python-logfmter icon indicating copy to clipboard operation
python-logfmter copied to clipboard

Enhancement: include extra fields in formatter

Open matteo-zanoni opened this issue 2 years ago • 4 comments

Currently, the formatter lacks the ability to include static extra fields in the log output. It is possible to include extra fields in each call (.info, .error...) but this is not always the most ergonomic way. This enhancement proposes adding support for including extra fields within the formatter configuration.

Proposed Solution

Introduce a new feature in the logging formatter that allows users to specify extra fields to be included in the log output. This could be achieved by extending the existing formatter configuration with an optional parameter in the form of a dictionary of key and value of the extra fields.

Example

import logging
import logfmter

logger = logging.getLogger(__name__)

handler = logging.StreamHandler()
formatter = logfmter.Logfmter(
    keys=["level", "module", "custom_field1", "custom_field2"],
    mapping={
        "level": "levelname",
        "module": "name",
    },
    default_extra: {
        "custom_field1": "aaa",
         "custom_field2": "bbb",
})
handler.setFormatter(formatter)
logger.addHandler(handler)

logger.info("the fields in default_extra will be included")

Impact

This enhancement is backward-compatible and does not affect existing code that does not utilize the new feature. It provides an opt-in mechanism for users to take advantage of the enhanced formatter.

Related Issues/PRs

#10

matteo-zanoni avatar Jan 23 '24 15:01 matteo-zanoni

@matteo-zanoni I added a guide section which talks about adding pairs via the log record factory. I am not totally convinced that's not the most pythonic way of solving this problem.

josheppinette avatar Jan 23 '24 20:01 josheppinette

@jteppinette I totally agree with you, using the log factory in this manner seems hacky to me too. If you think this proposal has a more elegant API I would be happy to implement it and open a PR. Let me know

matteo-zanoni avatar Jan 24 '24 07:01 matteo-zanoni

As commented on the suggested PR in https://github.com/jteppinette/python-logfmter/pull/12#issuecomment-2612197942 This can often be solved better by a filter or if they are dynamic a LoggerAdapter

tapetersen avatar Jan 24 '25 10:01 tapetersen