macos下 ragflow二次开发环境搭建

1. 参考官网链接
   https://ragflow.io/docs/dev/launch_ragflow_from_source
2. 虚拟环境

git clone https://github.com/infiniflow/ragflow.git
cd ragflow/
# if not pipx, please install it at first
pip3 install pipxpipx install uv
uv sync --python 3.10 --all-extras 

3. 安装 elk, redis, mysql等docker

docker compose -f docker/docker-compose-base.yml up -d

4. .env配置
   将 .env中主机指向docker host主机的IP
   去掉macos=1的注释
5. service_conf.yaml.template配置
   将5个资源节点的host指向docker host主机的IP，直接写IP地址
   如：
   mysql:
   name: ‘${MYSQ{L}_{D}BNAME:-ra{g}_{f}low}^{\prime }user{:}^{\prime }${MYSQL_USER:-root}’
   password: ‘${MYSQL_PASSWORD:-infini_rag_flow}’
   host: ‘101.15.113.113’
   port: 5455
6. pyprojct.toml修改
   因为macos下 torch包与linux要求不同。

 	  "xgboost==1.6.0", # 原始1.5.0"debugpy>=1.8.13","threadpoolctl>=3.6.0", //增加[project.optional-dependencies]
full = ["bcembedding==0.1.5","fastembed>=0.3.6,<0.4.0; sys_platform == 'darwin' or platform_machine != 'x86_64'","fastembed-gpu>=0.3.6,<0.4.0; sys_platform != 'darwin' and platform_machine == 'x86_64'","flagembedding==1.2.10","torch>=2.0.0,<2.2.2; sys_platform == 'darwin' or platform_machine != 'x86_64'","torch>=2.4.0,<2.5.1; sys_platform != 'darwin' and platform_machine == 'x86_64'","transformers>=4.35.0,<5.0.0"
]

7. 安装其它包，避免报错 Fatal Python error: Segmentation fault

source .venv/bin/activate
brew install libjemalloc
pip3 install aiohttp==3.11.13
pip3 install google-cloud-aiplatform==1.64.0
pip3 install akracer==0.0.13

8. 修改 docker/entrypoint.sh
   因为macos下不是.so动态库

function task_exe() {local consumer_id="$1"local host_id="$2"JEMALLOC_PATH="$(pkg-config --variable=libdir jemalloc)/libjemalloc.2.dylib"#JEMALLOC_PATH="$(pkg-config --variable=libdir jemalloc)/libjemalloc.so"while true; do#LD_PRELOAD="$JEMALLOC_PATH" \DYLD_INSERT_LIBRARIES "$JEMALLOC_PATH" \"$PY" rag/svr/task_executor.py "${host_id}_${consumer_id}"done
}

9. vscode 修改 lanuch.json
   新增 .vscode/launch.json

{"version": "0.2.0","configurations": [{"name": "Python: ragflow_server","type": "debugpy","request": "launch","program": "${workspaceFolder}/api/ragflow_server.py","justMyCode": true,"cwd": "${workspaceFolder}", // 设置工作目录为项目根目录"console": "integratedTerminal","env": {"HF_ENDPOINT": "https://hf-mirror.com","PYTHONPATH": "${workspaceFolder}","VIRTUAL_ENV": "${workspaceFolder}/.venv",  // 设定虚拟环境路径"PATH": "${workspaceFolder}/.venv/bin:${env:PATH}",  // 将虚拟环境的 bin 目录添加到 PATH},"envFile": "${workspaceFolder}/docker/.env","args": []},{"name": "Python: task_executor","type": "debugpy","request": "launch","program": "${workspaceFolder}/rag/svr/task_executor.py","justMyCode": true,"cwd": "${workspaceFolder}", // 设置工作目录为项目根目录"console": "integratedTerminal","env": {"HF_ENDPOINT": "https://hf-mirror.com","PYTHONPATH": "${workspaceFolder}","VIRTUAL_ENV": "${workspaceFolder}/.venv",  // 设定虚拟环境路径"PATH": "${workspaceFolder}/.venv/bin:${env:PATH}",  // 将虚拟环境的 bin 目录添加到 PATH//"LD_PRELOAD": "$(pkg-config --variable=libdir jemalloc)/libjemalloc.so", // linux"DYLD_INSERT_LIBRARIES": "/usr/local/Cellar/jemalloc/5.3.0/lib/libjemalloc.2.dylib", // macos},"envFile": "${workspaceFolder}/docker/.env","args": []}]}

10. 其它参考官网