1 :mod:`logging` --- Logging facility for Python
2 ==============================================
5 :synopsis: Flexible error logging system for applications.
8 .. moduleauthor:: Vinay Sajip <vinay_sajip@red-dove.com>
9 .. sectionauthor:: Vinay Sajip <vinay_sajip@red-dove.com>
12 .. index:: pair: Errors; logging
16 This module defines functions and classes which implement a flexible error
17 logging system for applications.
19 Logging is performed by calling methods on instances of the :class:`Logger`
20 class (hereafter called :dfn:`loggers`). Each instance has a name, and they are
21 conceptually arranged in a namespace hierarchy using dots (periods) as
22 separators. For example, a logger named "scan" is the parent of loggers
23 "scan.text", "scan.html" and "scan.pdf". Logger names can be anything you want,
24 and indicate the area of an application in which a logged message originates.
26 Logged messages also have levels of importance associated with them. The default
27 levels provided are :const:`DEBUG`, :const:`INFO`, :const:`WARNING`,
28 :const:`ERROR` and :const:`CRITICAL`. As a convenience, you indicate the
29 importance of a logged message by calling an appropriate method of
30 :class:`Logger`. The methods are :meth:`debug`, :meth:`info`, :meth:`warning`,
31 :meth:`error` and :meth:`critical`, which mirror the default levels. You are not
32 constrained to use these levels: you can specify your own and use a more general
33 :class:`Logger` method, :meth:`log`, which takes an explicit level argument.
39 The key benefit of having the logging API provided by a standard library module
40 is that all Python modules can participate in logging, so your application log
41 can include messages from third-party modules.
43 It is, of course, possible to log messages with different verbosity levels or to
44 different destinations. Support for writing log messages to files, HTTP
45 GET/POST locations, email via SMTP, generic sockets, or OS-specific logging
46 mechanisms are all supported by the standard module. You can also create your
47 own log destination class if you have special requirements not met by any of the
53 .. sectionauthor:: Doug Hellmann
54 .. (see <http://blog.doughellmann.com/2007/05/pymotw-logging.html>)
56 Most applications are probably going to want to log to a file, so let's start
57 with that case. Using the :func:`basicConfig` function, we can set up the
58 default handler so that debug messages are written to a file::
61 LOG_FILENAME = '/tmp/logging_example.out'
62 logging.basicConfig(filename=LOG_FILENAME,level=logging.DEBUG,)
64 logging.debug('This message should go to the log file')
66 And now if we open the file and look at what we have, we should find the log
69 DEBUG:root:This message should go to the log file
71 If you run the script repeatedly, the additional log messages are appended to
72 the file. To create a new file each time, you can pass a *filemode* argument to
73 :func:`basicConfig` with a value of ``'w'``. Rather than managing the file size
74 yourself, though, it is simpler to use a :class:`RotatingFileHandler`::
78 import logging.handlers
80 LOG_FILENAME = '/tmp/logging_rotatingfile_example.out'
82 # Set up a specific logger with our desired output level
83 my_logger = logging.getLogger('MyLogger')
84 my_logger.setLevel(logging.DEBUG)
86 # Add the log message handler to the logger
87 handler = logging.handlers.RotatingFileHandler(
88 LOG_FILENAME, maxBytes=20, backupCount=5)
90 my_logger.addHandler(handler)
94 my_logger.debug('i = %d' % i)
96 # See what files are created
97 logfiles = glob.glob('%s*' % LOG_FILENAME)
99 for filename in logfiles:
102 The result should be 6 separate files, each with part of the log history for the
105 /tmp/logging_rotatingfile_example.out
106 /tmp/logging_rotatingfile_example.out.1
107 /tmp/logging_rotatingfile_example.out.2
108 /tmp/logging_rotatingfile_example.out.3
109 /tmp/logging_rotatingfile_example.out.4
110 /tmp/logging_rotatingfile_example.out.5
112 The most current file is always :file:`/tmp/logging_rotatingfile_example.out`,
113 and each time it reaches the size limit it is renamed with the suffix
114 ``.1``. Each of the existing backup files is renamed to increment the suffix
115 (``.1`` becomes ``.2``, etc.) and the ``.6`` file is erased.
117 Obviously this example sets the log length much much too small as an extreme
118 example. You would want to set *maxBytes* to an appropriate value.
120 Another useful feature of the logging API is the ability to produce different
121 messages at different log levels. This allows you to instrument your code with
122 debug messages, for example, but turning the log level down so that those debug
123 messages are not written for your production system. The default levels are
124 ``CRITICAL``, ``ERROR``, ``WARNING``, ``INFO``, ``DEBUG`` and ``NOTSET``.
126 The logger, handler, and log message call each specify a level. The log message
127 is only emitted if the handler and logger are configured to emit messages of
128 that level or lower. For example, if a message is ``CRITICAL``, and the logger
129 is set to ``ERROR``, the message is emitted. If a message is a ``WARNING``, and
130 the logger is set to produce only ``ERROR``\s, the message is not emitted::
135 LEVELS = {'debug': logging.DEBUG,
136 'info': logging.INFO,
137 'warning': logging.WARNING,
138 'error': logging.ERROR,
139 'critical': logging.CRITICAL}
141 if len(sys.argv) > 1:
142 level_name = sys.argv[1]
143 level = LEVELS.get(level_name, logging.NOTSET)
144 logging.basicConfig(level=level)
146 logging.debug('This is a debug message')
147 logging.info('This is an info message')
148 logging.warning('This is a warning message')
149 logging.error('This is an error message')
150 logging.critical('This is a critical error message')
152 Run the script with an argument like 'debug' or 'warning' to see which messages
153 show up at different levels::
155 $ python logging_level_example.py debug
156 DEBUG:root:This is a debug message
157 INFO:root:This is an info message
158 WARNING:root:This is a warning message
159 ERROR:root:This is an error message
160 CRITICAL:root:This is a critical error message
162 $ python logging_level_example.py info
163 INFO:root:This is an info message
164 WARNING:root:This is a warning message
165 ERROR:root:This is an error message
166 CRITICAL:root:This is a critical error message
168 You will notice that these log messages all have ``root`` embedded in them. The
169 logging module supports a hierarchy of loggers with different names. An easy
170 way to tell where a specific log message comes from is to use a separate logger
171 object for each of your modules. Each new logger "inherits" the configuration
172 of its parent, and log messages sent to a logger include the name of that
173 logger. Optionally, each logger can be configured differently, so that messages
174 from different modules are handled in different ways. Let's look at a simple
175 example of how to log from different modules so it is easy to trace the source
180 logging.basicConfig(level=logging.WARNING)
182 logger1 = logging.getLogger('package1.module1')
183 logger2 = logging.getLogger('package2.module2')
185 logger1.warning('This message comes from one module')
186 logger2.warning('And this message comes from another module')
190 $ python logging_modules_example.py
191 WARNING:package1.module1:This message comes from one module
192 WARNING:package2.module2:And this message comes from another module
194 There are many more options for configuring logging, including different log
195 message formatting options, having messages delivered to multiple destinations,
196 and changing the configuration of a long-running application on the fly using a
197 socket interface. All of these options are covered in depth in the library
198 module documentation.
203 The logging library takes a modular approach and offers the several categories
204 of components: loggers, handlers, filters, and formatters. Loggers expose the
205 interface that application code directly uses. Handlers send the log records to
206 the appropriate destination. Filters provide a finer grained facility for
207 determining which log records to send on to a handler. Formatters specify the
208 layout of the resultant log record.
210 :class:`Logger` objects have a threefold job. First, they expose several
211 methods to application code so that applications can log messages at runtime.
212 Second, logger objects determine which log messages to act upon based upon
213 severity (the default filtering facility) or filter objects. Third, logger
214 objects pass along relevant log messages to all interested log handlers.
216 The most widely used methods on logger objects fall into two categories:
217 configuration and message sending.
