user.id,
@@ -155,14 +204,15 @@ def add_llm():
         "model_type": req["model_type"],
         "llm_name": llm_name,
         "api_base": req.get("api_base", ""),
-        "api_key": api_key
+        "api_key": api_key,
+        "max_tokens": req.get("max_tokens")
     }
 
     msg = ""
     if llm["model_type"] == LLMType.EMBEDDING.value:
         mdl = EmbeddingModel[factory](
             key=llm['api_key'],
-            model_name=llm["llm_name"], 
+            model_name=llm["llm_name"],
             base_url=llm["api_base"])
         try:
             arr, tc = mdl.encode(["Test if the api key is available"])
@@ -178,7 +228,7 @@ def add_llm():
         )
         try:
             m, tc = mdl.chat(None, [{"role": "user", "content": "Hello! How are you doing!"}], {
-                             "temperature": 0.9})
+                "temperature": 0.9})
             if not tc:
                 raise Exception(m)
         except Exception as e:
@@ -186,12 +236,12 @@ def add_llm():
                 e)
     elif llm["model_type"] == LLMType.RERANK:
         mdl = RerankModel[factory](
-            key=llm["api_key"], 
-            model_name=llm["llm_name"], 
+            key=llm["api_key"],
+            model_name=llm["llm_name"],
             base_url=llm["api_base"]
         )
         try:
-            arr, tc = mdl.similarity("Hello~ Ragflower!", ["Hi, there!"])
+            arr, tc = mdl.similarity("Hello~ Ragflower!", ["Hi, there!", "Ohh, my friend!"])
             if len(arr) == 0 or tc == 0:
                 raise Exception("Not known.")
         except Exception as e:
@@ -199,8 +249,8 @@ def add_llm():
                 e)
     elif llm["model_type"] == LLMType.IMAGE2TEXT.value:
         mdl = CvModel[factory](
-            key=llm["api_key"], 
-            model_name=llm["llm_name"], 
+            key=llm["api_key"],
+            model_name=llm["llm_name"],
             base_url=llm["api_base"]
         )
         try:
@@ -218,15 +268,25 @@ def add_llm():
                 pass
         except Exception as e:
             msg += f"\nFail to access model({llm['llm_name']})." + str(e)
+    elif llm["model_type"] == LLMType.TTS:
+        mdl = TTSModel[factory](
+            key=llm["api_key"], model_name=llm["llm_name"], base_url=llm["api_base"]
+        )
+        try:
+            for resp in mdl.tts("Hello~ Ragflower!"):
+                pass
+        except RuntimeError as e:
+            msg += f"\nFail to access model({llm['llm_name']})." + str(e)
     else:
         # TODO: check other type of models
         pass
 
     if msg:
-        return get_data_error_result(retmsg=msg)
+        return get_data_error_result(message=msg)
 
     if not TenantLLMService.filter_update(
-            [TenantLLM.tenant_id == current_user.id, TenantLLM.llm_factory == factory, TenantLLM.llm_name == llm["llm_name"]], llm):
+            [TenantLLM.tenant_id == current_user.id, TenantLLM.llm_factory == factory,
+             TenantLLM.llm_name == llm["llm_name"]], llm):
         TenantLLMService.save(**llm)
 
     return get_json_result(data=True)
@@ -238,7 +298,18 @@ def add_llm():
 def delete_llm():
     req = request.json
     TenantLLMService.filter_delete(
-            [TenantLLM.tenant_id == current_user.id, TenantLLM.llm_factory == req["llm_factory"], TenantLLM.llm_name == req["llm_name"]])
+        [TenantLLM.tenant_id == current_user.id, TenantLLM.llm_factory == req["llm_factory"],
+         TenantLLM.llm_name == req["llm_name"]])
+    return get_json_result(data=True)
+
+
+@manager.route('/delete_factory', methods=['POST'])
+@login_required
+@validate_request("llm_factory")
+def delete_factory():
+    req = request.json
+    TenantLLMService.filter_delete(
+        [TenantLLM.tenant_id == current_user.id, TenantLLM.llm_factory == req["llm_factory"]])
     return get_json_result(data=True)
 
 
@@ -266,25 +337,27 @@ def my_llms():
 @manager.route('/list', methods=['GET'])
 @login_required
 def list_app():
+    self_deploied = ["Youdao", "FastEmbed", "BAAI", "Ollama", "Xinference", "LocalAI", "LM-Studio"]
+    weighted = ["Youdao", "FastEmbed", "BAAI"] if settings.LIGHTEN != 0 else []
     model_type = request.args.get("model_type")
     try:
         objs = TenantLLMService.query(tenant_id=current_user.id)
         facts = set([o.to_dict()["llm_factory"] for o in objs if o.api_key])
         llms = LLMService.get_all()
         llms = [m.to_dict()
-                for m in llms if m.status == StatusEnum.VALID.value]
+                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 ["Youdao","FastEmbed", "BAAI"]
+            m["available"] = m["fid"] in facts or m["llm_name"].lower() == "flag-embedding" or m["fid"] in self_deploied
 
-        llm_set = set([m["llm_name"] for m in llms])
+        llm_set = set([m["llm_name"] + "@" + m["fid"] for m in llms])
         for o in objs:
-            if not o.api_key:continue
-            if o.llm_name in llm_set:continue
+            if not o.api_key: continue
+            if o.llm_name + "@" + o.llm_factory in llm_set: continue
             llms.append({"llm_name": o.llm_name, "model_type": o.model_type, "fid": o.llm_factory, "available": True})
 
         res = {}
         for m in llms:
-            if model_type and m["model_type"].find(model_type)<0:
+            if model_type and m["model_type"].find(model_type) < 0:
                 continue
             if m["fid"] not in res:
                 res[m["fid"]] = []
@@ -292,4 +365,4 @@ def list_app():
 
         return get_json_result(data=res)
     except Exception as e:
-        return server_error_response(e)
+        return server_error_response(e)

api/apps/sdk/chat.py

@@ -0,0 +1,313 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+from flask import request
+from api import settings
+from api.db import StatusEnum
+from api.db.services.dialog_service import DialogService
+from api.db.services.knowledgebase_service import KnowledgebaseService
+from api.db.services.llm_service import  TenantLLMService
+from api.db.services.user_service import TenantService
+from api.utils import get_uuid
+from api.utils.api_utils import get_error_data_result, token_required
+from api.utils.api_utils import get_result
+
+
+
+@manager.route('/chats', methods=['POST'])
+@token_required
+def create(tenant_id):
+    req=request.json
+    ids= req.get("dataset_ids")
+    if not ids:
+        return get_error_data_result(message="`dataset_ids` is required")
+    for kb_id in ids:
+        kbs = KnowledgebaseService.accessible(kb_id=kb_id,user_id=tenant_id)
+        if not kbs:
+            return get_error_data_result(f"You don't own the dataset {kb_id}")
+        kbs = KnowledgebaseService.query(id=kb_id)
+        kb = kbs[0]
+        if kb.chunk_num == 0:
+            return get_error_data_result(f"The dataset {kb_id} doesn't own parsed file")
+    kbs = KnowledgebaseService.get_by_ids(ids)
+    embd_count = list(set([kb.embd_id for kb in kbs]))
+    if len(embd_count) != 1:
+        return get_result(message='Datasets use different embedding models."',code=settings.RetCode.AUTHENTICATION_ERROR)
+    req["kb_ids"] = ids
+    # llm
+    llm = req.get("llm")
+    if llm:
+        if "model_name" in llm:
+            req["llm_id"] = llm.pop("model_name")
+            if not TenantLLMService.query(tenant_id=tenant_id,llm_name=req["llm_id"],model_type="chat"):
+                return get_error_data_result(f"`model_name` {req.get('llm_id')} doesn't exist")
+        req["llm_setting"] = req.pop("llm")
+    e, tenant = TenantService.get_by_id(tenant_id)
+    if not e:
+        return get_error_data_result(message="Tenant not found!")
+    # prompt
+    prompt = req.get("prompt")
+    key_mapping = {"parameters": "variables",
+                   "prologue": "opener",
+                   "quote": "show_quote",
+                   "system": "prompt",
+                   "rerank_id": "rerank_model",
+                   "vector_similarity_weight": "keywords_similarity_weight"}
+    key_list = ["similarity_threshold", "vector_similarity_weight", "top_n", "rerank_id"]
+    if prompt:
+        for new_key, old_key in key_mapping.items():
+            if old_key in prompt:
+                prompt[new_key] = prompt.pop(old_key)
+        for key in key_list:
+            if key in prompt:
+                req[key] = prompt.pop(key)
+        req["prompt_config"] = req.pop("prompt")
+    # init
+    req["id"] = get_uuid()
+    req["description"] = req.get("description", "A helpful Assistant")
+    req["icon"] = req.get("avatar", "")
+    req["top_n"] = req.get("top_n", 6)
+    req["top_k"] = req.get("top_k", 1024)
+    req["rerank_id"] = req.get("rerank_id", "")
+    if req.get("rerank_id"):
+        if not TenantLLMService.query(tenant_id=tenant_id,llm_name=req.get("rerank_id"),model_type="rerank"):
+            return get_error_data_result(f"`rerank_model` {req.get('rerank_id')} doesn't exist")
+    if not req.get("llm_id"):
+        req["llm_id"] = tenant.llm_id
+    if not req.get("name"):
+        return get_error_data_result(message="`name` is required.")
+    if DialogService.query(name=req["name"], tenant_id=tenant_id, status=StatusEnum.VALID.value):
+        return get_error_data_result(message="Duplicated chat name in creating chat.")
+    # tenant_id
+    if req.get("tenant_id"):
+        return get_error_data_result(message="`tenant_id` must not be provided.")
+    req["tenant_id"] = tenant_id
+    # prompt more parameter
+    default_prompt = {
+        "system": """You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence "The answer you are looking for is not found in the knowledge base!" Answers need to consider chat history.
+      Here is the knowledge base:
+      {knowledge}
+      The above is the knowledge base.""",
+        "prologue": "Hi! I'm your assistant, what can I do for you?",
+        "parameters": [
+            {"key": "knowledge", "optional": False}
+        ],
+        "empty_response": "Sorry! No relevant content was found in the knowledge base!"
+    }
+    key_list_2 = ["system", "prologue", "parameters", "empty_response"]
+    if "prompt_config" not in req:
+        req['prompt_config'] = {}
+    for key in key_list_2:
+        temp = req['prompt_config'].get(key)
+        if (not temp and key == 'system') or (key not in req["prompt_config"]):
+            req['prompt_config'][key] = default_prompt[key]
+    for p in req['prompt_config']["parameters"]:
+        if p["optional"]:
+            continue
+        if req['prompt_config']["system"].find("{%s}" % p["key"]) < 0:
+            return get_error_data_result(
+                message="Parameter '{}' is not used".format(p["key"]))
+    # save
+    if not DialogService.save(**req):
+        return get_error_data_result(message="Fail to new a chat!")
+    # response
+    e, res = DialogService.get_by_id(req["id"])
+    if not e:
+        return get_error_data_result(message="Fail to new a chat!")
+    res = res.to_json()
+    renamed_dict = {}
+    for key, value in res["prompt_config"].items():
+        new_key = key_mapping.get(key, key)
+        renamed_dict[new_key] = value
+    res["prompt"] = renamed_dict
+    del res["prompt_config"]
+    new_dict = {"similarity_threshold": res["similarity_threshold"],
+                "keywords_similarity_weight": res["vector_similarity_weight"],
+                "top_n": res["top_n"],
+                "rerank_model": res['rerank_id']}
+    res["prompt"].update(new_dict)
+    for key in key_list:
+        del res[key]
+    res["llm"] = res.pop("llm_setting")
+    res["llm"]["model_name"] = res.pop("llm_id")
+    del res["kb_ids"]
+    res["dataset_ids"] = req["dataset_ids"]
+    res["avatar"] = res.pop("icon")
+    return get_result(data=res)
+
+@manager.route('/chats/<chat_id>', methods=['PUT'])
+@token_required
+def update(tenant_id,chat_id):
+    if not DialogService.query(tenant_id=tenant_id, id=chat_id, status=StatusEnum.VALID.value):
+        return get_error_data_result(message='You do not own the chat')
+    req =request.json
+    ids = req.get("dataset_ids")
+    if "show_quotation" in req:
+        req["do_refer"]=req.pop("show_quotation")
+    if "dataset_ids" in req:
+        if not ids:
+            return get_error_data_result("`datasets` can't be empty")
+        if ids:
+            for kb_id in ids:
+                kbs = KnowledgebaseService.accessible(kb_id=chat_id, user_id=tenant_id)
+                if not kbs:
+                    return get_error_data_result(f"You don't own the dataset {kb_id}")
+                kbs = KnowledgebaseService.query(id=kb_id)
+                kb = kbs[0]
+                if kb.chunk_num == 0:
+                    return get_error_data_result(f"The dataset {kb_id} doesn't own parsed file")
+            kbs = KnowledgebaseService.get_by_ids(ids)
+            embd_count=list(set([kb.embd_id for kb in kbs]))
+            if len(embd_count) != 1 :
+                return get_result(
+                    message='Datasets use different embedding models."',
+                    code=settings.RetCode.AUTHENTICATION_ERROR)
+            req["kb_ids"] = ids
+    llm = req.get("llm")
+    if llm:
+        if "model_name" in llm:
+            req["llm_id"] = llm.pop("model_name")
+            if not TenantLLMService.query(tenant_id=tenant_id,llm_name=req["llm_id"],model_type="chat"):
+                return get_error_data_result(f"`model_name` {req.get('llm_id')} doesn't exist")
+        req["llm_setting"] = req.pop("llm")
+    e, tenant = TenantService.get_by_id(tenant_id)
+    if not e:
+        return get_error_data_result(message="Tenant not found!")
+    if req.get("rerank_model"):
+        if not TenantLLMService.query(tenant_id=tenant_id,llm_name=req.get("rerank_model"),model_type="rerank"):
+            return get_error_data_result(f"`rerank_model` {req.get('rerank_model')} doesn't exist")
+    # prompt
+    prompt = req.get("prompt")
+    key_mapping = {"parameters": "variables",
+                   "prologue": "opener",
+                   "quote": "show_quote",
+                   "system": "prompt",
+                   "rerank_id": "rerank_model",
+                   "vector_similarity_weight": "keywords_similarity_weight"}
+    key_list = ["similarity_threshold", "vector_similarity_weight", "top_n", "rerank_id"]
+    if prompt:
+        for new_key, old_key in key_mapping.items():
+            if old_key in prompt:
+                prompt[new_key] = prompt.pop(old_key)
+        for key in key_list:
+            if key in prompt:
+                req[key] = prompt.pop(key)
+        req["prompt_config"] = req.pop("prompt")
+    e, res = DialogService.get_by_id(chat_id)
+    res = res.to_json()
+    if "name" in req:
+        if not req.get("name"):
+            return get_error_data_result(message="`name` is not empty.")
+        if req["name"].lower() != res["name"].lower() \
+                and len(
+            DialogService.query(name=req["name"], tenant_id=tenant_id, status=StatusEnum.VALID.value)) > 0:
+            return get_error_data_result(message="Duplicated chat name in updating dataset.")
+    if "prompt_config" in req:
+        res["prompt_config"].update(req["prompt_config"])
+        for p in res["prompt_config"]["parameters"]:
+            if p["optional"]:
+                continue
+            if res["prompt_config"]["system"].find("{%s}" % p["key"]) < 0:
+                return get_error_data_result(message="Parameter '{}' is not used".format(p["key"]))
+    if "llm_setting" in req:
+        res["llm_setting"].update(req["llm_setting"])
+    req["prompt_config"] = res["prompt_config"]
+    req["llm_setting"] = res["llm_setting"]
+    # avatar
+    if "avatar" in req:
+        req["icon"] = req.pop("avatar")
+    if "dataset_ids" in req:
+        req.pop("dataset_ids")
+    if not DialogService.update_by_id(chat_id, req):
+        return get_error_data_result(message="Chat not found!")
+    return get_result()
+
+
+@manager.route('/chats', methods=['DELETE'])
+@token_required
+def delete(tenant_id):
+    req = request.json
+    if not req:
+        ids=None
+    else:
+        ids=req.get("ids")
+    if not ids:
+        id_list = []
+        dias=DialogService.query(tenant_id=tenant_id,status=StatusEnum.VALID.value)
+        for dia in dias:
+            id_list.append(dia.id)
+    else:
+        id_list=ids
+    for id in id_list:
+        if not DialogService.query(tenant_id=tenant_id, id=id, status=StatusEnum.VALID.value):
+            return get_error_data_result(message=f"You don't own the chat {id}")
+        temp_dict = {"status": StatusEnum.INVALID.value}
+        DialogService.update_by_id(id, temp_dict)
+    return get_result()
+
+@manager.route('/chats', methods=['GET'])
+@token_required
+def list_chat(tenant_id):
+    id = request.args.get("id")
+    name = request.args.get("name")
+    chat = DialogService.query(id=id,name=name,status=StatusEnum.VALID.value,tenant_id=tenant_id)
+    if not chat:
+        return get_error_data_result(message="The chat doesn't exist")
+    page_number = int(request.args.get("page", 1))
+    items_per_page = int(request.args.get("page_size", 30))
+    orderby = request.args.get("orderby", "create_time")
+    if request.args.get("desc") == "False" or request.args.get("desc") == "false":
+        desc = False
+    else:
+        desc = True
+    chats = DialogService.get_list(tenant_id,page_number,items_per_page,orderby,desc,id,name)
+    if not chats:
+        return get_result(data=[])
+    list_assts = []
+    renamed_dict = {}
+    key_mapping = {"parameters": "variables",
+                   "prologue": "opener",
+                   "quote": "show_quote",
+                   "system": "prompt",
+                   "rerank_id": "rerank_model",
+                   "vector_similarity_weight": "keywords_similarity_weight",
+                   "do_refer":"show_quotation"}
+    key_list = ["similarity_threshold", "vector_similarity_weight", "top_n", "rerank_id"]
+    for res in chats:
+        for key, value in res["prompt_config"].items():
+            new_key = key_mapping.get(key, key)
+            renamed_dict[new_key] = value
+        res["prompt"] = renamed_dict
+        del res["prompt_config"]
+        new_dict = {"similarity_threshold": res["similarity_threshold"],
+                    "keywords_similarity_weight": res["vector_similarity_weight"],
+                    "top_n": res["top_n"],
+                    "rerank_model": res['rerank_id']}
+        res["prompt"].update(new_dict)
+        for key in key_list:
+            del res[key]
+        res["llm"] = res.pop("llm_setting")
+        res["llm"]["model_name"] = res.pop("llm_id")
+        kb_list = []
+        for kb_id in res["kb_ids"]:
+            kb = KnowledgebaseService.query(id=kb_id)
+            if not kb :
+                return get_error_data_result(message=f"Don't exist the kb {kb_id}")
+            kb_list.append(kb[0].to_json())
+        del res["kb_ids"]
+        res["datasets"] = kb_list
+        res["avatar"] = res.pop("icon")
+        list_assts.append(res)
+    return get_result(data=list_assts)

api/apps/sdk/dataset.py

@@ -0,0 +1,531 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+
+from flask import request
+from api.db import StatusEnum, FileSource
+from api.db.db_models import File
+from api.db.services.document_service import DocumentService
+from api.db.services.file2document_service import File2DocumentService
+from api.db.services.file_service import FileService
+from api.db.services.knowledgebase_service import KnowledgebaseService
+from api.db.services.llm_service import TenantLLMService, LLMService
+from api.db.services.user_service import TenantService
+from api import settings
+from api.utils import get_uuid
+from api.utils.api_utils import (
+    get_result,
+    token_required,
+    get_error_data_result,
+    valid,
+    get_parser_config,
+)
+
+
+@manager.route("/datasets", methods=["POST"])
+@token_required
+def create(tenant_id):
+    """
+    Create a new dataset.
+    ---
+    tags:
+      - Datasets
+    security:
+      - ApiKeyAuth: []
+    parameters:
+      - in: header
+        name: Authorization
+        type: string
+        required: true
+        description: Bearer token for authentication.
+      - in: body
+        name: body
+        description: Dataset creation parameters.
+        required: true
+        schema:
+          type: object
+          required:
+            - name
+          properties:
+            name:
+              type: string
+              description: Name of the dataset.
+            permission:
+              type: string
+              enum: ['me', 'team']
+              description: Dataset permission.
+            language:
+              type: string
+              enum: ['Chinese', 'English']
+              description: Language of the dataset.
+            chunk_method:
+              type: string
+              enum: ["naive", "manual", "qa", "table", "paper", "book", "laws",
+                     "presentation", "picture", "one", "knowledge_graph", "email"]
+              description: Chunking method.
+            parser_config:
+              type: object
+              description: Parser configuration.
+    responses:
+      200:
+        description: Successful operation.
+        schema:
+          type: object
+          properties:
+            data:
+              type: object
+    """
+    req = request.json
+    e, t = TenantService.get_by_id(tenant_id)
+    permission = req.get("permission")
+    language = req.get("language")
+    chunk_method = req.get("chunk_method")
+    parser_config = req.get("parser_config")
+    valid_permission = ["me", "team"]
+    valid_language = ["Chinese", "English"]
+    valid_chunk_method = [
+        "naive",
+        "manual",
+        "qa",
+        "table",
+        "paper",
+        "book",
+        "laws",
+        "presentation",
+        "picture",
+        "one",
+        "knowledge_graph",
+        "email",
+    ]
+    check_validation = valid(
+        permission,
+        valid_permission,
+        language,
+        valid_language,
+        chunk_method,
+        valid_chunk_method,
+    )
+    if check_validation:
+        return check_validation
+    req["parser_config"] = get_parser_config(chunk_method, parser_config)
+    if "tenant_id" in req:
+        return get_error_data_result(message="`tenant_id` must not be provided")
+    if "chunk_count" in req or "document_count" in req:
+        return get_error_data_result(
+            message="`chunk_count` or `document_count` must not be provided"
+        )
+    if "name" not in req:
+        return get_error_data_result(message="`name` is not empty!")
+    req["id"] = get_uuid()
+    req["name"] = req["name"].strip()
+    if req["name"] == "":
+        return get_error_data_result(message="`name` is not empty string!")
+    if KnowledgebaseService.query(
+        name=req["name"], tenant_id=tenant_id, status=StatusEnum.VALID.value
+    ):
+        return get_error_data_result(
+            message="Duplicated dataset name in creating dataset."
+        )
+    req["tenant_id"] = req["created_by"] = tenant_id
+    if not req.get("embedding_model"):
+        req["embedding_model"] = t.embd_id
+    else:
+        valid_embedding_models = [
+            "BAAI/bge-large-zh-v1.5",
+            "BAAI/bge-base-en-v1.5",
+            "BAAI/bge-large-en-v1.5",
+            "BAAI/bge-small-en-v1.5",
+            "BAAI/bge-small-zh-v1.5",
+            "jinaai/jina-embeddings-v2-base-en",
+            "jinaai/jina-embeddings-v2-small-en",
+            "nomic-ai/nomic-embed-text-v1.5",
+            "sentence-transformers/all-MiniLM-L6-v2",
+            "text-embedding-v2",
+            "text-embedding-v3",
+            "maidalun1020/bce-embedding-base_v1",
+        ]
+        embd_model = LLMService.query(
+            llm_name=req["embedding_model"], model_type="embedding"
+        )
+        if embd_model:
+            if req["embedding_model"] not in valid_embedding_models and not TenantLLMService.query(tenant_id=tenant_id,model_type="embedding",llm_name=req.get("embedding_model"),):
+                return get_error_data_result(f"`embedding_model` {req.get('embedding_model')} doesn't exist")
+        if not embd_model:
+            embd_model=TenantLLMService.query(tenant_id=tenant_id,model_type="embedding", llm_name=req.get("embedding_model"))
+        if not embd_model:
+            return get_error_data_result(
+                f"`embedding_model` {req.get('embedding_model')} doesn't exist"
+            )
+    key_mapping = {
+        "chunk_num": "chunk_count",
+        "doc_num": "document_count",
+        "parser_id": "chunk_method",
+        "embd_id": "embedding_model",
+    }
+    mapped_keys = {
+        new_key: req[old_key]
+        for new_key, old_key in key_mapping.items()
+        if old_key in req
+    }
+    req.update(mapped_keys)
+    if not KnowledgebaseService.save(**req):
+        return get_error_data_result(message="Create dataset error.(Database error)")
+    renamed_data = {}
+    e, k = KnowledgebaseService.get_by_id(req["id"])
+    for key, value in k.to_dict().items():
+        new_key = key_mapping.get(key, key)
+        renamed_data[new_key] = value
+    return get_result(data=renamed_data)
+
+
+@manager.route("/datasets", methods=["DELETE"])
+@token_required
+def delete(tenant_id):
+    """
+    Delete datasets.
+    ---
+    tags:
+      - Datasets
+    security:
+      - ApiKeyAuth: []
+    parameters:
+      - in: header
+        name: Authorization
+        type: string
+        required: true
+        description: Bearer token for authentication.
+      - in: body
+        name: body
+        description: Dataset deletion parameters.
+        required: true
+        schema:
+          type: object
+          properties:
+            ids:
+              type: array
+              items:
+                type: string
+              description: List of dataset IDs to delete.
+    responses:
+      200:
+        description: Successful operation.
+        schema:
+          type: object
+    """
+    req = request.json
+    if not req:
+        ids = None
+    else:
+        ids = req.get("ids")
+    if not ids:
+        id_list = []
+        kbs = KnowledgebaseService.query(tenant_id=tenant_id)
+        for kb in kbs:
+            id_list.append(kb.id)
+    else:
+        id_list = ids
+    for id in id_list:
+        kbs = KnowledgebaseService.query(id=id, tenant_id=tenant_id)
+        if not kbs:
+            return get_error_data_result(message=f"You don't own the dataset {id}")
+        for doc in DocumentService.query(kb_id=id):
+            if not DocumentService.remove_document(doc, tenant_id):
+                return get_error_data_result(
+                    message="Remove document error.(Database error)"
+                )
+            f2d = File2DocumentService.get_by_document_id(doc.id)
+            FileService.filter_delete(
+                [
+                    File.source_type == FileSource.KNOWLEDGEBASE,
+                    File.id == f2d[0].file_id,
+                ]
+            )
+            FileService.filter_delete(
+                [File.source_type == FileSource.KNOWLEDGEBASE, File.type == "folder", File.name == kbs[0].name])
+            File2DocumentService.delete_by_document_id(doc.id)
+        if not KnowledgebaseService.delete_by_id(id):
+            return get_error_data_result(message="Delete dataset error.(Database error)")
+    return get_result(code=settings.RetCode.SUCCESS)
+
+
+@manager.route("/datasets/<dataset_id>", methods=["PUT"])
+@token_required
+def update(tenant_id, dataset_id):
+    """
+    Update a dataset.
+    ---
+    tags:
+      - Datasets
+    security:
+      - ApiKeyAuth: []
+    parameters:
+      - in: path
+        name: dataset_id
+        type: string
+        required: true
+        description: ID of the dataset to update.
+      - in: header
+        name: Authorization
+        type: string
+        required: true
+        description: Bearer token for authentication.
+      - in: body
+        name: body
+        description: Dataset update parameters.
+        required: true
+        schema:
+          type: object
+          properties:
+            name:
+              type: string
+              description: New name of the dataset.
+            permission:
+              type: string
+              enum: ['me', 'team']
+              description: Updated permission.
+            language:
+              type: string
+              enum: ['Chinese', 'English']
+              description: Updated language.
+            chunk_method:
+              type: string
+              enum: ["naive", "manual", "qa", "table", "paper", "book", "laws",
+                     "presentation", "picture", "one", "knowledge_graph", "email"]
+              description: Updated chunking method.
+            parser_config:
+              type: object
+              description: Updated parser configuration.
+    responses:
+      200:
+        description: Successful operation.
+        schema:
+          type: object
+    """
+    if not KnowledgebaseService.query(id=dataset_id, tenant_id=tenant_id):
+        return get_error_data_result(message="You don't own the dataset")
+    req = request.json
+    e, t = TenantService.get_by_id(tenant_id)
+    invalid_keys = {"id", "embd_id", "chunk_num", "doc_num", "parser_id"}
+    if any(key in req for key in invalid_keys):
+        return get_error_data_result(message="The input parameters are invalid.")
+    permission = req.get("permission")
+    language = req.get("language")
+    chunk_method = req.get("chunk_method")
+    parser_config = req.get("parser_config")
+    valid_permission = ["me", "team"]
+    valid_language = ["Chinese", "English"]
+    valid_chunk_method = [
+        "naive",
+        "manual",
+        "qa",
+        "table",
+        "paper",
+        "book",
+        "laws",
+        "presentation",
+        "picture",
+        "one",
+        "knowledge_graph",
+        "email",
+    ]
+    check_validation = valid(
+        permission,
+        valid_permission,
+        language,
+        valid_language,
+        chunk_method,
+        valid_chunk_method,
+    )
+    if check_validation:
+        return check_validation
+    if "tenant_id" in req:
+        if req["tenant_id"] != tenant_id:
+            return get_error_data_result(message="Can't change `tenant_id`.")
+    e, kb = KnowledgebaseService.get_by_id(dataset_id)
+    if "parser_config" in req:
+        temp_dict = kb.parser_config
+        temp_dict.update(req["parser_config"])
+        req["parser_config"] = temp_dict
+    if "chunk_count" in req:
+        if req["chunk_count"] != kb.chunk_num:
+            return get_error_data_result(message="Can't change `chunk_count`.")
+        req.pop("chunk_count")
+    if "document_count" in req:
+        if req["document_count"] != kb.doc_num:
+            return get_error_data_result(message="Can't change `document_count`.")
+        req.pop("document_count")
+    if "chunk_method" in req:
+        if kb.chunk_num != 0 and req["chunk_method"] != kb.parser_id:
+            return get_error_data_result(
+                message="If `chunk_count` is not 0, `chunk_method` is not changeable."
+            )
+        req["parser_id"] = req.pop("chunk_method")
+        if req["parser_id"] != kb.parser_id:
+            if not req.get("parser_config"):
+                req["parser_config"] = get_parser_config(chunk_method, parser_config)
+    if "embedding_model" in req:
+        if kb.chunk_num != 0 and req["embedding_model"] != kb.embd_id:
+            return get_error_data_result(
+                message="If `chunk_count` is not 0, `embedding_model` is not changeable."
+            )
+        if not req.get("embedding_model"):
+            return get_error_data_result("`embedding_model` can't be empty")
+        valid_embedding_models = [
+            "BAAI/bge-large-zh-v1.5",
+            "BAAI/bge-base-en-v1.5",
+            "BAAI/bge-large-en-v1.5",
+            "BAAI/bge-small-en-v1.5",
+            "BAAI/bge-small-zh-v1.5",
+            "jinaai/jina-embeddings-v2-base-en",
+            "jinaai/jina-embeddings-v2-small-en",
+            "nomic-ai/nomic-embed-text-v1.5",
+            "sentence-transformers/all-MiniLM-L6-v2",
+            "text-embedding-v2",
+            "text-embedding-v3",
+            "maidalun1020/bce-embedding-base_v1",
+        ]
+        embd_model = LLMService.query(
+            llm_name=req["embedding_model"], model_type="embedding"
+        )
+        if embd_model:
+            if req["embedding_model"] not in valid_embedding_models and not TenantLLMService.query(tenant_id=tenant_id,model_type="embedding",llm_name=req.get("embedding_model"),):
+                return get_error_data_result(f"`embedding_model` {req.get('embedding_model')} doesn't exist")
+        if not embd_model:
+            embd_model=TenantLLMService.query(tenant_id=tenant_id,model_type="embedding", llm_name=req.get("embedding_model"))
+
+        if not embd_model:
+            return get_error_data_result(
+                f"`embedding_model` {req.get('embedding_model')} doesn't exist"
+            )
+        req["embd_id"] = req.pop("embedding_model")
+    if "name" in req:
+        req["name"] = req["name"].strip()
+        if (
+            req["name"].lower() != kb.name.lower()
+            and len(
+                KnowledgebaseService.query(
+                    name=req["name"], tenant_id=tenant_id, status=StatusEnum.VALID.value
+                )
+            )
+            > 0
+        ):
+            return get_error_data_result(
+                message="Duplicated dataset name in updating dataset."
+            )
+    if not KnowledgebaseService.update_by_id(kb.id, req):
+        return get_error_data_result(message="Update dataset error.(Database error)")
+    return get_result(code=settings.RetCode.SUCCESS)
+
+
+@manager.route("/datasets", methods=["GET"])
+@token_required
+def list(tenant_id):
+    """
+    List datasets.
+    ---
+    tags:
+      - Datasets
+    security:
+      - ApiKeyAuth: []
+    parameters:
+      - in: query
+        name: id
+        type: string
+        required: false
+        description: Dataset ID to filter.
+      - in: query
+        name: name
+        type: string
+        required: false
+        description: Dataset name to filter.
+      - in: query
+        name: page
+        type: integer
+        required: false
+        default: 1
+        description: Page number.
+      - in: query
+        name: page_size
+        type: integer
+        required: false
+        default: 1024
+        description: Number of items per page.
+      - in: query
+        name: orderby
+        type: string
+        required: false
+        default: "create_time"
+        description: Field to order by.
+      - in: query
+        name: desc
+        type: boolean
+        required: false
+        default: true
+        description: Order in descending.
+      - in: header
+        name: Authorization
+        type: string
+        required: true
+        description: Bearer token for authentication.
+    responses:
+      200:
+        description: Successful operation.
+        schema:
+          type: array
+          items:
+            type: object
+    """
+    id = request.args.get("id")
+    name = request.args.get("name")
+    if id:
+        kbs = KnowledgebaseService.get_kb_by_id(id,tenant_id)
+        if not kbs:
+            return get_error_data_result(f"You don't own the dataset {id}")
+    if name:
+        kbs = KnowledgebaseService.get_kb_by_name(name,tenant_id)
+        if not kbs:
+            return get_error_data_result(f"You don't own the dataset {name}")
+    page_number = int(request.args.get("page", 1))
+    items_per_page = int(request.args.get("page_size", 30))
+    orderby = request.args.get("orderby", "create_time")
+    if request.args.get("desc") == "False" or request.args.get("desc") == "false":
+        desc = False
+    else:
+        desc = True
+    tenants = TenantService.get_joined_tenants_by_user_id(tenant_id)
+    kbs = KnowledgebaseService.get_list(
+        [m["tenant_id"] for m in tenants],
+        tenant_id,
+        page_number,
+        items_per_page,
+        orderby,
+        desc,
+        id,
+        name,
+    )
+    renamed_list = []
+    for kb in kbs:
+        key_mapping = {
+            "chunk_num": "chunk_count",
+            "doc_num": "document_count",
+            "parser_id": "chunk_method",
+            "embd_id": "embedding_model",
+        }
+        renamed_data = {}
+        for key, value in kb.items():
+            new_key = key_mapping.get(key, key)
+            renamed_data[new_key] = value
+        renamed_list.append(renamed_data)
+    return get_result(data=renamed_list)

