Logging as the caller module in Python

Question

I have the following application structure:

./utils.py
def do_something(logger=None):
    if not logger:
        logger = logging.getLogger(__name__)

    print('hello')
    logger.debug('test')

./one.py
from utils import do_something
logger = logging.getLogger(__name__)
do_something(logger=logger)

./two.py
from utils import do_something
logger = logging.getLogger(__name__)
do_something(logger=logger)

Now, then this runs, the logging output will show the names of the respective modules that are using the functions (one and two) rather than utils. I then use this information (the logger's name) to filter the messages.

Is there a way to do this without having to pass the logger though as an argument? Would I have to basically introspect the caller function and initialize a logger based on that? What's the common solution for this type of recipe, when one function is used all over the code base but the loggers should be of those calling the function?

I created a gist to demonstrate. Only tested in Python 3.7 so far. gist.github.com/rgacote/f6a3381c71f6808ebf2893793e22f3ae — rgacote
– rgacote, Commented Apr 1, 2020 at 20:13

Elis Byberi · Accepted Answer · 2017-11-20 15:13:41Z

1

Variable logger being a global, can be accessed from inside do_something() function like this:

logger = logging.getLogger(__name__)
def do_something():
    x = logger

After reading this carefully:

What's the common solution for this type of recipe, when one function is used all over the code base but the loggers should be of those calling the function?

In simple english would be:

How to access global variable logger from imported function do_something()?

I conclude that there is no other way!
You have to pass logger as an argument for this particular case.

from utils import do_something
logger = logging.getLogger(__name__)

do_something()  # logger global is not visible in do_something()
do_something(logger=logger)  # the only way

edited Nov 20, 2017 at 15:13

answered Nov 20, 2017 at 13:16

Elis Byberi

1,4511 gold badge11 silver badges22 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

mart1n Over a year ago

Not sure I understand. I don't want do_something to be in one.py or two.py. At the same time, I want the records in logs to have the names of those files rather than utils.

Elis Byberi Over a year ago

It is not possible for an imported function to access global variables of the module where it [function] is imported.

Elis Byberi Over a year ago

@mart1n You have to pass logger as an argument.

mart1n Over a year ago

Ok, passing as an argument seems to be the only way. I thought there would be a more elegant solution provided by the logging module. Thanks for your help!

Stop harming Monica · Accepted Answer · 2017-11-20 14:28:14Z

1

Passing the logger as an argument looks fine to me. But if you have many calls to do_something then passing the same logger in each call is not only too verbose but prone to errors --you might pass the wrong logger by mistake. Also, while you do need to make do_something aware of the logger, doing it at the point where you are calling it probably does not make sense --you want to do something, not mess with loggers. OOP to the rescue: make do_something a method of a class and pass the logger when instantiating the class, then you do not need to pass it to the method.

# utils.py
class UsefulObject:
    def __init__(self, logger):
        self._logger = logger

    def do_something(self):
        print('hello')
        self._logger.debug('test')

Now in the client module you create an instance of UsefulObject and you can call .do_something on it without passing the logger every time.

# one.py
from utils import UsefulObject
logger = logging.getLogger(__name__)
useful_object = UsefulObject(logger)  # do this once
# ...
useful_object.do_something()  # do this many times, do not pass the logger

answered Nov 20, 2017 at 14:28

Stop harming Monica

12.7k1 gold badge40 silver badges63 bronze badges

1 Comment

mart1n Over a year ago

Great alternative! Thank you!

Alex Hall · Accepted Answer · 2019-02-26 18:37:03Z

0

If you don't like the existing solutions, here's a magic way to do it:

def do_something():
    logger = inspect.currentframe().f_back.f_globals['log']
    ...

The calling function must then have a global variable called log.

answered Feb 26, 2019 at 18:37

Alex Hall

36.2k5 gold badges63 silver badges98 bronze badges

Collectives™ on Stack Overflow

Logging as the caller module in Python

3 Answers 3

4 Comments

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

4 Comments

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related