219 * :meth:`Logger.setLevel` specifies the lowest-severity log message a logger
220 will handle, where debug is the lowest built-in severity level and critical is
221 the highest built-in severity. For example, if the severity level is info,
222 the logger will handle only info, warning, error, and critical messages and
223 will ignore debug messages.
225 * :meth:`Logger.addFilter` and :meth:`Logger.removeFilter` add and remove filter
226 objects from the logger object. This tutorial does not address filters.
228 With the logger object configured, the following methods create log messages:
230 * :meth:`Logger.debug`, :meth:`Logger.info`, :meth:`Logger.warning`,
231 :meth:`Logger.error`, and :meth:`Logger.critical` all create log records with
232 a message and a level that corresponds to their respective method names. The
233 message is actually a format string, which may contain the standard string
234 substitution syntax of :const:`%s`, :const:`%d`, :const:`%f`, and so on. The
235 rest of their arguments is a list of objects that correspond with the
236 substitution fields in the message. With regard to :const:`**kwargs`, the
237 logging methods care only about a keyword of :const:`exc_info` and use it to
238 determine whether to log exception information.
240 * :meth:`Logger.exception` creates a log message similar to
241 :meth:`Logger.error`. The difference is that :meth:`Logger.exception` dumps a
242 stack trace along with it. Call this method only from an exception handler.
244 * :meth:`Logger.log` takes a log level as an explicit argument. This is a
245 little more verbose for logging messages than using the log level convenience
246 methods listed above, but this is how to log at custom log levels.
248 :func:`getLogger` returns a reference to a logger instance with the specified
249 if it it is provided, or ``root`` if not. The names are period-separated
250 hierarchical structures. Multiple calls to :func:`getLogger` with the same name
251 will return a reference to the same logger object. Loggers that are further
252 down in the hierarchical list are children of loggers higher up in the list.
253 For example, given a logger with a name of ``foo``, loggers with names of
254 ``foo.bar``, ``foo.bar.baz``, and ``foo.bam`` are all children of ``foo``.
255 Child loggers propagate messages up to their parent loggers. Because of this,
256 it is unnecessary to define and configure all the loggers an application uses.
257 It is sufficient to configure a top-level logger and create child loggers as
264 :class:`Handler` objects are responsible for dispatching the appropriate log
265 messages (based on the log messages' severity) to the handler's specified
266 destination. Logger objects can add zero or more handler objects to themselves
267 with an :func:`addHandler` method. As an example scenario, an application may
268 want to send all log messages to a log file, all log messages of error or higher
269 to stdout, and all messages of critical to an email address. This scenario
270 requires three individual handlers where each handler is responsible for sending
271 messages of a specific severity to a specific location.
273 The standard library includes quite a few handler types; this tutorial uses only
274 :class:`StreamHandler` and :class:`FileHandler` in its examples.
276 There are very few methods in a handler for application developers to concern
277 themselves with. The only handler methods that seem relevant for application
278 developers who are using the built-in handler objects (that is, not creating
279 custom handlers) are the following configuration methods:
281 * The :meth:`Handler.setLevel` method, just as in logger objects, specifies the
282 lowest severity that will be dispatched to the appropriate destination. Why
283 are there two :func:`setLevel` methods? The level set in the logger
284 determines which severity of messages it will pass to its handlers. The level
285 set in each handler determines which messages that handler will send on.
286 :func:`setFormatter` selects a Formatter object for this handler to use.
288 * :func:`addFilter` and :func:`removeFilter` respectively configure and
289 deconfigure filter objects on handlers.
291 Application code should not directly instantiate and use handlers. Instead, the
292 :class:`Handler` class is a base class that defines the interface that all
293 Handlers should have and establishes some default behavior that child classes
294 can use (or override).
300 Formatter objects configure the final order, structure, and contents of the log
301 message. Unlike the base :class:`logging.Handler` class, application code may
302 instantiate formatter classes, although you could likely subclass the formatter
303 if your application needs special behavior. The constructor takes two optional
304 arguments: a message format string and a date format string. If there is no
305 message format string, the default is to use the raw message. If there is no
306 date format string, the default date format is::
310 with the milliseconds tacked on at the end.
312 The message format string uses ``%(<dictionary key>)s`` styled string
313 substitution; the possible keys are documented in :ref:`formatter`.
315 The following message format string will log the time in a human-readable
316 format, the severity of the message, and the contents of the message, in that
319 "%(asctime)s - %(levelname)s - %(message)s"
325 Programmers can configure logging either by creating loggers, handlers, and
326 formatters explicitly in a main module with the configuration methods listed
327 above (using Python code), or by creating a logging config file. The following
328 code is an example of configuring a very simple logger, a console handler, and a
329 simple formatter in a Python module::
334 logger = logging.getLogger("simple_example")
335 logger.setLevel(logging.DEBUG)
336 # create console handler and set level to debug
337 ch = logging.StreamHandler()
338 ch.setLevel(logging.DEBUG)
340 formatter = logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
341 # add formatter to ch
342 ch.setFormatter(formatter)
344 logger.addHandler(ch)
347 logger.debug("debug message")
348 logger.info("info message")
349 logger.warn("warn message")
350 logger.error("error message")
351 logger.critical("critical message")
353 Running this module from the command line produces the following output::
355 $ python simple_logging_module.py
356 2005-03-19 15:10:26,618 - simple_example - DEBUG - debug message
357 2005-03-19 15:10:26,620 - simple_example - INFO - info message
358 2005-03-19 15:10:26,695 - simple_example - WARNING - warn message
359 2005-03-19 15:10:26,697 - simple_example - ERROR - error message
360 2005-03-19 15:10:26,773 - simple_example - CRITICAL - critical message
362 The following Python module creates a logger, handler, and formatter nearly
363 identical to those in the example listed above, with the only difference being
364 the names of the objects::
367 import logging.config
369 logging.config.fileConfig("logging.conf")
372 logger = logging.getLogger("simpleExample")
375 logger.debug("debug message")
376 logger.info("info message")
377 logger.warn("warn message")
378 logger.error("error message")
379 logger.critical("critical message")
381 Here is the logging.conf file::
384 keys=root,simpleExample
394 handlers=consoleHandler
396 [logger_simpleExample]
398 handlers=consoleHandler
399 qualname=simpleExample
402 [handler_consoleHandler]
405 formatter=simpleFormatter
408 [formatter_simpleFormatter]
409 format=%(asctime)s - %(name)s - %(levelname)s - %(message)s
412 The output is nearly identical to that of the non-config-file-based example::
414 $ python simple_logging_config.py
415 2005-03-19 15:38:55,977 - simpleExample - DEBUG - debug message
416 2005-03-19 15:38:55,979 - simpleExample - INFO - info message
417 2005-03-19 15:38:56,054 - simpleExample - WARNING - warn message
418 2005-03-19 15:38:56,055 - simpleExample - ERROR - error message
419 2005-03-19 15:38:56,130 - simpleExample - CRITICAL - critical message
421 You can see that the config file approach has a few advantages over the Python
422 code approach, mainly separation of configuration and code and the ability of
423 noncoders to easily modify the logging properties.
427 Configuring Logging for a Library
428 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
430 When developing a library which uses logging, some consideration needs to be
431 given to its configuration. If the using application does not use logging, and
432 library code makes logging calls, then a one-off message "No handlers could be
433 found for logger X.Y.Z" is printed to the console. This message is intended
434 to catch mistakes in logging configuration, but will confuse an application
435 developer who is not aware of logging by the library.
437 In addition to documenting how a library uses logging, a good way to configure
438 library logging so that it does not cause a spurious message is to add a
439 handler which does nothing. This avoids the message being printed, since a
440 handler will be found: it just doesn't produce any output. If the library user
441 configures logging for application use, presumably that configuration will add
442 some handlers, and if levels are suitably configured then logging calls made
443 in library code will send output to those handlers, as normal.
