Derafu: http-ping ("A tank for a fly")

Perform HTTP requests (GET, POST, …) and get status code, response body and timing. Includes automatic retry with exponential backoff. Useful for scheduled triggers, health checks, and smoke tests.

Basic use case: scheduled trigger (e.g. Celery beat)

On AWS Elastic Beanstalk, Celery's built-in beat only runs on the leader instance. With auto-scaling, the leader can change and beat may not run, so scheduled tasks never fire.

Approach: Expose an HTTP endpoint that triggers your tasks. Call that endpoint on a schedule from AWS Lambda (e.g. via EventBridge). This project provides the Lambda code and a small http-ping module that performs the request and returns the result.

You keep responsibility for ensuring each logical task runs only once in your own code when handling the request.

Classes

HttpRequest

Value object holding the request configuration:

from http_ping import HttpRequest

request = HttpRequest(
    url="https://example.com/api/beat",
    method="POST",                    # default: "GET"
    headers={"Accept": "application/json"},
    body={"task": "run"},             # serialized as JSON; default: None
    timeout=30.0,                     # seconds; default: 30.0
    auth="Token abc123",              # full Authorization header value; default: None
)

HttpAuth

Helpers to build common Authorization header values:

from http_ping import HttpAuth

HttpAuth.token("abc123")             # → "Token abc123"
HttpAuth.bearer("abc123")            # → "Bearer abc123"
HttpAuth.basic("user", "pass")       # → "Basic dXNlcjpwYXNz"

Pass the result to HttpRequest(auth=...).

HttpPing

Executes an HttpRequest with optional retry and exponential backoff:

from http_ping import HttpAuth, HttpPing, HttpRequest

request = HttpRequest(
    url="https://example.com/api/beat",
    auth=HttpAuth.token("abc123"),
)
result = HttpPing(request, retries=3, backoff=1.0).run()

run() retries on network errors and 5xx responses. Backoff between retries: backoff, backoff*2, backoff*4, …

Returns a dict:

{
  "status_code": 200,
  "body": { "…": "…" },
  "elapsed_seconds": 0.234,
  "attempts": 1
}

body is parsed JSON if possible, otherwise a string.

HttpPingBatch

Executes a list of HttpRequest objects sequentially, sharing the same retry config. Individual failures are captured per URL without aborting the rest:

from http_ping import HttpAuth, HttpPingBatch, HttpRequest

requests = [
    HttpRequest(url="https://example.com/api/health", auth=HttpAuth.token("abc123")),
    HttpRequest(url="https://example.com/api/beat",   auth=HttpAuth.token("abc123")),
]
results = HttpPingBatch(requests, retries=3, backoff=1.0).run()

Returns a list of dicts, one per URL. Each dict includes a "url" key. On failure, "error" replaces "status_code"/"body"/"elapsed_seconds":

[
  {"url": "https://…/health", "status_code": 200, "body": {}, "elapsed_seconds": 0.1, "attempts": 1},
  {"url": "https://…/beat",   "error": "Connection timeout", "attempts": 3}
]

Console usage

Install the project first (e.g. python3 -m venv .venv && .venv/bin/pip install .), then:

# GET, no auth.
python -m http_ping "https://example.com/health"

# Multiple URLs (output is a list).
python -m http_ping "https://example.com/health" "https://example.com/api/beat"

# POST with bearer token.
HTTP_PING_METHOD=POST HTTP_PING_BEARER=abc123 python -m http_ping "https://example.com/api/run"

# Token auth via env var.
HTTP_PING_TOKEN=abc123 python -m http_ping "https://example.com/api/beat"

# Full Authorization header value (any scheme).
HTTP_PING_AUTH="Token abc123" python -m http_ping "https://example.com/api/beat"

# All options via env vars (space-separated URLs for multiple).
export HTTP_PING_URL="https://example.com/api/health https://example.com/api/beat"
export HTTP_PING_TOKEN="abc123"
export HTTP_PING_RETRIES=5
export HTTP_PING_BACKOFF=2.0
python -m http_ping

CLI arguments override env vars:

python -m http_ping --help
python -m http_ping "https://example.com/api/beat" --method POST --timeout 10 --retries 2 --backoff 0.5

Environment variables

HTTP_PING_AUTH takes precedence over HTTP_PING_TOKEN and HTTP_PING_BEARER.

Makefile

• make all – install the project into function/ and build http-ping-function.zip (for Lambda).
• make clean – remove the zip and everything in function/ except lambda_function.py.
• make dist – build sdist and wheel (python -m build).
• make upload – upload with twine (run after make dist).

Note on project structure

This project is intentionally over-structured for what it does ("a tank for a fly"). Its real purpose is to serve as a reference for real Python projects, following modern standards and conventions end to end:

• Package (http_ping/) — proper Python package with __init__.py exposing the public API, __main__.py as CLI entry point, and internal modules with clear separation of concerns: data (HttpRequest), helpers (HttpAuth), and logic (HttpPing, HttpPingBatch).
• Adapters (function/lambda_function.py, http_ping/__main__.py) — thin layers that read environment variables and wire up the core classes. The business logic knows nothing about env vars or deployment targets.
• Linting (pyproject.toml → [tool.ruff]) — ruff configured with an explicit rule set, enforcing style, correctness, complexity and docstring conventions.
• Tests (tests/) — pytest with coverage, showing both unit tests (pure functions, no I/O) and integration tests (real HTTP calls).
• CI (.github/workflows/ci.yml) — runs lint and tests on every push across multiple Python versions.
• Docs (docs/, .github/workflows/docs.yml, .readthedocs.yaml) — Sphinx with autodoc, deployed automatically to GitHub Pages and Read the Docs.
• Packaging (pyproject.toml, Makefile) — make lambda builds the deployable AWS Lambda zip; make dist / make upload build and publish to PyPI. These targets are kept as reference even though this project is not published.