MinerU Dify Plugin

GitHub

MinerU is a tool that converts PDFs into machine-readable formats (e.g., markdown, JSON), allowing for easy extraction into any format.

MinerU is a document parser that can parse complex document data for any downstream LLM use case (RAG, agents)

GitHub - opendatalab/MinerU: A high-quality tool for convert PDF to Markdown and JSON.

Key Features

• Remove headers, footers, footnotes, page numbers, etc., to ensure semantic coherence.
• Output text in human-readable order, suitable for single-column, multi-column, and complex layouts.
• Preserve the structure of the original document, including headings, paragraphs, lists, etc.
• Extract images, image descriptions, tables, table titles, and footnotes.
• Automatically recognize and convert formulas in the document to LaTeX format.
• Automatically recognize and convert tables in the document to HTML format.
• Automatically detect scanned PDFs and garbled PDFs and enable OCR functionality.
• OCR supports detection and recognition of 84 languages.
• Supports multiple output formats, such as multimodal and NLP Markdown, JSON sorted by reading order, and rich intermediate formats.
• Supports various visualization results, including layout visualization and span visualization, for efficient confirmation of output quality.
• Supports running in a pure CPU environment, and also supports GPU(CUDA)/NPU(CANN)/MPS acceleration
• Compatible with Windows, Linux, and Mac platforms.

What's New?

• Support the official API of MinerU

• Local Deploy corresponds to MinerU release 2.5

• The supported input file types have been increased to include PDF, DOC, DOCX, PPT, PPTX, PNG, JPG, and JPEG.

• Remove the "Replace Markdown Image Path" tool. Now, the image paths in the Markdown will be automatically replaced with previewable URLs (the validity period of the URL is determined by the FILES_ACCESS_TIMEOUT in dify.env).If you want to use this feature, please update the Dify's core code.

• Supports more export formats (HTML, DOC, LaTeX). The download links for the additional formats will be stored in the files of the output variables.

DEMO DSL

You can download the YAML file and import it into Dify. This demo includes a basic capability demonstration of the plugin.

demo_dsl.yml

Getting Started

1. Using the Official MinerU API

The version 0.0.2 can now support the official API of MinerU.

Configuration Steps

1. Log into your Dify platform.

2. Go to "Tools" -> "Plugin Market", search for "MinerU" plugin and add it.

3. Configure the MinerU plugin parameters:

   • Base URL of the MinerU API service: https://mineru.net
   • API token: Get your API token from MinerU
   • Service Type: Select "MinerU Official API"

4. Save your configuration.

Input Parameters

parameter	type	required	example	description
enable formula recognition	bool	false	true	Whether to enable formula recognition, the default is true
enable table recognition	bool	false	true	Whether to enable table recognition, the default is true
document language	string	false	ch	Specify the document language, the default ch, can be set to auto, when it is auto, the model will automatically recognize the document language, see the list of other optional values for details：PaddleOCR
enable ocr recognition	bool	false	true	Whether to start the ocr function, the default is false
extra export formats	[string]	false	["docx","html"]	Markdown and json are the default export formats without setting. This parameter only supports one or more of the three formats of docx, html, and latex.
model version	string	false	vlm	MinerU model version; options: pipeline or vlm

Output Variables

The plugin provides five types of output for each processed file:

text : The parsed Markdown text

files: The extra export formats files(html,docx,latex)

json: The parsed content list

full_zip_url: Only for Official API, the zip URL of the complete parsed result

images: The images extracted from the PDF

2. Using a Locally Deployed MinerU Service

Version 0.3.1 of the plugin corresponds to MinerU release 2.1.1.

Prerequisites

1. Get your local IP address: For Dify to correctly access the MinerU API, you need to use your LAN IP address (Do NOT use 127.0.0.1 or localhost). Get your IP address based on your operating system:

   • Windows: Open Command Prompt, run ipconfig, and look for "IPv4 Address".
   • macOS: Open Terminal, run:
     • Wireless network: ipconfig getifaddr en0
     • Wired network: ipconfig getifaddr en1
   • Linux: Open Terminal, run ip hostname -I or ip addr.

   Note your IP address, for example: 192.168.1.100

2. Deploy the MinerU Web API project: Follow the instructions here: MinerU/projects/web_api/README.md at magic_pdf-1.2.2-released · opendatalab/MinerU · GitHub

Configuration Steps

1. Log into your Dify platform.

2. Go to "Tools" -> "Plugin Market", search for "MinerU" plugin and add it.

3. Configure the MinerU plugin parameters:

   • Base URL of the locally deployed MinerU service: http://YOUR_LOCAL_IP:8888 (e.g., http://192.168.1.100:8888)
   • Token: Leave empty (not required for local deployment)
   • Service Type: Select "Local Deployment"

   Note: Ensure that the Dify service can access this base URL.

4. Save your configuration.

Input Parameters

Parameter	Type	Required	Default	Description
file	file	Yes	-	File to be parsed
parse_method	select	Yes	auto	Parsing method, can be auto, ocr, or txt.

**Note: Other parameters are invalid for the local deployment version. **

Output Variables

Same as the Official API output variables (see above).

3. Important: Dify Environment Configuration (FILES_URL)

To ensure the MinerU plugin can properly handle file uploads, you need to configure the FILES_URL setting in Dify:

1. Find your Dify deployment directory and edit the .env file.

2. Modify the FILES_URL configuration based on your deployment method:

   • Docker Compose deployment: Set FILES_URL to http://api:5001
   • Other deployment methods: Set FILES_URL to http://DIFY_HOST_IP:5001 (e.g., http://192.168.1.101:5001, where the IP is usually the machine running Dify)

3. Confirm that the Dify API service's port 5001 is exposed externally (check port mapping in the docker-compose.yaml file).

4. After saving the .env file, restart the Dify service for the configuration to take effect:

   docker compose down
   docker compose up -d
   

Tips:

• If you use your local IP for the FILES_URL, your IP address may change when your network environment changes (e.g., connecting to a different WiFi). When this happens, you'll need to:

  • Get your new local IP address
  • Update the MinerU plugin's Base URL configuration in Dify
  • If necessary (if Dify's FILES_URL is configured with an IP address rather than api), update Dify's .env file and restart the Dify service

• When using the Dify MinerU plugin, especially when processing file uploads, if you don't configure this step, you may encounter errors like Request URL is missing an 'http://' or 'https://' protocol. This usually occurs because Dify's api service cannot correctly access its own file service.

FAQs

1. What should I do if I encounter the following error during execution?

Run failed: Failed to transform tool message: PluginInvokeError: {"args":{},"error_type":"Exception","message":"Error extracting page from PDF: Request URL is missing an 'http://' or 'https://' protocol."}

---

Please follow the instructions above to configure the settings accordingly, and this issue will be resolved.

https://github.com/langgenius/dify/issues/16327

Credits

This plugin is powered by [MinerU](GitHub - opendatalab/MinerU: A high-quality tool for convert PDF to Markdown and JSON.)