Structlog seems to use __repr__
when creating log messages, which leads to date
and datetime
objects looking like 'datetime.datetime(2018, 9, 20, 10, 1, 52, 414750)'
where '2018-09-20 10:01:52.414750'
would be preferable.
I would have thought there would be an off the shelf processor to handle this, but I can't find one.
This is the solution I came up with, but I'm concerned about the performance given the highly recursive nature of it, and the unnecessary rebuilding of list
and dict
objects.
def _convert_dates(obj):
if isinstance(obj, datetime.date):
# all datetimes are also dates
return str(obj)
elif isinstance(obj, dict):
# Assume dates won't be keys
return {k: _convert_dates(v) for k, v in obj.items()}
elif isinstance(obj, list):
return [_convert_dates(v) for v in obj]
return obj
def dates_to_str_filter(_, __, event):
return _convert_dates(event)
Then include dates_to_str_filter
in processors
for the structlog.configure
call.