T-autolog 1.0.0

Zero-config auto-instrumentation logging for Python

autolog

Zero-config auto-instrumentation logging for Python. Drop one line into your entrypoint and every function call gets logged — inputs, outputs, duration, errors, and trace IDs. No changes to your business logic required.

---

Features

• Zero-touch mode — patch an entire package at import time with one line
• Decorator mode — opt-in per function or class
• Async-native — supports async def, generators, and async generators
• Sensitive data redaction — automatically masks passwords, tokens, API keys
• Trace IDs — async-safe per-request correlation via contextvars
• Structured output — consistent log blocks with inputs, result, duration, error
• FastAPI & Flask middleware — HTTP request/response logging out of the box
• Configurable via env vars — no code changes needed to tune behavior
• Windows-safe — UTF-8 output even on Windows terminals

---

Installation

pip install autolog

With optional framework support:

pip install "autolog[fastapi]"
pip install "autolog[flask]"

Install from source (development):

git clone https://github.com/YOUR_USERNAME/autolog.git
cd autolog
pip install -e ".[dev]"

---

Quick Start

Zero-touch — no changes to your code

Create a run.py next to your project and call autolog.install() before anything else:

import autolog
autolog.install(["myapp"])

# your normal entrypoint, completely untouched
from myapp.main import main
main()

Every function in the myapp package is now logged automatically.

Sample output:

[2026-05-07 14:23:01.042] [INFO ] [billing.services.customers.get_customer]
  ▸ trace    : 3f4a1b2c-...
  ▸ inputs   : {"customer_id": "cust_123"}
  ▸ result   : {"id": "cust_123", "name": "Alice"}
  ▸ duration : 0.0031 s
────────────────────────────────────────────────

---

Usage

Zero-touch (import hook)

import autolog

autolog.install(
    ["myapp", "utils"],   # list of packages to patch
    service="billing",    # optional: label shown in logs
    level="DEBUG",        # optional: minimum log level
    log_file="app.log",   # optional: mirror logs to a file
)

Must be called before importing the packages you want to log.

---

Decorator — opt-in per function

from autolog import log_function

@log_function
def calculate_tax(amount: float, rate: float) -> float:
    return amount * rate

@log_function
async def fetch_invoice(invoice_id: str):
    ...

# with options
@log_function(service="billing", level="DEBUG", log_file="billing.log")
def process_payment(payload: dict):
    ...

Works with sync, async, generators, and async generators.

---

Decorator — opt-in per class

from autolog import log_class

@log_class
class InvoiceService:
    def create(self, data): ...
    async def fetch(self, invoice_id): ...
    def _internal(self): ...     # skipped — underscore methods are excluded

Wraps all public methods. Handles @staticmethod, @classmethod, and @property correctly.

---

Skip specific functions

from autolog import no_log

@no_log
def hot_path():
    """This function will never be wrapped, even inside an auto-patched package."""
    ...

---

Trace IDs

Correlate all log entries for a single request:

from autolog import new_trace_id, get_trace_id

# at the start of a request / task
trace_id = new_trace_id()   # generates UUID4, stores in ContextVar

# anywhere downstream
trace_id = get_trace_id()   # returns current ID or "-"

Trace IDs are async-safe — each coroutine/task has its own context.

---

Framework Middleware

FastAPI

from fastapi import FastAPI
from autolog.middleware.fastapi import AutoLogMiddleware

app = FastAPI()
app.add_middleware(AutoLogMiddleware, service="my-api")

Logs method, path, status code, inputs, and response for every HTTP request. Uses BackgroundTask so logging never blocks the response.

---

Flask

from flask import Flask
from autolog.middleware.flask import init_autolog

app = Flask(__name__)
init_autolog(app, service="my-api")

---

Configuration

All options can be set via environment variables — no code changes needed:

Env var	Default	Description
AUTOLOG_LEVEL	INFO	Minimum log level (DEBUG, INFO, WARNING, ERROR)
AUTOLOG_FILE	—	Path to mirror logs to a file
AUTOLOG_SERVICE	package name	Service label shown in log lines
AUTOLOG_DISABLE	—	Set to 1 to disable all wrapping (e.g. in tests)
NO_COLOR	—	Set to 1 to disable ANSI color output

Example:

AUTOLOG_LEVEL=DEBUG AUTOLOG_FILE=app.log python run.py

---

Sensitive Data Redaction

The following argument names are automatically redacted in both inputs and outputs:

password, passwd, token, secret, key, auth, credential, api_key, apikey, jwt, bearer

def login(username, password):
    ...

# logged as:
# ▸ inputs   : {"username": "alice", "password": "[REDACTED]"}

Redaction is recursive — works inside nested dicts, lists, and tuples.

---

Log Output Format

Every log entry is a structured block:

[TIMESTAMP] [LEVEL] [module.function]
  ▸ trace    : <uuid>
  ▸ inputs   : {"arg": value, ...}
  ▸ result   : <value>           (on success)
  ▸ error    : <exception repr>  (on failure)
  ▸ duration : 0.0031 s
────────────────────────────────────────────────

---

Running Tests

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

---

License

MIT