api/apps/sdk/dify_retrieval.py

@@ -0,0 +1,76 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+from flask import request, jsonify
+
+from api.db import LLMType, ParserType
+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 validate_request, build_error_result, apikey_required
+
+
+@manager.route('/dify/retrieval', methods=['POST'])
+@apikey_required
+@validate_request("knowledge_id", "query")
+def retrieval(tenant_id):
+    req = request.json
+    question = req["query"]
+    kb_id = req["knowledge_id"]
+    retrieval_setting = req.get("retrieval_setting", {})
+    similarity_threshold = float(retrieval_setting.get("score_threshold", 0.0))
+    top = int(retrieval_setting.get("top_k", 1024))
+
+    try:
+
+        e, kb = KnowledgebaseService.get_by_id(kb_id)
+        if not e:
+            return build_error_result(message="Knowledgebase not found!", code=settings.RetCode.NOT_FOUND)
+
+        if kb.tenant_id != tenant_id:
+            return build_error_result(message="Knowledgebase not found!", code=settings.RetCode.NOT_FOUND)
+
+        embd_mdl = LLMBundle(kb.tenant_id, LLMType.EMBEDDING.value, llm_name=kb.embd_id)
+
+        retr = settings.retrievaler if kb.parser_id != ParserType.KG else settings.kg_retrievaler
+        ranks = retr.retrieval(
+            question,
+            embd_mdl,
+            kb.tenant_id,
+            [kb_id],
+            page=1,
+            page_size=top,
+            similarity_threshold=similarity_threshold,
+            vector_similarity_weight=0.3,
+            top=top
+        )
+        records = []
+        for c in ranks["chunks"]:
+            c.pop("vector", None)
+            records.append({
+                "content": c["content_ltks"],
+                "score": c["similarity"],
+                "title": c["docnm_kwd"],
+                "metadata": {}
+            })
+
+        return jsonify({"records": records})
+    except Exception as e:
+        if str(e).find("not_found") > 0:
+            return build_error_result(
+                message='No chunk found! Check the chunk status please!',
+                code=settings.RetCode.NOT_FOUND
+            )
+        return build_error_result(message=str(e), code=settings.RetCode.SERVER_ERROR)

api/apps/sdk/session.py

@@ -0,0 +1,531 @@
+#
+#  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 re
+import json
+from copy import deepcopy
+from uuid import uuid4
+from api.db import LLMType
+from flask import request, Response
+from api.db.services.dialog_service import ask
+from agent.canvas import Canvas
+from api.db import StatusEnum
+from api.db.db_models import API4Conversation
+from api.db.services.api_service import API4ConversationService
+from api.db.services.canvas_service import UserCanvasService
+from api.db.services.dialog_service import DialogService, ConversationService, chat
+from api.db.services.knowledgebase_service import KnowledgebaseService
+from api.utils import get_uuid
+from api.utils.api_utils import get_error_data_result
+from api.utils.api_utils import get_result, token_required
+from api.db.services.llm_service import LLMBundle
+
+
+@manager.route('/chats/<chat_id>/sessions', methods=['POST'])
+@token_required
+def create(tenant_id,chat_id):
+    req = request.json
+    req["dialog_id"] = chat_id
+    dia = DialogService.query(tenant_id=tenant_id, id=req["dialog_id"], status=StatusEnum.VALID.value)
+    if not dia:
+        return get_error_data_result(message="You do not own the assistant.")
+    conv = {
+        "id": get_uuid(),
+        "dialog_id": req["dialog_id"],
+        "name": req.get("name", "New session"),
+        "message": [{"role": "assistant", "content": "Hi! I am your assistant，can I help you?"}]
+    }
+    if not conv.get("name"):
+        return get_error_data_result(message="`name` can not be empty.")
+    ConversationService.save(**conv)
+    e, conv = ConversationService.get_by_id(conv["id"])
+    if not e:
+        return get_error_data_result(message="Fail to create a session!")
+    conv = conv.to_dict()
+    conv['messages'] = conv.pop("message")
+    conv["chat_id"] = conv.pop("dialog_id")
+    del conv["reference"]
+    return get_result(data=conv)
+
+
+@manager.route('/agents/<agent_id>/sessions', methods=['POST'])
+@token_required
+def create_agent_session(tenant_id, agent_id):
+    req = request.json
+    e, cvs = UserCanvasService.get_by_id(agent_id)
+    if not e:
+        return get_error_data_result("Agent not found.")
+    if cvs.user_id != tenant_id:
+        return get_error_data_result(message="You do not own the agent.")
+
+    if not isinstance(cvs.dsl, str):
+        cvs.dsl = json.dumps(cvs.dsl, ensure_ascii=False)
+
+    canvas = Canvas(cvs.dsl, tenant_id)
+    conv = {
+        "id": get_uuid(),
+        "dialog_id": cvs.id,
+        "user_id": req.get("usr_id","") if isinstance(req, dict) else "",
+        "message": [{"role": "assistant", "content": canvas.get_prologue()}],
+        "source": "agent"
+    }
+    API4ConversationService.save(**conv)
+    conv["agent_id"] = conv.pop("dialog_id")
+    return get_result(data=conv)
+
+
+@manager.route('/chats/<chat_id>/sessions/<session_id>', methods=['PUT'])
+@token_required
+def update(tenant_id,chat_id,session_id):
+    req = request.json
+    req["dialog_id"] = chat_id
+    conv_id = session_id
+    conv = ConversationService.query(id=conv_id,dialog_id=chat_id)
+    if not conv:
+        return get_error_data_result(message="Session does not exist")
+    if not DialogService.query(id=chat_id, tenant_id=tenant_id, status=StatusEnum.VALID.value):
+        return get_error_data_result(message="You do not own the session")
+    if "message" in req or "messages" in req:
+        return get_error_data_result(message="`message` can not be change")
+    if "reference" in req:
+        return get_error_data_result(message="`reference` can not be change")
+    if "name" in req and not req.get("name"):
+        return get_error_data_result(message="`name` can not be empty.")
+    if not ConversationService.update_by_id(conv_id, req):
+        return get_error_data_result(message="Session updates error")
+    return get_result()
+
+
+@manager.route('/chats/<chat_id>/completions', methods=['POST'])
+@token_required
+def completion(tenant_id, chat_id):
+    req = request.json
+    if not req.get("session_id"):
+        conv = {
+            "id": get_uuid(),
+            "dialog_id": chat_id,
+            "name": req.get("name", "New session"),
+            "message": [{"role": "assistant", "content": "Hi! I am your assistant，can I help you?"}]
+        }
+        if not conv.get("name"):
+            return get_error_data_result(message="`name` can not be empty.")
+        ConversationService.save(**conv)
+        e, conv = ConversationService.get_by_id(conv["id"])
+        session_id=conv.id
+    else:
+        session_id = req.get("session_id")
+    if not req.get("question"):
+        return get_error_data_result(message="Please input your question.")
+    conv = ConversationService.query(id=session_id,dialog_id=chat_id)
+    if not conv:
+        return get_error_data_result(message="Session does not exist")
+    conv = conv[0]
+    if not DialogService.query(id=chat_id, tenant_id=tenant_id, status=StatusEnum.VALID.value):
+        return get_error_data_result(message="You do not own the chat")
+    msg = []
+    question = {
+        "content": req.get("question"),
+        "role": "user",
+        "id": str(uuid4())
+    }
+    conv.message.append(question)
+    for m in conv.message:
+        if m["role"] == "system": continue
+        if m["role"] == "assistant" and not msg: continue
+        msg.append(m)
+    message_id = msg[-1].get("id")
+    e, dia = DialogService.get_by_id(conv.dialog_id)
+
+    if not conv.reference:
+        conv.reference = []
+    conv.message.append({"role": "assistant", "content": "", "id": message_id})
+    conv.reference.append({"chunks": [], "doc_aggs": []})
+
+    def fillin_conv(ans):
+        reference = ans["reference"]
+        temp_reference = deepcopy(ans["reference"])
+        nonlocal conv, message_id
+        if not conv.reference:
+            conv.reference.append(temp_reference)
+        else:
+            conv.reference[-1] = temp_reference
+        conv.message[-1] = {"role": "assistant", "content": ans["answer"],
+                            "id": message_id, "prompt": ans.get("prompt", "")}
+        if "chunks" in reference:
+            chunks = reference.get("chunks")
+            chunk_list = []
+            for chunk in chunks:
+                new_chunk = {
+                    "id": chunk["chunk_id"],
+                    "content": chunk["content_with_weight"],
+                    "document_id": chunk["doc_id"],
+                    "document_name": chunk["docnm_kwd"],
+                    "dataset_id": chunk["kb_id"],
+                    "image_id": chunk.get("image_id", ""),
+                    "similarity": chunk["similarity"],
+                    "vector_similarity": chunk["vector_similarity"],
+                    "term_similarity": chunk["term_similarity"],
+                    "positions": chunk.get("positions", []),
+                }
+                chunk_list.append(new_chunk)
+            reference["chunks"] = chunk_list
+        ans["id"] = message_id
+        ans["session_id"]=session_id
+
+    def stream():
+        nonlocal dia, msg, req, conv
+        try:
+            for ans in chat(dia, msg, **req):
+                fillin_conv(ans)
+                yield "data:" + json.dumps({"code": 0,  "data": ans}, ensure_ascii=False) + "\n\n"
+            ConversationService.update_by_id(conv.id, conv.to_dict())
+        except Exception as e:
+            yield "data:" + json.dumps({"code": 500, "message": str(e),
+                                        "data": {"answer": "**ERROR**: " + str(e),"reference": []}},
+                                       ensure_ascii=False) + "\n\n"
+        yield "data:" + json.dumps({"code": 0, "data": True}, ensure_ascii=False) + "\n\n"
+
+    if req.get("stream", True):
+        resp = Response(stream(), mimetype="text/event-stream")
+        resp.headers.add_header("Cache-control", "no-cache")
+        resp.headers.add_header("Connection", "keep-alive")
+        resp.headers.add_header("X-Accel-Buffering", "no")
+        resp.headers.add_header("Content-Type", "text/event-stream; charset=utf-8")
+
+        return resp
+
+    else:
+        answer = None
+        for ans in chat(dia, msg, **req):
+            answer = ans
+            fillin_conv(ans)
+            ConversationService.update_by_id(conv.id, conv.to_dict())
+            break
+        return get_result(data=answer)
+
+
+@manager.route('/agents/<agent_id>/completions', methods=['POST'])
+@token_required
+def agent_completion(tenant_id, agent_id):
+    req = request.json
+
+    e, cvs = UserCanvasService.get_by_id(agent_id)
+    if not e:
+        return get_error_data_result("Agent not found.")
+    if cvs.user_id != tenant_id:
+        return get_error_data_result(message="You do not own the agent.")
+    if not isinstance(cvs.dsl, str):
+        cvs.dsl = json.dumps(cvs.dsl, ensure_ascii=False)
+    canvas = Canvas(cvs.dsl, tenant_id)
+
+    if not req.get("session_id"):
+        session_id = get_uuid()
+        conv = {
+            "id": session_id,
+            "dialog_id": cvs.id,
+            "user_id": req.get("user_id",""),
+            "message": [{"role": "assistant", "content": canvas.get_prologue()}],
+            "source": "agent"
+        }
+        API4ConversationService.save(**conv)
+        conv = API4Conversation(**conv)
+    else:
+        session_id = req.get("session_id")
+        e, conv = API4ConversationService.get_by_id(req["session_id"])
+        if not e:
+            return get_error_data_result(message="Session not found!")
+
+    messages = conv.message
+    question = req.get("question")
+    if not question:
+        return get_error_data_result("`question` is required.")
+    question={
+        "role":"user",
+        "content":question,
+        "id": str(uuid4())
+    }
+    messages.append(question)
+    msg = []
+    for m in messages:
+        if m["role"] == "system":
+            continue
+        if m["role"] == "assistant" and not msg:
+            continue
+        msg.append(m)
+    if not msg[-1].get("id"): msg[-1]["id"] = get_uuid()
+    message_id = msg[-1]["id"]
+
+    if "quote" not in req: req["quote"] = False
+    stream = req.get("stream", True)
+
+    def fillin_conv(ans):
+        reference = ans["reference"]
+        temp_reference = deepcopy(ans["reference"])
+        nonlocal conv, message_id
+        if not conv.reference:
+            conv.reference.append(temp_reference)
+        else:
+            conv.reference[-1] = temp_reference
+        conv.message[-1] = {"role": "assistant", "content": ans["answer"], "id": message_id}
+        if "chunks" in reference:
+            chunks = reference.get("chunks")
+            chunk_list = []
+            for chunk in chunks:
+                new_chunk = {
+                    "id": chunk["chunk_id"],
+                    "content": chunk["content_with_weight"],
+                    "document_id": chunk["doc_id"],
+                    "document_name": chunk["docnm_kwd"],
+                    "dataset_id": chunk["kb_id"],
+                    "image_id": chunk["image_id"],
+                    "similarity": chunk["similarity"],
+                    "vector_similarity": chunk["vector_similarity"],
+                    "term_similarity": chunk["term_similarity"],
+                    "positions": chunk["positions"],
+                }
+                chunk_list.append(new_chunk)
+            reference["chunks"] = chunk_list
+        ans["id"] = message_id
+        ans["session_id"] = session_id
+
+    def rename_field(ans):
+        reference = ans['reference']
+        if not isinstance(reference, dict):
+            return
+        for chunk_i in reference.get('chunks', []):
+            if 'docnm_kwd' in chunk_i:
+                chunk_i['doc_name'] = chunk_i['docnm_kwd']
+                chunk_i.pop('docnm_kwd')
+    conv.message.append(msg[-1])
+
+    if not conv.reference:
+        conv.reference = []
+    conv.message.append({"role": "assistant", "content": "", "id": message_id})
+    conv.reference.append({"chunks": [], "doc_aggs": []})
+
+    final_ans = {"reference": [], "content": ""}
+
+    canvas.messages.append(msg[-1])
+    canvas.add_user_input(msg[-1]["content"])
+
+    if stream:
+        def sse():
+            nonlocal answer, cvs
+            try:
+                for ans in canvas.run(stream=True):
+                    if ans.get("running_status"):
+                        yield "data:" + json.dumps({"code": 0, "message": "",
+                                                    "data": {"answer": ans["content"],
+                                                             "running_status": True}},
+                                                   ensure_ascii=False) + "\n\n"
+                        continue
+                    for k in ans.keys():
+                        final_ans[k] = ans[k]
+                    ans = {"answer": ans["content"], "reference": ans.get("reference", [])}
+                    fillin_conv(ans)
+                    rename_field(ans)
+                    yield "data:" + json.dumps({"code": 0, "message": "", "data": ans},
+                                               ensure_ascii=False) + "\n\n"
+
+                canvas.messages.append({"role": "assistant", "content": final_ans["content"], "id": message_id})
+                canvas.history.append(("assistant", final_ans["content"]))
+                if final_ans.get("reference"):
+                    canvas.reference.append(final_ans["reference"])
+                cvs.dsl = json.loads(str(canvas))
+                API4ConversationService.append_message(conv.id, conv.to_dict())
+            except Exception as e:
+                cvs.dsl = json.loads(str(canvas))
+                API4ConversationService.append_message(conv.id, conv.to_dict())
+                yield "data:" + json.dumps({"code": 500, "message": str(e),
+                                            "data": {"answer": "**ERROR**: " + str(e), "reference": []}},
+                                           ensure_ascii=False) + "\n\n"
+            yield "data:" + json.dumps({"code": 0, "message": "", "data": True}, ensure_ascii=False) + "\n\n"
+
+        resp = Response(sse(), mimetype="text/event-stream")
+        resp.headers.add_header("Cache-control", "no-cache")
+        resp.headers.add_header("Connection", "keep-alive")
+        resp.headers.add_header("X-Accel-Buffering", "no")
+        resp.headers.add_header("Content-Type", "text/event-stream; charset=utf-8")
+        return resp
+
+    for answer in canvas.run(stream=False):
+        if answer.get("running_status"): continue
+        final_ans["content"] = "\n".join(answer["content"]) if "content" in answer else ""
+        canvas.messages.append({"role": "assistant", "content": final_ans["content"], "id": message_id})
+        if final_ans.get("reference"):
+            canvas.reference.append(final_ans["reference"])
+        cvs.dsl = json.loads(str(canvas))
+
+        result = {"answer": final_ans["content"], "reference": final_ans.get("reference", [])}
+        fillin_conv(result)
+        API4ConversationService.append_message(conv.id, conv.to_dict())
+        rename_field(result)
+        return get_result(data=result)
+
+
+@manager.route('/chats/<chat_id>/sessions', methods=['GET'])
+@token_required
+def list_session(chat_id,tenant_id):
+    if not DialogService.query(tenant_id=tenant_id, id=chat_id, status=StatusEnum.VALID.value):
+        return get_error_data_result(message=f"You don't own the assistant {chat_id}.")
+    id = request.args.get("id")
+    name = request.args.get("name")
+    page_number = int(request.args.get("page", 1))
+    items_per_page = int(request.args.get("page_size", 30))
+    orderby = request.args.get("orderby", "create_time")
+    if request.args.get("desc") == "False" or request.args.get("desc") == "false":
+        desc = False
+    else:
+        desc = True
+    convs = ConversationService.get_list(chat_id,page_number,items_per_page,orderby,desc,id,name)
+    if not convs:
+        return get_result(data=[])
+    for conv in convs:
+        conv['messages'] = conv.pop("message")
+        infos = conv["messages"]
+        for info in infos:
+            if "prompt" in info:
+                info.pop("prompt")
+        conv["chat_id"] = conv.pop("dialog_id")
+        if conv["reference"]:
+            messages = conv["messages"]
+            message_num = 0
+            chunk_num = 0
+            while message_num < len(messages):
+                if message_num != 0 and messages[message_num]["role"] != "user":
+                    chunk_list = []
+                    if "chunks" in conv["reference"][chunk_num]:
+                        chunks = conv["reference"][chunk_num]["chunks"]
+                        for chunk in chunks:
+                            new_chunk = {
+                                "id": chunk["chunk_id"],
+                                "content": chunk["content_with_weight"],
+                                "document_id": chunk["doc_id"],
+                                "document_name": chunk["docnm_kwd"],
+                                "dataset_id": chunk["kb_id"],
+                                "image_id": chunk["image_id"],
+                                "similarity": chunk["similarity"],
+                                "vector_similarity": chunk["vector_similarity"],
+                                "term_similarity": chunk["term_similarity"],
+                                "positions": chunk["positions"],
+                            }
+                            chunk_list.append(new_chunk)
+                    chunk_num += 1
+                    messages[message_num]["reference"] = chunk_list
+                message_num += 1
+        del conv["reference"]
+    return get_result(data=convs)
+
+
+@manager.route('/chats/<chat_id>/sessions', methods=["DELETE"])
+@token_required
+def delete(tenant_id,chat_id):
+    if not DialogService.query(id=chat_id, tenant_id=tenant_id, status=StatusEnum.VALID.value):
+        return get_error_data_result(message="You don't own the chat")
+    req = request.json
+    convs = ConversationService.query(dialog_id=chat_id)
+    if not req:
+        ids = None
+    else:
+        ids=req.get("ids")
+
+    if not ids:
+        conv_list = []
+        for conv in convs:
+            conv_list.append(conv.id)
+    else:
+        conv_list=ids
+    for id in conv_list:
+        conv = ConversationService.query(id=id,dialog_id=chat_id)
+        if not conv:
+            return get_error_data_result(message="The chat doesn't own the session")
+        ConversationService.delete_by_id(id)
+    return get_result()
+
+@manager.route('/sessions/ask', methods=['POST'])
+@token_required
+def ask_about(tenant_id):
+    req = request.json
+    if not req.get("question"):
+        return get_error_data_result("`question` is required.")
+    if not req.get("dataset_ids"):
+        return get_error_data_result("`dataset_ids` is required.")
+    if not isinstance(req.get("dataset_ids"),list):
+        return get_error_data_result("`dataset_ids` should be a list.")
+    req["kb_ids"]=req.pop("dataset_ids")
+    for kb_id in req["kb_ids"]:
+        if not KnowledgebaseService.accessible(kb_id,tenant_id):
+            return get_error_data_result(f"You don't own the dataset {kb_id}.")
+        kbs = KnowledgebaseService.query(id=kb_id)
+        kb = kbs[0]
+        if kb.chunk_num == 0:
+            return get_error_data_result(f"The dataset {kb_id} doesn't own parsed file")
+    uid = tenant_id
+    def stream():
+        nonlocal req, uid
+        try:
+            for ans in ask(req["question"], req["kb_ids"], uid):
+                yield "data:" + json.dumps({"code": 0, "message": "", "data": ans}, ensure_ascii=False) + "\n\n"
+        except Exception as e:
+            yield "data:" + json.dumps({"code": 500, "message": str(e),
+                                        "data": {"answer": "**ERROR**: " + str(e), "reference": []}},
+                                       ensure_ascii=False) + "\n\n"
+        yield "data:" + json.dumps({"code": 0, "message": "", "data": True}, ensure_ascii=False) + "\n\n"
+
+    resp = Response(stream(), mimetype="text/event-stream")
+    resp.headers.add_header("Cache-control", "no-cache")
+    resp.headers.add_header("Connection", "keep-alive")
+    resp.headers.add_header("X-Accel-Buffering", "no")
+    resp.headers.add_header("Content-Type", "text/event-stream; charset=utf-8")
+    return resp
+
+
+@manager.route('/sessions/related_questions', methods=['POST'])
+@token_required
+def related_questions(tenant_id):
+    req = request.json
+    if not req.get("question"):
+        return get_error_data_result("`question` is required.")
+    question = req["question"]
+    chat_mdl = LLMBundle(tenant_id, LLMType.CHAT)
+    prompt = """
+Objective: To generate search terms related to the user's search keywords, helping users find more valuable information.
+Instructions:
+ - Based on the keywords provided by the user, generate 5-10 related search terms.
+ - Each search term should be directly or indirectly related to the keyword, guiding the user to find more valuable information.
+ - Use common, general terms as much as possible, avoiding obscure words or technical jargon.
+ - Keep the term length between 2-4 words, concise and clear.
+ - DO NOT translate, use the language of the original keywords.
+
+### Example:
+Keywords: Chinese football
+Related search terms:
+1. Current status of Chinese football
+2. Reform of Chinese football
+3. Youth training of Chinese football
+4. Chinese football in the Asian Cup
+5. Chinese football in the World Cup
+
+Reason:
+ - When searching, users often only use one or two keywords, making it difficult to fully express their information needs.
+ - Generating related search terms can help users dig deeper into relevant information and improve search efficiency. 
+ - At the same time, related terms can also help search engines better understand user needs and return more accurate search results.
+
+"""
+    ans = chat_mdl.chat(prompt, [{"role": "user", "content": f"""
+Keywords: {question}
+Related search terms:
+    """}], {"temperature": 0.9})
+    return get_result(data=[re.sub(r"^[0-9]\. ", "", a) for a in ans.split("\n") if re.match(r"^[0-9]\. ", a)])

api/apps/system_app.py

@@ -13,77 +13,281 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License
 #
+import logging
+from datetime import datetime
 import json
 
