Add console logger

[samet-akcay]

Description

• Add python console logger and replace print

• Fixes Add proper logging and remove print statements with logger via logger = logging.getLogger("pytorch_lightning") #54

Changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

Checklist

• My code follows the pre-commit style and check guidelines of this project.
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing tests pass locally with my changes

[ashwinvaidya17]

Looks good! Just two minor comments.

[ashwinvaidya17]

Is this log-level passed to the console logger somewhere?

[samet-akcay]

Not yet, it's a bit more involved than we thought. It's something I was going to bring up. This could be either addressed in a separate PR, or this one needs to be modified a little bit.

[djdameln]

Maybe we could rename this configure_logger and instead of having it return the logger, just define logger = logging.getLogger("anomalib") in train.py and the other entrypoitns. That way the logging in those entrypoint files would be more in line with the other submodules.

[samet-akcay]

the problem is that we already have a logger defined for experiment loggers. This is the main reason for not being able to use logger name. You could refer to this line, for further details.

I like the name configure_logger, but again in line 68, we already have get_logger

[samet-akcay]

This would work if the existing logger were renamed to something like experiment_logger or similar

[djdameln]

I think we could rename the experiment logger to experiment_logger. This would make sense since the function is already renamed to get_experiment_logger in your PR.

[samet-akcay]

Do you mean something like the following in train.py and other entrypoints?

...
configure_logger()
logger = logging.getLogger("anomalib")
...

If yes, this would cost us two lines for the sake of following the standard logging definition.

We could maybe rename this to get_logger instead, which would be a one liner in this case. And the experiment logger would also be renamed accordingly.

...
logger = get_logger("anomalib", level=logging.INFO)
...
experiment_logger = get_experiment_logger(config)

[djdameln]

If yes, this would cost us two lines for the sake of following the standard logging definition.

It would add a single line logger = logging.getLogger("anomalib"), but this line would be placed at the top of the file, right under the imports. So it may be an additional line, but it does not clutter the train function.

Also this same line would be removed from the get_logger/configure_logger function, which would end up a lot cleaner, because it would also only require a single argument (log_level), no logic for setting the logger name, and no return type.

[djdameln]

Thanks!