Python logging module is printing lines multiple times

后端 未结 4 711
南旧
南旧 2021-01-01 17:53

I have the following code:

import logging
class A(object):
    def __init__(self):
        self._l = self._get_logger()

    def _get_logger(self):
        l         


        
相关标签:
4条回答
  • 2021-01-01 18:22

    logger is created once, but multiple handlers are created.

    Create A once.

    a = A()
    for msg in ["hey", "there"]:
        a.p(msg)
    

    Or change _get_logger as follow:

    def _get_logger(self):
        loglevel = logging.INFO
        l = logging.getLogger(__name__)
        if not getattr(l, 'handler_set', None):
            l.setLevel(loglevel)
            h = logging.StreamHandler()
            f = logging.Formatter('%(asctime)s %(levelname)s %(message)s')
            h.setFormatter(f)
            l.addHandler(h)
            l.setLevel(loglevel)
            l.handler_set = True
        return l  
    

    UPDATE

    Since Python 3.2, you can use logging.Logger.hasHandlers to see if this logger has any handlers configured. (thanks @toom)

    def _get_logger(self):
        loglevel = logging.INFO
        l = logging.getLogger(__name__)
        if not l.hasHandlers():
            ...
        return l
    
    0 讨论(0)
  • 2021-01-01 18:28

    In my case , the root loggers handler were also being called , All I did was to set propagate attribute of logger instance to False.

    import logging
    logger = logging.getLogger("MyLogger")
    
    # stop propagting to root logger
    logger.propagate = False
    
    # other log configuration stuff
    # ....
    
    0 讨论(0)
  • 2021-01-01 18:28

    Best to set up the logger at the module level outside all classes and functions to avoid having it repeatedly setting up the handler.

    For use cases where that's unavoidable, check the number of handlers already attached to the named logger, or better still check presence of the handler in the list. In Python 2.7.6, Logger's class attribute handlers is a list of the handlers set up on the Logger class instance. Don't attach a handler that's already in the list. E.g.

    >>> import logging
    >>> import logging.handlers # lib of log record handlers
    >>> dir(logging.handlers)   # all the handlers from the lib
    ['BaseRotatingHandler', 'BufferingHandler', 'DEFAULT_HTTP_LOGGING_PORT', 'DEFAULT_SOAP_LOGGING_PORT', 'DEFAULT_TCP_LOGGING_PORT', 'DEFAULT_UDP_LOGGING_PORT', 'DatagramHandler', 'HTTPHandler', 'MemoryHandler', 'NTEventLogHandler', 'RotatingFileHandler', 'SMTPHandler', 'ST_DEV', 'ST_INO', 'ST_MTIME', 'SYSLOG_TCP_PORT', 'SYSLOG_UDP_PORT', 'SocketHandler', 'SysLogHandler', 'TimedRotatingFileHandler', 'WatchedFileHandler', '_MIDNIGHT', '__builtins__', '__doc__', '__file__', '__name__', '__package__', '_unicode', 'cPickle', 'codecs', 'errno', 'logging', 'os', 're', 'socket', 'struct', 'time']
    >>> l = logging.getLogger()   # root logger
    >>> l.addHandler(logging.handlers.TimedRotatingFileHandler)
    >>> l.addHandler(logging.handlers.WatchedFileHandler)
    >>> l.handlers  # handlers set up on the logger instance
    [<class 'logging.handlers.TimedRotatingFileHandler'>, <class 'logging.handlers.WatchedFileHandler'>]
    >>> logging.handlers.TimedRotatingFileHandler in l.handlers # check - Yes
    True
    >>> logging.handlers.WatchedFileHandler in l.handlers # check - Yes
    True
    >>> logging.handlers.HTTPHandler in l.handlers # check - No
    False
    >>> 
    
    0 讨论(0)
  • 2021-01-01 18:35

    Since python 3.2 and newer:

    Consider using hasHandlers() to check if a logger has handlers or not.

    https://docs.python.org/3/library/logging.html#logging.Logger.hasHandlers

    0 讨论(0)
提交回复
热议问题