445 A do-nothing handler can be simply defined as follows::
449 class NullHandler(logging.Handler):
450 def emit(self, record):
453 An instance of this handler should be added to the top-level logger of the
454 logging namespace used by the library. If all logging by a library *foo* is
455 done using loggers with names matching "foo.x.y", then the code::
460 logging.getLogger("foo").addHandler(h)
462 should have the desired effect. If an organisation produces a number of
463 libraries, then the logger name specified can be "orgname.foo" rather than
466 .. versionadded:: 2.7
468 The :class:`NullHandler` class was not present in previous versions, but is now
469 included, so that it need not be defined in library code.
476 The numeric values of logging levels are given in the following table. These are
477 primarily of interest if you want to define your own levels, and need them to
478 have specific values relative to the predefined levels. If you define a level
479 with the same numeric value, it overwrites the predefined value; the predefined
482 +--------------+---------------+
483 | Level | Numeric value |
484 +==============+===============+
485 | ``CRITICAL`` | 50 |
486 +--------------+---------------+
488 +--------------+---------------+
490 +--------------+---------------+
492 +--------------+---------------+
494 +--------------+---------------+
496 +--------------+---------------+
498 Levels can also be associated with loggers, being set either by the developer or
499 through loading a saved logging configuration. When a logging method is called
500 on a logger, the logger compares its own level with the level associated with
501 the method call. If the logger's level is higher than the method call's, no
502 logging message is actually generated. This is the basic mechanism controlling
503 the verbosity of logging output.
505 Logging messages are encoded as instances of the :class:`LogRecord` class. When
506 a logger decides to actually log an event, a :class:`LogRecord` instance is
507 created from the logging message.
509 Logging messages are subjected to a dispatch mechanism through the use of
510 :dfn:`handlers`, which are instances of subclasses of the :class:`Handler`
511 class. Handlers are responsible for ensuring that a logged message (in the form
512 of a :class:`LogRecord`) ends up in a particular location (or set of locations)
513 which is useful for the target audience for that message (such as end users,
514 support desk staff, system administrators, developers). Handlers are passed
515 :class:`LogRecord` instances intended for particular destinations. Each logger
516 can have zero, one or more handlers associated with it (via the
517 :meth:`addHandler` method of :class:`Logger`). In addition to any handlers
518 directly associated with a logger, *all handlers associated with all ancestors
519 of the logger* are called to dispatch the message.
521 Just as for loggers, handlers can have levels associated with them. A handler's
522 level acts as a filter in the same way as a logger's level does. If a handler
523 decides to actually dispatch an event, the :meth:`emit` method is used to send
524 the message to its destination. Most user-defined subclasses of :class:`Handler`
525 will need to override this :meth:`emit`.
530 In addition to the base :class:`Handler` class, many useful subclasses are
533 #. :ref:`stream-handler` instances send error messages to streams (file-like
536 #. :ref:`file-handler` instances send error messages to disk files.
538 #. :class:`BaseRotatingHandler` is the base class for handlers that
539 rotate log files at a certain point. It is not meant to be instantiated
540 directly. Instead, use :ref:`rotating-file-handler` or
541 :ref:`timed-rotating-file-handler`.
543 #. :ref:`rotating-file-handler` instances send error messages to disk
544 files, with support for maximum log file sizes and log file rotation.
546 #. :ref:`timed-rotating-file-handler` instances send error messages to
547 disk files, rotating the log file at certain timed intervals.
549 #. :ref:`socket-handler` instances send error messages to TCP/IP
552 #. :ref:`datagram-handler` instances send error messages to UDP
555 #. :ref:`smtp-handler` instances send error messages to a designated
558 #. :ref:`syslog-handler` instances send error messages to a Unix
559 syslog daemon, possibly on a remote machine.
561 #. :ref:`nt-eventlog-handler` instances send error messages to a
562 Windows NT/2000/XP event log.
564 #. :ref:`memory-handler` instances send error messages to a buffer
565 in memory, which is flushed whenever specific criteria are met.
567 #. :ref:`http-handler` instances send error messages to an HTTP
568 server using either ``GET`` or ``POST`` semantics.
570 #. :ref:`watched-file-handler` instances watch the file they are
571 logging to. If the file changes, it is closed and reopened using the file
572 name. This handler is only useful on Unix-like systems; Windows does not
573 support the underlying mechanism used.
575 #. :ref:`null-handler` instances do nothing with error messages. They are used
576 by library developers who want to use logging, but want to avoid the "No
577 handlers could be found for logger XXX" message which can be displayed if
578 the library user has not configured logging. See :ref:`library-config` for
581 .. versionadded:: 2.7
583 The :class:`NullHandler` class was not present in previous versions.
585 The :class:`NullHandler`, :class:`StreamHandler` and :class:`FileHandler`
586 classes are defined in the core logging package. The other handlers are
587 defined in a sub- module, :mod:`logging.handlers`. (There is also another
588 sub-module, :mod:`logging.config`, for configuration functionality.)
590 Logged messages are formatted for presentation through instances of the
591 :class:`Formatter` class. They are initialized with a format string suitable for
592 use with the % operator and a dictionary.
594 For formatting multiple messages in a batch, instances of
595 :class:`BufferingFormatter` can be used. In addition to the format string (which
596 is applied to each message in the batch), there is provision for header and
597 trailer format strings.
599 When filtering based on logger level and/or handler level is not enough,
600 instances of :class:`Filter` can be added to both :class:`Logger` and
601 :class:`Handler` instances (through their :meth:`addFilter` method). Before
602 deciding to process a message further, both loggers and handlers consult all
603 their filters for permission. If any filter returns a false value, the message
604 is not processed further.
606 The basic :class:`Filter` functionality allows filtering by specific logger
607 name. If this feature is used, messages sent to the named logger and its
608 children are allowed through the filter, and all others dropped.
610 Module-Level Functions
611 ----------------------
613 In addition to the classes described above, there are a number of module- level
617 .. function:: getLogger([name])
619 Return a logger with the specified name or, if no name is specified, return a
620 logger which is the root logger of the hierarchy. If specified, the name is
621 typically a dot-separated hierarchical name like *"a"*, *"a.b"* or *"a.b.c.d"*.
622 Choice of these names is entirely up to the developer who is using logging.
624 All calls to this function with a given name return the same logger instance.
625 This means that logger instances never need to be passed between different parts
629 .. function:: getLoggerClass()
631 Return either the standard :class:`Logger` class, or the last class passed to
632 :func:`setLoggerClass`. This function may be called from within a new class
633 definition, to ensure that installing a customised :class:`Logger` class will
634 not undo customisations already applied by other code. For example::
636 class MyLogger(logging.getLoggerClass()):
637 # ... override behaviour here
640 .. function:: debug(msg[, *args[, **kwargs]])
642 Logs a message with level :const:`DEBUG` on the root logger. The *msg* is the
643 message format string, and the *args* are the arguments which are merged into
644 *msg* using the string formatting operator. (Note that this means that you can
645 use keywords in the format string, together with a single dictionary argument.)
647 There are two keyword arguments in *kwargs* which are inspected: *exc_info*
648 which, if it does not evaluate as false, causes exception information to be
649 added to the logging message. If an exception tuple (in the format returned by
650 :func:`sys.exc_info`) is provided, it is used; otherwise, :func:`sys.exc_info`
651 is called to get the exception information.
653 The other optional keyword argument is *extra* which can be used to pass a
654 dictionary which is used to populate the __dict__ of the LogRecord created for
655 the logging event with user-defined attributes. These custom attributes can then
656 be used as you like. For example, they could be incorporated into logged
657 messages. For example::
659 FORMAT = "%(asctime)-15s %(clientip)s %(user)-8s %(message)s"
660 logging.basicConfig(format=FORMAT)
661 d = {'clientip': '192.168.0.1', 'user': 'fbloggs'}
662 logging.warning("Protocol problem: %s", "connection reset", extra=d)
664 would print something like ::
666 2006-02-08 22:20:02,165 192.168.0.1 fbloggs Protocol problem: connection reset
668 The keys in the dictionary passed in *extra* should not clash with the keys used
669 by the logging system. (See the :class:`Formatter` documentation for more
670 information on which keys are used by the logging system.)
