pygofastproxy 1.1.0

A blazing-fast HTTP proxy for Python, powered by Go’s fasthttp library

pygofastproxy

A high-performance HTTP reverse proxy for Python, powered by Go's fasthttp. Three lines of Python to put a Go-speed proxy with built-in security in front of your app.

from pygofastproxy import run_proxy

run_proxy(target="http://localhost:4000", port=8080)

┌────────┐       ┌──────────────────────┐       ┌──────────┐
│        │       │    pygofastproxy     │       │          │
│ Client │──────▶│    (Go binary)       │──────▶│ Backend  │
│        │◀──────│                      │◀──────│ (Flask,  │
└────────┘       │  + Security headers  │       │  Django, │
                 │  + Per-IP rate limit  │       │  FastAPI)│
                 │  + CORS              │       └──────────┘
                 │  + TLS termination   │
                 └──────────────────────┘

Why pygofastproxy?

Python web frameworks are great for building apps, but they weren't designed to be internet-facing reverse proxies. The typical solution is deploying Nginx or Caddy alongside your app — but that means extra infrastructure, config files, and moving parts.

pygofastproxy gives you Go-level proxy performance from a pip install:

	pygofastproxy	proxy.py	Nginx	mitmproxy
pip install + 3 lines of code	Yes	No (plugin system)	No (separate install)	No (inspection tool)
Performance	Go (fasthttp)	Python	C	Python
Security headers by default	Yes	No	No (manual config)	No
Per-IP rate limiting built-in	Yes	No	Module	No
CORS handling built-in	Yes	No	No	No
TLS termination	Yes	Yes	Yes	Yes
Zero Python dependencies	Yes	Yes	N/A	No

Best for: Python developers who want a fast, secure reverse proxy without deploying separate infrastructure.

Not designed for: Traffic inspection/debugging (use mitmproxy), load balancing across multiple backends (use Nginx/Traefik), or HTTP/2/WebSocket proxying (fasthttp limitation).

Quick Start

1. Install:

   pip install pygofastproxy
   

2. Start your backend (e.g., Flask on port 4000)

3. Run the proxy:

   from pygofastproxy import run_proxy
   
   run_proxy(target="http://localhost:4000", port=8080)
   

4. Send requests to http://localhost:8080 — they'll be proxied to your backend with security headers, rate limiting, and CORS handling added automatically.

Requirements: Python 3.8+ and Go (for building the proxy binary on first run). The binary is cached and only rebuilt when source changes.

Configuration

Python API

run_proxy(
    target="http://localhost:4000",
    port=8080,
    rate_limit_rps=5000,
    allowed_origins="https://yourdomain.com,https://app.yourdomain.com",
    tls_cert_file="/path/to/cert.pem",
    tls_key_file="/path/to/key.pem",
)

Parameter	Type	Default	Description
target	str	"http://localhost:4000"	Backend URL to proxy to
port	int	8080	Proxy listen port
rate_limit_rps	int	1000	Per-IP requests/sec (0 = unlimited)
max_conns_per_host	int	1000	Max concurrent backend connections
read_timeout	str	"10s"	Read timeout (e.g., "30s", "1m")
write_timeout	str	"10s"	Write timeout
max_idle_conn_duration	str	"60s"	Idle connection TTL
max_request_body_size	int	10485760	Max request body in bytes (10MB)
allowed_origins	str	None	Comma-separated CORS origins
cors_allow_credentials	bool	False	Enable Access-Control-Allow-Credentials
tls_cert_file	str	None	TLS certificate path (enables HTTPS)
tls_key_file	str	None	TLS private key path

Environment Variables

All settings can also be configured via environment variables (useful for Docker and CLI usage):

