pengjunjie/ragflow
1
0
Fork 1
mirror of https://github.com/infiniflow/ragflow.git synced 2025-12-08 20:42:30 +08:00
Code Issues Packages Projects Releases Wiki Activity

Feat: supports MinerU http-client/server method (#10961)

Browse Source 
### What problem does this PR solve?

Add support for MinerU http-client/server method.

To use MinerU with vLLM server:

1. Set up a vLLM server running MinerU:
   ```bash
   mineru-vllm-server --port 30000
   ```

2. Configure the following environment variables:
- `MINERU_EXECUTABLE=/ragflow/uv_tools/.venv/bin/mineru` (or the path to
your MinerU executable)
   - `MINERU_BACKEND="vlm-http-client"`
   - `MINERU_SERVER_URL="http://your-vllm-server-ip:30000"`

3. Follow the standard MinerU setup steps as described above.

With this configuration, RAGFlow will connect to your vLLM server to
perform document parsing, which can significantly improve parsing
performance for complex documents while reducing the resource
requirements on your RAGFlow server.



![1](https://github.com/user-attachments/assets/46624a0c-0f3b-423e-ace8-81801e97a27d)

![2](https://github.com/user-attachments/assets/66ccc004-a598-47d4-93cb-fe176834f83b)


### Type of change

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

---------

Co-authored-by: writinwaters <cai.keith@gmail.com>
This commit is contained in:
Yongteng Lei
2025-11-04 16:03:30 +08:00
committed by GitHub
parent 03038c7d3d
commit 2677617f93
4 changed files with 120 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
docs/faq.mdx
View File

@ -540,11 +540,38 @@ uv pip install -U "mineru[core]" -i https://mirrors.aliyun.com/pypi/simple
### How to configure MinerU-specific settings?
1. Set `MINERU_EXECUTABLE` (default: `mineru`) to the path of the MinerU executable.
1. Set `MINERU_EXECUTABLE` (default: `mineru`) to the path to the MinerU executable.
2. Set `MINERU_DELETE_OUTPUT` to `0` to keep MinerU's output. (Default: `1`, which deletes temporary output)
3. Set `MINERU_OUTPUT_DIR` to specify the output directory for MinerU.
4. Set `MINERU_BACKEND` to `"pipeline"`. (Options: `"pipeline"` (default) | `"vlm-transformers"`)
4. Set `MINERU_BACKEND` to specify a parsing backend:
- `"pipeline"` (default): The traditional multimodel pipeline.
- `"vlm-transformers"`: A vision-language model using HuggingFace Transformers.
- `"vlm-vllm-engine"`: A vision-language model using local vLLM engine (requires a local GPU).
- `"vlm-http-client"`: A vision-language model via HTTP client to remote vLLM server (RAGFlow only requires CPU).
5. If using the `"vlm-http-client"` backend, you must also set `MINERU_SERVER_URL` to the URL of your vLLM server.
:::tip NOTE
For information about other environment variables natively supported by MinerU, see [here](https://opendatalab.github.io/MinerU/usage/cli_tools/#environment-variables-description).
:::
---
### How to use MinerU with a vLLM server for document parsing?
RAGFlow supports MinerU's `vlm-http-client` backend, enabling you to delegate document parsing tasks to a remote vLLM server. With this configuration, RAGFlow will connect to your remote vLLM server as a client and use its powerful GPU resources for document parsing. This significantly improves performance for parsing complex documents while reducing the resources required on your RAGFlow server. To configure MinerU with a vLLM server:
1. Set up a vLLM server running MinerU:
```bash
mineru-vllm-server --port 30000
```
2. Configure the following environment variables in your **docker/.env** file:
- `MINERU_EXECUTABLE=/ragflow/uv_tools/.venv/bin/mineru` (or the path to your MinerU executable)
- `MINERU_BACKEND="vlm-http-client"`
- `MINERU_SERVER_URL="http://your-vllm-server-ip:30000"`
3. Complete the rest standard MinerU setup steps as described [here](#how-to-configure-mineru-specific-settings).
:::tip NOTE
When using the `vlm-http-client` backend, the RAGFlow server requires no GPU, only network connectivity. This enables cost-effective distributed deployment with multiple RAGFlow instances sharing one remote vLLM server.
:::