Python Logger module with a class and rotating log handling

Question

I was creating a set of .py files into a 'package' that I wanted all to import a standardized 'logger' that would allow all modules to log to a single file.

class BaseLogger():
def __init__(self, module_name, specific_task_name):
    logging.basicConfig(
        filename="app." + specific_task_name + "." + module_name + ".log",
        format="%(levelname)-10s;%(asctime)s;%(module)s;%(funcName)s;%(message)s;%(thread)s;%(threadName)s",
        #level=logging.DEBUG
    )
    self.baselogger = logging.getLogger("app." + module_name + "." + specific_task_name )
    self.baselogger.setLevel('DEBUG')

So the idea is that all my .py files, import BaseLogger and instantiate that object and it'll find a current logger by same name, therefore LOG to one file...

Problem is i'm now trying to create a rotating log structure and just lost... I've added to .basicConfig [failed to iterable], I'm thinking I should do a self.baselogger.addHandler .. but that makes the log file go empty....

How can I create a rotating log file with my above 'classed' implementation... or have I been doing it wrong?...

Thanks

Answer 1

You do not need a class wrapper as logging module is pretty awesome by design. What you need to do is to register/get a logger and configure it with a format and file handler.

Here is an example with a wrapper:

class BaseLogger(object):

    def __init__(self, module_name, specific_task_name):
        filename = "app." + specific_task_name + "." + module_name + ".log",
        format = ("%(levelname)-10s;%(asctime)s;%(module)s;%(funcName)s;"
                  "%(message)s;%(thread)s;%(threadName)s"),
        # you do not have to keep the reference
        # it is possible to get logger from anywhere using getLogger
        self.logger = logging.getLogger(specific_task_name)
        self.logger.setLevel(logging.INFO)

        # make sure to add RotatingFileHandler only once
        # you might need to update condition if you have more than 1 handler
        if not self.logger.handlers:
            fh = handlers.RotatingFileHandler(filename,
                                              mode='a',
                                              maxBytes=MAX_LOG_FILE_SIZE,
                                              backupCount=2,
                                              encoding=None,
                                              delay=0)
            fh.setLevel(logging.INFO)
            formatter = logging.Formatter(format)
            fh.setFormatter(formatter)
            self.logger.addHandler(fh)

I would drop a class and go with a fabric function:

def get_logger(name):
    # setup new logger here if it does not exist 
    # return instance if it is there
    pass