OpenAI API Proxy with SOCKS5 Support

This is a Flask-based proxy server that forwards requests to the OpenAI API through a SOCKS5 proxy. It supports both regular and streaming requests and preserves request headers and authorization tokens.

Features

• Proxies all OpenAI API requests under /v1/* paths.
• Supports GET and POST methods (including streaming via Server-Sent Events).
• Forwards client Authorization and other headers.
• Routes traffic through a configurable SOCKS5 proxy (with optional authentication).
• Adds request IDs and client IP info to logs for tracing.
• Simple setup using environment variables.

---

Requirements

• Python 3.7+
• pip packages:
  • Flask
  • Requests with SOCKS support
  • python-dotenv

Install dependencies with:

pip install flask requests[socks] python-dotenv

---

Configuration

Create a .env file in the project root with the following configurations:

# OpenAI API base URL (default is official OpenAI API)
OPENAI_BASE_URL=https://api.openai.com/v1

# SOCKS5 proxy settings (required)
SOCKS5_PROXY_HOST=your.socks5.proxy.host
SOCKS5_PROXY_PORT=1080

# SOCKS5 proxy authentication (optional)
SOCKS5_PROXY_USERNAME=your_proxy_username
SOCKS5_PROXY_PASSWORD=your_proxy_password

# Flask server settings
FLASK_HOST=0.0.0.0
FLASK_PORT=8868

# Debug mode (set to True or False)
FLASK_DEBUG=False

• SOCKS5_PROXY_USERNAME and SOCKS5_PROXY_PASSWORD can be omitted if your SOCKS5 proxy does not require authentication.
• Adjust FLASK_HOST and FLASK_PORT for your deployment environment.

---

Running the Proxy

Run the proxy using:

python openai_proxy.py

For production deployment, it is recommended to run with a WSGI server such as Gunicorn:

gunicorn -w 4 -b 0.0.0.0:8868 openai_proxy:app

---

Usage

Send your OpenAI API requests to this proxy instead of directly to api.openai.com. For example:

Request URL:

http://localhost:8868/v1/chat/completions

Headers:

Authorization: Bearer YOUR_OPENAI_API_KEY
Content-Type: application/json

The proxy will forward the request through the configured SOCKS5 proxy and return the OpenAI API response transparently.

---

Streaming Support

The proxy supports OpenAI streaming completions:

• When you include "stream": true in the request JSON body, the proxy will stream the response back to you as Server-Sent Events (SSE).
• The HTTP response will have the Content-Type: text/event-stream header.
• Clients should read chunks progressively to see partial generation results.

---

Logging

• Each request is logged with a unique Request ID and the client IP address.
• Logs include info-level messages for proxied requests and warnings/errors for failures.

---

Notes

• Ensure your SOCKS5 proxy is reachable and properly configured.
• The script requires the requests[socks] package and underlying PySocks.
• This proxy assumes clients provide the OpenAI Authorization header.
• The proxy does not currently implement rate limiting or additional authentication.

---

License

This project is provided as-is without warranty. Adapt and use at your own risk.

---

Feel free to open issues or contribute improvements!