672 If you choose to use these attributes in logged messages, you need to exercise
673 some care. In the above example, for instance, the :class:`Formatter` has been
674 set up with a format string which expects 'clientip' and 'user' in the attribute
675 dictionary of the LogRecord. If these are missing, the message will not be
676 logged because a string formatting exception will occur. So in this case, you
677 always need to pass the *extra* dictionary with these keys.
679 While this might be annoying, this feature is intended for use in specialized
680 circumstances, such as multi-threaded servers where the same code executes in
681 many contexts, and interesting conditions which arise are dependent on this
682 context (such as remote client IP address and authenticated user name, in the
683 above example). In such circumstances, it is likely that specialized
684 :class:`Formatter`\ s would be used with particular :class:`Handler`\ s.
686 .. versionchanged:: 2.5
690 .. function:: info(msg[, *args[, **kwargs]])
692 Logs a message with level :const:`INFO` on the root logger. The arguments are
693 interpreted as for :func:`debug`.
696 .. function:: warning(msg[, *args[, **kwargs]])
698 Logs a message with level :const:`WARNING` on the root logger. The arguments are
699 interpreted as for :func:`debug`.
702 .. function:: error(msg[, *args[, **kwargs]])
704 Logs a message with level :const:`ERROR` on the root logger. The arguments are
705 interpreted as for :func:`debug`.
708 .. function:: critical(msg[, *args[, **kwargs]])
710 Logs a message with level :const:`CRITICAL` on the root logger. The arguments
711 are interpreted as for :func:`debug`.
714 .. function:: exception(msg[, *args])
716 Logs a message with level :const:`ERROR` on the root logger. The arguments are
717 interpreted as for :func:`debug`. Exception info is added to the logging
718 message. This function should only be called from an exception handler.
721 .. function:: log(level, msg[, *args[, **kwargs]])
723 Logs a message with level *level* on the root logger. The other arguments are
724 interpreted as for :func:`debug`.
727 .. function:: disable(lvl)
729 Provides an overriding level *lvl* for all loggers which takes precedence over
730 the logger's own level. When the need arises to temporarily throttle logging
731 output down across the whole application, this function can be useful.
734 .. function:: addLevelName(lvl, levelName)
736 Associates level *lvl* with text *levelName* in an internal dictionary, which is
737 used to map numeric levels to a textual representation, for example when a
738 :class:`Formatter` formats a message. This function can also be used to define
739 your own levels. The only constraints are that all levels used must be
740 registered using this function, levels should be positive integers and they
741 should increase in increasing order of severity.
744 .. function:: getLevelName(lvl)
746 Returns the textual representation of logging level *lvl*. If the level is one
747 of the predefined levels :const:`CRITICAL`, :const:`ERROR`, :const:`WARNING`,
748 :const:`INFO` or :const:`DEBUG` then you get the corresponding string. If you
749 have associated levels with names using :func:`addLevelName` then the name you
750 have associated with *lvl* is returned. If a numeric value corresponding to one
751 of the defined levels is passed in, the corresponding string representation is
752 returned. Otherwise, the string "Level %s" % lvl is returned.
755 .. function:: makeLogRecord(attrdict)
757 Creates and returns a new :class:`LogRecord` instance whose attributes are
758 defined by *attrdict*. This function is useful for taking a pickled
759 :class:`LogRecord` attribute dictionary, sent over a socket, and reconstituting
760 it as a :class:`LogRecord` instance at the receiving end.
763 .. function:: basicConfig([**kwargs])
765 Does basic configuration for the logging system by creating a
766 :class:`StreamHandler` with a default :class:`Formatter` and adding it to the
767 root logger. The function does nothing if any handlers have been defined for
768 the root logger. The functions :func:`debug`, :func:`info`, :func:`warning`,
769 :func:`error` and :func:`critical` will call :func:`basicConfig` automatically
770 if no handlers are defined for the root logger.
772 This function does nothing if the root logger already has handlers configured.
774 .. versionchanged:: 2.4
775 Formerly, :func:`basicConfig` did not take any keyword arguments.
777 The following keyword arguments are supported.
779 +--------------+---------------------------------------------+
780 | Format | Description |
781 +==============+=============================================+
782 | ``filename`` | Specifies that a FileHandler be created, |
783 | | using the specified filename, rather than a |
785 +--------------+---------------------------------------------+
786 | ``filemode`` | Specifies the mode to open the file, if |
787 | | filename is specified (if filemode is |
788 | | unspecified, it defaults to 'a'). |
789 +--------------+---------------------------------------------+
790 | ``format`` | Use the specified format string for the |
792 +--------------+---------------------------------------------+
793 | ``datefmt`` | Use the specified date/time format. |
794 +--------------+---------------------------------------------+
795 | ``level`` | Set the root logger level to the specified |
797 +--------------+---------------------------------------------+
798 | ``stream`` | Use the specified stream to initialize the |
799 | | StreamHandler. Note that this argument is |
800 | | incompatible with 'filename' - if both are |
801 | | present, 'stream' is ignored. |
802 +--------------+---------------------------------------------+
805 .. function:: shutdown()
807 Informs the logging system to perform an orderly shutdown by flushing and
808 closing all handlers. This should be called at application exit and no
809 further use of the logging system should be made after this call.
812 .. function:: setLoggerClass(klass)
814 Tells the logging system to use the class *klass* when instantiating a logger.
815 The class should define :meth:`__init__` such that only a name argument is
816 required, and the :meth:`__init__` should call :meth:`Logger.__init__`. This
817 function is typically called before any loggers are instantiated by applications
818 which need to use custom logger behavior.
823 :pep:`282` - A Logging System
824 The proposal which described this feature for inclusion in the Python standard
827 `Original Python logging package <http://www.red-dove.com/python_logging.html>`_
828 This is the original source for the :mod:`logging` package. The version of the
829 package available from this site is suitable for use with Python 1.5.2, 2.1.x
830 and 2.2.x, which do not include the :mod:`logging` package in the standard
838 Loggers have the following attributes and methods. Note that Loggers are never
839 instantiated directly, but always through the module-level function
840 ``logging.getLogger(name)``.
843 .. attribute:: Logger.propagate
845 If this evaluates to false, logging messages are not passed by this logger or by
846 child loggers to higher level (ancestor) loggers. The constructor sets this
850 .. method:: Logger.setLevel(lvl)
852 Sets the threshold for this logger to *lvl*. Logging messages which are less
853 severe than *lvl* will be ignored. When a logger is created, the level is set to
854 :const:`NOTSET` (which causes all messages to be processed when the logger is
855 the root logger, or delegation to the parent when the logger is a non-root
856 logger). Note that the root logger is created with level :const:`WARNING`.
858 The term "delegation to the parent" means that if a logger has a level of
859 NOTSET, its chain of ancestor loggers is traversed until either an ancestor with
860 a level other than NOTSET is found, or the root is reached.
862 If an ancestor is found with a level other than NOTSET, then that ancestor's
863 level is treated as the effective level of the logger where the ancestor search
864 began, and is used to determine how a logging event is handled.
866 If the root is reached, and it has a level of NOTSET, then all messages will be
867 processed. Otherwise, the root's level will be used as the effective level.
870 .. method:: Logger.isEnabledFor(lvl)
872 Indicates if a message of severity *lvl* would be processed by this logger.
873 This method checks first the module-level level set by
874 ``logging.disable(lvl)`` and then the logger's effective level as determined
875 by :meth:`getEffectiveLevel`.
878 .. method:: Logger.getEffectiveLevel()
880 Indicates the effective level for this logger. If a value other than
881 :const:`NOTSET` has been set using :meth:`setLevel`, it is returned. Otherwise,
882 the hierarchy is traversed towards the root until a value other than
883 :const:`NOTSET` is found, and that value is returned.
886 .. method:: Logger.debug(msg[, *args[, **kwargs]])
888 Logs a message with level :const:`DEBUG` on this logger. The *msg* is the
889 message format string, and the *args* are the arguments which are merged into
890 *msg* using the string formatting operator. (Note that this means that you can
891 use keywords in the format string, together with a single dictionary argument.)
