python多个日志模块怎么隔离_带有多个模块的Python日志记录

I have got various modules in which I use Python logging heavily. When I import them into the main module like in the Python documentation and try to run it, I don't get any output from the logging. Has anyone got any idea what is happening?

Logging is called in the module imported by the public module imported below (the piece of code is too large to be put up here). The piece of code below is where the whole program is run and the logging is initialized:

import logging

from bottle import run, debug

import public

logging.basicConfig(level=logging.DEBUG)

if __name__ == '__main__':

logging.info('Started')

debug(mode=True)

run(host='localhost', port = 8080, reloader=True)

logging.info('Finished')

解决方案

Your problem is probably being caused by the import public statement making a call to logging.debug(...) or similar. What then happens is this:

You import public. As a side-effect, this calls e.g. logging.debug or similar, which automatically calls basicConfig, which adds a StreamHandler to the root logger, but doesn't change the level.

You then call basicConfig, but as the root logger already has a handler, it does nothing (as documented).

Since the default logging level is WARNING, your info and debug calls produce no output.

You really should avoid side-effects on import: for example, your call to basicConfig should be in the if __name__ == '__main__' clause. With this public.py:

import logging

def main():

logging.debug('Hello from public')

and this main.py:

import logging

from bottle import run, debug

import public

def main():

logging.basicConfig(level=logging.DEBUG)

logging.info('Started')

debug(mode=True)

public.main()

run(host='localhost', port = 8080, reloader=True)

logging.info('Finished')

if __name__ == '__main__':

main()

You get the following output:

$ python main.py

INFO:root:Started

DEBUG:root:Hello from public

INFO:root:Started

DEBUG:root:Hello from public

Bottle server starting up (using WSGIRefServer())...

Listening on http://localhost:8080/

Hit Ctrl-C to quit.

^CINFO:root:Finished

$ Shutdown...

INFO:root:Finished

You'll see from this that Bottle actually re-runs the script in a separate process, which accounts for the doubling up of messages. You can illustrate this by using a format string which shows the process ID: if you use

logging.basicConfig(level=logging.DEBUG,

format='%(process)s %(levelname)s %(message)s')

then you get output like

$ python main.py

13839 INFO Started

13839 DEBUG Hello from public

13840 INFO Started

13840 DEBUG Hello from public

Bottle server starting up (using WSGIRefServer())...

Listening on http://localhost:8080/

Hit Ctrl-C to quit.

^C13839 INFO Finished

$ Shutdown...

13840 INFO Finished

Note that if you add a side-effect producing statement to public.py like this:

logging.debug('Side-effect from public')

at the module level, then you get no logging output at all:

$ python main.py

Bottle server starting up (using WSGIRefServer())...

Listening on http://localhost:8080/

Hit Ctrl-C to quit.

^C$ Shutdown...

which appears to confirm the above analysis.