I've just released logot - a log capture and assertion plugin 🚀

[etianen]

Hello! 👋 I've just released logot, which as the title says, is a log capture and assertion plugin.

logot has a few unique things, such as being logging-framework-agnostic and having support for testing highly concurrent code using threads or async. But those things are relatively niche. What I'd like to show here are a few examples of how it can be a nice caplog replacement, even in "normal" synchronous code.

As a caplog replacement

Here's a really simple example testing that a piece of code logs as expected:

from logot import Logot, logged

def test_something(logot: Logot) -> None:
    do_something()
    logot.assert_logged(logged.info("Something was done"))

You'll see a couple of things here. A logot fixture, and a logged API. Use these together to make neat little log assertions. The equivalent code using caplog would be:

def test_something(caplog: pytest.LogCaptureFixture) -> None:
    do_something()
    assert any(
        record.levelno == logging.INFO and record.message == "Something was done"
        for record in caplog.records
    )

I think the logot code is clearer, and hopefully you do too! 🤗

Log message matching

One of logots more useful features is the ability to match log messages using %-style placeholders rather than regex. This syntax was chosen to be as close as possible to the % placeholders used by the stdlib logging library.

from logot import Logot, logged

def test_something(logot: Logot) -> None:
    do_something()
    # Match a string placeholder with `%s`.
    logot.assert_logged(logged.info("Something %s done"))

The equivalent using caplog gets pretty verbose:

import re

def test_something(caplog: pytest.LogCaptureFixture) -> None:
    do_something()
    assert any(
        record.levelno == logging.INFO and re.fullmatch("Something .*? done", record.message, re.DOTALL)
        for record in caplog.records
    )

If your message contained everyday punctuation like ., you have to start worrying about regex escaping too! I hope that %-style message matching gives a clearer, more loggy way of matching log messages.

Log pattern matching

This feature is generally aimed towards testing code using threads or async, where messages can arrive out-of-order. But it's also useful for testing synchronous code.

from logot import Logot, logged

def test_app(logot: Logot) -> None:
   do_something()
   logot.wait_for(logged.info("Something happened") | logged.error("Something broke!"))

This example tests whether the INFO log "Something happened" or the ERROR "Something broke!" was emitted, and passes on either. The equivalent using caplog gets quite long:

def test_something(caplog: pytest.LogCaptureFixture) -> None:
    do_something()
    assert any(
        (record.levelno == logging.INFO and record.message == "Something happened")
        or (record.levelno == logging.ERROR and record.message == "Something broke!")
        for record in caplog.records
    )

I hope you like it! ❤️

This is only a v1 release, but it's building on a lot of ideas I've been developing in different projects for a while now. I hope you like it, and find it useful.

The project documentation and pytest integeration guide are there if you'd like to find out more. 🙇