893 There are two keyword arguments in *kwargs* which are inspected: *exc_info*
894 which, if it does not evaluate as false, causes exception information to be
895 added to the logging message. If an exception tuple (in the format returned by
896 :func:`sys.exc_info`) is provided, it is used; otherwise, :func:`sys.exc_info`
897 is called to get the exception information.
899 The other optional keyword argument is *extra* which can be used to pass a
900 dictionary which is used to populate the __dict__ of the LogRecord created for
901 the logging event with user-defined attributes. These custom attributes can then
902 be used as you like. For example, they could be incorporated into logged
903 messages. For example::
905 FORMAT = "%(asctime)-15s %(clientip)s %(user)-8s %(message)s"
906 logging.basicConfig(format=FORMAT)
907 d = { 'clientip' : '192.168.0.1', 'user' : 'fbloggs' }
908 logger = logging.getLogger("tcpserver")
909 logger.warning("Protocol problem: %s", "connection reset", extra=d)
911 would print something like ::
913 2006-02-08 22:20:02,165 192.168.0.1 fbloggs Protocol problem: connection reset
915 The keys in the dictionary passed in *extra* should not clash with the keys used
916 by the logging system. (See the :class:`Formatter` documentation for more
917 information on which keys are used by the logging system.)
919 If you choose to use these attributes in logged messages, you need to exercise
920 some care. In the above example, for instance, the :class:`Formatter` has been
921 set up with a format string which expects 'clientip' and 'user' in the attribute
922 dictionary of the LogRecord. If these are missing, the message will not be
923 logged because a string formatting exception will occur. So in this case, you
924 always need to pass the *extra* dictionary with these keys.
926 While this might be annoying, this feature is intended for use in specialized
927 circumstances, such as multi-threaded servers where the same code executes in
928 many contexts, and interesting conditions which arise are dependent on this
929 context (such as remote client IP address and authenticated user name, in the
930 above example). In such circumstances, it is likely that specialized
931 :class:`Formatter`\ s would be used with particular :class:`Handler`\ s.
933 .. versionchanged:: 2.5
937 .. method:: Logger.info(msg[, *args[, **kwargs]])
939 Logs a message with level :const:`INFO` on this logger. The arguments are
940 interpreted as for :meth:`debug`.
943 .. method:: Logger.warning(msg[, *args[, **kwargs]])
945 Logs a message with level :const:`WARNING` on this logger. The arguments are
946 interpreted as for :meth:`debug`.
949 .. method:: Logger.error(msg[, *args[, **kwargs]])
951 Logs a message with level :const:`ERROR` on this logger. The arguments are
952 interpreted as for :meth:`debug`.
955 .. method:: Logger.critical(msg[, *args[, **kwargs]])
957 Logs a message with level :const:`CRITICAL` on this logger. The arguments are
958 interpreted as for :meth:`debug`.
961 .. method:: Logger.log(lvl, msg[, *args[, **kwargs]])
963 Logs a message with integer level *lvl* on this logger. The other arguments are
964 interpreted as for :meth:`debug`.
967 .. method:: Logger.exception(msg[, *args])
969 Logs a message with level :const:`ERROR` on this logger. The arguments are
970 interpreted as for :meth:`debug`. Exception info is added to the logging
971 message. This method should only be called from an exception handler.
974 .. method:: Logger.addFilter(filt)
976 Adds the specified filter *filt* to this logger.
979 .. method:: Logger.removeFilter(filt)
981 Removes the specified filter *filt* from this logger.
984 .. method:: Logger.filter(record)
986 Applies this logger's filters to the record and returns a true value if the
987 record is to be processed.
990 .. method:: Logger.addHandler(hdlr)
992 Adds the specified handler *hdlr* to this logger.
995 .. method:: Logger.removeHandler(hdlr)
997 Removes the specified handler *hdlr* from this logger.
1000 .. method:: Logger.findCaller()
1002 Finds the caller's source filename and line number. Returns the filename, line
1003 number and function name as a 3-element tuple.
1005 .. versionchanged:: 2.4
1006 The function name was added. In earlier versions, the filename and line number
1007 were returned as a 2-element tuple..
1010 .. method:: Logger.handle(record)
1012 Handles a record by passing it to all handlers associated with this logger and
1013 its ancestors (until a false value of *propagate* is found). This method is used
1014 for unpickled records received from a socket, as well as those created locally.
1015 Logger-level filtering is applied using :meth:`~Logger.filter`.
1018 .. method:: Logger.makeRecord(name, lvl, fn, lno, msg, args, exc_info [, func, extra])
1020 This is a factory method which can be overridden in subclasses to create
1021 specialized :class:`LogRecord` instances.
1023 .. versionchanged:: 2.5
1024 *func* and *extra* were added.
1027 .. _minimal-example:
1032 .. versionchanged:: 2.4
1033 formerly :func:`basicConfig` did not take any keyword arguments.
1035 The :mod:`logging` package provides a lot of flexibility, and its configuration
1036 can appear daunting. This section demonstrates that simple use of the logging
1037 package is possible.
1039 The simplest example shows logging to the console::
1043 logging.debug('A debug message')
1044 logging.info('Some information')
1045 logging.warning('A shot across the bows')
1047 If you run the above script, you'll see this::
1049 WARNING:root:A shot across the bows
1051 Because no particular logger was specified, the system used the root logger. The
1052 debug and info messages didn't appear because by default, the root logger is
1053 configured to only handle messages with a severity of WARNING or above. The
1054 message format is also a configuration default, as is the output destination of
1055 the messages - ``sys.stderr``. The severity level, the message format and
1056 destination can be easily changed, as shown in the example below::
1060 logging.basicConfig(level=logging.DEBUG,
1061 format='%(asctime)s %(levelname)s %(message)s',
1062 filename='/tmp/myapp.log',
1064 logging.debug('A debug message')
1065 logging.info('Some information')
1066 logging.warning('A shot across the bows')
1068 The :meth:`basicConfig` method is used to change the configuration defaults,
1069 which results in output (written to ``/tmp/myapp.log``) which should look
1070 something like the following::
1072 2004-07-02 13:00:08,743 DEBUG A debug message
1073 2004-07-02 13:00:08,743 INFO Some information
1074 2004-07-02 13:00:08,743 WARNING A shot across the bows
1076 This time, all messages with a severity of DEBUG or above were handled, and the
1077 format of the messages was also changed, and output went to the specified file
1078 rather than the console.
1080 Formatting uses standard Python string formatting - see section
1081 :ref:`string-formatting`. The format string takes the following common
1082 specifiers. For a complete list of specifiers, consult the :class:`Formatter`
1085 +-------------------+-----------------------------------------------+
1086 | Format | Description |
1087 +===================+===============================================+
1088 | ``%(name)s`` | Name of the logger (logging channel). |
1089 +-------------------+-----------------------------------------------+
1090 | ``%(levelname)s`` | Text logging level for the message |
1091 | | (``'DEBUG'``, ``'INFO'``, ``'WARNING'``, |
1092 | | ``'ERROR'``, ``'CRITICAL'``). |
1093 +-------------------+-----------------------------------------------+
1094 | ``%(asctime)s`` | Human-readable time when the |
1095 | | :class:`LogRecord` was created. By default |
1096 | | this is of the form "2003-07-08 16:49:45,896" |
1097 | | (the numbers after the comma are millisecond |
1098 | | portion of the time). |
1099 +-------------------+-----------------------------------------------+
1100 | ``%(message)s`` | The logged message. |
1101 +-------------------+-----------------------------------------------+
1103 To change the date/time format, you can pass an additional keyword parameter,
1104 *datefmt*, as in the following::
1108 logging.basicConfig(level=logging.DEBUG,
1109 format='%(asctime)s %(levelname)-8s %(message)s',
1110 datefmt='%a, %d %b %Y %H:%M:%S',
1111 filename='/temp/myapp.log',
1113 logging.debug('A debug message')
1114 logging.info('Some information')
1115 logging.warning('A shot across the bows')
1117 which would result in output like ::
1119 Fri, 02 Jul 2004 13:06:18 DEBUG A debug message
1120 Fri, 02 Jul 2004 13:06:18 INFO Some information
1121 Fri, 02 Jul 2004 13:06:18 WARNING A shot across the bows
1123 The date format string follows the requirements of :func:`strftime` - see the
1124 documentation for the :mod:`time` module.