-from flask_login import login_required
+from flask_login import login_required, current_user
 
+from api.db.db_models import APIToken
+from api.db.services.api_service import APITokenService
 from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.utils.api_utils import get_json_result
-from api.versions import get_rag_version
-from rag.settings import SVR_QUEUE_NAME
-from rag.utils.es_conn import ELASTICSEARCH
-from rag.utils.minio_conn import MINIO
+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,
+    server_error_response,
+    generate_confirmation_token,
+)
+from api.versions import get_ragflow_version
+from rag.utils.storage_factory import STORAGE_IMPL, STORAGE_IMPL_TYPE
 from timeit import default_timer as timer
 
 from rag.utils.redis_conn import REDIS_CONN
 
 
-@manager.route('/version', methods=['GET'])
+@manager.route("/version", methods=["GET"])
 @login_required
 def version():
-    return get_json_result(data=get_rag_version())
+    """
+    Get the current version of the application.
+    ---
+    tags:
+      - System
+    security:
+      - ApiKeyAuth: []
+    responses:
+      200:
+        description: Version retrieved successfully.
+        schema:
+          type: object
+          properties:
+            version:
+              type: string
+              description: Version number.
+    """
+    return get_json_result(data=get_ragflow_version())
 
 
-@manager.route('/status', methods=['GET'])
+@manager.route("/status", methods=["GET"])
 @login_required
 def status():
+    """
+    Get the system status.
+    ---
+    tags:
+      - System
+    security:
+      - ApiKeyAuth: []
+    responses:
+      200:
+        description: System is operational.
+        schema:
+          type: object
+          properties:
+            es:
+              type: object
+              description: Elasticsearch status.
+            storage:
+              type: object
+              description: Storage status.
+            database:
+              type: object
+              description: Database status.
+      503:
+        description: Service unavailable.
+        schema:
+          type: object
+          properties:
+            error:
+              type: string
+              description: Error message.
+    """
     res = {}
     st = timer()
     try:
-        res["es"] = ELASTICSEARCH.health()
-        res["es"]["elapsed"] = "{:.1f}".format((timer() - st)*1000.)
+        res["doc_engine"] = settings.docStoreConn.health()
+        res["doc_engine"]["elapsed"] = "{:.1f}".format((timer() - st) * 1000.0)
     except Exception as e:
-        res["es"] = {"status": "red", "elapsed": "{:.1f}".format((timer() - st)*1000.), "error": str(e)}
+        res["doc_engine"] = {
+            "type": "unknown",
+            "status": "red",
+            "elapsed": "{:.1f}".format((timer() - st) * 1000.0),
+            "error": str(e),
+        }
 
     st = timer()
     try:
-        MINIO.health()
-        res["minio"] = {"status": "green", "elapsed": "{:.1f}".format((timer() - st)*1000.)}
+        STORAGE_IMPL.health()
+        res["storage"] = {
+            "storage": STORAGE_IMPL_TYPE.lower(),
+            "status": "green",
+            "elapsed": "{:.1f}".format((timer() - st) * 1000.0),
+        }
     except Exception as e:
-        res["minio"] = {"status": "red", "elapsed": "{:.1f}".format((timer() - st)*1000.), "error": str(e)}
+        res["storage"] = {
+            "storage": STORAGE_IMPL_TYPE.lower(),
+            "status": "red",
+            "elapsed": "{:.1f}".format((timer() - st) * 1000.0),
+            "error": str(e),
+        }
 
     st = timer()
     try:
         KnowledgebaseService.get_by_id("x")
-        res["mysql"] = {"status": "green", "elapsed": "{:.1f}".format((timer() - st)*1000.)}
+        res["database"] = {
+            "database": settings.DATABASE_TYPE.lower(),
+            "status": "green",
+            "elapsed": "{:.1f}".format((timer() - st) * 1000.0),
+        }
     except Exception as e:
-        res["mysql"] = {"status": "red", "elapsed": "{:.1f}".format((timer() - st)*1000.), "error": str(e)}
+        res["database"] = {
+            "database": settings.DATABASE_TYPE.lower(),
+            "status": "red",
+            "elapsed": "{:.1f}".format((timer() - st) * 1000.0),
+            "error": str(e),
+        }
 
     st = timer()
     try:
         if not REDIS_CONN.health():
             raise Exception("Lost connection!")
-        res["redis"] = {"status": "green", "elapsed": "{:.1f}".format((timer() - st)*1000.)}
+        res["redis"] = {
+            "status": "green",
+            "elapsed": "{:.1f}".format((timer() - st) * 1000.0),
+        }
     except Exception as e:
-        res["redis"] = {"status": "red", "elapsed": "{:.1f}".format((timer() - st)*1000.), "error": str(e)}
+        res["redis"] = {
+            "status": "red",
+            "elapsed": "{:.1f}".format((timer() - st) * 1000.0),
+            "error": str(e),
+        }
 
+    task_executor_heartbeats = {}
     try:
-        v = REDIS_CONN.get("TASKEXE")
-        if not v:
-            raise Exception("No task executor running!")
-        obj = json.loads(v)
-        color = "green"
-        for id in obj.keys():
-            arr = obj[id]
-            if len(arr) == 1:
-                obj[id] = [0]
-            else:
-                obj[id] = [arr[i+1]-arr[i] for i in range(len(arr)-1)]
-            elapsed = max(obj[id])
-            if elapsed > 50: color = "yellow"
-            if elapsed > 120: color = "red"
-        res["task_executor"] = {"status": color, "elapsed": obj}
-    except Exception as e:
-        res["task_executor"] = {"status": "red", "error": str(e)}
+        task_executors = REDIS_CONN.smembers("TASKEXE")
+        now = datetime.now().timestamp()
+        for task_executor_id in task_executors:
+            heartbeats = REDIS_CONN.zrangebyscore(task_executor_id, now - 60*30, now)
+            heartbeats = [json.loads(heartbeat) for heartbeat in heartbeats]
+            task_executor_heartbeats[task_executor_id] = heartbeats
+    except Exception:
+        logging.exception("get task executor heartbeats failed!")
+    res["task_executor_heartbeats"] = task_executor_heartbeats
 
     return get_json_result(data=res)
+
+
+@manager.route("/new_token", methods=["POST"])
+@login_required
+def new_token():
+    """
+    Generate a new API token.
+    ---
+    tags:
+      - API Tokens
+    security:
+      - ApiKeyAuth: []
+    parameters:
+      - in: query
+        name: name
+        type: string
+        required: false
+        description: Name of the token.
+    responses:
+      200:
+        description: Token generated successfully.
+        schema:
+          type: object
+          properties:
+            token:
+              type: string
+              description: The generated API token.
+    """
+    try:
+        tenants = UserTenantService.query(user_id=current_user.id)
+        if not tenants:
+            return get_data_error_result(message="Tenant not found!")
+
+        tenant_id = tenants[0].tenant_id
+        obj = {
+            "tenant_id": tenant_id,
+            "token": generate_confirmation_token(tenant_id),
+            "create_time": current_timestamp(),
+            "create_date": datetime_format(datetime.now()),
+            "update_time": None,
+            "update_date": None,
+        }
+
+        if not APITokenService.save(**obj):
+            return get_data_error_result(message="Fail to new a dialog!")
+
+        return get_json_result(data=obj)
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route("/token_list", methods=["GET"])
+@login_required
+def token_list():
+    """
+    List all API tokens for the current user.
+    ---
+    tags:
+      - API Tokens
+    security:
+      - ApiKeyAuth: []
+    responses:
+      200:
+        description: List of API tokens.
+        schema:
+          type: object
+          properties:
+            tokens:
+              type: array
+              items:
+                type: object
+                properties:
+                  token:
+                    type: string
+                    description: The API token.
+                  name:
+                    type: string
+                    description: Name of the token.
+                  create_time:
+                    type: string
+                    description: Token creation time.
+    """
+    try:
+        tenants = UserTenantService.query(user_id=current_user.id)
+        if not tenants:
+            return get_data_error_result(message="Tenant not found!")
+
+        objs = APITokenService.query(tenant_id=tenants[0].tenant_id)
+        return get_json_result(data=[o.to_dict() for o in objs])
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route("/token/<token>", methods=["DELETE"])
+@login_required
+def rm(token):
+    """
+    Remove an API token.
+    ---
+    tags:
+      - API Tokens
+    security:
+      - ApiKeyAuth: []
+    parameters:
+      - in: path
+        name: token
+        type: string
+        required: true
+        description: The API token to remove.
+    responses:
+      200:
+        description: Token removed successfully.
+        schema:
+          type: object
+          properties:
+            success:
+              type: boolean
+              description: Deletion status.
+    """
+    APITokenService.filter_delete(
+        [APIToken.tenant_id == current_user.id, APIToken.token == token]
+    )
+    return get_json_result(data=True)

api/apps/tenant_app.py

@@ -0,0 +1,118 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+
+from flask import request
+from flask_login import login_required, current_user
+
+from api import settings
+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.api_utils import get_json_result, validate_request, server_error_response, get_data_error_result
+
+
+@manager.route("/<tenant_id>/user/list", methods=["GET"])
+@login_required
+def user_list(tenant_id):
+    if current_user.id != tenant_id:
+        return get_json_result(
+            data=False,
+            message='No authorization.',
+            code=settings.RetCode.AUTHENTICATION_ERROR)
+
+    try:
+        users = UserTenantService.get_by_tenant_id(tenant_id)
+        for u in users:
+            u["delta_seconds"] = delta_seconds(str(u["update_date"]))
+        return get_json_result(data=users)
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route('/<tenant_id>/user', methods=['POST'])
+@login_required
+@validate_request("email")
+def create(tenant_id):
+    if current_user.id != tenant_id:
+        return get_json_result(
+            data=False,
+            message='No authorization.',
+            code=settings.RetCode.AUTHENTICATION_ERROR)
+
+    req = request.json
+    usrs = UserService.query(email=req["email"])
+    if not usrs:
+        return get_data_error_result(message="User not found.")
+
+    user_id = usrs[0].id
+    user_tenants = UserTenantService.query(user_id=user_id, tenant_id=tenant_id)
+    if user_tenants:
+        if user_tenants[0].status == UserTenantRole.NORMAL.value:
+            return get_data_error_result(message="This user is in the team already.")
+        return get_data_error_result(message="Invitation notification is sent.")
+
+    UserTenantService.save(
+        id=get_uuid(),
+        user_id=user_id,
+        tenant_id=tenant_id,
+        invited_by=current_user.id,
+        role=UserTenantRole.INVITE,
+        status=StatusEnum.VALID.value)
+
+    usr = usrs[0].to_dict()
+    usr = {k: v for k, v in usr.items() if k in ["id", "avatar", "email", "nickname"]}
+
+    return get_json_result(data=usr)
+
+
+@manager.route('/<tenant_id>/user/<user_id>', methods=['DELETE'])
+@login_required
+def rm(tenant_id, user_id):
+    if current_user.id != tenant_id and current_user.id != user_id:
+        return get_json_result(
+            data=False,
+            message='No authorization.',
+            code=settings.RetCode.AUTHENTICATION_ERROR)
+
+    try:
+        UserTenantService.filter_delete([UserTenant.tenant_id == tenant_id, UserTenant.user_id == user_id])
+        return get_json_result(data=True)
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route("/list", methods=["GET"])
+@login_required
+def tenant_list():
+    try:
+        users = UserTenantService.get_tenants_by_user_id(current_user.id)
+        for u in users:
+            u["delta_seconds"] = delta_seconds(str(u["update_date"]))
+        return get_json_result(data=users)
+    except Exception as e:
+        return server_error_response(e)
+
+
+@manager.route("/agree/<tenant_id>", methods=["PUT"])
+@login_required
+def agree(tenant_id):
+    try:
+        UserTenantService.filter_update([UserTenant.tenant_id == tenant_id, UserTenant.user_id == current_user.id], {"role": UserTenantRole.NORMAL})
+        return get_json_result(data=True)
+    except Exception as e:
+        return server_error_response(e)

api/apps/user_app.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 import json
 import re
 from datetime import datetime
@@ -23,65 +24,127 @@ from flask_login import login_required, current_user, login_user, logout_user
 
 from api.db.db_models import TenantLLM
 from api.db.services.llm_service import TenantLLMService, LLMService
-from api.utils.api_utils import server_error_response, validate_request
-from api.utils import get_uuid, get_format_time, decrypt, download_img, current_timestamp, datetime_format
-from api.db import UserTenantRole, LLMType, FileType
-from api.settings import RetCode, GITHUB_OAUTH, FEISHU_OAUTH, CHAT_MDL, EMBEDDING_MDL, ASR_MDL, IMAGE2TEXT_MDL, PARSERS, \
-    API_KEY, \
-    LLM_FACTORY, LLM_BASE_URL, RERANK_MDL
+from api.utils.api_utils import (
+    server_error_response,
+    validate_request,
+    get_data_error_result,
+)
+from api.utils import (
+    get_uuid,
+    get_format_time,
+    decrypt,
+    download_img,
+    current_timestamp,
+    datetime_format,
+)
+from api.db import UserTenantRole, FileType
+from api import settings
 from api.db.services.user_service import UserService, TenantService, UserTenantService
 from api.db.services.file_service import FileService
-from api.settings import stat_logger
 from api.utils.api_utils import get_json_result, construct_response
 
 
-@manager.route('/login', methods=['POST', 'GET'])
+@manager.route("/login", methods=["POST", "GET"])
 def login():
+    """
+    User login endpoint.
+    ---
+    tags:
+      - User
+    parameters:
+      - in: body
+        name: body
+        description: Login credentials.
+        required: true
+        schema:
+          type: object
+          properties:
+            email:
+              type: string
+              description: User email.
+            password:
+              type: string
+              description: User password.
+    responses:
+      200:
+        description: Login successful.
+        schema:
+          type: object
+      401:
+        description: Authentication failed.
+        schema:
+          type: object
+    """
     if not request.json:
-        return get_json_result(data=False,
-                               retcode=RetCode.AUTHENTICATION_ERROR,
-                               retmsg='Unauthorized!')
+        return get_json_result(
+            data=False, code=settings.RetCode.AUTHENTICATION_ERROR, message="Unauthorized!"
+        )
 
-    email = request.json.get('email', "")
+    email = request.json.get("email", "")
     users = UserService.query(email=email)
     if not users:
-        return get_json_result(data=False,
-                               retcode=RetCode.AUTHENTICATION_ERROR,
-                               retmsg=f'Email: {email} is not registered!')
+        return get_json_result(
+            data=False,
+            code=settings.RetCode.AUTHENTICATION_ERROR,
+            message=f"Email: {email} is not registered!",
+        )
 
-    password = request.json.get('password')
+    password = request.json.get("password")
     try:
         password = decrypt(password)
     except BaseException:
-        return get_json_result(data=False,
-                               retcode=RetCode.SERVER_ERROR,
-                               retmsg='Fail to crypt password')
+        return get_json_result(
+            data=False, code=settings.RetCode.SERVER_ERROR, message="Fail to crypt password"
+        )
 
     user = UserService.query_user(email, password)
     if user:
         response_data = user.to_json()
         user.access_token = get_uuid()
         login_user(user)
-        user.update_time = current_timestamp(),
-        user.update_date = datetime_format(datetime.now()),
+        user.update_time = (current_timestamp(),)
+        user.update_date = (datetime_format(datetime.now()),)
         user.save()
         msg = "Welcome back!"
-        return construct_response(data=response_data, auth=user.get_id(), retmsg=msg)
+        return construct_response(data=response_data, auth=user.get_id(), message=msg)
     else:
-        return get_json_result(data=False,
-                               retcode=RetCode.AUTHENTICATION_ERROR,
-                               retmsg='Email and password do not match!')
+        return get_json_result(
+            data=False,
+            code=settings.RetCode.AUTHENTICATION_ERROR,
+            message="Email and password do not match!",
+        )
 
 
-@manager.route('/github_callback', methods=['GET'])
+@manager.route("/github_callback", methods=["GET"])
 def github_callback():
+    """
+    GitHub OAuth callback endpoint.
+    ---
+    tags:
+      - OAuth
+    parameters:
+      - in: query
+        name: code
+        type: string
+        required: true
+        description: Authorization code from GitHub.
+    responses:
+      200:
+        description: Authentication successful.
+        schema:
+          type: object
+    """
     import requests
-    res = requests.post(GITHUB_OAUTH.get("url"),
-                        data={
-                            "client_id": GITHUB_OAUTH.get("client_id"),
-                            "client_secret": GITHUB_OAUTH.get("secret_key"),
-                            "code": request.args.get('code')},
-                        headers={"Accept": "application/json"})
+
+    res = requests.post(
+        settings.GITHUB_OAUTH.get("url"),
+        data={
+            "client_id": settings.GITHUB_OAUTH.get("client_id"),
+            "client_secret": settings.GITHUB_OAUTH.get("secret_key"),
+            "code": request.args.get("code"),
+        },
+        headers={"Accept": "application/json"},
+    )
     res = res.json()
     if "error" in res:
         return redirect("/?error=%s" % res["error_description"])
@@ -101,21 +164,24 @@ def github_callback():
             try:
                 avatar = download_img(user_info["avatar_url"])
             except Exception as e:
-                stat_logger.exception(e)
+                logging.exception(e)
                 avatar = ""
-            users = user_register(user_id, {
-                "access_token": session["access_token"],
-                "email": email_address,
-                "avatar": avatar,
-                "nickname": user_info["login"],
-                "login_channel": "github",
-                "last_login_time": get_format_time(),
-                "is_superuser": False,
-            })
+            users = user_register(
+                user_id,
+                {
+                    "access_token": session["access_token"],
+                    "email": email_address,
+                    "avatar": avatar,
+                    "nickname": user_info["login"],
+                    "login_channel": "github",
+                    "last_login_time": get_format_time(),
+                    "is_superuser": False,
+                },
+            )
             if not users:
-                raise Exception(f'Fail to register {email_address}.')
+                raise Exception(f"Fail to register {email_address}.")
             if len(users) > 1:
-                raise Exception(f'Same email: {email_address} exists!')
+                raise Exception(f"Same email: {email_address} exists!")
 
             # Try to log in
             user = users[0]
@@ -123,7 +189,7 @@ def github_callback():
             return redirect("/?auth=%s" % user.get_id())
         except Exception as e:
             rollback_user_registration(user_id)
-            stat_logger.exception(e)
+            logging.exception(e)
             return redirect("/?error=%s" % str(e))
 
     # User has already registered, try to log in
@@ -134,30 +200,56 @@ def github_callback():
     return redirect("/?auth=%s" % user.get_id())
 
 
-@manager.route('/feishu_callback', methods=['GET'])
+@manager.route("/feishu_callback", methods=["GET"])
 def feishu_callback():
+    """
+    Feishu OAuth callback endpoint.
+    ---
+    tags:
+      - OAuth
+    parameters:
+      - in: query
+        name: code
+        type: string
+        required: true
+        description: Authorization code from Feishu.
+    responses:
+      200:
+        description: Authentication successful.
+        schema:
+          type: object
+    """
     import requests
-    app_access_token_res = requests.post(FEISHU_OAUTH.get("app_access_token_url"),
-                                         data=json.dumps({
-                                             "app_id": FEISHU_OAUTH.get("app_id"),
-                                             "app_secret": FEISHU_OAUTH.get("app_secret")
-                                         }),
-                                         headers={"Content-Type": "application/json; charset=utf-8"})
+
+    app_access_token_res = requests.post(
+        settings.FEISHU_OAUTH.get("app_access_token_url"),
+        data=json.dumps(
+            {
+                "app_id": settings.FEISHU_OAUTH.get("app_id"),
+                "app_secret": settings.FEISHU_OAUTH.get("app_secret"),
+            }
+        ),
+        headers={"Content-Type": "application/json; charset=utf-8"},
+    )
     app_access_token_res = app_access_token_res.json()
-    if app_access_token_res['code'] != 0:
+    if app_access_token_res["code"] != 0:
         return redirect("/?error=%s" % app_access_token_res)
 
-    res = requests.post(FEISHU_OAUTH.get("user_access_token_url"),
-                        data=json.dumps({
-                            "grant_type": FEISHU_OAUTH.get("grant_type"),
-                            "code": request.args.get('code')
-                        }),
-                        headers={
-                            "Content-Type": "application/json; charset=utf-8",
-                            'Authorization': f"Bearer {app_access_token_res['app_access_token']}"
-                        })
+    res = requests.post(
+        settings.FEISHU_OAUTH.get("user_access_token_url"),
+        data=json.dumps(
+            {
+                "grant_type": settings.FEISHU_OAUTH.get("grant_type"),
+                "code": request.args.get("code"),
+            }
+        ),
+        headers={
+            "Content-Type": "application/json; charset=utf-8",
+            "Authorization": f"Bearer {app_access_token_res['app_access_token']}",
+        },
+    )
     res = res.json()
-    if res['code'] != 0:
+    if res["code"] != 0:
         return redirect("/?error=%s" % res["message"])
 
     if "contact:user.email:readonly" not in res["data"]["scope"].split(" "):
@@ -174,21 +266,24 @@ def feishu_callback():
             try:
                 avatar = download_img(user_info["avatar_url"])
             except Exception as e:
-                stat_logger.exception(e)
+                logging.exception(e)
                 avatar = ""
-            users = user_register(user_id, {
-                "access_token": session["access_token"],
-                "email": email_address,
-                "avatar": avatar,
-                "nickname": user_info["en_name"],
-                "login_channel": "feishu",
-                "last_login_time": get_format_time(),
-                "is_superuser": False,
-            })
+            users = user_register(
+                user_id,
+                {
+                    "access_token": session["access_token"],
+                    "email": email_address,
+                    "avatar": avatar,
+                    "nickname": user_info["en_name"],
+                    "login_channel": "feishu",
+                    "last_login_time": get_format_time(),
+                    "is_superuser": False,
+                },
+            )
             if not users:
-                raise Exception(f'Fail to register {email_address}.')
+                raise Exception(f"Fail to register {email_address}.")
             if len(users) > 1:
-                raise Exception(f'Same email: {email_address} exists!')
+                raise Exception(f"Same email: {email_address} exists!")
 
