Installation

This library is available on PyPI and can be installed with

python3 -m pip install log-with-context

Usage

This library provides a wrapped Python logging.Logger that adds a shared context to each logging message, passed as the extra parameter.

You will need an additional library (like JSON-log-formatter) to actually output the logging messages. We avoided putting this functionality in this library to keep it lightweight and flexible. We assumed that you already have a preferred way to format your logging messages.

import logging
import logging.config

from log_with_context import add_logging_context, Logger

logging.config.dictConfig({
    "version": 1,
    "disable_existing_loggers": True,
    "formatters": {
        "json": {"()": "json_log_formatter.JSONFormatter"},
    },
    "handlers": {
        "console": {
            "formatter": "json",
            "class": "logging.StreamHandler",
        }
    },
    "loggers": {
        "": {"handlers": ["console"], "level": "INFO"},
    },
})

LOGGER = Logger(__name__)

LOGGER.info("First message. No context")

with add_logging_context(current_request="hi"):
    LOGGER.info("Level 1")

    with add_logging_context(more_info="this"):
        LOGGER.warning("Level 2")

    LOGGER.info("Back to level 1")

LOGGER.error("No context at all...")

The above program logs the following messages to standard error:

{"message": "First message. No context", "time": "2021-04-08T16:37:23.126099"}
{"current_request": "hi", "message": "Level 1", "time": "2021-04-08T16:37:23.126336"}
{"current_request": "hi", "more_info": "this", "message": "Level 2", "time": "2021-04-08T16:37:23.126389"}
{"current_request": "hi", "message": "Back to level 1", "time": "2021-04-08T16:37:23.126457"}
{"message": "No context at all...", "time": "2021-04-08T16:37:23.126514"}

This example may look trivial, but it is very handy to maintain a logging context up and down a Python call stack without having to pass additional variables to the functions and methods that you call.

Implementation details

Logging contexts are stored as thread-local variables. If you want to share information between threads, you must create a Logging context in each thread with the same information.

Similarly, logging contexts are deliberately not copied when creating subprocesses. This is done to minimize bugs and make sure that log-with-context behaves in the exact same manner across operating systems.