1126 If, instead of sending logging output to the console or a file, you'd rather use
1127 a file-like object which you have created separately, you can pass it to
1128 :func:`basicConfig` using the *stream* keyword argument. Note that if both
1129 *stream* and *filename* keyword arguments are passed, the *stream* argument is
1132 Of course, you can put variable information in your output. To do this, simply
1133 have the message be a format string and pass in additional arguments containing
1134 the variable information, as in the following example::
1138 logging.basicConfig(level=logging.DEBUG,
1139 format='%(asctime)s %(levelname)-8s %(message)s',
1140 datefmt='%a, %d %b %Y %H:%M:%S',
1141 filename='/temp/myapp.log',
1143 logging.error('Pack my box with %d dozen %s', 5, 'liquor jugs')
1145 which would result in ::
1147 Wed, 21 Jul 2004 15:35:16 ERROR Pack my box with 5 dozen liquor jugs
1150 .. _multiple-destinations:
1152 Logging to multiple destinations
1153 --------------------------------
1155 Let's say you want to log to console and file with different message formats and
1156 in differing circumstances. Say you want to log messages with levels of DEBUG
1157 and higher to file, and those messages at level INFO and higher to the console.
1158 Let's also assume that the file should contain timestamps, but the console
1159 messages should not. Here's how you can achieve this::
1163 # set up logging to file - see previous section for more details
1164 logging.basicConfig(level=logging.DEBUG,
1165 format='%(asctime)s %(name)-12s %(levelname)-8s %(message)s',
1166 datefmt='%m-%d %H:%M',
1167 filename='/temp/myapp.log',
1169 # define a Handler which writes INFO messages or higher to the sys.stderr
1170 console = logging.StreamHandler()
1171 console.setLevel(logging.INFO)
1172 # set a format which is simpler for console use
1173 formatter = logging.Formatter('%(name)-12s: %(levelname)-8s %(message)s')
1174 # tell the handler to use this format
1175 console.setFormatter(formatter)
1176 # add the handler to the root logger
1177 logging.getLogger('').addHandler(console)
1179 # Now, we can log to the root logger, or any other logger. First the root...
1180 logging.info('Jackdaws love my big sphinx of quartz.')
1182 # Now, define a couple of other loggers which might represent areas in your
1185 logger1 = logging.getLogger('myapp.area1')
1186 logger2 = logging.getLogger('myapp.area2')
1188 logger1.debug('Quick zephyrs blow, vexing daft Jim.')
1189 logger1.info('How quickly daft jumping zebras vex.')
1190 logger2.warning('Jail zesty vixen who grabbed pay from quack.')
1191 logger2.error('The five boxing wizards jump quickly.')
1193 When you run this, on the console you will see ::
1195 root : INFO Jackdaws love my big sphinx of quartz.
1196 myapp.area1 : INFO How quickly daft jumping zebras vex.
1197 myapp.area2 : WARNING Jail zesty vixen who grabbed pay from quack.
1198 myapp.area2 : ERROR The five boxing wizards jump quickly.
1200 and in the file you will see something like ::
1202 10-22 22:19 root INFO Jackdaws love my big sphinx of quartz.
1203 10-22 22:19 myapp.area1 DEBUG Quick zephyrs blow, vexing daft Jim.
1204 10-22 22:19 myapp.area1 INFO How quickly daft jumping zebras vex.
1205 10-22 22:19 myapp.area2 WARNING Jail zesty vixen who grabbed pay from quack.
1206 10-22 22:19 myapp.area2 ERROR The five boxing wizards jump quickly.
1208 As you can see, the DEBUG message only shows up in the file. The other messages
1209 are sent to both destinations.
1211 This example uses console and file handlers, but you can use any number and
1212 combination of handlers you choose.
1217 Adding contextual information to your logging output
1218 ----------------------------------------------------
1220 Sometimes you want logging output to contain contextual information in
1221 addition to the parameters passed to the logging call. For example, in a
1222 networked application, it may be desirable to log client-specific information
1223 in the log (e.g. remote client's username, or IP address). Although you could
1224 use the *extra* parameter to achieve this, it's not always convenient to pass
1225 the information in this way. While it might be tempting to create
1226 :class:`Logger` instances on a per-connection basis, this is not a good idea
1227 because these instances are not garbage collected. While this is not a problem
1228 in practice, when the number of :class:`Logger` instances is dependent on the
1229 level of granularity you want to use in logging an application, it could
1230 be hard to manage if the number of :class:`Logger` instances becomes
1231 effectively unbounded.
1233 An easy way in which you can pass contextual information to be output along
1234 with logging event information is to use the :class:`LoggerAdapter` class.
1235 This class is designed to look like a :class:`Logger`, so that you can call
1236 :meth:`debug`, :meth:`info`, :meth:`warning`, :meth:`error`,
1237 :meth:`exception`, :meth:`critical` and :meth:`log`. These methods have the
1238 same signatures as their counterparts in :class:`Logger`, so you can use the
1239 two types of instances interchangeably.
1241 When you create an instance of :class:`LoggerAdapter`, you pass it a
1242 :class:`Logger` instance and a dict-like object which contains your contextual
1243 information. When you call one of the logging methods on an instance of
1244 :class:`LoggerAdapter`, it delegates the call to the underlying instance of
1245 :class:`Logger` passed to its constructor, and arranges to pass the contextual
1246 information in the delegated call. Here's a snippet from the code of
1247 :class:`LoggerAdapter`::
1249 def debug(self, msg, *args, **kwargs):
1251 Delegate a debug call to the underlying logger, after adding
1252 contextual information from this adapter instance.
1254 msg, kwargs = self.process(msg, kwargs)
1255 self.logger.debug(msg, *args, **kwargs)
1257 The :meth:`process` method of :class:`LoggerAdapter` is where the contextual
1258 information is added to the logging output. It's passed the message and
1259 keyword arguments of the logging call, and it passes back (potentially)
1260 modified versions of these to use in the call to the underlying logger. The
1261 default implementation of this method leaves the message alone, but inserts
1262 an "extra" key in the keyword argument whose value is the dict-like object
1263 passed to the constructor. Of course, if you had passed an "extra" keyword
1264 argument in the call to the adapter, it will be silently overwritten.
1266 The advantage of using "extra" is that the values in the dict-like object are
1267 merged into the :class:`LogRecord` instance's __dict__, allowing you to use
1268 customized strings with your :class:`Formatter` instances which know about
1269 the keys of the dict-like object. If you need a different method, e.g. if you
1270 want to prepend or append the contextual information to the message string,
1271 you just need to subclass :class:`LoggerAdapter` and override :meth:`process`
1272 to do what you need. Here's an example script which uses this class, which
1273 also illustrates what dict-like behaviour is needed from an arbitrary
1274 "dict-like" object for use in the constructor::
1280 An example class which shows how an arbitrary class can be used as
1281 the 'extra' context information repository passed to a LoggerAdapter.
1284 def __getitem__(self, name):
1286 To allow this instance to look like a dict.
1288 from random import choice
1290 result = choice(["127.0.0.1", "192.168.0.1"])
1291 elif name == "user":
1292 result = choice(["jim", "fred", "sheila"])
1294 result = self.__dict__.get(name, "?")
1299 To allow iteration over keys, which will be merged into
1300 the LogRecord dict before formatting and output.