             # Try to log in
             user = users[0]
@@ -196,7 +291,7 @@ def feishu_callback():
             return redirect("/?auth=%s" % user.get_id())
         except Exception as e:
             rollback_user_registration(user_id)
-            stat_logger.exception(e)
+            logging.exception(e)
             return redirect("/?error=%s" % str(e))
 
     # User has already registered, try to log in
@@ -209,11 +304,14 @@ def feishu_callback():
 
 def user_info_from_feishu(access_token):
     import requests
-    headers = {"Content-Type": "application/json; charset=utf-8",
-               'Authorization': f"Bearer {access_token}"}
+
+    headers = {
+        "Content-Type": "application/json; charset=utf-8",
+        "Authorization": f"Bearer {access_token}",
+    }
     res = requests.get(
-        f"https://open.feishu.cn/open-apis/authen/v1/user_info",
-        headers=headers)
+        "https://open.feishu.cn/open-apis/authen/v1/user_info", headers=headers
+    )
     user_info = res.json()["data"]
     user_info["email"] = None if user_info.get("email") == "" else user_info["email"]
     return user_info
@@ -221,24 +319,38 @@ def user_info_from_feishu(access_token):
 
 def user_info_from_github(access_token):
     import requests
-    headers = {"Accept": "application/json",
-               'Authorization': f"token {access_token}"}
+
+    headers = {"Accept": "application/json", "Authorization": f"token {access_token}"}
     res = requests.get(
-        f"https://api.github.com/user?access_token={access_token}",
-        headers=headers)
+        f"https://api.github.com/user?access_token={access_token}", headers=headers
+    )
     user_info = res.json()
     email_info = requests.get(
         f"https://api.github.com/user/emails?access_token={access_token}",
-        headers=headers).json()
+        headers=headers,
+    ).json()
     user_info["email"] = next(
-        (email for email in email_info if email['primary'] == True),
-        None)["email"]
+        (email for email in email_info if email["primary"] == True), None
+    )["email"]
     return user_info
 
 
-@manager.route("/logout", methods=['GET'])
+@manager.route("/logout", methods=["GET"])
 @login_required
 def log_out():
+    """
+    User logout endpoint.
+    ---
+    tags:
+      - User
+    security:
+      - ApiKeyAuth: []
+    responses:
+      200:
+        description: Logout successful.
+        schema:
+          type: object
+    """
     current_user.access_token = ""
     current_user.save()
     logout_user()
@@ -248,19 +360,62 @@ def log_out():
 @manager.route("/setting", methods=["POST"])
 @login_required
 def setting_user():
+    """
+    Update user settings.
+    ---
+    tags:
+      - User
+    security:
+      - ApiKeyAuth: []
+    parameters:
+      - in: body
+        name: body
+        description: User settings to update.
+        required: true
+        schema:
+          type: object
+          properties:
+            nickname:
+              type: string
+              description: New nickname.
+            email:
+              type: string
+              description: New email.
+    responses:
+      200:
+        description: Settings updated successfully.
+        schema:
+          type: object
+    """
     update_dict = {}
     request_data = request.json
     if request_data.get("password"):
         new_password = request_data.get("new_password")
         if not check_password_hash(
-                current_user.password, decrypt(request_data["password"])):
-            return get_json_result(data=False, retcode=RetCode.AUTHENTICATION_ERROR, retmsg='Password error!')
+                current_user.password, decrypt(request_data["password"])
+        ):
+            return get_json_result(
+                data=False,
+                code=settings.RetCode.AUTHENTICATION_ERROR,
+                message="Password error!",
+            )
 
         if new_password:
             update_dict["password"] = generate_password_hash(decrypt(new_password))
 
     for k in request_data.keys():
-        if k in ["password", "new_password"]:
+        if k in [
+            "password",
+            "new_password",
+            "email",
+            "status",
+            "is_superuser",
+            "login_channel",
+            "is_anonymous",
+            "is_active",
+            "is_authenticated",
+            "last_login_time",
+        ]:
             continue
         update_dict[k] = request_data[k]
 
@@ -268,34 +423,59 @@ def setting_user():
         UserService.update_by_id(current_user.id, update_dict)
         return get_json_result(data=True)
     except Exception as e:
-        stat_logger.exception(e)
-        return get_json_result(data=False, retmsg='Update failure!', retcode=RetCode.EXCEPTION_ERROR)
+        logging.exception(e)
+        return get_json_result(
+            data=False, message="Update failure!", code=settings.RetCode.EXCEPTION_ERROR
+        )
 
 
 @manager.route("/info", methods=["GET"])
 @login_required
 def user_profile():
+    """
+    Get user profile information.
+    ---
+    tags:
+      - User
+    security:
+      - ApiKeyAuth: []
+    responses:
+      200:
+        description: User profile retrieved successfully.
+        schema:
+          type: object
+          properties:
+            id:
+              type: string
+              description: User ID.
+            nickname:
+              type: string
+              description: User nickname.
+            email:
+              type: string
+              description: User email.
+    """
     return get_json_result(data=current_user.to_dict())
 
 
 def rollback_user_registration(user_id):
     try:
         UserService.delete_by_id(user_id)
-    except Exception as e:
+    except Exception:
         pass
     try:
         TenantService.delete_by_id(user_id)
-    except Exception as e:
+    except Exception:
         pass
     try:
         u = UserTenantService.query(tenant_id=user_id)
         if u:
             UserTenantService.delete_by_id(u[0].id)
-    except Exception as e:
+    except Exception:
         pass
     try:
         TenantLLM.delete().where(TenantLLM.tenant_id == user_id).execute()
-    except Exception as e:
+    except Exception:
         pass
 
 
@@ -304,18 +484,18 @@ def user_register(user_id, user):
     tenant = {
         "id": user_id,
         "name": user["nickname"] + "‘s Kingdom",
-        "llm_id": CHAT_MDL,
-        "embd_id": EMBEDDING_MDL,
-        "asr_id": ASR_MDL,
-        "parser_ids": PARSERS,
-        "img2txt_id": IMAGE2TEXT_MDL,
-        "rerank_id": RERANK_MDL
+        "llm_id": settings.CHAT_MDL,
+        "embd_id": settings.EMBEDDING_MDL,
+        "asr_id": settings.ASR_MDL,
+        "parser_ids": settings.PARSERS,
+        "img2txt_id": settings.IMAGE2TEXT_MDL,
+        "rerank_id": settings.RERANK_MDL,
     }
     usr_tenant = {
         "tenant_id": user_id,
         "user_id": user_id,
         "invited_by": user_id,
-        "role": UserTenantRole.OWNER
+        "role": UserTenantRole.OWNER,
     }
     file_id = get_uuid()
     file = {
@@ -329,14 +509,18 @@ def user_register(user_id, user):
         "location": "",
     }
     tenant_llm = []
-    for llm in LLMService.query(fid=LLM_FACTORY):
-        tenant_llm.append({"tenant_id": user_id,
-                           "llm_factory": LLM_FACTORY,
-                           "llm_name": llm.llm_name,
-                           "model_type": llm.model_type,
-                           "api_key": API_KEY,
-                           "api_base": LLM_BASE_URL
-                           })
+    for llm in LLMService.query(fid=settings.LLM_FACTORY):
+        tenant_llm.append(
+            {
+                "tenant_id": user_id,
+                "llm_factory": settings.LLM_FACTORY,
+                "llm_name": llm.llm_name,
+                "model_type": llm.model_type,
+                "api_key": settings.API_KEY,
+                "api_base": settings.LLM_BASE_URL,
+                "max_tokens": llm.max_tokens if llm.max_tokens else 8192
+            }
+        )
 
     if not UserService.save(**user):
         return
@@ -350,21 +534,52 @@ def user_register(user_id, user):
 @manager.route("/register", methods=["POST"])
 @validate_request("nickname", "email", "password")
 def user_add():
+    """
+    Register a new user.
+    ---
+    tags:
+      - User
+    parameters:
+      - in: body
+        name: body
+        description: Registration details.
+        required: true
+        schema:
+          type: object
+          properties:
+            nickname:
+              type: string
+              description: User nickname.
+            email:
+              type: string
+              description: User email.
+            password:
+              type: string
+              description: User password.
+    responses:
+      200:
+        description: Registration successful.
+        schema:
+          type: object
+    """
     req = request.json
     email_address = req["email"]
 
     # Validate the email address
-    if not re.match(r"^[\w\._-]+@([\w_-]+\.)+[\w-]{2,4}$", email_address):
-        return get_json_result(data=False,
-                               retmsg=f'Invalid email address: {email_address}!',
-                               retcode=RetCode.OPERATING_ERROR)
+    if not re.match(r"^[\w\._-]+@([\w_-]+\.)+[\w-]{2,5}$", email_address):
+        return get_json_result(
+            data=False,
+            message=f"Invalid email address: {email_address}!",
+            code=settings.RetCode.OPERATING_ERROR,
+        )
 
     # Check if the email address is already used
     if UserService.query(email=email_address):
         return get_json_result(
             data=False,
-            retmsg=f'Email: {email_address} has already registered!',
-            retcode=RetCode.OPERATING_ERROR)
+            message=f"Email: {email_address} has already registered!",
+            code=settings.RetCode.OPERATING_ERROR,
+        )
 
     # Construct user info data
     nickname = req["nickname"]
@@ -382,28 +597,60 @@ def user_add():
     try:
         users = user_register(user_id, user_dict)
         if not users:
-            raise Exception(f'Fail to register {email_address}.')
+            raise Exception(f"Fail to register {email_address}.")
         if len(users) > 1:
-            raise Exception(f'Same email: {email_address} exists!')
+            raise Exception(f"Same email: {email_address} exists!")
         user = users[0]
         login_user(user)
-        return construct_response(data=user.to_json(),
-                                  auth=user.get_id(),
-                                  retmsg=f"{nickname}, welcome aboard!")
+        return construct_response(
+            data=user.to_json(),
+            auth=user.get_id(),
+            message=f"{nickname}, welcome aboard!",
+        )
     except Exception as e:
         rollback_user_registration(user_id)
-        stat_logger.exception(e)
-        return get_json_result(data=False,
-                               retmsg=f'User registration failure, error: {str(e)}',
-                               retcode=RetCode.EXCEPTION_ERROR)
+        logging.exception(e)
+        return get_json_result(
+            data=False,
+            message=f"User registration failure, error: {str(e)}",
+            code=settings.RetCode.EXCEPTION_ERROR,
+        )
 
 
 @manager.route("/tenant_info", methods=["GET"])
 @login_required
 def tenant_info():
+    """
+    Get tenant information.
+    ---
+    tags:
+      - Tenant
+    security:
+      - ApiKeyAuth: []
+    responses:
+      200:
+        description: Tenant information retrieved successfully.
+        schema:
+          type: object
+          properties:
+            tenant_id:
+              type: string
+              description: Tenant ID.
+            name:
+              type: string
+              description: Tenant name.
+            llm_id:
+              type: string
+              description: LLM ID.
+            embd_id:
+              type: string
+              description: Embedding model ID.
+    """
     try:
-        tenants = TenantService.get_by_user_id(current_user.id)[0]
-        return get_json_result(data=tenants)
+        tenants = TenantService.get_info_by(current_user.id)
+        if not tenants:
+            return get_data_error_result(message="Tenant not found!")
+        return get_json_result(data=tenants[0])
     except Exception as e:
         return server_error_response(e)
 
@@ -412,10 +659,45 @@ def tenant_info():
 @login_required
 @validate_request("tenant_id", "asr_id", "embd_id", "img2txt_id", "llm_id")
 def set_tenant_info():
+    """
+    Update tenant information.
+    ---
+    tags:
+      - Tenant
+    security:
+      - ApiKeyAuth: []
+    parameters:
+      - in: body
+        name: body
+        description: Tenant information to update.
+        required: true
+        schema:
+          type: object
+          properties:
+            tenant_id:
+              type: string
+              description: Tenant ID.
+            llm_id:
+              type: string
+              description: LLM ID.
+            embd_id:
+              type: string
+              description: Embedding model ID.
+            asr_id:
+              type: string
+              description: ASR model ID.
+            img2txt_id:
+              type: string
+              description: Image to Text model ID.
+    responses:
+      200:
+        description: Tenant information updated successfully.
+        schema:
+          type: object
+    """
     req = request.json
     try:
-        tid = req["tenant_id"]
-        del req["tenant_id"]
+        tid = req.pop("tenant_id")
         TenantService.update_by_id(tid, req)
         return get_json_result(data=True)
     except Exception as e:

api/constants.py

@@ -13,4 +13,13 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 
-NAME_LENGTH_LIMIT = 2 ** 10
+NAME_LENGTH_LIMIT = 2 ** 10
+
+IMG_BASE64_PREFIX = 'data:image/png;base64,'
+
+SERVICE_CONF = "service_conf.yaml"
+
+API_VERSION = "v1"
+RAG_FLOW_SERVICE_NAME = "ragflow"
+REQUEST_WAIT_SEC = 2
+REQUEST_MAX_WAIT_SEC = 300

api/db/__init__.py

@@ -27,6 +27,7 @@ class UserTenantRole(StrEnum):
     OWNER = 'owner'
     ADMIN = 'admin'
     NORMAL = 'normal'
+    INVITE = 'invite'
 
 
 class TenantPermission(StrEnum):
@@ -55,6 +56,7 @@ class LLMType(StrEnum):
     SPEECH2TEXT = 'speech2text'
     IMAGE2TEXT = 'image2text'
     RERANK = 'rerank'
+    TTS    = 'tts'
 
 
 class ChatStyle(StrEnum):

api/db/db_models.py

@@ -13,27 +13,28 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 import inspect
 import os
 import sys
 import typing
 import operator
+from enum import Enum
 from functools import wraps
 from itsdangerous.url_safe import URLSafeTimedSerializer as Serializer
 from flask_login import UserMixin
-from playhouse.migrate import MySQLMigrator, migrate
+from playhouse.migrate import MySQLMigrator, PostgresqlMigrator, migrate
 from peewee import (
     BigIntegerField, BooleanField, CharField,
     CompositeKey, IntegerField, TextField, FloatField, DateTimeField,
     Field, Model, Metadata
 )
-from playhouse.pool import PooledMySQLDatabase
-from api.db import SerializedType, ParserType
-from api.settings import DATABASE, stat_logger, SECRET_KEY
-from api.utils.log_utils import getLogger
-from api import utils
+from playhouse.pool import PooledMySQLDatabase, PooledPostgresqlDatabase
 
-LOGGER = getLogger()
+
+from api.db import SerializedType, ParserType
+from api import settings
+from api import utils
 
 
 def singleton(cls, *args, **kw):
@@ -58,8 +59,13 @@ AUTO_DATE_TIMESTAMP_FIELD_PREFIX = {
     "write_access"}
 
 
+class TextFieldType(Enum):
+    MYSQL = 'LONGTEXT'
+    POSTGRES = 'TEXT'
+
+
 class LongTextField(TextField):
-    field_type = 'LONGTEXT'
+    field_type = TextFieldType[settings.DATABASE_TYPE.upper()].value
 
 
 class JSONField(LongTextField):
@@ -267,17 +273,71 @@ class JsonSerializedField(SerializedField):
                                                   object_pairs_hook=object_pairs_hook, **kwargs)
 
 
+class PooledDatabase(Enum):
+    MYSQL = PooledMySQLDatabase
+    POSTGRES = PooledPostgresqlDatabase
+
+
+class DatabaseMigrator(Enum):
+    MYSQL = MySQLMigrator
+    POSTGRES = PostgresqlMigrator
+
+
 @singleton
 class BaseDataBase:
     def __init__(self):
-        database_config = DATABASE.copy()
+        database_config = settings.DATABASE.copy()
         db_name = database_config.pop("name")
-        self.database_connection = PooledMySQLDatabase(
-            db_name, **database_config)
-        stat_logger.info('init mysql database on cluster mode successfully')
+        self.database_connection = PooledDatabase[settings.DATABASE_TYPE.upper()].value(db_name, **database_config)
+        logging.info('init database on cluster mode successfully')
 
 
-class DatabaseLock:
+class PostgresDatabaseLock:
+    def __init__(self, lock_name, timeout=10, db=None):
+        self.lock_name = lock_name
+        self.timeout = int(timeout)
+        self.db = db if db else DB
+
+    def lock(self):
+        cursor = self.db.execute_sql("SELECT pg_try_advisory_lock(%s)", self.timeout)
+        ret = cursor.fetchone()
+        if ret[0] == 0:
+            raise Exception(f'acquire postgres lock {self.lock_name} timeout')
+        elif ret[0] == 1:
+            return True
+        else:
+            raise Exception(f'failed to acquire lock {self.lock_name}')
+
+    def unlock(self):
+        cursor = self.db.execute_sql("SELECT pg_advisory_unlock(%s)", self.timeout)
+        ret = cursor.fetchone()
+        if ret[0] == 0:
+            raise Exception(
+                f'postgres lock {self.lock_name} was not established by this thread')
+        elif ret[0] == 1:
+            return True
+        else:
+            raise Exception(f'postgres lock {self.lock_name} does not exist')
+
+    def __enter__(self):
+        if isinstance(self.db, PostgresDatabaseLock):
+            self.lock()
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        if isinstance(self.db, PostgresDatabaseLock):
+            self.unlock()
+
+    def __call__(self, func):
+        @wraps(func)
+        def magic(*args, **kwargs):
+            with self:
+                return func(*args, **kwargs)
+
+        return magic
+
+
+class MysqlDatabaseLock:
     def __init__(self, lock_name, timeout=10, db=None):
         self.lock_name = lock_name
         self.timeout = int(timeout)
@@ -325,8 +385,13 @@ class DatabaseLock:
         return magic
 
 
+class DatabaseLock(Enum):
+    MYSQL = MysqlDatabaseLock
+    POSTGRES = PostgresDatabaseLock
+
+
 DB = BaseDataBase().database_connection
-DB.lock = DatabaseLock
+DB.lock = DatabaseLock[settings.DATABASE_TYPE.upper()].value
 
 
 def close_connection():
@@ -334,7 +399,7 @@ def close_connection():
         if DB:
             DB.close_stale(age=30)
     except Exception as e:
-        LOGGER.exception(e)
+        logging.exception(e)
 
 
 class DataBaseModel(BaseModel):
@@ -350,15 +415,15 @@ def init_database_tables(alter_fields=[]):
     for name, obj in members:
         if obj != DataBaseModel and issubclass(obj, DataBaseModel):
             table_objs.append(obj)
-            LOGGER.info(f"start create table {obj.__name__}")
+            logging.debug(f"start create table {obj.__name__}")
             try:
                 obj.create_table()
-                LOGGER.info(f"create table success: {obj.__name__}")
+                logging.debug(f"create table success: {obj.__name__}")
             except Exception as e:
-                LOGGER.exception(e)
+                logging.exception(e)
                 create_failed_list.append(obj.__name__)
     if create_failed_list:
-        LOGGER.info(f"create tables failed: {create_failed_list}")
+        logging.error(f"create tables failed: {create_failed_list}")
         raise Exception(f"create tables failed: {create_failed_list}")
     migrate_db()
 
@@ -408,7 +473,7 @@ class User(DataBaseModel, UserMixin):
     status = CharField(
         max_length=1,
         null=True,
-        help_text="is it validate(0: wasted，1: validate)",
+        help_text="is it validate(0: wasted, 1: validate)",
         default="1",
         index=True)
     is_superuser = BooleanField(null=True, help_text="is root", default=False, index=True)
@@ -417,7 +482,7 @@ class User(DataBaseModel, UserMixin):
         return self.email
 
     def get_id(self):
-        jwt = Serializer(secret_key=SECRET_KEY)
+        jwt = Serializer(secret_key=settings.SECRET_KEY)
         return jwt.dumps(str(self.access_token))
 
     class Meta:
@@ -449,6 +514,11 @@ class Tenant(DataBaseModel):
         null=False,
         help_text="default rerank model ID",
         index=True)
+    tts_id = CharField(
+        max_length=256,
+        null=True,
+        help_text="default tts model ID",
+        index=True)
     parser_ids = CharField(
         max_length=256,
         null=False,
@@ -458,7 +528,7 @@ class Tenant(DataBaseModel):
     status = CharField(
         max_length=1,
         null=True,
-        help_text="is it validate(0: wasted，1: validate)",
+        help_text="is it validate(0: wasted, 1: validate)",
         default="1",
         index=True)
 
@@ -475,7 +545,7 @@ class UserTenant(DataBaseModel):
     status = CharField(
         max_length=1,
         null=True,
-        help_text="is it validate(0: wasted，1: validate)",
+        help_text="is it validate(0: wasted, 1: validate)",
         default="1",
         index=True)
 
@@ -492,7 +562,7 @@ class InvitationCode(DataBaseModel):
     status = CharField(
         max_length=1,
         null=True,
-        help_text="is it validate(0: wasted，1: validate)",
+        help_text="is it validate(0: wasted, 1: validate)",
         default="1",
         index=True)
 
@@ -515,7 +585,7 @@ class LLMFactories(DataBaseModel):
     status = CharField(
         max_length=1,
         null=True,
-        help_text="is it validate(0: wasted，1: validate)",
+        help_text="is it validate(0: wasted, 1: validate)",
         default="1",
         index=True)
 
@@ -549,7 +619,7 @@ class LLM(DataBaseModel):
     status = CharField(
         max_length=1,
         null=True,
-        help_text="is it validate(0: wasted，1: validate)",
+        help_text="is it validate(0: wasted, 1: validate)",
         default="1",
         index=True)
 
@@ -581,7 +651,7 @@ class TenantLLM(DataBaseModel):
         index=True)
     api_key = CharField(max_length=1024, null=True, help_text="API KEY", index=True)
     api_base = CharField(max_length=255, null=True, help_text="API Base")
-
+    max_tokens = IntegerField(default=8192, index=True)
     used_tokens = IntegerField(default=0, index=True)
 
     def __str__(self):
@@ -636,7 +706,7 @@ class Knowledgebase(DataBaseModel):
     status = CharField(
         max_length=1,
         null=True,
-        help_text="is it validate(0: wasted，1: validate)",
+        help_text="is it validate(0: wasted, 1: validate)",
         default="1",
         index=True)
 
@@ -700,7 +770,7 @@ class Document(DataBaseModel):
     status = CharField(
         max_length=1,
         null=True,
-        help_text="is it validate(0: wasted，1: validate)",
+        help_text="is it validate(0: wasted, 1: validate)",
         default="1",
         index=True)
 
@@ -773,7 +843,7 @@ class Task(DataBaseModel):
     doc_id = CharField(max_length=32, null=False, index=True)
     from_page = IntegerField(default=0)
 
-    to_page = IntegerField(default=-1)
+    to_page = IntegerField(default=100000000)
 
     begin_at = DateTimeField(null=True, index=True)
     process_duation = FloatField(default=0)
@@ -783,6 +853,7 @@ class Task(DataBaseModel):
         null=True,
         help_text="process message",
         default="")
+    retry_count = IntegerField(default=0)
 
 
 class Dialog(DataBaseModel):
@@ -811,8 +882,10 @@ class Dialog(DataBaseModel):
         default="simple",
         help_text="simple|advanced",
         index=True)
-    prompt_config = JSONField(null=False, default={"system": "", "prologue": "您好，我是您的助手小樱，长得可爱又善良，can I help you?",
-                                                   "parameters": [], "empty_response": "Sorry! 知识库中未找到相关内容！"})
+    prompt_config = JSONField(null=False,
+                              default={"system": "", "prologue": "Hi! I'm your assistant, what can I do for you?",
+                                       "parameters": [],
+                                       "empty_response": "Sorry! No relevant content was found in the knowledge base!"})
 
     similarity_threshold = FloatField(default=0.2)
     vector_similarity_weight = FloatField(default=0.3)
@@ -824,8 +897,9 @@ class Dialog(DataBaseModel):
     do_refer = CharField(
         max_length=1,
         null=False,
+        default="1",
         help_text="it needs to insert reference index into answer or not")
-    
+
     rerank_id = CharField(
         max_length=128,
         null=False,
@@ -835,7 +909,7 @@ class Dialog(DataBaseModel):
     status = CharField(
         max_length=1,
         null=True,
-        help_text="is it validate(0: wasted，1: validate)",
+        help_text="is it validate(0: wasted, 1: validate)",
         default="1",
         index=True)
 
@@ -911,14 +985,14 @@ class CanvasTemplate(DataBaseModel):
 
 def migrate_db():
     with DB.transaction():
-        migrator = MySQLMigrator(DB)
+        migrator = DatabaseMigrator[settings.DATABASE_TYPE.upper()].value(DB)
         try:
             migrate(
                 migrator.add_column('file', 'source_type', CharField(max_length=128, null=False, default="",
                                                                      help_text="where dose this document come from",
                                                                      index=True))
             )
-        except Exception as e:
+        except Exception:
             pass
         try:
             migrate(
@@ -927,7 +1001,7 @@ def migrate_db():
                                               help_text="default rerank model ID"))
 
             )
-        except Exception as e:
+        except Exception:
             pass
         try:
             migrate(
@@ -935,38 +1009,64 @@ def migrate_db():
                                                                      help_text="default rerank model ID"))
 
             )
-        except Exception as e:
+        except Exception:
             pass
         try:
             migrate(
                 migrator.add_column('dialog', 'top_k', IntegerField(default=1024))
 
             )
-        except Exception as e:
+        except Exception:
             pass
         try:
             migrate(
                 migrator.alter_column_type('tenant_llm', 'api_key',
                                            CharField(max_length=1024, null=True, help_text="API KEY", index=True))
             )
-        except Exception as e:
+        except Exception:
             pass
         try:
             migrate(
                 migrator.add_column('api_token', 'source',
                                     CharField(max_length=16, null=True, help_text="none|agent|dialog", index=True))
             )
-        except Exception as e:
+        except Exception:
+            pass
+        try:
+            migrate(
+                migrator.add_column("tenant", "tts_id",
+                                    CharField(max_length=256, null=True, help_text="default tts model ID", index=True))
+            )
+        except Exception:
             pass
         try:
             migrate(
                 migrator.add_column('api_4_conversation', 'source',
                                     CharField(max_length=16, null=True, help_text="none|agent|dialog", index=True))
             )
-        except Exception as e:
+        except Exception:
             pass
         try:
             DB.execute_sql('ALTER TABLE llm DROP PRIMARY KEY;')
             DB.execute_sql('ALTER TABLE llm ADD PRIMARY KEY (llm_name,fid);')
-        except Exception as e:
+        except Exception:
+            pass
+        try:
+            migrate(
+                migrator.add_column('task', 'retry_count', IntegerField(default=0))
+            )
+        except Exception:
+            pass
+        try:
+            migrate(
+                migrator.alter_column_type('api_token', 'dialog_id',
+                                           CharField(max_length=32, null=True, index=True))
+            )
+        except Exception:
+            pass
+        try:
+            migrate(
+                migrator.add_column("tenant_llm","max_tokens",IntegerField(default=8192,index=True))
+            )
+        except Exception:
             pass

api/db/db_utils.py

@@ -15,17 +15,12 @@
 #
 import operator
 from functools import reduce
-from typing import Dict, Type, Union
+
+from playhouse.pool import PooledMySQLDatabase
 
 from api.utils import current_timestamp, timestamp_to_date
 
 from api.db.db_models import DB, DataBaseModel
-from api.db.runtime_config import RuntimeConfig
-from api.utils.log_utils import getLogger
-from enum import Enum
-
-
-LOGGER = getLogger()
 
 
 @DB.connection_context()
@@ -49,7 +44,10 @@ def bulk_insert_into_db(model, data_source, replace_on_conflict=False):
         with DB.atomic():
             query = model.insert_many(data_source[i:i + batch_size])
             if replace_on_conflict:
-                query = query.on_conflict(preserve=preserve)
+                if isinstance(DB, PooledMySQLDatabase):
+                    query = query.on_conflict(preserve=preserve)
+                else:
+                    query = query.on_conflict(conflict_target="id", preserve=preserve)
             query.execute()
 
 
@@ -88,7 +86,7 @@ supported_operators = {
 
 
 def query_dict2expression(
-        model: Type[DataBaseModel], query: Dict[str, Union[bool, int, str, list, tuple]]):
+        model: type[DataBaseModel], query: dict[str, bool | int | str | list | tuple]):
     expression = []
 
     for field, value in query.items():
@@ -106,8 +104,8 @@ def query_dict2expression(
     return reduce(operator.iand, expression)
 
 
-def query_db(model: Type[DataBaseModel], limit: int = 0, offset: int = 0,
-             query: dict = None, order_by: Union[str, list, tuple] = None):
+def query_db(model: type[DataBaseModel], limit: int = 0, offset: int = 0,
+             query: dict = None, order_by: str | list | tuple | None = None):
     data = model.select()
     if query:
         data = data.where(query_dict2expression(model, query))

api/db/init_data.py

@@ -13,6 +13,8 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
+import base64
 import json
 import os
 import time
@@ -27,14 +29,19 @@ from api.db.services.document_service import DocumentService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMFactoriesService, LLMService, TenantLLMService, LLMBundle
 from api.db.services.user_service import TenantService, UserTenantService
-from api.settings import CHAT_MDL, EMBEDDING_MDL, ASR_MDL, IMAGE2TEXT_MDL, PARSERS, LLM_FACTORY, API_KEY, LLM_BASE_URL
+from api import settings
 from api.utils.file_utils import get_project_base_directory
 
 
+def encode_to_base64(input_string):
+    base64_encoded = base64.b64encode(input_string.encode('utf-8'))
+    return base64_encoded.decode('utf-8')
+
+
 def init_superuser():
     user_info = {
         "id": uuid.uuid1().hex,
-        "password": "admin",
+        "password": encode_to_base64("admin"),
         "nickname": "admin",
         "is_superuser": True,
         "email": "admin@ragflow.io",
@@ -44,11 +51,11 @@ def init_superuser():
     tenant = {
         "id": user_info["id"],
         "name": user_info["nickname"] + "‘s Kingdom",
-        "llm_id": CHAT_MDL,
-        "embd_id": EMBEDDING_MDL,
-        "asr_id": ASR_MDL,
-        "parser_ids": PARSERS,
-        "img2txt_id": IMAGE2TEXT_MDL
+        "llm_id": settings.CHAT_MDL,
+        "embd_id": settings.EMBEDDING_MDL,
+        "asr_id": settings.ASR_MDL,
+        "parser_ids": settings.PARSERS,
+        "img2txt_id": settings.IMAGE2TEXT_MDL
     }
     usr_tenant = {
         "tenant_id": user_info["id"],
@@ -57,42 +64,43 @@ def init_superuser():
         "role": UserTenantRole.OWNER
     }
     tenant_llm = []
-    for llm in LLMService.query(fid=LLM_FACTORY):
+    for llm in LLMService.query(fid=settings.LLM_FACTORY):
         tenant_llm.append(
-            {"tenant_id": user_info["id"], "llm_factory": LLM_FACTORY, "llm_name": llm.llm_name, "model_type": llm.model_type,
-             "api_key": API_KEY, "api_base": LLM_BASE_URL})
+            {"tenant_id": user_info["id"], "llm_factory": settings.LLM_FACTORY, "llm_name": llm.llm_name,
+             "model_type": llm.model_type,
+             "api_key": settings.API_KEY, "api_base": settings.LLM_BASE_URL})
 
     if not UserService.save(**user_info):
-        print("\033[93m【ERROR】\033[0mcan't init admin.")
+        logging.error("can't init admin.")
         return
     TenantService.insert(**tenant)
     UserTenantService.insert(**usr_tenant)
     TenantLLMService.insert_many(tenant_llm)
-    print(
-        "【INFO】Super user initialized. \033[93memail: admin@ragflow.io, password: admin\033[0m. Changing the password after logining is strongly recomanded.")
+    logging.info(
+        "Super user initialized. email: admin@ragflow.io, password: admin. Changing the password after login is strongly recommended.")
 