PY_BACKEND_TARGET=http://localhost:4000
PY_BACKEND_PORT=8080
PROXY_RATE_LIMIT_RPS=5000
PROXY_MAX_CONNS_PER_HOST=2000
PROXY_MAX_CONN_WAIT_TIMEOUT=5s
PROXY_READ_TIMEOUT=30s
PROXY_WRITE_TIMEOUT=30s
PROXY_MAX_IDLE_CONN_DURATION=60s
PROXY_READ_BUFFER_SIZE=16384
PROXY_WRITE_BUFFER_SIZE=16384
PROXY_MAX_REQUEST_BODY_SIZE=20971520
ALLOWED_ORIGINS=https://yourdomain.com
PROXY_CORS_ALLOW_CREDENTIALS=false
PROXY_TLS_CERT_FILE=/path/to/cert.pem
PROXY_TLS_KEY_FILE=/path/to/key.pem

Run via CLI: python -m pygofastproxy

Security Features

All security features are enabled by default with zero configuration:

• Per-IP rate limiting — Token bucket algorithm, 1000 req/s per IP by default. Stale buckets are cleaned up automatically.
• Security headers — Every response gets X-Content-Type-Options: nosniff, X-Frame-Options: DENY, X-XSS-Protection: 1; mode=block. Cache-Control: no-store is added only if the backend didn't set its own.
• Request size limits — Rejects bodies over 10MB by default (configurable).
• Hop-by-hop header stripping — Removes Connection, Proxy-Authorization, Transfer-Encoding, and any additional headers listed in the Connection header value (RFC 7230).
• X-Forwarded headers — Always overwrites X-Forwarded-For, X-Forwarded-Proto, and X-Forwarded-Host to prevent client spoofing.
• CORS protection — When allowed_origins is set, only listed origins get CORS headers. The null origin is always rejected. Access-Control-Allow-Credentials is off by default.
• No information leakage — Internal backend URLs and error details are never exposed to clients.
• TLS termination — Optional HTTPS support via cert/key configuration.
• Private target warnings — Logs a warning if the backend target is a private/loopback address.
• Connection backpressure — MaxConnWaitTimeout prevents goroutine accumulation when the backend is slow.

Benchmarks

Run the included benchmark to measure throughput on your hardware:

pip install pygofastproxy
python benchmarks/benchmark.py

You can also use external tools for more rigorous testing:

# Using hey (https://github.com/rakyll/hey)
hey -n 10000 -c 50 http://localhost:8080/

# Using wrk (https://github.com/wg/wrk)
wrk -t4 -c50 -d10s http://localhost:8080/

The proxy uses Go's fasthttp library, which is designed for high throughput with minimal allocations. All header byte slices are pre-allocated, the rate limiter uses per-bucket locks instead of a global mutex, and CORS origins are refreshed on a timer rather than per-request.

Docker

docker compose up --build

The Dockerfile uses a multi-stage build (Go builder + slim Python runtime) and runs as a non-root user. Configure via environment variables in .env or docker-compose.yml.

Testing

pip install -e ".[test]"
pytest tests/ -v

The test suite covers proxy forwarding, HTTP methods, security headers, CORS, rate limiting, health endpoint, and configuration validation.

Architecture

pygofastproxy is a hybrid Python+Go project:

• Python layer (pygofastproxy/) — Package API, CLI, Go binary lifecycle management. Python never touches HTTP traffic.
• Go layer (pygofastproxy/go/) — The actual proxy, built on fasthttp. Handles all request forwarding, security, rate limiting, and CORS.

The Go binary is compiled automatically on first use via runner.py and cached. It's only rebuilt when .go, go.mod, or go.sum files change. Go is only required at build time — if the binary already exists (e.g., in Docker), Go is not needed.

Roadmap

• Pre-compiled wheels — Ship platform-specific wheels with the Go binary already built, removing the Go requirement entirely. Pure pip install with no build step. (Track progress)

Contributing

See CONTRIBUTING.md for guidelines on reporting bugs, suggesting features, and submitting pull requests.

License

MIT — Copyright 2026 Laurent Zogaj

Powered by fasthttp by valyala.