1302 keys = ["ip", "user"]
1303 keys.extend(self.__dict__.keys())
1304 return keys.__iter__()
1306 if __name__ == "__main__":
1307 from random import choice
1308 levels = (logging.DEBUG, logging.INFO, logging.WARNING, logging.ERROR, logging.CRITICAL)
1309 a1 = logging.LoggerAdapter(logging.getLogger("a.b.c"),
1310 { "ip" : "123.231.231.123", "user" : "sheila" })
1311 logging.basicConfig(level=logging.DEBUG,
1312 format="%(asctime)-15s %(name)-5s %(levelname)-8s IP: %(ip)-15s User: %(user)-8s %(message)s")
1313 a1.debug("A debug message")
1314 a1.info("An info message with %s", "some parameters")
1315 a2 = logging.LoggerAdapter(logging.getLogger("d.e.f"), ConnInfo())
1317 lvl = choice(levels)
1318 lvlname = logging.getLevelName(lvl)
1319 a2.log(lvl, "A message at %s level with %d %s", lvlname, 2, "parameters")
1321 When this script is run, the output should look something like this::
1323 2008-01-18 14:49:54,023 a.b.c DEBUG IP: 123.231.231.123 User: sheila A debug message
1324 2008-01-18 14:49:54,023 a.b.c INFO IP: 123.231.231.123 User: sheila An info message with some parameters
1325 2008-01-18 14:49:54,023 d.e.f CRITICAL IP: 192.168.0.1 User: jim A message at CRITICAL level with 2 parameters
1326 2008-01-18 14:49:54,033 d.e.f INFO IP: 192.168.0.1 User: jim A message at INFO level with 2 parameters
1327 2008-01-18 14:49:54,033 d.e.f WARNING IP: 192.168.0.1 User: sheila A message at WARNING level with 2 parameters
1328 2008-01-18 14:49:54,033 d.e.f ERROR IP: 127.0.0.1 User: fred A message at ERROR level with 2 parameters
1329 2008-01-18 14:49:54,033 d.e.f ERROR IP: 127.0.0.1 User: sheila A message at ERROR level with 2 parameters
1330 2008-01-18 14:49:54,033 d.e.f WARNING IP: 192.168.0.1 User: sheila A message at WARNING level with 2 parameters
1331 2008-01-18 14:49:54,033 d.e.f WARNING IP: 192.168.0.1 User: jim A message at WARNING level with 2 parameters
1332 2008-01-18 14:49:54,033 d.e.f INFO IP: 192.168.0.1 User: fred A message at INFO level with 2 parameters
1333 2008-01-18 14:49:54,033 d.e.f WARNING IP: 192.168.0.1 User: sheila A message at WARNING level with 2 parameters
1334 2008-01-18 14:49:54,033 d.e.f WARNING IP: 127.0.0.1 User: jim A message at WARNING level with 2 parameters
1336 .. versionadded:: 2.6
1338 The :class:`LoggerAdapter` class was not present in previous versions.
1340 .. _multiple-processes:
1342 Logging to a single file from multiple processes
1343 ------------------------------------------------
1345 Although logging is thread-safe, and logging to a single file from multiple
1346 threads in a single process *is* supported, logging to a single file from
1347 *multiple processes* is *not* supported, because there is no standard way to
1348 serialize access to a single file across multiple processes in Python. If you
1349 need to log to a single file from multiple processes, the best way of doing
1350 this is to have all the processes log to a :class:`SocketHandler`, and have a
1351 separate process which implements a socket server which reads from the socket
1352 and logs to file. (If you prefer, you can dedicate one thread in one of the
1353 existing processes to perform this function.) The following section documents
1354 this approach in more detail and includes a working socket receiver which can
1355 be used as a starting point for you to adapt in your own applications.
1357 If you are using a recent version of Python which includes the
1358 :mod:`multiprocessing` module, you can write your own handler which uses the
1359 :class:`Lock` class from this module to serialize access to the file from
1360 your processes. The existing :class:`FileHandler` and subclasses do not make
1361 use of :mod:`multiprocessing` at present, though they may do so in the future.
1363 .. _network-logging:
1365 Sending and receiving logging events across a network
1366 -----------------------------------------------------
1368 Let's say you want to send logging events across a network, and handle them at
1369 the receiving end. A simple way of doing this is attaching a
1370 :class:`SocketHandler` instance to the root logger at the sending end::
1372 import logging, logging.handlers
1374 rootLogger = logging.getLogger('')
1375 rootLogger.setLevel(logging.DEBUG)
1376 socketHandler = logging.handlers.SocketHandler('localhost',
1377 logging.handlers.DEFAULT_TCP_LOGGING_PORT)
1378 # don't bother with a formatter, since a socket handler sends the event as
1379 # an unformatted pickle
1380 rootLogger.addHandler(socketHandler)
1382 # Now, we can log to the root logger, or any other logger. First the root...
1383 logging.info('Jackdaws love my big sphinx of quartz.')
1385 # Now, define a couple of other loggers which might represent areas in your
1388 logger1 = logging.getLogger('myapp.area1')
1389 logger2 = logging.getLogger('myapp.area2')
1391 logger1.debug('Quick zephyrs blow, vexing daft Jim.')
1392 logger1.info('How quickly daft jumping zebras vex.')
1393 logger2.warning('Jail zesty vixen who grabbed pay from quack.')
1394 logger2.error('The five boxing wizards jump quickly.')
1396 At the receiving end, you can set up a receiver using the :mod:`SocketServer`
1397 module. Here is a basic working example::
1401 import logging.handlers
1406 class LogRecordStreamHandler(SocketServer.StreamRequestHandler):
1407 """Handler for a streaming logging request.
1409 This basically logs the record using whatever logging policy is
1415 Handle multiple requests - each expected to be a 4-byte length,
1416 followed by the LogRecord in pickle format. Logs the record
1417 according to whatever policy is configured locally.
1420 chunk = self.connection.recv(4)
1423 slen = struct.unpack(">L", chunk)[0]
1424 chunk = self.connection.recv(slen)
1425 while len(chunk) < slen:
1426 chunk = chunk + self.connection.recv(slen - len(chunk))
1427 obj = self.unPickle(chunk)
1428 record = logging.makeLogRecord(obj)
1429 self.handleLogRecord(record)
1431 def unPickle(self, data):
1432 return cPickle.loads(data)
1434 def handleLogRecord(self, record):
1435 # if a name is specified, we use the named logger rather than the one
1436 # implied by the record.
1437 if self.server.logname is not None:
1438 name = self.server.logname
1441 logger = logging.getLogger(name)
1442 # N.B. EVERY record gets logged. This is because Logger.handle
1443 # is normally called AFTER logger-level filtering. If you want
1444 # to do filtering, do it at the client end to save wasting
1445 # cycles and network bandwidth!
1446 logger.handle(record)
1448 class LogRecordSocketReceiver(SocketServer.ThreadingTCPServer):
1449 """simple TCP socket-based logging receiver suitable for testing.
1452 allow_reuse_address = 1
1454 def __init__(self, host='localhost',
1455 port=logging.handlers.DEFAULT_TCP_LOGGING_PORT,
1456 handler=LogRecordStreamHandler):
1457 SocketServer.ThreadingTCPServer.__init__(self, (host, port), handler)
1462 def serve_until_stopped(self):
1466 rd, wr, ex = select.select([self.socket.fileno()],
1470 self.handle_request()
1474 logging.basicConfig(
1475 format="%(relativeCreated)5d %(name)-15s %(levelname)-8s %(message)s")
1476 tcpserver = LogRecordSocketReceiver()
1477 print "About to start TCP server..."
1478 tcpserver.serve_until_stopped()
1480 if __name__ == "__main__":
1483 First run the server, and then the client. On the client side, nothing is
1484 printed on the console; on the server side, you should see something like::
1486 About to start TCP server...
1487 59 root INFO Jackdaws love my big sphinx of quartz.
1488 59 myapp.area1 DEBUG Quick zephyrs blow, vexing daft Jim.
1489 69 myapp.area1 INFO How quickly daft jumping zebras vex.
1490 69 myapp.area2 WARNING Jail zesty vixen who grabbed pay from quack.
1491 69 myapp.area2 ERROR The five boxing wizards jump quickly.