     chat_mdl = LLMBundle(tenant["id"], LLMType.CHAT, tenant["llm_id"])
     msg = chat_mdl.chat(system="", history=[
-                        {"role": "user", "content": "Hello!"}], gen_conf={})
+        {"role": "user", "content": "Hello!"}], gen_conf={})
     if msg.find("ERROR: ") == 0:
-        print(
-            "\33[91m【ERROR】\33[0m: ",
+        logging.error(
             "'{}' dosen't work. {}".format(
                 tenant["llm_id"],
                 msg))
     embd_mdl = LLMBundle(tenant["id"], LLMType.EMBEDDING, tenant["embd_id"])
     v, c = embd_mdl.encode(["Hello!"])
     if c == 0:
-        print(
-            "\33[91m【ERROR】\33[0m:",
-            " '{}' dosen't work!".format(
+        logging.error(
+            "'{}' dosen't work!".format(
                 tenant["embd_id"]))
 
 
 def init_llm_factory():
     try:
         LLMService.filter_delete([(LLM.fid == "MiniMax" or LLM.fid == "Minimax")])
-    except Exception as e:
+        LLMService.filter_delete([(LLM.fid == "cohere")])
+        LLMFactoriesService.filter_delete([LLMFactories.name == "cohere"])
+    except Exception:
         pass
 
     factory_llm_infos = json.load(
@@ -105,14 +113,14 @@ def init_llm_factory():
         llm_infos = factory_llm_info.pop("llm")
         try:
             LLMFactoriesService.save(**factory_llm_info)
-        except Exception as e:
+        except Exception:
             pass
         LLMService.filter_delete([LLM.fid == factory_llm_info["name"]])
         for llm_info in llm_infos:
             llm_info["fid"] = factory_llm_info["name"]
             try:
                 LLMService.save(**llm_info)
-            except Exception as e:
+            except Exception:
                 pass
 
     LLMFactoriesService.filter_delete([LLMFactories.name == "Local"])
@@ -123,10 +131,11 @@ def init_llm_factory():
     LLMFactoriesService.filter_delete([LLMFactoriesService.model.name == "QAnything"])
     LLMService.filter_delete([LLMService.model.fid == "QAnything"])
     TenantLLMService.filter_update([TenantLLMService.model.llm_factory == "QAnything"], {"llm_factory": "Youdao"})
+    TenantLLMService.filter_update([TenantLLMService.model.llm_factory == "cohere"], {"llm_factory": "Cohere"})
     TenantService.filter_update([1 == 1], {
         "parser_ids": "naive:General,qa:Q&A,resume:Resume,manual:Manual,table:Table,paper:Paper,book:Book,laws:Laws,presentation:Presentation,picture:Picture,one:One,audio:Audio,knowledge_graph:Knowledge Graph,email:Email"})
     ## insert openai two embedding models to the current openai user.
-    print("Start to insert 2 OpenAI embedding models...")
+    # print("Start to insert 2 OpenAI embedding models...")
     tenant_ids = set([row["tenant_id"] for row in TenantLLMService.get_openai_models()])
     for tid in tenant_ids:
         for row in TenantLLMService.query(llm_factory="OpenAI", tenant_id=tid):
@@ -139,7 +148,7 @@ def init_llm_factory():
                 row = deepcopy(row)
                 row["llm_name"] = "text-embedding-3-large"
                 TenantLLMService.save(**row)
-            except Exception as e:
+            except Exception:
                 pass
             break
     for kb_id in KnowledgebaseService.get_all_ids():
@@ -163,20 +172,19 @@ def add_graph_templates():
                 CanvasTemplateService.save(**cnvs)
             except:
                 CanvasTemplateService.update_by_id(cnvs["id"], cnvs)
-        except Exception as e:
-            print("Add graph templates error: ", e)
-            print("------------", flush=True)
+        except Exception:
+            logging.exception("Add graph templates error: ")
 
 
 def init_web_data():
     start_time = time.time()
 
     init_llm_factory()
-    if not UserService.get_all().count():
-        init_superuser()
+    # if not UserService.get_all().count():
+    #    init_superuser()
 
     add_graph_templates()
-    print("init web data success:{}".format(time.time() - start_time))
+    logging.info("init web data success:{}".format(time.time() - start_time))
 
 
 if __name__ == '__main__':

api/db/runtime_config.py

@@ -13,7 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-from api.versions import get_versions
+from api.versions import get_ragflow_version
 from .reload_config_base import ReloadConfigBase
 
 
@@ -35,7 +35,7 @@ class RuntimeConfig(ReloadConfigBase):
 
     @classmethod
     def init_env(cls):
-        cls.ENV.update(get_versions())
+        cls.ENV.update({"version": get_ragflow_version()})
 
     @classmethod
     def load_config_manager(cls):

api/db/services/api_service.py

@@ -14,7 +14,9 @@
 #  limitations under the License.
 #
 from datetime import datetime
+
 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
@@ -41,7 +43,7 @@ class API4ConversationService(CommonService):
     @DB.connection_context()
     def append_message(cls, id, conversation):
         cls.update_by_id(id, conversation)
-        return cls.model.update(round=cls.model.round + 1).where(cls.model.id==id).execute()
+        return cls.model.update(round=cls.model.round + 1).where(cls.model.id == id).execute()
 
     @classmethod
     @DB.connection_context()
@@ -61,7 +63,7 @@ class API4ConversationService(CommonService):
                 cls.model.round).alias("round"),
             peewee.fn.SUM(
                 cls.model.thumb_up).alias("thumb_up")
-        ).join(Dialog, on=(cls.model.dialog_id == Dialog.id & Dialog.tenant_id == tenant_id)).where(
+        ).join(Dialog, on=((cls.model.dialog_id == Dialog.id) & (Dialog.tenant_id == tenant_id))).where(
             cls.model.create_date >= from_date,
             cls.model.create_date <= to_date,
             cls.model.source == source

api/db/services/canvas_service.py

@@ -22,5 +22,6 @@ from api.db.services.common_service import CommonService
 class CanvasTemplateService(CommonService):
     model = CanvasTemplate
 
+
 class UserCanvasService(CommonService):
     model = UserCanvas

api/db/services/dialog_service.py

@@ -13,19 +13,22 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
+import binascii
 import os
 import json
 import re
 from copy import deepcopy
-
-from api.db import LLMType, ParserType
-from api.db.db_models import Dialog, Conversation
+from timeit import default_timer as timer
+import datetime
+from datetime import timedelta
+from api.db import LLMType, ParserType,StatusEnum
+from api.db.db_models import Dialog, Conversation,DB
 from api.db.services.common_service import CommonService
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMService, TenantLLMService, LLMBundle
-from api.settings import chat_logger, retrievaler, kg_retrievaler
+from api import settings
 from rag.app.resume import forbidden_select_fields4resume
-from rag.nlp import keyword_extraction
 from rag.nlp.search import index_name
 from rag.utils import rmSpace, num_tokens_from_string, encoder
 from api.utils.file_utils import get_project_base_directory
@@ -34,10 +37,49 @@ from api.utils.file_utils import get_project_base_directory
 class DialogService(CommonService):
     model = Dialog
 
+    @classmethod
+    @DB.connection_context()
+    def get_list(cls, tenant_id,
+                 page_number, items_per_page, orderby, desc, id , name):
+        chats = cls.model.select()
+        if id:
+            chats = chats.where(cls.model.id == id)
+        if name:
+            chats = chats.where(cls.model.name == name)
+        chats = chats.where(
+              (cls.model.tenant_id == tenant_id)
+            & (cls.model.status == StatusEnum.VALID.value)
+        )
+        if desc:
+            chats = chats.order_by(cls.model.getter_by(orderby).desc())
+        else:
+            chats = chats.order_by(cls.model.getter_by(orderby).asc())
+
+        chats = chats.paginate(page_number, items_per_page)
+
+        return list(chats.dicts())
+
 
 class ConversationService(CommonService):
     model = Conversation
 
+    @classmethod
+    @DB.connection_context()
+    def get_list(cls,dialog_id,page_number, items_per_page, orderby, desc, id , name):
+        sessions = cls.model.select().where(cls.model.dialog_id ==dialog_id)
+        if id:
+            sessions = sessions.where(cls.model.id == id)
+        if name:
+            sessions = sessions.where(cls.model.name == name)
+        if desc:
+            sessions = sessions.order_by(cls.model.getter_by(orderby).desc())
+        else:
+            sessions = sessions.order_by(cls.model.getter_by(orderby).asc())
+
+        sessions = sessions.paginate(page_number, items_per_page)
+
+        return list(sessions.dicts())
+
 
 def message_fit_in(msg, max_length=4000):
     def count():
@@ -56,7 +98,8 @@ def message_fit_in(msg, max_length=4000):
         return c, msg
 
     msg_ = [m for m in msg[:-1] if m["role"] == "system"]
-    msg_.append(msg[-1])
+    if len(msg) > 1:
+        msg_.append(msg[-1])
     msg = msg_
     c = count()
     if c < max_length:
@@ -77,19 +120,27 @@ def message_fit_in(msg, max_length=4000):
 
 
 def llm_id2llm_type(llm_id):
+    llm_id = llm_id.split("@")[0]
     fnm = os.path.join(get_project_base_directory(), "conf")
     llm_factories = json.load(open(os.path.join(fnm, "llm_factories.json"), "r"))
     for llm_factory in llm_factories["factory_llm_infos"]:
         for llm in llm_factory["llm"]:
             if llm_id == llm["llm_name"]:
                 return llm["model_type"].strip(",")[-1]
-                
+
 
 def chat(dialog, messages, stream=True, **kwargs):
     assert messages[-1]["role"] == "user", "The last content of this conversation is not from user."
-    llm = LLMService.query(llm_name=dialog.llm_id)
+    st = timer()
+    tmp = dialog.llm_id.split("@")
+    fid = None
+    llm_id = tmp[0]
+    if len(tmp)>1: fid = tmp[1]
+
+    llm = LLMService.query(llm_name=llm_id) if not fid else LLMService.query(llm_name=llm_id, fid=fid)
     if not llm:
-        llm = TenantLLMService.query(tenant_id=dialog.tenant_id, llm_name=dialog.llm_id)
+        llm = TenantLLMService.query(tenant_id=dialog.tenant_id, llm_name=llm_id) if not fid else \
+            TenantLLMService.query(tenant_id=dialog.tenant_id, llm_name=llm_id, llm_factory=fid)
         if not llm:
             raise LookupError("LLM(%s) not found" % dialog.llm_id)
         max_tokens = 8192
@@ -102,7 +153,7 @@ def chat(dialog, messages, stream=True, **kwargs):
         return {"answer": "**ERROR**: Knowledge bases use different embedding models.", "reference": []}
 
     is_kg = all([kb.parser_id == ParserType.KG for kb in kbs])
-    retr = retrievaler if not is_kg else kg_retrievaler
+    retr = settings.retrievaler if not is_kg else settings.kg_retrievaler
 
     questions = [m["content"] for m in messages if m["role"] == "user"][-3:]
     attachments = kwargs["doc_ids"].split(",") if "doc_ids" in kwargs else None
@@ -113,6 +164,9 @@ def chat(dialog, messages, stream=True, **kwargs):
                 attachments.extend(m["doc_ids"])
 
     embd_mdl = LLMBundle(dialog.tenant_id, LLMType.EMBEDDING, embd_nms[0])
+    if not embd_mdl:
+        raise LookupError("Embedding model(%s) not found" % embd_nms[0])
+
     if llm_id2llm_type(dialog.llm_id) == "image2text":
         chat_mdl = LLMBundle(dialog.tenant_id, LLMType.IMAGE2TEXT, dialog.llm_id)
     else:
@@ -120,9 +174,12 @@ def chat(dialog, messages, stream=True, **kwargs):
 
     prompt_config = dialog.prompt_config
     field_map = KnowledgebaseService.get_field_map(dialog.kb_ids)
+    tts_mdl = None
+    if prompt_config.get("tts"):
+        tts_mdl = LLMBundle(dialog.tenant_id, LLMType.TTS)
     # try to use sql if field mapping is good to go
     if field_map:
-        chat_logger.info("Use SQL to retrieval:{}".format(questions[-1]))
+        logging.debug("Use SQL to retrieval:{}".format(questions[-1]))
         ans = use_sql(questions[-1], field_map, dialog.tenant_id, chat_mdl, prompt_config.get("quote", True))
         if ans:
             yield ans
@@ -137,6 +194,13 @@ def chat(dialog, messages, stream=True, **kwargs):
             prompt_config["system"] = prompt_config["system"].replace(
                 "{%s}" % p["key"], " ")
 
+    if len(questions) > 1 and prompt_config.get("refine_multiturn"):
+        questions = [full_question(dialog.tenant_id, dialog.llm_id, messages)]
+    else:
+        questions = questions[-1:]
+    refineQ_tm = timer()
+    keyword_tm = timer()
+
     rerank_mdl = None
     if dialog.rerank_id:
         rerank_mdl = LLMBundle(dialog.tenant_id, LLMType.RERANK, dialog.rerank_id)
@@ -148,30 +212,25 @@ def chat(dialog, messages, stream=True, **kwargs):
     else:
         if prompt_config.get("keyword", False):
             questions[-1] += keyword_extraction(chat_mdl, questions[-1])
-        kbinfos = retr.retrieval(" ".join(questions), embd_mdl, dialog.tenant_id, dialog.kb_ids, 1, dialog.top_n,
+            keyword_tm = timer()
+
+        tenant_ids = list(set([kb.tenant_id for kb in kbs]))
+        kbinfos = retr.retrieval(" ".join(questions), embd_mdl, tenant_ids, dialog.kb_ids, 1, dialog.top_n,
                                         dialog.similarity_threshold,
                                         dialog.vector_similarity_weight,
                                         doc_ids=attachments,
                                         top=dialog.top_k, aggs=False, rerank_mdl=rerank_mdl)
     knowledges = [ck["content_with_weight"] for ck in kbinfos["chunks"]]
-    #self-rag
-    if dialog.prompt_config.get("self_rag") and not relevant(dialog.tenant_id, dialog.llm_id, questions[-1], knowledges):
-        questions[-1] = rewrite(dialog.tenant_id, dialog.llm_id, questions[-1])
-        kbinfos = retr.retrieval(" ".join(questions), embd_mdl, dialog.tenant_id, dialog.kb_ids, 1, dialog.top_n,
-                                        dialog.similarity_threshold,
-                                        dialog.vector_similarity_weight,
-                                        doc_ids=attachments,
-                                        top=dialog.top_k, aggs=False, rerank_mdl=rerank_mdl)
-        knowledges = [ck["content_with_weight"] for ck in kbinfos["chunks"]]
-
-    chat_logger.info(
+    logging.debug(
         "{}->{}".format(" ".join(questions), "\n->".join(knowledges)))
+    retrieval_tm = timer()
 
     if not knowledges and prompt_config.get("empty_response"):
-        yield {"answer": prompt_config["empty_response"], "reference": kbinfos}
+        empty_res = prompt_config["empty_response"]
+        yield {"answer": empty_res, "reference": kbinfos, "audio_binary": tts(tts_mdl, empty_res)}
         return {"answer": prompt_config["empty_response"], "reference": kbinfos}
 
-    kwargs["knowledge"] = "\n".join(knowledges)
+    kwargs["knowledge"] = "\n\n------\n\n".join(knowledges)
     gen_conf = dialog.llm_setting
 
     msg = [{"role": "system", "content": prompt_config["system"].format(**kwargs)}]
@@ -179,6 +238,8 @@ def chat(dialog, messages, stream=True, **kwargs):
                 for m in messages if m["role"] != "system"])
     used_token_count, msg = message_fit_in(msg, int(max_tokens * 0.97))
     assert len(msg) >= 2, f"message_fit_in has bug: {msg}"
+    prompt = msg[0]["content"]
+    prompt += "\n\n### Query:\n%s" % " ".join(questions)
 
     if "max_tokens" in gen_conf:
         gen_conf["max_tokens"] = min(
@@ -186,7 +247,7 @@ def chat(dialog, messages, stream=True, **kwargs):
             max_tokens - used_token_count)
 
     def decorate_answer(answer):
-        nonlocal prompt_config, knowledges, kwargs, kbinfos
+        nonlocal prompt_config, knowledges, kwargs, kbinfos, prompt, retrieval_tm
         refs = []
         if knowledges and (prompt_config.get("quote", True) and kwargs.get("quote", True)):
             answer, idx = retr.insert_citations(answer,
@@ -210,20 +271,33 @@ def chat(dialog, messages, stream=True, **kwargs):
 
         if answer.lower().find("invalid key") >= 0 or answer.lower().find("invalid api") >= 0:
             answer += " Please set LLM API-Key in 'User Setting -> Model Providers -> API-Key'"
-        return {"answer": answer, "reference": refs}
+        done_tm = timer()
+        prompt += "\n\n### Elapsed\n  - Refine Question: %.1f ms\n  - Keywords: %.1f ms\n  - Retrieval: %.1f ms\n  - LLM: %.1f ms" % (
+            (refineQ_tm - st) * 1000, (keyword_tm - refineQ_tm) * 1000, (retrieval_tm - keyword_tm) * 1000,
+            (done_tm - retrieval_tm) * 1000)
+        return {"answer": answer, "reference": refs, "prompt": prompt}
 
     if stream:
+        last_ans = ""
         answer = ""
-        for ans in chat_mdl.chat_streamly(msg[0]["content"], msg[1:], gen_conf):
+        for ans in chat_mdl.chat_streamly(prompt, msg[1:], gen_conf):
             answer = ans
-            yield {"answer": answer, "reference": {}}
+            delta_ans = ans[len(last_ans):]
+            if num_tokens_from_string(delta_ans) < 16:
+                continue
+            last_ans = answer
+            yield {"answer": answer, "reference": {}, "audio_binary": tts(tts_mdl, delta_ans)}
+        delta_ans = answer[len(last_ans):]
+        if delta_ans:
+            yield {"answer": answer, "reference": {}, "audio_binary": tts(tts_mdl, delta_ans)}
         yield decorate_answer(answer)
     else:
-        answer = chat_mdl.chat(
-            msg[0]["content"], msg[1:], gen_conf)
-        chat_logger.info("User: {}|Assistant: {}".format(
+        answer = chat_mdl.chat(prompt, msg[1:], gen_conf)
+        logging.debug("User: {}|Assistant: {}".format(
             msg[-1]["content"], answer))
-        yield decorate_answer(answer)
+        res = decorate_answer(answer)
+        res["audio_binary"] = tts(tts_mdl, answer)
+        yield res
 
 
 def use_sql(question, field_map, tenant_id, chat_mdl, quota=True):
@@ -247,8 +321,7 @@ def use_sql(question, field_map, tenant_id, chat_mdl, quota=True):
         nonlocal sys_prompt, user_promt, question, tried_times
         sql = chat_mdl.chat(sys_prompt, [{"role": "user", "content": user_promt}], {
             "temperature": 0.06})
-        print(user_promt, sql)
-        chat_logger.info(f"“{question}”==>{user_promt} get SQL: {sql}")
+        logging.debug(f"{question} ==> {user_promt} get SQL: {sql}")
         sql = re.sub(r"[\r\n]+", " ", sql.lower())
         sql = re.sub(r".*select ", "select ", sql.lower())
         sql = re.sub(r" +", " ", sql)
@@ -268,11 +341,9 @@ def use_sql(question, field_map, tenant_id, chat_mdl, quota=True):
                     flds.append(k)
                 sql = "select doc_id,docnm_kwd," + ",".join(flds) + sql[8:]
 
-        print(f"“{question}” get SQL(refined): {sql}")
-
-        chat_logger.info(f"“{question}” get SQL(refined): {sql}")
+        logging.debug(f"{question} get SQL(refined): {sql}")
         tried_times += 1
-        return retrievaler.sql_retrieval(sql, format="json"), sql
+        return settings.retrievaler.sql_retrieval(sql, format="json"), sql
 
     tbl, sql = get_table()
     if tbl is None:
@@ -299,10 +370,9 @@ def use_sql(question, field_map, tenant_id, chat_mdl, quota=True):
             question, sql, tbl["error"]
         )
         tbl, sql = get_table()
-        chat_logger.info("TRY it again: {}".format(sql))
+        logging.debug("TRY it again: {}".format(sql))
 
-    chat_logger.info("GET table: {}".format(tbl))
-    print(tbl)
+    logging.debug("GET table: {}".format(tbl))
     if tbl.get("error") or len(tbl["rows"]) == 0:
         return None
 
@@ -324,6 +394,7 @@ def use_sql(question, field_map, tenant_id, chat_mdl, quota=True):
     rows = ["|" +
             "|".join([rmSpace(str(r[i])) for i in clmn_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)])
     else:
@@ -331,10 +402,11 @@ def use_sql(question, field_map, tenant_id, chat_mdl, quota=True):
     rows = re.sub(r"T[0-9]{2}:[0-9]{2}:[0-9]{2}(\.[0-9]+Z)?\|", "|", rows)
 
     if not docid_idx or not docnm_idx:
-        chat_logger.warning("SQL missing field: " + sql)
+        logging.warning("SQL missing field: " + sql)
         return {
             "answer": "\n".join([clmns, line, rows]),
-            "reference": {"chunks": [], "doc_aggs": []}
+            "reference": {"chunks": [], "doc_aggs": []},
+            "prompt": sys_prompt
         }
 
     docid_idx = list(docid_idx)[0]
@@ -348,7 +420,8 @@ def use_sql(question, field_map, tenant_id, chat_mdl, quota=True):
         "answer": "\n".join([clmns, line, rows]),
         "reference": {"chunks": [{"doc_id": r[docid_idx], "docnm_kwd": r[docnm_idx]} for r in tbl["rows"]],
                       "doc_aggs": [{"doc_id": did, "doc_name": d["doc_name"], "count": d["count"]} for did, d in
-                                   doc_aggs.items()]}
+                                   doc_aggs.items()]},
+        "prompt": sys_prompt
     }
 
 
@@ -390,3 +463,202 @@ def rewrite(tenant_id, llm_id, question):
     """
     ans = chat_mdl.chat(prompt, [{"role": "user", "content": question}], {"temperature": 0.8})
     return ans
+
+
+def keyword_extraction(chat_mdl, content, topn=3):
+    prompt = f"""
+Role: You're a text analyzer. 
+Task: extract the most important keywords/phrases of a given piece of text content.
+Requirements: 
+  - Summarize the text content, and give top {topn} important keywords/phrases.
+  - The keywords MUST be in language of the given piece of text content.
+  - The keywords are delimited by ENGLISH COMMA.
+  - Keywords ONLY in output.
+
+### Text Content 
+{content}
+
+"""
+    msg = [
+        {"role": "system", "content": prompt},
+        {"role": "user", "content": "Output: "}
+    ]
+    _, msg = message_fit_in(msg, chat_mdl.max_length)
+    kwd = chat_mdl.chat(prompt, msg[1:], {"temperature": 0.2})
+    if isinstance(kwd, tuple): kwd = kwd[0]
+    if kwd.find("**ERROR**") >=0: return ""
+    return kwd
+
+
+def question_proposal(chat_mdl, content, topn=3):
+    prompt = f"""
+Role: You're a text analyzer. 
+Task:  propose {topn} questions about a given piece of text content.
+Requirements: 
+  - Understand and summarize the text content, and propose top {topn} important questions.
+  - The questions SHOULD NOT have overlapping meanings.
+  - The questions SHOULD cover the main content of the text as much as possible.
+  - The questions MUST be in language of the given piece of text content.
+  - One question per line.
+  - Question ONLY in output.
+
+### Text Content 
+{content}
+
+"""
+    msg = [
+        {"role": "system", "content": prompt},
+        {"role": "user", "content": "Output: "}
+    ]
+    _, msg = message_fit_in(msg, chat_mdl.max_length)
+    kwd = chat_mdl.chat(prompt, msg[1:], {"temperature": 0.2})
+    if isinstance(kwd, tuple): kwd = kwd[0]
+    if kwd.find("**ERROR**") >= 0: return ""
+    return kwd
+
+
+def full_question(tenant_id, llm_id, messages):
+    if llm_id2llm_type(llm_id) == "image2text":
+        chat_mdl = LLMBundle(tenant_id, LLMType.IMAGE2TEXT, llm_id)
+    else:
+        chat_mdl = LLMBundle(tenant_id, LLMType.CHAT, llm_id)
+    conv = []
+    for m in messages:
+        if m["role"] not in ["user", "assistant"]: continue
+        conv.append("{}: {}".format(m["role"].upper(), m["content"]))
+    conv = "\n".join(conv)
+    today = datetime.date.today().isoformat()
+    yesterday = (datetime.date.today() - timedelta(days=1)).isoformat()
+    tomorrow = (datetime.date.today() + timedelta(days=1)).isoformat()
+    prompt = f"""
+Role: A helpful assistant
+
+Task and steps: 
+    1. Generate a full user question that would follow the conversation.
+    2. If the user's question involves relative date, you need to convert it into absolute date based on the current date, which is {today}. For example: 'yesterday' would be converted to {yesterday}.
+    
+Requirements & Restrictions:
+  - Text generated MUST be in the same language of the original user's question.
+  - If the user's latest question is completely, don't do anything, just return the original question.
+  - DON'T generate anything except a refined question.
+
+######################
+-Examples-
+######################
+
+# Example 1
+## Conversation
+USER: What is the name of Donald Trump's father?
+ASSISTANT:  Fred Trump.
+USER: And his mother?
+###############
+Output: What's the name of Donald Trump's mother?
+
+------------
+# Example 2
+## Conversation
+USER: What is the name of Donald Trump's father?
+ASSISTANT:  Fred Trump.
+USER: And his mother?
+ASSISTANT:  Mary Trump.
+User: What's her full name?
+###############
+Output: What's the full name of Donald Trump's mother Mary Trump?
+
+------------
+# Example 3
+## Conversation
+USER: What's the weather today in London?
+ASSISTANT:  Cloudy.
+USER: What's about tomorrow in Rochester?
+###############
+Output: What's the weather in Rochester on {tomorrow}?
+######################
+
+# Real Data
+## Conversation
+{conv}
+###############
+    """
+    ans = chat_mdl.chat(prompt, [{"role": "user", "content": "Output: "}], {"temperature": 0.2})
+    return ans if ans.find("**ERROR**") < 0 else messages[-1]["content"]
+
+
+def tts(tts_mdl, text):
+    if not tts_mdl or not text: return
+    bin = b""
+    for chunk in tts_mdl.tts(text):
+        bin += chunk
+    return binascii.hexlify(bin).decode("utf-8")
+
+
+def ask(question, kb_ids, tenant_id):
+    kbs = KnowledgebaseService.get_by_ids(kb_ids)
+    tenant_ids = [kb.tenant_id for kb in kbs]
+    embd_nms = list(set([kb.embd_id for kb in kbs]))
+
+    is_kg = all([kb.parser_id == ParserType.KG for kb in kbs])
+    retr = settings.retrievaler if not is_kg else settings.kg_retrievaler
+
+    embd_mdl = LLMBundle(tenant_id, LLMType.EMBEDDING, embd_nms[0])
+    chat_mdl = LLMBundle(tenant_id, LLMType.CHAT)
+    max_tokens = chat_mdl.max_length
+
+    kbinfos = retr.retrieval(question, embd_mdl, tenant_ids, kb_ids, 1, 12, 0.1, 0.3, aggs=False)
+    knowledges = [ck["content_with_weight"] for ck in kbinfos["chunks"]]
+
+    used_token_count = 0
+    for i, c in enumerate(knowledges):
+        used_token_count += num_tokens_from_string(c)
+        if max_tokens * 0.97 < used_token_count:
+            knowledges = knowledges[:i]
+            break
+
+    prompt = """
+    Role: You're a smart assistant. Your name is Miss R.
+    Task: Summarize the information from knowledge bases and answer user's question.
+    Requirements and restriction:
+      - DO NOT make things up, especially for numbers.
+      - If the information from knowledge is irrelevant with user's question, JUST SAY: Sorry, no relevant information provided.
+      - Answer with markdown format text.
+      - Answer in language of user's question.
+      - DO NOT make things up, especially for numbers.
+      
+    ### Information from knowledge bases
+    %s
+    
+    The above is information from knowledge bases.
+     
+    """%"\n".join(knowledges)
+    msg = [{"role": "user", "content": question}]
+
+    def decorate_answer(answer):
+        nonlocal knowledges, kbinfos, prompt
+        answer, idx = retr.insert_citations(answer,
+                                           [ck["content_ltks"]
+                                            for ck in kbinfos["chunks"]],
+                                           [ck["vector"]
+                                            for ck in kbinfos["chunks"]],
+                                           embd_mdl,
+                                           tkweight=0.7,
+                                           vtweight=0.3)
+        idx = set([kbinfos["chunks"][int(i)]["doc_id"] for i in idx])
+        recall_docs = [
+            d for d in kbinfos["doc_aggs"] if d["doc_id"] in idx]
+        if not recall_docs: recall_docs = kbinfos["doc_aggs"]
+        kbinfos["doc_aggs"] = recall_docs
+        refs = deepcopy(kbinfos)
+        for c in refs["chunks"]:
+            if c.get("vector"):
+                del c["vector"]
+
+        if answer.lower().find("invalid key") >= 0 or answer.lower().find("invalid api") >= 0:
+            answer += " Please set LLM API-Key in 'User Setting -> Model Providers -> API-Key'"
+        return {"answer": answer, "reference": refs}
+
+    answer = ""
+    for ans in chat_mdl.chat_streamly(prompt, msg, {"temperature": 0.1}):
+        answer = ans
+        yield {"answer": answer, "reference": {}}
+    yield decorate_answer(answer)
+

api/db/services/document_service.py

@@ -13,32 +13,28 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 import hashlib
 import json
-import os
 import random
 import re
-import traceback
 from concurrent.futures import ThreadPoolExecutor
 from copy import deepcopy
 from datetime import datetime
 from io import BytesIO
 
