RetryHTTP

1.4.0 · active · verified Fri Apr 10

RetryHTTP is a Python library designed to simplify retrying potentially transient HTTP errors. It extends the `tenacity` library with custom retry and wait strategies, offering sensible defaults while remaining fully customizable. The library provides decorators to automatically retry HTTP requests in applications using `requests`, `httpx`, or `aiohttp`.

Warnings

gotcha Be cautious when retrying non-idempotent HTTP methods like POST, PUT, or DELETE by default. While `retryhttp` has sensible defaults for many scenarios, blindly retrying these can lead to unintended side effects (e.g., duplicate resource creation). Always understand your API's idempotency guarantees.
Fix: Carefully configure `retry` decorator parameters (`retry_exceptions`, `retry_on_exception`, `retry_codes`) or wrap non-idempotent calls with more restrictive retry policies if there's any doubt about idempotency.
gotcha For full integration and client-specific error handling, remember to install `retryhttp` with the relevant optional dependencies (e.g., `pip install retryhttp[httpx]` for HTTPX support). Without these, `retryhttp` may still catch generic exceptions but won't be able to leverage client-specific error types for more granular retry logic.
Fix: Install the necessary client-specific extras: `pip install retryhttp[httpx]`, `pip install retryhttp[requests]`, or `pip install retryhttp[aiohttp]` depending on your chosen HTTP client library.
gotcha If customizing retry behavior, be aware that `retryhttp` extends `tenacity`. While it provides its own decorator, deeper customizations might involve understanding `tenacity`'s retry strategies, wait strategies, and stop conditions. Ensure your custom logic correctly integrates with `retryhttp`'s opinionated defaults.
Fix: Refer to both `retryhttp`'s documentation for its provided decorators and `tenacity`'s documentation for advanced customization options when creating complex retry policies.

Install

pip install retryhttp Full support for HTTPX, requests, and aiohttp
pip install retryhttp[httpx] HTTPX-only support
pip install retryhttp[requests] requests-only support
pip install retryhttp[aiohttp] aiohttp-only support

Imports

retry
```
from retryhttp import retry
```

Quickstart

This quickstart demonstrates how to use the `@retry` decorator with `httpx`. The decorated function will automatically retry on transient HTTP status codes (429, 500, 502, 503, 504), network errors, and timeouts, with appropriate backoff strategies. The `raise_for_status()` call is crucial to convert HTTP error responses into exceptions that `retryhttp` can intercept and retry.

import httpx
from retryhttp import retry

@retry
def fetch_data_with_retries():
    """Fetches data from example.com, retrying on transient HTTP errors."""
    print("Attempting to fetch data...")
    try:
        response = httpx.get("https://example.com/")
        response.raise_for_status() # Raise an exception for HTTP errors (4xx or 5xx)
        print("Successfully fetched data!")
        return response.text
    except httpx.HTTPStatusError as e:
        print(f"HTTP error occurred: {e.response.status_code}")
        raise # Re-raise to trigger retry logic if it's a retryable status
    except httpx.RequestError as e:
        print(f"Request error occurred: {e}")
        raise # Re-raise to trigger retry logic on network errors/timeouts

if __name__ == "__main__":
    data = fetch_data_with_retries()
    # In a real scenario, you'd process 'data' here
    # For this example, we just show successful fetch or final failure

view raw JSON →