Blog2MD

English | 中文文档

Blog2MD is an intelligent tool designed to convert web articles (from platforms like WeChat, Toutiao, Juejin, and more) into clean, local Markdown files.

Unlike traditional converters that rely solely on regex or simple HTML parsing, Blog2MD leverages Large Language Models (LLMs) (such as OpenAI GPT-4, DeepSeek, Qwen) to:

1. Intelligently format the content into high-quality Markdown.
2. Automatically categorize articles based on their content.
3. Preserve and download images locally for offline access.

It comes with a modern Web Interface for easy management, monitoring, and reading.

(Replace with actual screenshot if available)

• 🤖 LLM-Powered Conversion: Uses AI to strip clutter and format text perfectly.
• 📂 Auto-Categorization: Automatically organizes articles into folders like "Technology", "Finance", "News", etc.
• 🖼️ Image Archiving: Downloads all images to a local assets folder and updates links, ensuring your markdown is truly offline-ready.
• 🌍 Multi-Platform Support:
  • WeChat Official Accounts (微信公众号)
  • Toutiao (今日头条) - Supports mobile/short links & redirects
  • Juejin (掘金)
  • Generic (Fallbacks for other websites)
• 🖥️ Modern Web UI:
  • Built with FastAPI & Tailwind CSS.
  • Dark Mode support.
  • Real-time Monitoring of conversion tasks via WebSockets.
  • PDF Export: Convert your archived Markdown articles to PDF with proper font support.
  • Management: View, read, and delete categories or articles directly from the browser.

• Python 3.8 or higher.
• An API Key for an LLM provider (OpenAI, DeepSeek, DashScope/Qwen, etc.).

1. Clone the repository:

   git clone https://github.com/your-username/blog2md.git
   cd blog2md

2. Install dependencies:

   pip install -r requirements.txt

3. Configure Environment: Copy the example environment file and edit it:

   cp .env.example .env

   Open .env and set your LLM provider and API keys. Example:

   # Options: openai, deepseek, qwen, kimi, doubao, custom
   LLM_PROVIDER=openai
   
   # OpenAI Configuration
   OPENAI_API_KEY=sk-your-key-here
   OPENAI_MODEL=gpt-4o-mini
   
   # Or for DeepSeek
   # LLM_PROVIDER=deepseek
   # DEEPSEEK_API_KEY=sk-your-deepseek-key
   # DEEPSEEK_MODEL=deepseek-chat

1. Build and Run with Docker Compose (Recommended):

   docker-compose up -d

2. Access the application: http://localhost:8000

3. Custom Configuration:

   • Port: Change the port by setting the PORT environment variable.

     PORT=9000 docker-compose up -d

     Then access at http://localhost:9000
   • Output Mapping: The output directory is automatically mapped to ./output in your project root.
   • Fonts: If you need custom fonts, you can uncomment the fonts volume mapping in docker-compose.yml.

Start the web server:

python server.py

Open your browser and navigate to: http://localhost:8000

• Click "Fetch URL" to add a new article.
• View your knowledge base in the home page.
• Monitor background tasks in the "Monitor" tab.

Easily send articles from your phone to Blog2MD while on the same Wi-Fi.

Note: Replace 192.168.1.82 in the examples below with your computer's actual LAN IP address.

1. Open the Shortcuts app on your iPhone.
2. Tap "+" to create a new shortcut.
3. Add action "Get Contents of URL".
   • URL: http://192.168.1.82:8000/api/quick_add?url=
   • After the =, select "Shortcut Input" from the variable bar.
4. Tap the "i" icon (or shortcut name) and enable "Show in Share Sheet".
5. Tap "Receive" settings at the top:
   • Set to receive "Safari web pages" or "URLs".
6. Name it "Save to Blog2MD".

Usage: In Safari/Apps, tap Share -> "Save to Blog2MD".

1. Install HTTP Shortcuts app.
2. Create a new shortcut.
3. Basic Settings:
   • Name: Save to Blog2MD
   • URL: http://192.168.1.82:8000/api/quick_add?url={share_text}
4. Request Method: GET
5. Enable "Show in Share Sheet".

Create a bookmark in your mobile browser with the following URL (Address):

javascript:(function(){window.location.href='http://192.168.1.82:8000/api/quick_add?url='+encodeURIComponent(window.location.href);})();

Usage: While viewing an article, tap the address bar and select this bookmark.

You can also use the tool directly from the terminal:

python main.py https://mp.weixin.qq.com/s/some-article-url

Or simply run python main.py and paste the URL when prompted.

All converted articles are saved in the output directory:

output/
├── Technology/
│   ├── Understanding_LLMs/
│   │   ├── index.md
│   │   └── assets/
│   │       ├── image1.jpg
│   │       └── image2.png
├── Finance/
│   ├── Market_Trends_2025/
│   │   ├── index.md
│   │   └── assets/
│       ...

• Backend: Python, FastAPI, Uvicorn
• Frontend: HTML5, Jinja2 Templates, Tailwind CSS, JavaScript
• AI/LLM: LangChain (integrating OpenAI, DeepSeek, DashScope, etc.)
• Scraping: Requests, BeautifulSoup4
• Utilities: Markdown, xhtml2pdf

MIT License