-from elasticsearch_dsl import Q
 from peewee import fn
 
 from api.db.db_utils import bulk_insert_into_db
-from api.settings import stat_logger
+from api import settings
 from api.utils import current_timestamp, get_format_time, get_uuid
-from api.utils.file_utils import get_project_base_directory
 from graphrag.mind_map_extractor import MindMapExtractor
 from rag.settings import SVR_QUEUE_NAME
-from rag.utils.es_conn import ELASTICSEARCH
-from rag.utils.minio_conn import MINIO
+from rag.utils.storage_factory import STORAGE_IMPL
 from rag.nlp import search, rag_tokenizer
 
 from api.db import FileType, TaskStatus, ParserType, LLMType
-from api.db.db_models import DB, Knowledgebase, Tenant, Task
+from api.db.db_models import DB, Knowledgebase, Tenant, Task, UserTenant
 from api.db.db_models import Document
 from api.db.services.common_service import CommonService
 from api.db.services.knowledgebase_service import KnowledgebaseService
@@ -49,6 +45,31 @@ from rag.utils.redis_conn import REDIS_CONN
 class DocumentService(CommonService):
     model = Document
 
+    @classmethod
+    @DB.connection_context()
+    def get_list(cls, kb_id, page_number, items_per_page,
+                 orderby, desc, keywords, id, name):
+        docs = cls.model.select().where(cls.model.kb_id == kb_id)
+        if id:
+            docs = docs.where(
+                cls.model.id == id)
+        if name:
+            docs = docs.where(
+                cls.model.name == name
+            )
+        if keywords:
+            docs = docs.where(
+                fn.LOWER(cls.model.name).contains(keywords.lower())
+            )
+        if desc:
+            docs = docs.order_by(cls.model.getter_by(orderby).desc())
+        else:
+            docs = docs.order_by(cls.model.getter_by(orderby).asc())
+
+        docs = docs.paginate(page_number, items_per_page)
+        count = docs.count()
+        return list(docs.dicts()), count
+
     @classmethod
     @DB.connection_context()
     def get_by_kb_id(cls, kb_id, page_number, items_per_page,
@@ -70,35 +91,6 @@ class DocumentService(CommonService):
 
         return list(docs.dicts()), count
 
-    @classmethod
-    @DB.connection_context()
-    def list_documents_in_dataset(cls, dataset_id, offset, count, order_by, descend, keywords):
-        if keywords:
-            docs = cls.model.select().where(
-                (cls.model.kb_id == dataset_id),
-                (fn.LOWER(cls.model.name).contains(keywords.lower()))
-            )
-        else:
-            docs = cls.model.select().where(cls.model.kb_id == dataset_id)
-
-        total = docs.count()
-
-        if descend == 'True':
-            docs = docs.order_by(cls.model.getter_by(order_by).desc())
-        if descend == 'False':
-            docs = docs.order_by(cls.model.getter_by(order_by).asc())
-
-        docs = list(docs.dicts())
-        docs_length = len(docs)
-
-        if offset < 0 or offset > docs_length:
-            raise IndexError("Offset is out of the valid range.")
-
-        if count == -1:
-            return docs[offset:], total
-
-        return docs[offset:offset + count], total
-
     @classmethod
     @DB.connection_context()
     def insert(cls, doc):
@@ -116,8 +108,7 @@ class DocumentService(CommonService):
     @classmethod
     @DB.connection_context()
     def remove_document(cls, doc, tenant_id):
-        ELASTICSEARCH.deleteByQuery(
-                Q("match", doc_id=doc.id), idxnm=search.index_name(tenant_id))
+        settings.docStoreConn.delete({"doc_id": doc.id}, search.index_name(tenant_id), doc.kb_id)
         cls.clear_chunk_num(doc.id)
         return cls.delete_by_id(doc.id)
 
@@ -140,26 +131,27 @@ class DocumentService(CommonService):
             cls.model.update_time]
         docs = cls.model.select(*fields) \
             .join(Knowledgebase, on=(cls.model.kb_id == Knowledgebase.id)) \
-            .join(Tenant, on=(Knowledgebase.tenant_id == Tenant.id))\
+            .join(Tenant, on=(Knowledgebase.tenant_id == Tenant.id)) \
             .where(
-                cls.model.status == StatusEnum.VALID.value,
-                ~(cls.model.type == FileType.VIRTUAL.value),
-                cls.model.progress == 0,
-                cls.model.update_time >= current_timestamp() - 1000 * 600,
-                cls.model.run == TaskStatus.RUNNING.value)\
+            cls.model.status == StatusEnum.VALID.value,
+            ~(cls.model.type == FileType.VIRTUAL.value),
+            cls.model.progress == 0,
+            cls.model.update_time >= current_timestamp() - 1000 * 600,
+            cls.model.run == TaskStatus.RUNNING.value) \
             .order_by(cls.model.update_time.asc())
         return list(docs.dicts())
 
     @classmethod
     @DB.connection_context()
     def get_unfinished_docs(cls):
-        fields = [cls.model.id, cls.model.process_begin_at, cls.model.parser_config, cls.model.progress_msg, cls.model.run]
+        fields = [cls.model.id, cls.model.process_begin_at, cls.model.parser_config, cls.model.progress_msg,
+                  cls.model.run]
         docs = cls.model.select(*fields) \
             .where(
-                cls.model.status == StatusEnum.VALID.value,
-                ~(cls.model.type == FileType.VIRTUAL.value),
-                cls.model.progress < 1,
-                cls.model.progress > 0)
+            cls.model.status == StatusEnum.VALID.value,
+            ~(cls.model.type == FileType.VIRTUAL.value),
+            cls.model.progress < 1,
+            cls.model.progress > 0)
         return list(docs.dicts())
 
     @classmethod
@@ -174,12 +166,12 @@ class DocumentService(CommonService):
                 "Document not found which is supposed to be there")
         num = Knowledgebase.update(
             token_num=Knowledgebase.token_num +
-            token_num,
+                      token_num,
             chunk_num=Knowledgebase.chunk_num +
-            chunk_num).where(
+                      chunk_num).where(
             Knowledgebase.id == kb_id).execute()
         return num
-    
+
     @classmethod
     @DB.connection_context()
     def decrement_chunk_num(cls, doc_id, kb_id, token_num, chunk_num, duation):
@@ -192,13 +184,13 @@ class DocumentService(CommonService):
                 "Document not found which is supposed to be there")
         num = Knowledgebase.update(
             token_num=Knowledgebase.token_num -
-            token_num,
+                      token_num,
             chunk_num=Knowledgebase.chunk_num -
-            chunk_num
+                      chunk_num
         ).where(
             Knowledgebase.id == kb_id).execute()
         return num
-    
+
     @classmethod
     @DB.connection_context()
     def clear_chunk_num(cls, doc_id):
@@ -207,10 +199,10 @@ class DocumentService(CommonService):
 
         num = Knowledgebase.update(
             token_num=Knowledgebase.token_num -
-            doc.token_num,
+                      doc.token_num,
             chunk_num=Knowledgebase.chunk_num -
-            doc.chunk_num,
-            doc_num=Knowledgebase.doc_num-1
+                      doc.chunk_num,
+            doc_num=Knowledgebase.doc_num - 1
         ).where(
             Knowledgebase.id == doc.kb_id).execute()
         return num
@@ -221,13 +213,22 @@ class DocumentService(CommonService):
         docs = cls.model.select(
             Knowledgebase.tenant_id).join(
             Knowledgebase, on=(
-                Knowledgebase.id == cls.model.kb_id)).where(
-                cls.model.id == doc_id, Knowledgebase.status == StatusEnum.VALID.value)
+                    Knowledgebase.id == cls.model.kb_id)).where(
+            cls.model.id == doc_id, Knowledgebase.status == StatusEnum.VALID.value)
         docs = docs.dicts()
         if not docs:
             return
         return docs[0]["tenant_id"]
 
+    @classmethod
+    @DB.connection_context()
+    def get_knowledgebase_id(cls, doc_id):
+        docs = cls.model.select(cls.model.kb_id).where(cls.model.id == doc_id)
+        docs = docs.dicts()
+        if not docs:
+            return
+        return docs[0]["kb_id"]
+
     @classmethod
     @DB.connection_context()
     def get_tenant_id_by_name(cls, name):
@@ -241,19 +242,46 @@ class DocumentService(CommonService):
             return
         return docs[0]["tenant_id"]
 
+    @classmethod
+    @DB.connection_context()
+    def accessible(cls, doc_id, user_id):
+        docs = cls.model.select(
+            cls.model.id).join(
+            Knowledgebase, on=(
+                    Knowledgebase.id == cls.model.kb_id)
+        ).join(UserTenant, on=(UserTenant.tenant_id == Knowledgebase.tenant_id)
+               ).where(cls.model.id == doc_id, UserTenant.user_id == user_id).paginate(0, 1)
+        docs = docs.dicts()
+        if not docs:
+            return False
+        return True
+
+    @classmethod
+    @DB.connection_context()
+    def accessible4deletion(cls, doc_id, user_id):
+        docs = cls.model.select(
+            cls.model.id).join(
+            Knowledgebase, on=(
+                    Knowledgebase.id == cls.model.kb_id)
+        ).where(cls.model.id == doc_id, Knowledgebase.created_by == user_id).paginate(0, 1)
+        docs = docs.dicts()
+        if not docs:
+            return False
+        return True
+
     @classmethod
     @DB.connection_context()
     def get_embd_id(cls, doc_id):
         docs = cls.model.select(
             Knowledgebase.embd_id).join(
             Knowledgebase, on=(
-                Knowledgebase.id == cls.model.kb_id)).where(
-                cls.model.id == doc_id, Knowledgebase.status == StatusEnum.VALID.value)
+                    Knowledgebase.id == cls.model.kb_id)).where(
+            cls.model.id == doc_id, Knowledgebase.status == StatusEnum.VALID.value)
         docs = docs.dicts()
         if not docs:
             return
         return docs[0]["embd_id"]
-    
+
     @classmethod
     @DB.connection_context()
     def get_doc_id_by_doc_name(cls, doc_name):
@@ -268,7 +296,7 @@ class DocumentService(CommonService):
     @classmethod
     @DB.connection_context()
     def get_thumbnails(cls, docids):
-        fields = [cls.model.id, cls.model.thumbnail]
+        fields = [cls.model.id, cls.model.kb_id, cls.model.thumbnail]
         return list(cls.model.select(
             *fields).where(cls.model.id.in_(docids)).dicts())
 
@@ -289,6 +317,7 @@ class DocumentService(CommonService):
                     dfs_update(old[k], v)
                 else:
                     old[k] = v
+
         dfs_update(d.parser_config, config)
         cls.update_by_id(id, {"parser_config": d.parser_config})
 
@@ -305,7 +334,7 @@ class DocumentService(CommonService):
     def begin2parse(cls, docid):
         cls.update_by_id(
             docid, {"progress": random.random() * 1 / 100.,
-                    "progress_msg": "Task dispatched...",
+                    "progress_msg": "Task is queued...",
                     "process_begin_at": get_format_time()
                     })
 
@@ -323,7 +352,7 @@ class DocumentService(CommonService):
                 finished = True
                 bad = 0
                 e, doc = DocumentService.get_by_id(d["id"])
-                status = doc.run#TaskStatus.RUNNING.value
+                status = doc.run  # TaskStatus.RUNNING.value
                 for t in tsks:
                     if 0 <= t.progress < 1:
                         finished = False
@@ -337,9 +366,10 @@ class DocumentService(CommonService):
                     prg = -1
                     status = TaskStatus.FAIL.value
                 elif finished:
-                    if d["parser_config"].get("raptor", {}).get("use_raptor") and d["progress_msg"].lower().find(" raptor")<0:
+                    if d["parser_config"].get("raptor", {}).get("use_raptor") and d["progress_msg"].lower().find(
+                            " raptor") < 0:
                         queue_raptor_tasks(d)
-                        prg *= 0.98
+                        prg = 0.98 * len(tsks) / (len(tsks) + 1)
                         msg.append("------ RAPTOR -------")
                     else:
                         status = TaskStatus.DONE.value
@@ -356,7 +386,8 @@ class DocumentService(CommonService):
                     info["progress_msg"] = msg
                 cls.update_by_id(d["id"], info)
             except Exception as e:
-                stat_logger.error("fetch task exception:" + str(e))
+                if str(e).find("'0'") < 0:
+                    logging.exception("fetch task exception")
 
     @classmethod
     @DB.connection_context()
@@ -364,14 +395,13 @@ class DocumentService(CommonService):
         return len(cls.model.select(cls.model.id).where(
             cls.model.kb_id == kb_id).dicts())
 
-
     @classmethod
     @DB.connection_context()
     def do_cancel(cls, doc_id):
         try:
             _, doc = DocumentService.get_by_id(doc_id)
             return doc.run == TaskStatus.CANCEL.value or doc.progress < 0
-        except Exception as e:
+        except Exception:
             pass
         return False
 
@@ -384,7 +414,7 @@ def queue_raptor_tasks(doc):
             "doc_id": doc["id"],
             "from_page": 0,
             "to_page": -1,
-            "progress_msg": "Start to do RAPTOR (Recursive Abstractive Processing For Tree-Organized Retrieval)."
+            "progress_msg": "Start to do RAPTOR (Recursive Abstractive Processing for Tree-Organized Retrieval)."
         }
 
     task = new_task()
@@ -412,11 +442,6 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
     if not e:
         raise LookupError("Can't find this knowledgebase!")
 
-    idxnm = search.index_name(kb.tenant_id)
-    if not ELASTICSEARCH.indexExist(idxnm):
-        ELASTICSEARCH.createIdx(idxnm, json.load(
-            open(os.path.join(get_project_base_directory(), "conf", "mapping.json"), "r")))
-
     embd_mdl = LLMBundle(kb.tenant_id, LLMType.EMBEDDING, llm_name=kb.embd_id, lang=kb.language)
 
     err, files = FileService.upload_document(kb, file_objs, user_id)
@@ -460,7 +485,7 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
             md5 = hashlib.md5()
             md5.update((ck["content_with_weight"] +
                         str(d["doc_id"])).encode("utf-8"))
-            d["_id"] = md5.hexdigest()
+            d["id"] = md5.hexdigest()
             d["create_time"] = str(datetime.now()).replace("T", " ")[:19]
             d["create_timestamp_flt"] = datetime.now().timestamp()
             if not d.get("image"):
@@ -473,9 +498,9 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
             else:
                 d["image"].save(output_buffer, format='JPEG')
 
-            MINIO.put(kb.id, d["_id"], output_buffer.getvalue())
-            d["img_id"] = "{}-{}".format(kb.id, d["_id"])
-            del d["image"]
+            STORAGE_IMPL.put(kb.id, d["id"], output_buffer.getvalue())
+            d["img_id"] = "{}-{}".format(kb.id, d["id"])
+            d.pop("image", None)
             docs.append(d)
 
     parser_ids = {d["id"]: d["parser_id"] for d, _ in files}
@@ -494,6 +519,9 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
             token_counts[doc_id] += c
         return vects
 
+    idxnm = search.index_name(kb.tenant_id)
+    try_create_idx = True
+
     _, tenant = TenantService.get_by_id(kb.tenant_id)
     llm_bdl = LLMBundle(kb.tenant_id, LLMType.CHAT, tenant.llm_id)
     for doc_id in docids:
@@ -516,7 +544,7 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
                     "knowledge_graph_kwd": "mind_map"
                 })
             except Exception as e:
-                stat_logger.error("Mind map generation error:", traceback.format_exc())
+                logging.exception("Mind map generation error")
 
         vects = embedding(doc_id, [c["content_with_weight"] for c in cks])
         assert len(cks) == len(vects)
@@ -524,9 +552,13 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
             v = vects[i]
             d["q_%d_vec" % len(v)] = v
         for b in range(0, len(cks), es_bulk_size):
-            ELASTICSEARCH.bulk(cks[b:b + es_bulk_size], idxnm)
+            if try_create_idx:
+                if not settings.docStoreConn.indexExist(idxnm, kb_id):
+                    settings.docStoreConn.createIdx(idxnm, kb_id, len(vects[0]))
+                try_create_idx = False
+            settings.docStoreConn.insert(cks[b:b + es_bulk_size], idxnm, kb_id)
 
         DocumentService.increment_chunk_num(
             doc_id, kb.id, token_counts[doc_id], chunk_counts[doc_id], 0)
 
-    return [d["id"] for d,_ in files]
+    return [d["id"] for d, _ in files]

api/db/services/file2document_service.py

@@ -69,14 +69,14 @@ class File2DocumentService(CommonService):
 
     @classmethod
     @DB.connection_context()
-    def get_minio_address(cls, doc_id=None, file_id=None):
+    def get_storage_address(cls, doc_id=None, file_id=None):
         if doc_id:
             f2d = cls.get_by_document_id(doc_id)
         else:
             f2d = cls.get_by_file_id(file_id)
         if f2d:
             file = File.get_by_id(f2d[0].file_id)
-            if file.source_type == FileSource.LOCAL:
+            if not file.source_type or file.source_type == FileSource.LOCAL:
                 return file.parent_id, file.location
             doc_id = f2d[0].document_id
 

api/db/services/file_service.py

@@ -13,8 +13,11 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 import re
 import os
+from concurrent.futures import ThreadPoolExecutor
+
 from flask_login import current_user
 from peewee import fn
 
@@ -26,8 +29,8 @@ from api.db.services.common_service import CommonService
 from api.db.services.document_service import DocumentService
 from api.db.services.file2document_service import File2DocumentService
 from api.utils import get_uuid
-from api.utils.file_utils import filename_type, thumbnail
-from rag.utils.minio_conn import MINIO
+from api.utils.file_utils import filename_type, thumbnail_img
+from rag.utils.storage_factory import STORAGE_IMPL
 
 
 class FileService(CommonService):
@@ -272,8 +275,8 @@ class FileService(CommonService):
                 cls.delete_folder_by_pf_id(user_id, file.id)
             return cls.model.delete().where((cls.model.tenant_id == user_id)
                                             & (cls.model.id == folder_id)).execute(),
-        except Exception as e:
-            print(e)
+        except Exception:
+            logging.exception("delete_folder_by_pf_id")
             raise RuntimeError("Database error (File retrieval)!")
 
     @classmethod
@@ -321,8 +324,8 @@ class FileService(CommonService):
     def move_file(cls, file_ids, folder_id):
         try:
             cls.filter_update((cls.model.id << file_ids, ), { 'parent_id': folder_id })
-        except Exception as e:
-            print(e)
+        except Exception:
+            logging.exception("move_file")
             raise RuntimeError("Database error (File move)!")
 
     @classmethod
@@ -350,30 +353,31 @@ class FileService(CommonService):
                     raise RuntimeError("This type of file has not been supported yet!")
 
                 location = filename
-                while MINIO.obj_exist(kb.id, location):
+                while STORAGE_IMPL.obj_exist(kb.id, location):
                     location += "_"
                 blob = file.read()
-                MINIO.put(kb.id, location, blob)
+                STORAGE_IMPL.put(kb.id, location, blob)
+
+                doc_id = get_uuid()
+
+                img = thumbnail_img(filename, blob)
+                thumbnail_location = ''
+                if img is not None:
+                    thumbnail_location = f'thumbnail_{doc_id}.png'
+                    STORAGE_IMPL.put(kb.id, thumbnail_location, img)
+
                 doc = {
-                    "id": get_uuid(),
+                    "id": doc_id,
                     "kb_id": kb.id,
-                    "parser_id": kb.parser_id,
+                    "parser_id": self.get_parser(filetype, filename, kb.parser_id),
                     "parser_config": kb.parser_config,
                     "created_by": user_id,
                     "type": filetype,
                     "name": filename,
                     "location": location,
                     "size": len(blob),
-                    "thumbnail": thumbnail(filename, blob)
+                    "thumbnail": thumbnail_location
                 }
-                if doc["type"] == FileType.VISUAL:
-                    doc["parser_id"] = ParserType.PICTURE.value
-                if doc["type"] == FileType.AURAL:
-                    doc["parser_id"] = ParserType.AUDIO.value
-                if re.search(r"\.(ppt|pptx|pages)$", filename):
-                    doc["parser_id"] = ParserType.PRESENTATION.value
-                if re.search(r"\.(eml)$", filename):
-                    doc["parser_id"] = ParserType.EMAIL.value
                 DocumentService.insert(doc)
 
                 FileService.add_file_from_kb(doc, kb_folder["id"], kb.tenant_id)
@@ -381,4 +385,51 @@ class FileService(CommonService):
             except Exception as e:
                 err.append(file.filename + ": " + str(e))
 
-        return err, files
+        return err, files
+
+    @staticmethod
+    def parse_docs(file_objs, user_id):
+        from rag.app import presentation, picture, naive, audio, email
+
+        def dummy(prog=None, msg=""):
+            pass
+
+        FACTORY = {
+            ParserType.PRESENTATION.value: presentation,
+            ParserType.PICTURE.value: picture,
+            ParserType.AUDIO.value: audio,
+            ParserType.EMAIL.value: email
+        }
+        parser_config = {"chunk_token_num": 16096, "delimiter": "\n!?;。；！？", "layout_recognize": False}
+        exe = ThreadPoolExecutor(max_workers=12)
+        threads = []
+        for file in file_objs:
+            kwargs = {
+                "lang": "English",
+                "callback": dummy,
+                "parser_config": parser_config,
+                "from_page": 0,
+                "to_page": 100000,
+                "tenant_id": user_id
+            }
+            filetype = filename_type(file.filename)
+            blob = file.read()
+            threads.append(exe.submit(FACTORY.get(FileService.get_parser(filetype, file.filename, ""), naive).chunk, file.filename, blob, **kwargs))
+
+        res = []
+        for th in threads:
+            res.append("\n".join([ck["content_with_weight"] for ck in th.result()]))
+
+        return "\n\n".join(res)
+
+    @staticmethod
+    def get_parser(doc_type, filename, default):
+        if doc_type == FileType.VISUAL:
+            return ParserType.PICTURE.value
+        if doc_type == FileType.AURAL:
+            return ParserType.AUDIO.value
+        if re.search(r"\.(ppt|pptx|pages)$", filename):
+            return ParserType.PRESENTATION.value
+        if re.search(r"\.(eml)$", filename):
+            return ParserType.EMAIL.value
+        return default

api/db/services/knowledgebase_service.py

@@ -14,21 +14,47 @@
 #  limitations under the License.
 #
 from api.db import StatusEnum, TenantPermission
-from api.db.db_models import Knowledgebase, DB, Tenant
+from api.db.db_models import Knowledgebase, DB, Tenant, User, UserTenant,Document
 from api.db.services.common_service import CommonService
 
 
 class KnowledgebaseService(CommonService):
     model = Knowledgebase
 
+    @classmethod
+    @DB.connection_context()
+    def list_documents_by_ids(cls,kb_ids):
+        doc_ids=cls.model.select(Document.id.alias("document_id")).join(Document,on=(cls.model.id == Document.kb_id)).where(
+            cls.model.id.in_(kb_ids)
+        )
+        doc_ids =list(doc_ids.dicts())
+        doc_ids = [doc["document_id"] for doc in doc_ids]
+        return doc_ids
+
     @classmethod
     @DB.connection_context()
     def get_by_tenant_ids(cls, joined_tenant_ids, user_id,
                           page_number, items_per_page, orderby, desc):
-        kbs = cls.model.select().where(
+        fields = [
+            cls.model.id,
+            cls.model.avatar,
+            cls.model.name,
+            cls.model.language,
+            cls.model.description,
+            cls.model.permission,
+            cls.model.doc_num,
+            cls.model.token_num,
+            cls.model.chunk_num,
+            cls.model.parser_id,
+            cls.model.embd_id,
+            User.nickname,
+            User.avatar.alias('tenant_avatar'),
+            cls.model.update_time
+        ]
+        kbs = cls.model.select(*fields).join(User, on=(cls.model.tenant_id == User.id)).where(
             ((cls.model.tenant_id.in_(joined_tenant_ids) & (cls.model.permission ==
                                                             TenantPermission.TEAM.value)) | (
-                         cls.model.tenant_id == user_id))
+                     cls.model.tenant_id == user_id))
             & (cls.model.status == StatusEnum.VALID.value)
         )
         if desc:
@@ -42,35 +68,20 @@ class KnowledgebaseService(CommonService):
 
     @classmethod
     @DB.connection_context()
-    def get_by_tenant_ids_by_offset(cls, joined_tenant_ids, user_id, offset, count, orderby, desc):
-        kbs = cls.model.select().where(
-            ((cls.model.tenant_id.in_(joined_tenant_ids) & (cls.model.permission ==
-                                                            TenantPermission.TEAM.value)) | (
-                     cls.model.tenant_id == user_id))
-            & (cls.model.status == StatusEnum.VALID.value)
-        )
-        if desc:
-            kbs = kbs.order_by(cls.model.getter_by(orderby).desc())
-        else:
-            kbs = kbs.order_by(cls.model.getter_by(orderby).asc())
-
-        kbs = list(kbs.dicts())
-
-        kbs_length = len(kbs)
-        if offset < 0 or offset > kbs_length:
-            raise IndexError("Offset is out of the valid range.")
-
-        if count == -1:
-            return kbs[offset:]
-
-        return kbs[offset:offset+count]
+    def get_kb_ids(cls, tenant_id):
+        fields = [
+            cls.model.id,
+        ]
+        kbs = cls.model.select(*fields).where(cls.model.tenant_id == tenant_id)
+        kb_ids = [kb.id for kb in kbs]
+        return kb_ids
 
     @classmethod
     @DB.connection_context()
     def get_detail(cls, kb_id):
         fields = [
             cls.model.id,
-            #Tenant.embd_id,
+            # Tenant.embd_id,
             cls.model.embd_id,
             cls.model.avatar,
             cls.model.name,
@@ -83,14 +94,14 @@ class KnowledgebaseService(CommonService):
             cls.model.parser_id,
             cls.model.parser_config]
         kbs = cls.model.select(*fields).join(Tenant, on=(
-                    (Tenant.id == cls.model.tenant_id) & (Tenant.status == StatusEnum.VALID.value))).where(
+                (Tenant.id == cls.model.tenant_id) & (Tenant.status == StatusEnum.VALID.value))).where(
             (cls.model.id == kb_id),
             (cls.model.status == StatusEnum.VALID.value)
         )
         if not kbs:
             return
         d = kbs[0].to_dict()
-        #d["embd_id"] = kbs[0].tenant.embd_id
+        # d["embd_id"] = kbs[0].tenant.embd_id
         return d
 
