Alter python logger stack level

Question

Very often when writing framework code, I would prefer the caller's line number and file name to be logged. For example, if I detect improper use of a framerwork level API call, I would like to log that.... not as an in-framework error but as "caller error".

This only comes into play when writing low level libraries and systems that use introspectionn.

Is there any way to get the logger to log "one level up"? Can I create a custom LogRecord and then modify it and use it within the installed loggers somehow? Trying to figure this out.

For example something like this:

def set(self, x):
   if x not in self._cols:
      log.error("Invalid attribute for set", stack_level=-1)

Answer 1

This is easy to find on google now since it was added to python 3.8, and it was mentioned in the other answer but here is a better explanation. You were close with the stack_level argument, but it should be stacklevel and the value should be 2. For example you can do:

import logging

def fun():
    logging.basicConfig(format='line %(lineno)s : %(message)s')
    log = logging.getLogger(__name__)
    log.error('Something is wrong', stacklevel=2)

if __name__ == '__main__':
    fun()

This will output:

line 9 : Something is wrong

If the stacklevel was set to the default of 1, it would output:

line 6 : Something is wrong

From the documentation: https://docs.python.org/3/library/logging.html#logging.Logger.debug

The third optional keyword argument is stacklevel, which defaults to 1. If greater than 1, the corresponding number of stack frames are skipped when computing the line number and function name set in the LogRecord created for the logging event. This can be used in logging helpers so that the function name, filename and line number recorded are not the information for the helper function/method, but rather its caller. The name of this parameter mirrors the equivalent one in the warnings module.