     @classmethod
@@ -142,3 +153,65 @@ class KnowledgebaseService(CommonService):
     @DB.connection_context()
     def get_all_ids(cls):
         return [m["id"] for m in cls.model.select(cls.model.id).dicts()]
+
+    @classmethod
+    @DB.connection_context()
+    def get_list(cls, joined_tenant_ids, user_id,
+                 page_number, items_per_page, orderby, desc, id, name):
+        kbs = cls.model.select()
+        if id:
+            kbs = kbs.where(cls.model.id == id)
+        if name:
+            kbs = kbs.where(cls.model.name == name)
+        kbs = kbs.where(
+            ((cls.model.tenant_id.in_(joined_tenant_ids) & (cls.model.permission ==
+                                                            TenantPermission.TEAM.value)) | (
+                     cls.model.tenant_id == user_id))
+            & (cls.model.status == StatusEnum.VALID.value)
+        )
+        if desc:
+            kbs = kbs.order_by(cls.model.getter_by(orderby).desc())
+        else:
+            kbs = kbs.order_by(cls.model.getter_by(orderby).asc())
+
+        kbs = kbs.paginate(page_number, items_per_page)
+
+        return list(kbs.dicts())
+
+    @classmethod
+    @DB.connection_context()
+    def accessible(cls, kb_id, user_id):
+        docs = cls.model.select(
+            cls.model.id).join(UserTenant, on=(UserTenant.tenant_id == Knowledgebase.tenant_id)
+            ).where(cls.model.id == kb_id, UserTenant.user_id == user_id).paginate(0, 1)
+        docs = docs.dicts()
+        if not docs:
+            return False
+        return True
+
+    @classmethod
+    @DB.connection_context()
+    def get_kb_by_id(cls, kb_id, user_id):
+        kbs = cls.model.select().join(UserTenant, on=(UserTenant.tenant_id == Knowledgebase.tenant_id)
+            ).where(cls.model.id == kb_id, UserTenant.user_id == user_id).paginate(0, 1)
+        kbs = kbs.dicts()
+        return list(kbs)
+
+    @classmethod
+    @DB.connection_context()
+    def get_kb_by_name(cls, kb_name, user_id):
+        kbs = cls.model.select().join(UserTenant, on=(UserTenant.tenant_id == Knowledgebase.tenant_id)
+            ).where(cls.model.name == kb_name, UserTenant.user_id == user_id).paginate(0, 1)
+        kbs = kbs.dicts()
+        return list(kbs)
+
+    @classmethod
+    @DB.connection_context()
+    def accessible4deletion(cls, kb_id, user_id):
+        docs = cls.model.select(
+            cls.model.id).where(cls.model.id == kb_id, cls.model.created_by == user_id).paginate(0, 1)
+        docs = docs.dicts()
+        if not docs:
+            return False
+        return True
+

api/db/services/llm_service.py

@@ -13,11 +13,11 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
 from api.db.services.user_service import TenantService
-from api.settings import database_logger
-from rag.llm import EmbeddingModel, CvModel, ChatModel, RerankModel, Seq2txtModel
+from rag.llm import EmbeddingModel, CvModel, ChatModel, RerankModel, Seq2txtModel, TTSModel
 from api.db import LLMType
-from api.db.db_models import DB, UserTenant
+from api.db.db_models import DB
 from api.db.db_models import LLMFactories, LLM, TenantLLM
 from api.db.services.common_service import CommonService
 
@@ -36,7 +36,11 @@ class TenantLLMService(CommonService):
     @classmethod
     @DB.connection_context()
     def get_api_key(cls, tenant_id, model_name):
-        objs = cls.query(tenant_id=tenant_id, llm_name=model_name)
+        arr = model_name.split("@")
+        if len(arr) < 2:
+            objs = cls.query(tenant_id=tenant_id, llm_name=model_name)
+        else:
+            objs = cls.query(tenant_id=tenant_id, llm_name=arr[0], llm_factory=arr[1])
         if not objs:
             return
         return objs[0]
@@ -75,18 +79,23 @@ class TenantLLMService(CommonService):
             mdlnm = tenant.llm_id if not llm_name else llm_name
         elif llm_type == LLMType.RERANK:
             mdlnm = tenant.rerank_id if not llm_name else llm_name
+        elif llm_type == LLMType.TTS:
+            mdlnm = tenant.tts_id if not llm_name else llm_name
         else:
             assert False, "LLM type error"
 
         model_config = cls.get_api_key(tenant_id, mdlnm)
+        tmp = mdlnm.split("@")
+        fid = None if len(tmp) < 2 else tmp[1]
+        mdlnm = tmp[0]
         if model_config: model_config = model_config.to_dict()
         if not model_config:
             if llm_type in [LLMType.EMBEDDING, LLMType.RERANK]:
-                llm = LLMService.query(llm_name=llm_name if llm_name else mdlnm)
+                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": llm_name if llm_name else mdlnm, "api_base": ""}
+                    model_config = {"llm_factory": llm[0].fid, "api_key":"", "llm_name": mdlnm, "api_base": ""}
             if not model_config:
-                if llm_name == "flag-embedding":
+                if mdlnm == "flag-embedding":
                     model_config = {"llm_factory": "Tongyi-Qianwen", "api_key": "",
                                 "llm_name": llm_name, "api_base": ""}
                 else:
@@ -124,9 +133,18 @@ class TenantLLMService(CommonService):
             if model_config["llm_factory"] not in Seq2txtModel:
                 return
             return Seq2txtModel[model_config["llm_factory"]](
-                model_config["api_key"], model_config["llm_name"], lang,
+                key=model_config["api_key"], model_name=model_config["llm_name"],
+                lang=lang,
                 base_url=model_config["api_base"]
             )
+        if llm_type == LLMType.TTS:
+            if model_config["llm_factory"] not in TTSModel:
+                return
+            return TTSModel[model_config["llm_factory"]](
+                model_config["api_key"],
+                model_config["llm_name"],
+                base_url=model_config["api_base"],
+            )
 
     @classmethod
     @DB.connection_context()
@@ -144,15 +162,19 @@ class TenantLLMService(CommonService):
         elif llm_type == LLMType.CHAT.value:
             mdlnm = tenant.llm_id if not llm_name else llm_name
         elif llm_type == LLMType.RERANK:
-            mdlnm = tenant.llm_id if not llm_name else llm_name
+            mdlnm = tenant.rerank_id if not llm_name else llm_name
+        elif llm_type == LLMType.TTS:
+            mdlnm = tenant.tts_id if not llm_name else llm_name
         else:
             assert False, "LLM type error"
 
+        llm_name = mdlnm.split("@")[0] if "@" in mdlnm else mdlnm
+
         num = 0
         try:
-            for u in cls.query(tenant_id = tenant_id, llm_name=mdlnm):
-                num += cls.model.update(used_tokens = u.used_tokens + used_tokens)\
-                    .where(cls.model.tenant_id == tenant_id, cls.model.llm_name == mdlnm)\
+            for u in cls.query(tenant_id=tenant_id, llm_name=llm_name):
+                num += cls.model.update(used_tokens=u.used_tokens + used_tokens)\
+                    .where(cls.model.tenant_id == tenant_id, cls.model.llm_name == llm_name)\
                     .execute()
         except Exception as e:
             pass
@@ -176,59 +198,69 @@ class LLMBundle(object):
         self.llm_name = llm_name
         self.mdl = TenantLLMService.model_instance(
             tenant_id, llm_type, llm_name, lang=lang)
-        assert self.mdl, "Can't find mole for {}/{}/{}".format(
+        assert self.mdl, "Can't find model for {}/{}/{}".format(
             tenant_id, llm_type, llm_name)
-        self.max_length = 512
+        self.max_length = 8192
         for lm in LLMService.query(llm_name=llm_name):
             self.max_length = lm.max_tokens
             break
-
+    
     def encode(self, texts: list, batch_size=32):
         emd, used_tokens = self.mdl.encode(texts, batch_size)
         if not TenantLLMService.increase_usage(
                 self.tenant_id, self.llm_type, used_tokens):
-            database_logger.error(
-                "Can't update token usage for {}/EMBEDDING".format(self.tenant_id))
+            logging.error(
+                "LLMBundle.encode can't update token usage for {}/EMBEDDING used_tokens: {}".format(self.tenant_id, used_tokens))
         return emd, used_tokens
 
     def encode_queries(self, query: str):
         emd, used_tokens = self.mdl.encode_queries(query)
         if not TenantLLMService.increase_usage(
                 self.tenant_id, self.llm_type, used_tokens):
-            database_logger.error(
-                "Can't update token usage for {}/EMBEDDING".format(self.tenant_id))
+            logging.error(
+                "LLMBundle.encode_queries can't update token usage for {}/EMBEDDING used_tokens: {}".format(self.tenant_id, used_tokens))
         return emd, used_tokens
 
     def similarity(self, query: str, texts: list):
         sim, used_tokens = self.mdl.similarity(query, texts)
         if not TenantLLMService.increase_usage(
                 self.tenant_id, self.llm_type, used_tokens):
-            database_logger.error(
-                "Can't update token usage for {}/RERANK".format(self.tenant_id))
+            logging.error(
+                "LLMBundle.similarity can't update token usage for {}/RERANK used_tokens: {}".format(self.tenant_id, used_tokens))
         return sim, used_tokens
 
     def describe(self, image, max_tokens=300):
         txt, used_tokens = self.mdl.describe(image, max_tokens)
         if not TenantLLMService.increase_usage(
                 self.tenant_id, self.llm_type, used_tokens):
-            database_logger.error(
-                "Can't update token usage for {}/IMAGE2TEXT".format(self.tenant_id))
+            logging.error(
+                "LLMBundle.describe can't update token usage for {}/IMAGE2TEXT used_tokens: {}".format(self.tenant_id, used_tokens))
         return txt
 
     def transcription(self, audio):
         txt, used_tokens = self.mdl.transcription(audio)
         if not TenantLLMService.increase_usage(
                 self.tenant_id, self.llm_type, used_tokens):
-            database_logger.error(
-                "Can't update token usage for {}/SEQUENCE2TXT".format(self.tenant_id))
+            logging.error(
+                "LLMBundle.transcription can't update token usage for {}/SEQUENCE2TXT used_tokens: {}".format(self.tenant_id, used_tokens))
         return txt
 
+    def tts(self, text):
+        for chunk in self.mdl.tts(text):
+            if isinstance(chunk,int):
+                if not TenantLLMService.increase_usage(
+                    self.tenant_id, self.llm_type, chunk, self.llm_name):
+                        logging.error(
+                            "LLMBundle.tts can't update token usage for {}/TTS".format(self.tenant_id))
+                return
+            yield chunk     
+
     def chat(self, system, history, gen_conf):
         txt, used_tokens = self.mdl.chat(system, history, gen_conf)
-        if not TenantLLMService.increase_usage(
+        if isinstance(txt, int) and not TenantLLMService.increase_usage(
                 self.tenant_id, self.llm_type, used_tokens, self.llm_name):
-            database_logger.error(
-                "Can't update token usage for {}/CHAT".format(self.tenant_id))
+            logging.error(
+                "LLMBundle.chat can't update token usage for {}/CHAT llm_name: {}, used_tokens: {}".format(self.tenant_id, self.llm_name, used_tokens))
         return txt
 
     def chat_streamly(self, system, history, gen_conf):
@@ -236,7 +268,7 @@ class LLMBundle(object):
             if isinstance(txt, int):
                 if not TenantLLMService.increase_usage(
                         self.tenant_id, self.llm_type, txt, self.llm_name):
-                    database_logger.error(
-                        "Can't update token usage for {}/CHAT".format(self.tenant_id))
+                    logging.error(
+                        "LLMBundle.chat_streamly can't update token usage for {}/CHAT llm_name: {}, content: {}".format(self.tenant_id, self.llm_name, txt))
                 return
             yield txt

api/db/services/task_service.py

@@ -27,7 +27,7 @@ from api.db.services.document_service import DocumentService
 from api.utils import current_timestamp, get_uuid
 from deepdoc.parser.excel_parser import RAGFlowExcelParser
 from rag.settings import SVR_QUEUE_NAME
-from rag.utils.minio_conn import MINIO
+from rag.utils.storage_factory import STORAGE_IMPL
 from rag.utils.redis_conn import REDIS_CONN
 
 
@@ -36,12 +36,13 @@ class TaskService(CommonService):
 
     @classmethod
     @DB.connection_context()
-    def get_tasks(cls, task_id):
+    def get_task(cls, task_id):
         fields = [
             cls.model.id,
             cls.model.doc_id,
             cls.model.from_page,
             cls.model.to_page,
+            cls.model.retry_count,
             Document.kb_id,
             Document.parser_id,
             Document.parser_config,
@@ -62,12 +63,23 @@ class TaskService(CommonService):
             .join(Tenant, on=(Knowledgebase.tenant_id == Tenant.id)) \
             .where(cls.model.id == task_id)
         docs = list(docs.dicts())
-        if not docs: return []
+        if not docs: return None
 
-        cls.model.update(progress_msg=cls.model.progress_msg + "\n" + "Task has been received.",
-                         progress=random.random() / 10.).where(
+        msg = "\nTask has been received."
+        prog = random.random() / 10.
+        if docs[0]["retry_count"] >= 3:
+            msg = "\nERROR: Task is abandoned after 3 times attempts."
+            prog = -1
+
+        cls.model.update(progress_msg=cls.model.progress_msg + msg,
+                         progress=prog,
+                         retry_count=docs[0]["retry_count"]+1
+                         ).where(
             cls.model.id == docs[0]["id"]).execute()
-        return docs
+
+        if docs[0]["retry_count"] >= 3: return None
+
+        return docs[0]
 
     @classmethod
     @DB.connection_context()
@@ -96,7 +108,7 @@ class TaskService(CommonService):
             task = cls.model.get_by_id(id)
             _, doc = DocumentService.get_by_id(task.doc_id)
             return doc.run == TaskStatus.CANCEL.value or doc.progress < 0
-        except Exception as e:
+        except Exception:
             pass
         return False
 
@@ -121,9 +133,8 @@ class TaskService(CommonService):
                     cls.model.id == id).execute()
 
 
-def queue_tasks(doc, bucket, name):
+def queue_tasks(doc: dict, bucket: str, name: str):
     def new_task():
-        nonlocal doc
         return {
             "id": get_uuid(),
             "doc_id": doc["id"]
@@ -131,21 +142,15 @@ def queue_tasks(doc, bucket, name):
     tsks = []
 
     if doc["type"] == FileType.PDF.value:
-        file_bin = MINIO.get(bucket, name)
+        file_bin = STORAGE_IMPL.get(bucket, name)
         do_layout = doc["parser_config"].get("layout_recognize", True)
         pages = PdfParser.total_page_number(doc["name"], file_bin)
         page_size = doc["parser_config"].get("task_page_size", 12)
         if doc["parser_id"] == "paper":
             page_size = doc["parser_config"].get("task_page_size", 22)
-        if doc["parser_id"] == "one":
-            page_size = 1000000000
-        if doc["parser_id"] == "knowledge_graph":
-            page_size = 1000000000
-        if not do_layout:
-            page_size = 1000000000
-        page_ranges = doc["parser_config"].get("pages")
-        if not page_ranges:
-            page_ranges = [(1, 100000)]
+        if doc["parser_id"] in ["one", "knowledge_graph"] or not do_layout:
+            page_size = 10 ** 9
+        page_ranges = doc["parser_config"].get("pages") or [(1, 10 ** 5)]
         for s, e in page_ranges:
             s -= 1
             s = max(0, s)
@@ -157,9 +162,8 @@ def queue_tasks(doc, bucket, name):
                 tsks.append(task)
 
     elif doc["parser_id"] == "table":
-        file_bin = MINIO.get(bucket, name)
-        rn = RAGFlowExcelParser.row_number(
-            doc["name"], file_bin)
+        file_bin = STORAGE_IMPL.get(bucket, name)
+        rn = RAGFlowExcelParser.row_number(doc["name"], file_bin)
         for i in range(0, rn, 3000):
             task = new_task()
             task["from_page"] = i

api/db/services/user_service.py

@@ -87,7 +87,7 @@ class TenantService(CommonService):
 
     @classmethod
     @DB.connection_context()
-    def get_by_user_id(cls, user_id):
+    def get_info_by(cls, user_id):
         fields = [
             cls.model.id.alias("tenant_id"),
             cls.model.name,
@@ -96,10 +96,11 @@ class TenantService(CommonService):
             cls.model.rerank_id,
             cls.model.asr_id,
             cls.model.img2txt_id,
+            cls.model.tts_id,
             cls.model.parser_ids,
             UserTenant.role]
         return list(cls.model.select(*fields)
-                    .join(UserTenant, on=((cls.model.id == UserTenant.tenant_id) & (UserTenant.user_id == user_id) & (UserTenant.status == StatusEnum.VALID.value)))
+                    .join(UserTenant, on=((cls.model.id == UserTenant.tenant_id) & (UserTenant.user_id == user_id) & (UserTenant.status == StatusEnum.VALID.value) & (UserTenant.role == UserTenantRole.OWNER)))
                     .where(cls.model.status == StatusEnum.VALID.value).dicts())
 
     @classmethod
@@ -114,7 +115,7 @@ class TenantService(CommonService):
             cls.model.img2txt_id,
             UserTenant.role]
         return list(cls.model.select(*fields)
-                    .join(UserTenant, on=((cls.model.id == UserTenant.tenant_id) & (UserTenant.user_id == user_id) & (UserTenant.status == StatusEnum.VALID.value) & (UserTenant.role == UserTenantRole.NORMAL.value)))
+                    .join(UserTenant, on=((cls.model.id == UserTenant.tenant_id) & (UserTenant.user_id == user_id) & (UserTenant.status == StatusEnum.VALID.value) & (UserTenant.role == UserTenantRole.NORMAL)))
                     .where(cls.model.status == StatusEnum.VALID.value).dicts())
 
     @classmethod
@@ -136,3 +137,39 @@ class UserTenantService(CommonService):
             kwargs["id"] = get_uuid()
         obj = cls.model(**kwargs).save(force_insert=True)
         return obj
+
+    @classmethod
+    @DB.connection_context()
+    def get_by_tenant_id(cls, tenant_id):
+        fields = [
+            cls.model.user_id,
+            cls.model.status,
+            cls.model.role,
+            User.nickname,
+            User.email,
+            User.avatar,
+            User.is_authenticated,
+            User.is_active,
+            User.is_anonymous,
+            User.status,
+            User.update_date,
+            User.is_superuser]
+        return list(cls.model.select(*fields)
+                    .join(User, on=((cls.model.user_id == User.id) & (cls.model.status == StatusEnum.VALID.value) & (cls.model.role != UserTenantRole.OWNER)))
+                    .where(cls.model.tenant_id == tenant_id)
+                    .dicts())
+
+    @classmethod
+    @DB.connection_context()
+    def get_tenants_by_user_id(cls, user_id):
+        fields = [
+            cls.model.tenant_id,
+            cls.model.role,
+            User.nickname,
+            User.email,
+            User.avatar,
+            User.update_date
+        ]
+        return list(cls.model.select(*fields)
+                    .join(User, on=((cls.model.tenant_id == User.id) & (UserTenant.user_id == user_id) & (UserTenant.status == StatusEnum.VALID.value)))
+                    .where(cls.model.status == StatusEnum.VALID.value).dicts())

api/ragflow_server.py

@@ -14,7 +14,21 @@
 #  limitations under the License.
 #
 
+# from beartype import BeartypeConf
+# from beartype.claw import beartype_all  # <-- you didn't sign up for this
+# beartype_all(conf=BeartypeConf(violation_type=UserWarning))    # <-- emit warnings from all code
+
 import logging
+from api.utils.log_utils import initRootLogger
+initRootLogger("ragflow_server")
+for module in ["pdfminer"]:
+    module_logger = logging.getLogger(module)
+    module_logger.setLevel(logging.WARNING)
+for module in ["peewee"]:
+    module_logger = logging.getLogger(module)
+    module_logger.handlers.clear()
+    module_logger.propagate = True
+
 import os
 import signal
 import sys
@@ -23,78 +37,84 @@ import traceback
 from concurrent.futures import ThreadPoolExecutor
 
 from werkzeug.serving import run_simple
+from api import settings
 from api.apps import app
 from api.db.runtime_config import RuntimeConfig
 from api.db.services.document_service import DocumentService
-from api.settings import (
-    HOST, HTTP_PORT, access_logger, database_logger, stat_logger,
-)
 from api import utils
 
 from api.db.db_models import init_database_tables as init_web_db
 from api.db.init_data import init_web_data
-from api.versions import get_versions
+from api.versions import get_ragflow_version
+from api.utils import show_configs
 
 
 def update_progress():
     while True:
-        time.sleep(1)
+        time.sleep(3)
         try:
             DocumentService.update_progress()
-        except Exception as e:
-            stat_logger.error("update_progress exception:" + str(e))
+        except Exception:
+            logging.exception("update_progress exception")
 
 
 if __name__ == '__main__':
-    print("""
-    ____                 ______ __               
-   / __ \ ____ _ ____ _ / ____// /____  _      __
-  / /_/ // __ `// __ `// /_   / // __ \| | /| / /
- / _, _// /_/ // /_/ // __/  / // /_/ /| |/ |/ / 
-/_/ |_| \__,_/ \__, //_/    /_/ \____/ |__/|__/  
-              /____/                             
+    logging.info(r"""
+        ____   ___    ______ ______ __               
+       / __ \ /   |  / ____// ____// /____  _      __
+      / /_/ // /| | / / __ / /_   / // __ \| | /| / /
+     / _, _// ___ |/ /_/ // __/  / // /_/ /| |/ |/ / 
+    /_/ |_|/_/  |_|\____//_/    /_/ \____/ |__/|__/                             
 
-    """, flush=True)
-    stat_logger.info(
+    """)
+    logging.info(
+        f'RAGFlow version: {get_ragflow_version()}'
+    )
+    logging.info(
         f'project base: {utils.file_utils.get_project_base_directory()}'
     )
+    show_configs()
+    settings.init_settings()
 
     # init db
     init_web_db()
     init_web_data()
     # init runtime config
     import argparse
+
     parser = argparse.ArgumentParser()
-    parser.add_argument('--version', default=False, help="rag flow version", action='store_true')
-    parser.add_argument('--debug', default=False, help="debug mode", action='store_true')
+    parser.add_argument(
+        "--version", default=False, help="RAGFlow version", action="store_true"
+    )
+    parser.add_argument(
+        "--debug", default=False, help="debug mode", action="store_true"
+    )
     args = parser.parse_args()
     if args.version:
-        print(get_versions())
+        print(get_ragflow_version())
         sys.exit(0)
 
     RuntimeConfig.DEBUG = args.debug
     if RuntimeConfig.DEBUG:
-        stat_logger.info("run on debug mode")
+        logging.info("run on debug mode")
 
     RuntimeConfig.init_env()
-    RuntimeConfig.init_config(JOB_SERVER_HOST=HOST, HTTP_PORT=HTTP_PORT)
+    RuntimeConfig.init_config(JOB_SERVER_HOST=settings.HOST_IP, HTTP_PORT=settings.HOST_PORT)
 
-    peewee_logger = logging.getLogger('peewee')
-    peewee_logger.propagate = False
-    # rag_arch.common.log.ROpenHandler
-    peewee_logger.addHandler(database_logger.handlers[0])
-    peewee_logger.setLevel(database_logger.level)
-
-    thr = ThreadPoolExecutor(max_workers=1)
-    thr.submit(update_progress)
+    thread = ThreadPoolExecutor(max_workers=1)
+    thread.submit(update_progress)
 
     # start http server
     try:
-        stat_logger.info("RAG Flow http server start...")
-        werkzeug_logger = logging.getLogger("werkzeug")
-        for h in access_logger.handlers:
-            werkzeug_logger.addHandler(h)
-        run_simple(hostname=HOST, port=HTTP_PORT, application=app, threaded=True, use_reloader=RuntimeConfig.DEBUG, use_debugger=RuntimeConfig.DEBUG)
+        logging.info("RAGFlow HTTP server start...")
+        run_simple(
+            hostname=settings.HOST_IP,
+            port=settings.HOST_PORT,
+            application=app,
+            threaded=True,
+            use_reloader=RuntimeConfig.DEBUG,
+            use_debugger=RuntimeConfig.DEBUG,
+        )
     except Exception:
         traceback.print_exc()
-        os.kill(os.getpid(), signal.SIGKILL)
+        os.kill(os.getpid(), signal.SIGKILL)

api/settings.py

@@ -14,198 +14,164 @@
 #  limitations under the License.
 #
 import os
+from datetime import date
 from enum import IntEnum, Enum
-from api.utils.file_utils import get_project_base_directory
-from api.utils.log_utils import LoggerFactory, getLogger
+import rag.utils.es_conn
+import rag.utils.infinity_conn
 
-# Logger
-LoggerFactory.set_directory(
-    os.path.join(
-        get_project_base_directory(),
-        "logs",
-        "api"))
-# {CRITICAL: 50, FATAL:50, ERROR:40, WARNING:30, WARN:30, INFO:20, DEBUG:10, NOTSET:0}
-LoggerFactory.LEVEL = 30
-
-stat_logger = getLogger("stat")
-access_logger = getLogger("access")
-database_logger = getLogger("database")
-chat_logger = getLogger("chat")
-
-from rag.utils.es_conn import ELASTICSEARCH
+import rag.utils
 from rag.nlp import search
 from graphrag import search as kg_search
 from api.utils import get_base_config, decrypt_database_config
+from api.constants import RAG_FLOW_SERVICE_NAME
 
-API_VERSION = "v1"
-RAG_FLOW_SERVICE_NAME = "ragflow"
-SERVER_MODULE = "rag_flow_server.py"
-TEMP_DIRECTORY = os.path.join(get_project_base_directory(), "temp")
-RAG_FLOW_CONF_PATH = os.path.join(get_project_base_directory(), "conf")
+LIGHTEN = int(os.environ.get('LIGHTEN', "0"))
 
-SUBPROCESS_STD_LOG_NAME = "std.log"
+LLM = None
+LLM_FACTORY = None
+LLM_BASE_URL = None
+CHAT_MDL = ""
+EMBEDDING_MDL = ""
+RERANK_MDL = ""
+ASR_MDL = ""
+IMAGE2TEXT_MDL = ""
+API_KEY = None
+PARSERS = None
+HOST_IP = None
+HOST_PORT = None
+SECRET_KEY = None
 
-ERROR_REPORT = True
-ERROR_REPORT_WITH_PATH = False
-
-MAX_TIMESTAMP_INTERVAL = 60
-SESSION_VALID_PERIOD = 7 * 24 * 60 * 60
-
-REQUEST_TRY_TIMES = 3
-REQUEST_WAIT_SEC = 2
-REQUEST_MAX_WAIT_SEC = 300
-
-USE_REGISTRY = get_base_config("use_registry")
-
-default_llm = {
-    "Tongyi-Qianwen": {
-        "chat_model": "qwen-plus",
-        "embedding_model": "text-embedding-v2",
-        "image2text_model": "qwen-vl-max",
-        "asr_model": "paraformer-realtime-8k-v1",
-    },
-    "OpenAI": {
-        "chat_model": "gpt-3.5-turbo",
-        "embedding_model": "text-embedding-ada-002",
-        "image2text_model": "gpt-4-vision-preview",
-        "asr_model": "whisper-1",
-    },
-    "Azure-OpenAI": {
-        "chat_model": "azure-gpt-35-turbo",
-        "embedding_model": "azure-text-embedding-ada-002",
-        "image2text_model": "azure-gpt-4-vision-preview",
-        "asr_model": "azure-whisper-1",
-    },
-    "ZHIPU-AI": {
-        "chat_model": "glm-3-turbo",
-        "embedding_model": "embedding-2",
-        "image2text_model": "glm-4v",
-        "asr_model": "",
-    },
-    "Ollama": {
-        "chat_model": "qwen-14B-chat",
-        "embedding_model": "flag-embedding",
-        "image2text_model": "",
-        "asr_model": "",
-    },
-    "Moonshot": {
-        "chat_model": "moonshot-v1-8k",
-        "embedding_model": "",
-        "image2text_model": "",
-        "asr_model": "",
-    },
-    "DeepSeek": {
-        "chat_model": "deepseek-chat",
-        "embedding_model": "",
-        "image2text_model": "",
-        "asr_model": "",
-    },
-    "VolcEngine": {
-        "chat_model": "",
-        "embedding_model": "",
-        "image2text_model": "",
-        "asr_model": "",
-    },
-    "BAAI": {
-        "chat_model": "",
-        "embedding_model": "BAAI/bge-large-zh-v1.5",
-        "image2text_model": "",
-        "asr_model": "",
-        "rerank_model": "BAAI/bge-reranker-v2-m3",
-    }
-}
-LLM = get_base_config("user_default_llm", {})
-LLM_FACTORY = LLM.get("factory", "Tongyi-Qianwen")
-LLM_BASE_URL = LLM.get("base_url")
-
-if LLM_FACTORY not in default_llm:
-    print(
-        "\33[91m【ERROR】\33[0m:",
-        f"LLM factory {LLM_FACTORY} has not supported yet, switch to 'Tongyi-Qianwen/QWen' automatically, and please check the API_KEY in service_conf.yaml.")
-    LLM_FACTORY = "Tongyi-Qianwen"
-CHAT_MDL = default_llm[LLM_FACTORY]["chat_model"]
-EMBEDDING_MDL = default_llm["BAAI"]["embedding_model"]
-RERANK_MDL = default_llm["BAAI"]["rerank_model"]
-ASR_MDL = default_llm[LLM_FACTORY]["asr_model"]
-IMAGE2TEXT_MDL = default_llm[LLM_FACTORY]["image2text_model"]
-
-API_KEY = LLM.get("api_key", "")
-PARSERS = LLM.get(
-    "parsers",
-    "naive:General,qa:Q&A,resume:Resume,manual:Manual,table:Table,paper:Paper,book:Book,laws:Laws,presentation:Presentation,picture:Picture,one:One,audio:Audio,knowledge_graph:Knowledge Graph,email:Email")
-
-# distribution
-DEPENDENT_DISTRIBUTION = get_base_config("dependent_distribution", False)
-RAG_FLOW_UPDATE_CHECK = False
-
-HOST = get_base_config(RAG_FLOW_SERVICE_NAME, {}).get("host", "127.0.0.1")
-HTTP_PORT = get_base_config(RAG_FLOW_SERVICE_NAME, {}).get("http_port")
-
-SECRET_KEY = get_base_config(
-    RAG_FLOW_SERVICE_NAME,
-    {}).get(
-        "secret_key",
-    "infiniflow")
-TOKEN_EXPIRE_IN = get_base_config(
-    RAG_FLOW_SERVICE_NAME, {}).get(
-        "token_expires_in", 3600)
-
-NGINX_HOST = get_base_config(
-    RAG_FLOW_SERVICE_NAME, {}).get(
-        "nginx", {}).get("host") or HOST
-NGINX_HTTP_PORT = get_base_config(
-    RAG_FLOW_SERVICE_NAME, {}).get(
-        "nginx", {}).get("http_port") or HTTP_PORT
-
-RANDOM_INSTANCE_ID = get_base_config(
-    RAG_FLOW_SERVICE_NAME, {}).get(
-        "random_instance_id", False)
-
-PROXY = get_base_config(RAG_FLOW_SERVICE_NAME, {}).get("proxy")
-PROXY_PROTOCOL = get_base_config(RAG_FLOW_SERVICE_NAME, {}).get("protocol")
-
-DATABASE = decrypt_database_config(name="mysql")
-
-# Switch
-# upload
-UPLOAD_DATA_FROM_CLIENT = True
+DATABASE_TYPE = os.getenv("DB_TYPE", 'mysql')
+DATABASE = decrypt_database_config(name=DATABASE_TYPE)
 
 # authentication
-AUTHENTICATION_CONF = get_base_config("authentication", {})
+AUTHENTICATION_CONF = None
 
 # client
-CLIENT_AUTHENTICATION = AUTHENTICATION_CONF.get(
-    "client", {}).get(
+CLIENT_AUTHENTICATION = None
+HTTP_APP_KEY = None
+GITHUB_OAUTH = None
+FEISHU_OAUTH = None
+
+DOC_ENGINE = None
+docStoreConn = None
+
+retrievaler = None
+kg_retrievaler = None
+
+
+def init_settings():
+    global LLM, LLM_FACTORY, LLM_BASE_URL, LIGHTEN, DATABASE_TYPE, DATABASE
+    LIGHTEN = int(os.environ.get('LIGHTEN', "0"))
+    DATABASE_TYPE = os.getenv("DB_TYPE", 'mysql')
+    DATABASE = decrypt_database_config(name=DATABASE_TYPE)
+    LLM = get_base_config("user_default_llm", {})
+    LLM_FACTORY = LLM.get("factory", "Tongyi-Qianwen")
+    LLM_BASE_URL = LLM.get("base_url")
+
+    global CHAT_MDL, EMBEDDING_MDL, RERANK_MDL, ASR_MDL, IMAGE2TEXT_MDL
+    if not LIGHTEN:
+        default_llm = {
+            "Tongyi-Qianwen": {
+                "chat_model": "qwen-plus",
+                "embedding_model": "text-embedding-v2",
+                "image2text_model": "qwen-vl-max",
+                "asr_model": "paraformer-realtime-8k-v1",
+            },
+            "OpenAI": {
+                "chat_model": "gpt-3.5-turbo",
+                "embedding_model": "text-embedding-ada-002",
+                "image2text_model": "gpt-4-vision-preview",
+                "asr_model": "whisper-1",
+            },
+            "Azure-OpenAI": {
+                "chat_model": "gpt-35-turbo",
+                "embedding_model": "text-embedding-ada-002",
+                "image2text_model": "gpt-4-vision-preview",
+                "asr_model": "whisper-1",
+            },
+            "ZHIPU-AI": {
+                "chat_model": "glm-3-turbo",
+                "embedding_model": "embedding-2",
+                "image2text_model": "glm-4v",
+                "asr_model": "",
+            },
+            "Ollama": {
+                "chat_model": "qwen-14B-chat",
+                "embedding_model": "flag-embedding",
+                "image2text_model": "",
+                "asr_model": "",
+            },
+            "Moonshot": {
+                "chat_model": "moonshot-v1-8k",
+                "embedding_model": "",
+                "image2text_model": "",
+                "asr_model": "",
+            },
+            "DeepSeek": {
+                "chat_model": "deepseek-chat",
+                "embedding_model": "",
+                "image2text_model": "",
+                "asr_model": "",
+            },
+            "VolcEngine": {
+                "chat_model": "",
+                "embedding_model": "",
+                "image2text_model": "",
+                "asr_model": "",
+            },
+            "BAAI": {
+                "chat_model": "",
+                "embedding_model": "BAAI/bge-large-zh-v1.5",
+                "image2text_model": "",
+                "asr_model": "",
+                "rerank_model": "BAAI/bge-reranker-v2-m3",
+            }
+        }
+
+        if LLM_FACTORY:
+            CHAT_MDL = default_llm[LLM_FACTORY]["chat_model"] + f"@{LLM_FACTORY}"
+            ASR_MDL = default_llm[LLM_FACTORY]["asr_model"] + f"@{LLM_FACTORY}"
+            IMAGE2TEXT_MDL = default_llm[LLM_FACTORY]["image2text_model"] + f"@{LLM_FACTORY}"
+        EMBEDDING_MDL = default_llm["BAAI"]["embedding_model"] + "@BAAI"
+        RERANK_MDL = default_llm["BAAI"]["rerank_model"] + "@BAAI"
+
+    global API_KEY, PARSERS, HOST_IP, HOST_PORT, SECRET_KEY
+    API_KEY = LLM.get("api_key", "")
+    PARSERS = LLM.get(
+        "parsers",
+        "naive:General,qa:Q&A,resume:Resume,manual:Manual,table:Table,paper:Paper,book:Book,laws:Laws,presentation:Presentation,picture:Picture,one:One,audio:Audio,knowledge_graph:Knowledge Graph,email:Email")
+
+    HOST_IP = get_base_config(RAG_FLOW_SERVICE_NAME, {}).get("host", "127.0.0.1")
+    HOST_PORT = get_base_config(RAG_FLOW_SERVICE_NAME, {}).get("http_port")
+
+    SECRET_KEY = get_base_config(
+        RAG_FLOW_SERVICE_NAME,
+        {}).get("secret_key", str(date.today()))
+
+    global AUTHENTICATION_CONF, CLIENT_AUTHENTICATION, HTTP_APP_KEY, GITHUB_OAUTH, FEISHU_OAUTH
+    # authentication
+    AUTHENTICATION_CONF = get_base_config("authentication", {})
+
+    # client
+    CLIENT_AUTHENTICATION = AUTHENTICATION_CONF.get(
+        "client", {}).get(
         "switch", False)
-HTTP_APP_KEY = AUTHENTICATION_CONF.get("client", {}).get("http_app_key")
-GITHUB_OAUTH = get_base_config("oauth", {}).get("github")
-FEISHU_OAUTH = get_base_config("oauth", {}).get("feishu")
-WECHAT_OAUTH = get_base_config("oauth", {}).get("wechat")
+    HTTP_APP_KEY = AUTHENTICATION_CONF.get("client", {}).get("http_app_key")
+    GITHUB_OAUTH = get_base_config("oauth", {}).get("github")
+    FEISHU_OAUTH = get_base_config("oauth", {}).get("feishu")
 
-# site
-SITE_AUTHENTICATION = AUTHENTICATION_CONF.get("site", {}).get("switch", False)
+    global DOC_ENGINE, docStoreConn, retrievaler, kg_retrievaler
+    DOC_ENGINE = os.environ.get('DOC_ENGINE', "elasticsearch")
+    if DOC_ENGINE == "elasticsearch":
+        docStoreConn = rag.utils.es_conn.ESConnection()
+    elif DOC_ENGINE == "infinity":
+        docStoreConn = rag.utils.infinity_conn.InfinityConnection()
+    else:
+        raise Exception(f"Not supported doc engine: {DOC_ENGINE}")
 
-# permission
-PERMISSION_CONF = get_base_config("permission", {})
-PERMISSION_SWITCH = PERMISSION_CONF.get("switch")
-COMPONENT_PERMISSION = PERMISSION_CONF.get("component")
-DATASET_PERMISSION = PERMISSION_CONF.get("dataset")
-
-HOOK_MODULE = get_base_config("hook_module")
-HOOK_SERVER_NAME = get_base_config("hook_server_name")
-
-ENABLE_MODEL_STORE = get_base_config('enable_model_store', False)
-# authentication
-USE_AUTHENTICATION = False
-USE_DATA_AUTHENTICATION = False
-AUTOMATIC_AUTHORIZATION_OUTPUT_DATA = True
-USE_DEFAULT_TIMEOUT = False
-AUTHENTICATION_DEFAULT_TIMEOUT = 7 * 24 * 60 * 60  # s
-PRIVILEGE_COMMAND_WHITELIST = []
-CHECK_NODES_IDENTITY = False
-
-retrievaler = search.Dealer(ELASTICSEARCH)
-kg_retrievaler = kg_search.KGSearch(ELASTICSEARCH)
+    retrievaler = search.Dealer(docStoreConn)
+    kg_retrievaler = kg_search.KGSearch(docStoreConn)
 
 
 class CustomEnum(Enum):
@@ -226,16 +192,6 @@ class CustomEnum(Enum):
         return [member.name for member in cls.__members__.values()]
 
 
-class PythonDependenceName(CustomEnum):
-    Rag_Source_Code = "python"
-    Python_Env = "miniconda"
-
-
-class ModelStorage(CustomEnum):
-    REDIS = "redis"
-    MYSQL = "mysql"
-
-
 class RetCode(IntEnum, CustomEnum):
     SUCCESS = 0
     NOT_EFFECTIVE = 10
@@ -249,3 +205,5 @@ class RetCode(IntEnum, CustomEnum):
     AUTHENTICATION_ERROR = 109
     UNAUTHORIZED = 401
     SERVER_ERROR = 500
+    FORBIDDEN = 403
+    NOT_FOUND = 404

api/utils/__init__.py

@@ -23,56 +23,64 @@ import socket
 import time
 import uuid
 import requests
+import logging
 from enum import Enum, IntEnum
 import importlib
 from Cryptodome.PublicKey import RSA
 from Cryptodome.Cipher import PKCS1_v1_5 as Cipher_pkcs1_v1_5
-
 from filelock import FileLock
+from api.constants import SERVICE_CONF
 
 from . import file_utils
 
-SERVICE_CONF = "service_conf.yaml"
-
 
 def conf_realpath(conf_name):
     conf_path = f"conf/{conf_name}"
     return os.path.join(file_utils.get_project_base_directory(), conf_path)
 
 
-def get_base_config(key, default=None, conf_name=SERVICE_CONF) -> dict:
+def read_config(conf_name=SERVICE_CONF):
     local_config = {}
     local_path = conf_realpath(f'local.{conf_name}')
-    if default is None:
-        default = os.environ.get(key.upper())
 
+    # load local config file
     if os.path.exists(local_path):
         local_config = file_utils.load_yaml_conf(local_path)
         if not isinstance(local_config, dict):
             raise ValueError(f'Invalid config file: "{local_path}".')
 
-        if key is not None and key in local_config:
-            return local_config[key]
+    global_config_path = conf_realpath(conf_name)
+    global_config = file_utils.load_yaml_conf(global_config_path)
 
-    config_path = conf_realpath(conf_name)
-    config = file_utils.load_yaml_conf(config_path)
+    if not isinstance(global_config, dict):
+        raise ValueError(f'Invalid config file: "{global_config_path}".')
 
-    if not isinstance(config, dict):
-        raise ValueError(f'Invalid config file: "{config_path}".')
+    global_config.update(local_config)
+    return global_config
 
-    config.update(local_config)
-    return config.get(key, default) if key is not None else config
+
+CONFIGS = read_config()
+
+
+def show_configs():
+    msg = f"Current configs, from {conf_realpath(SERVICE_CONF)}:"
+    for k, v in CONFIGS.items():
+        msg += f"\n\t{k}: {v}"
+    logging.info(msg)
+
+
+def get_base_config(key, default=None):
+    if key is None:
+        return None
+    if default is None:
+        default = os.environ.get(key.upper())
+    return CONFIGS.get(key, default)
 
 
 use_deserialize_safe_module = get_base_config(
     'use_deserialize_safe_module', False)
 
 
-class CoordinationCommunicationProtocol(object):
-    HTTP = "http"
-    GRPC = "grpc"
-
-
 class BaseType:
     def to_dict(self):
         return dict([(k.lstrip("_"), v) for k, v in self.__dict__.items()])
@@ -98,6 +106,7 @@ class BaseType:
                 data = obj
             return {"type": obj.__class__.__name__,
                     "data": data, "module": module}
+
         return _dict(self)
 
 
@@ -245,7 +254,7 @@ def get_lan_ip():
             try:
                 ip = get_interface_ip(ifname)
                 break
-            except IOError as e:
+            except IOError:
                 pass
     return ip or ''
 
@@ -342,5 +351,10 @@ def download_img(url):
         return ""
     response = requests.get(url)
     return "data:" + \
-           response.headers.get('Content-Type', 'image/jpg') + ";" + \
-           "base64," + base64.b64encode(response.content).decode("utf-8")
+        response.headers.get('Content-Type', 'image/jpg') + ";" + \
+        "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()

api/utils/api_utils.py

@@ -13,30 +13,33 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import logging
+import functools
 import json
 import random
 import time
+from base64 import b64encode
 from functools import wraps
+from hmac import HMAC
 from io import BytesIO
+from urllib.parse import quote, urlencode
+from uuid import uuid1
+
+import requests
 from flask import (
     Response, jsonify, send_file, make_response,
     request as flask_request,
 )
+from itsdangerous import URLSafeTimedSerializer
 from werkzeug.http import HTTP_STATUS_CODES
 
+from api.db.db_models import APIToken
+from api import settings
+
+from api import settings
+from api.utils import CustomJSONEncoder, get_uuid
 from api.utils import json_dumps
-from api.settings import RetCode
-from api.settings import (
-    REQUEST_MAX_WAIT_SEC, REQUEST_WAIT_SEC,
-    stat_logger, CLIENT_AUTHENTICATION, HTTP_APP_KEY, SECRET_KEY
-)
-import requests
-import functools
-from api.utils import CustomJSONEncoder
-from uuid import uuid1
-from base64 import b64encode
-from hmac import HMAC
-from urllib.parse import quote, urlencode
+from api.constants import REQUEST_WAIT_SEC, REQUEST_MAX_WAIT_SEC
 
 requests.models.complexjson.dumps = functools.partial(
     json.dumps, cls=CustomJSONEncoder)
@@ -50,18 +53,18 @@ def request(**kwargs):
         k.replace(
             '_',
             '-').upper(): v for k,
-        v in kwargs.get(
+                                v in kwargs.get(
             'headers',
             {}).items()}
     prepped = requests.Request(**kwargs).prepare()
 
-    if CLIENT_AUTHENTICATION and HTTP_APP_KEY and SECRET_KEY:
+    if settings.CLIENT_AUTHENTICATION and settings.HTTP_APP_KEY and settings.SECRET_KEY:
         timestamp = str(round(time() * 1000))
         nonce = str(uuid1())
-        signature = b64encode(HMAC(SECRET_KEY.encode('ascii'), b'\n'.join([
+        signature = b64encode(HMAC(settings.SECRET_KEY.encode('ascii'), b'\n'.join([
             timestamp.encode('ascii'),
             nonce.encode('ascii'),
-            HTTP_APP_KEY.encode('ascii'),
+            settings.HTTP_APP_KEY.encode('ascii'),
             prepped.path_url.encode('ascii'),
             prepped.body if kwargs.get('json') else b'',
             urlencode(
@@ -75,7 +78,7 @@ def request(**kwargs):
         prepped.headers.update({
             'TIMESTAMP': timestamp,
             'NONCE': nonce,
-            'APP-KEY': HTTP_APP_KEY,
+            'APP-KEY': settings.HTTP_APP_KEY,
             'SIGNATURE': signature,
         })
 
@@ -94,40 +97,19 @@ def get_exponential_backoff_interval(retries, full_jitter=False):
     return max(0, countdown)
 
 
-def get_json_result(retcode=RetCode.SUCCESS, retmsg='success',
-                    data=None, job_id=None, meta=None):
+def get_data_error_result(code=settings.RetCode.DATA_ERROR,
+                          message='Sorry! Data missing!'):
     import re
     result_dict = {
-        "retcode": retcode,
-        "retmsg": retmsg,
-        # "retmsg": re.sub(r"rag", "seceum", retmsg, flags=re.IGNORECASE),
-        "data": data,
-        "jobId": job_id,
-        "meta": meta,
-    }
-
-    response = {}
-    for key, value in result_dict.items():
-        if value is None and key != "retcode":
-            continue
-        else:
-            response[key] = value
-    return jsonify(response)
-
-
-def get_data_error_result(retcode=RetCode.DATA_ERROR,
-                          retmsg='Sorry! Data missing!'):
-    import re
-    result_dict = {
-        "retcode": retcode,
-        "retmsg": re.sub(
+        "code": code,
+        "message": re.sub(
             r"rag",
             "seceum",
-            retmsg,
+            message,
             flags=re.IGNORECASE)}
     response = {}
     for key, value in result_dict.items():
-        if value is None and key != "retcode":
+        if value is None and key != "code":
             continue
         else:
             response[key] = value
@@ -135,28 +117,25 @@ def get_data_error_result(retcode=RetCode.DATA_ERROR,
 
 
 def server_error_response(e):
-    stat_logger.exception(e)
+    logging.exception(e)
     try:
         if e.code == 401:
-            return get_json_result(retcode=401, retmsg=repr(e))
+            return get_json_result(code=401, message=repr(e))
     except BaseException:
         pass
     if len(e.args) > 1:
         return get_json_result(
-            retcode=RetCode.EXCEPTION_ERROR, retmsg=repr(e.args[0]), data=e.args[1])
-    if repr(e).find("index_not_found_exception") >= 0:
-        return get_json_result(retcode=RetCode.EXCEPTION_ERROR, retmsg="No chunk found, please upload file and parse it.")
-
-    return get_json_result(retcode=RetCode.EXCEPTION_ERROR, retmsg=repr(e))
+            code=settings.RetCode.EXCEPTION_ERROR, message=repr(e.args[0]), data=e.args[1])
+    return get_json_result(code=settings.RetCode.EXCEPTION_ERROR, message=repr(e))
 
 
-def error_response(response_code, retmsg=None):
-    if retmsg is None:
-        retmsg = HTTP_STATUS_CODES.get(response_code, 'Unknown Error')
+def error_response(response_code, message=None):
+    if message is None:
+        message = HTTP_STATUS_CODES.get(response_code, 'Unknown Error')
 
     return Response(json.dumps({
-        'retmsg': retmsg,
-        'retcode': response_code,
+        'message': message,
+        'code': response_code,
     }), status=response_code, mimetype='application/json')
 
 
@@ -188,9 +167,11 @@ def validate_request(*args, **kwargs):
                     error_string += "required argument values: {}".format(
                         ",".join(["{}={}".format(a[0], a[1]) for a in error_arguments]))
                 return get_json_result(
-                    retcode=RetCode.ARGUMENT_ERROR, retmsg=error_string)
+                    code=settings.RetCode.ARGUMENT_ERROR, message=error_string)
             return func(*_args, **_kwargs)
+
         return decorated_function
+
     return wrapper
 
 
@@ -211,17 +192,38 @@ def send_file_in_mem(data, filename):
     return send_file(f, as_attachment=True, attachment_filename=filename)
 
 
-def get_json_result(retcode=RetCode.SUCCESS, retmsg='success', data=None):
-    response = {"retcode": retcode, "retmsg": retmsg, "data": data}
+def get_json_result(code=settings.RetCode.SUCCESS, message='success', data=None):
+    response = {"code": code, "message": message, "data": data}
     return jsonify(response)
 
+def apikey_required(func):
+    @wraps(func)
+    def decorated_function(*args, **kwargs):
+        token = flask_request.headers.get('Authorization').split()[1]
+        objs = APIToken.query(token=token)
+        if not objs:
+            return build_error_result(
+                message='API-KEY is invalid!', code=settings.RetCode.FORBIDDEN
+            )
+        kwargs['tenant_id'] = objs[0].tenant_id
+        return func(*args, **kwargs)
 
-def construct_response(retcode=RetCode.SUCCESS,
-                 retmsg='success', data=None, auth=None):
-    result_dict = {"retcode": retcode, "retmsg": retmsg, "data": data}
+    return decorated_function
+
+
+def build_error_result(code=settings.RetCode.FORBIDDEN, message='success'):
+    response = {"code": code, "message": message}
+    response = jsonify(response)
+    response.status_code = code
+    return response
+
+
+def construct_response(code=settings.RetCode.SUCCESS,
+                       message='success', data=None, auth=None):
+    result_dict = {"code": code, "message": message, "data": data}
     response_dict = {}
     for key, value in result_dict.items():
-        if value is None and key != "retcode":
+        if value is None and key != "code":
             continue
         else:
             response_dict[key] = value
@@ -235,7 +237,8 @@ def construct_response(retcode=RetCode.SUCCESS,
     response.headers["Access-Control-Expose-Headers"] = "Authorization"
     return response
 
-def construct_result(code=RetCode.DATA_ERROR, message='data is missing'):
+
+def construct_result(code=settings.RetCode.DATA_ERROR, message='data is missing'):
     import re
     result_dict = {"code": code, "message": re.sub(r"rag", "seceum", message, flags=re.IGNORECASE)}
     response = {}
@@ -247,7 +250,7 @@ def construct_result(code=RetCode.DATA_ERROR, message='data is missing'):
     return jsonify(response)
 
 
-def construct_json_result(code=RetCode.SUCCESS, message='success', data=None):
+def construct_json_result(code=settings.RetCode.SUCCESS, message='success', data=None):
     if data is None:
         return jsonify({"code": code, "message": message})
     else:
@@ -255,15 +258,99 @@ def construct_json_result(code=RetCode.SUCCESS, message='success', data=None):
 
 
 def construct_error_response(e):
-    stat_logger.exception(e)
+    logging.exception(e)
     try:
         if e.code == 401:
-            return construct_json_result(code=RetCode.UNAUTHORIZED, message=repr(e))
+            return construct_json_result(code=settings.RetCode.UNAUTHORIZED, message=repr(e))
     except BaseException:
         pass
     if len(e.args) > 1:
-        return construct_json_result(code=RetCode.EXCEPTION_ERROR, message=repr(e.args[0]), data=e.args[1])
-    if repr(e).find("index_not_found_exception") >=0:
-        return construct_json_result(code=RetCode.EXCEPTION_ERROR, message="No chunk found, please upload file and parse it.")
+        return construct_json_result(code=settings.RetCode.EXCEPTION_ERROR, message=repr(e.args[0]), data=e.args[1])
+    return construct_json_result(code=settings.RetCode.EXCEPTION_ERROR, message=repr(e))
 
-    return construct_json_result(code=RetCode.EXCEPTION_ERROR, message=repr(e))
+
+def token_required(func):
+    @wraps(func)
+    def decorated_function(*args, **kwargs):
+        authorization_list=flask_request.headers.get('Authorization').split()
+        if len(authorization_list) < 2:
+            return get_json_result(data=False,message="Please check your authorization format.")
+        token = authorization_list[1]
+        objs = APIToken.query(token=token)
+        if not objs:
+            return get_json_result(
+                data=False, message='Token is not valid!', code=settings.RetCode.AUTHENTICATION_ERROR
+            )
+        kwargs['tenant_id'] = objs[0].tenant_id
+        return func(*args, **kwargs)
+
+    return decorated_function
+
+
+def get_result(code=settings.RetCode.SUCCESS, message="", data=None):
+    if code == 0:
+        if data is not None:
+            response = {"code": code, "data": data}
+        else:
+            response = {"code": code}
+    else:
+        response = {"code": code, "message": message}
+    return jsonify(response)
+
+
+def get_error_data_result(message='Sorry! Data missing!', code=settings.RetCode.DATA_ERROR,
+                          ):
+    import re
+    result_dict = {
+        "code": code,
+        "message": re.sub(
+            r"rag",
+            "seceum",
+            message,
+            flags=re.IGNORECASE)}
+    response = {}
+    for key, value in result_dict.items():
+        if value is None and key != "code":
+            continue
+        else:
+            response[key] = value
+    return jsonify(response)
+
+
+def generate_confirmation_token(tenent_id):
+    serializer = URLSafeTimedSerializer(tenent_id)
+    return "ragflow-" + serializer.dumps(get_uuid(), salt=tenent_id)[2:34]
+
+
+def valid(permission,valid_permission,language,valid_language,chunk_method,valid_chunk_method):
+    if valid_parameter(permission,valid_permission):
+        return valid_parameter(permission,valid_permission)
+    if valid_parameter(language,valid_language):
+        return valid_parameter(language,valid_language)
+    if valid_parameter(chunk_method,valid_chunk_method):
+        return valid_parameter(chunk_method,valid_chunk_method)
+
+def valid_parameter(parameter,valid_values):
+    if parameter and parameter not in valid_values:
+       return get_error_data_result(f"'{parameter}' is not in {valid_values}")
+
+def get_parser_config(chunk_method,parser_config):
+    if parser_config:
+        return parser_config
+    if not chunk_method:
+        chunk_method = "naive"
+    key_mapping={"naive":{"chunk_token_num": 128, "delimiter": "\\n!?;。；！？", "html4excel": False,"layout_recognize": True, "raptor": {"use_raptor": False}},
+                 "qa":{"raptor":{"use_raptor":False}},
+                 "resume":None,
+                 "manual":{"raptor":{"use_raptor":False}},
+                 "table":None,
+                 "paper":{"raptor":{"use_raptor":False}},
+                 "book":{"raptor":{"use_raptor":False}},
+                 "laws":{"raptor":{"use_raptor":False}},
+                 "presentation":{"raptor":{"use_raptor":False}},
+                 "one":None,
+                 "knowledge_graph":{"chunk_token_num":8192,"delimiter":"\\n!?;。；！？","entity_types":["organization","person","location","event","time"]},
+                 "email":None,
+                 "picture":None}
+    parser_config=key_mapping[chunk_method]
+    return parser_config