1499 Handlers have the following attributes and methods. Note that :class:`Handler`
1500 is never instantiated directly; this class acts as a base for more useful
1501 subclasses. However, the :meth:`__init__` method in subclasses needs to call
1502 :meth:`Handler.__init__`.
1505 .. method:: Handler.__init__(level=NOTSET)
1507 Initializes the :class:`Handler` instance by setting its level, setting the list
1508 of filters to the empty list and creating a lock (using :meth:`createLock`) for
1509 serializing access to an I/O mechanism.
1512 .. method:: Handler.createLock()
1514 Initializes a thread lock which can be used to serialize access to underlying
1515 I/O functionality which may not be threadsafe.
1518 .. method:: Handler.acquire()
1520 Acquires the thread lock created with :meth:`createLock`.
1523 .. method:: Handler.release()
1525 Releases the thread lock acquired with :meth:`acquire`.
1528 .. method:: Handler.setLevel(lvl)
1530 Sets the threshold for this handler to *lvl*. Logging messages which are less
1531 severe than *lvl* will be ignored. When a handler is created, the level is set
1532 to :const:`NOTSET` (which causes all messages to be processed).
1535 .. method:: Handler.setFormatter(form)
1537 Sets the :class:`Formatter` for this handler to *form*.
1540 .. method:: Handler.addFilter(filt)
1542 Adds the specified filter *filt* to this handler.
1545 .. method:: Handler.removeFilter(filt)
1547 Removes the specified filter *filt* from this handler.
1550 .. method:: Handler.filter(record)
1552 Applies this handler's filters to the record and returns a true value if the
1553 record is to be processed.
1556 .. method:: Handler.flush()
1558 Ensure all logging output has been flushed. This version does nothing and is
1559 intended to be implemented by subclasses.
1562 .. method:: Handler.close()
1564 Tidy up any resources used by the handler. This version does no output but
1565 removes the handler from an internal list of handlers which is closed when
1566 :func:`shutdown` is called. Subclasses should ensure that this gets called
1567 from overridden :meth:`close` methods.
1570 .. method:: Handler.handle(record)
1572 Conditionally emits the specified logging record, depending on filters which may
1573 have been added to the handler. Wraps the actual emission of the record with
1574 acquisition/release of the I/O thread lock.
1577 .. method:: Handler.handleError(record)
1579 This method should be called from handlers when an exception is encountered
1580 during an :meth:`emit` call. By default it does nothing, which means that
1581 exceptions get silently ignored. This is what is mostly wanted for a logging
1582 system - most users will not care about errors in the logging system, they are
1583 more interested in application errors. You could, however, replace this with a
1584 custom handler if you wish. The specified record is the one which was being
1585 processed when the exception occurred.
1588 .. method:: Handler.format(record)
1590 Do formatting for a record - if a formatter is set, use it. Otherwise, use the
1591 default formatter for the module.
1594 .. method:: Handler.emit(record)
1596 Do whatever it takes to actually log the specified logging record. This version
1597 is intended to be implemented by subclasses and so raises a
1598 :exc:`NotImplementedError`.
1606 The :class:`StreamHandler` class, located in the core :mod:`logging` package,
1607 sends logging output to streams such as *sys.stdout*, *sys.stderr* or any
1608 file-like object (or, more precisely, any object which supports :meth:`write`
1609 and :meth:`flush` methods).
1612 .. class:: StreamHandler([strm])
1614 Returns a new instance of the :class:`StreamHandler` class. If *strm* is
1615 specified, the instance will use it for logging output; otherwise, *sys.stderr*
1619 .. method:: emit(record)
1621 If a formatter is specified, it is used to format the record. The record
1622 is then written to the stream with a trailing newline. If exception
1623 information is present, it is formatted using
1624 :func:`traceback.print_exception` and appended to the stream.
1629 Flushes the stream by calling its :meth:`flush` method. Note that the
1630 :meth:`close` method is inherited from :class:`Handler` and so does
1631 no output, so an explicit :meth:`flush` call may be needed at times.
1639 The :class:`FileHandler` class, located in the core :mod:`logging` package,
1640 sends logging output to a disk file. It inherits the output functionality from
1641 :class:`StreamHandler`.
1644 .. class:: FileHandler(filename[, mode[, encoding[, delay]]])
1646 Returns a new instance of the :class:`FileHandler` class. The specified file is
1647 opened and used as the stream for logging. If *mode* is not specified,
1648 :const:`'a'` is used. If *encoding* is not *None*, it is used to open the file
1649 with that encoding. If *delay* is true, then file opening is deferred until the
1650 first call to :meth:`emit`. By default, the file grows indefinitely.
1652 .. versionchanged:: 2.6
1660 .. method:: emit(record)
1662 Outputs the record to the file.
1669 .. versionadded:: 2.7
1671 The :class:`NullHandler` class, located in the core :mod:`logging` package,
1672 does not do any formatting or output. It is essentially a "no-op" handler
1673 for use by library developers.
1676 .. class:: NullHandler()
1678 Returns a new instance of the :class:`NullHandler` class.
1681 .. method:: emit(record)
1683 This method does nothing.
1685 See :ref:`library-config` for more information on how to use
1686 :class:`NullHandler`.
1688 .. _watched-file-handler:
1693 .. versionadded:: 2.6
1695 .. currentmodule:: logging.handlers
1697 The :class:`WatchedFileHandler` class, located in the :mod:`logging.handlers`
1698 module, is a :class:`FileHandler` which watches the file it is logging to. If
1699 the file changes, it is closed and reopened using the file name.
1701 A file change can happen because of usage of programs such as *newsyslog* and
1702 *logrotate* which perform log file rotation. This handler, intended for use
1703 under Unix/Linux, watches the file to see if it has changed since the last emit.
1704 (A file is deemed to have changed if its device or inode have changed.) If the
1705 file has changed, the old file stream is closed, and the file opened to get a
1708 This handler is not appropriate for use under Windows, because under Windows
1709 open log files cannot be moved or renamed - logging opens the files with
1710 exclusive locks - and so there is no need for such a handler. Furthermore,
1711 *ST_INO* is not supported under Windows; :func:`stat` always returns zero for
1715 .. class:: WatchedFileHandler(filename[,mode[, encoding[, delay]]])
1717 Returns a new instance of the :class:`WatchedFileHandler` class. The specified
1718 file is opened and used as the stream for logging. If *mode* is not specified,
1719 :const:`'a'` is used. If *encoding* is not *None*, it is used to open the file
1720 with that encoding. If *delay* is true, then file opening is deferred until the
1721 first call to :meth:`emit`. By default, the file grows indefinitely.
1723 .. versionchanged:: 2.6
1727 .. method:: emit(record)
1729 Outputs the record to the file, but first checks to see if the file has
1730 changed. If it has, the existing stream is flushed and closed and the
1731 file opened again, before outputting the record to the file.
1733 .. _rotating-file-handler:
1738 The :class:`RotatingFileHandler` class, located in the :mod:`logging.handlers`
1739 module, supports rotation of disk log files.
1742 .. class:: RotatingFileHandler(filename[, mode[, maxBytes[, backupCount[, encoding[, delay]]]]])
1744 Returns a new instance of the :class:`RotatingFileHandler` class. The specified
1745 file is opened and used as the stream for logging. If *mode* is not specified,
1746 ``'a'`` is used. If *encoding* is not *None*, it is used to open the file
1747 with that encoding. If *delay* is true, then file opening is deferred until the
1748 first call to :meth:`emit`. By default, the file grows indefinitely.
1750 You can use the *maxBytes* and *backupCount* values to allow the file to
1751 :dfn:`rollover` at a predetermined size. When the size is about to be exceeded,
1752 the file is closed and a new file is silently opened for output. Rollover occurs
1753 whenever the current log file is nearly *maxBytes* in length; if *maxBytes* is
1754 zero, rollover never occurs. If *backupCount* is non-zero, the system will save
1755 old log files by appending the extensions ".1", ".2" etc., to the filename. For
1756 example, with a *backupCount* of 5 and a base file name of :file:`app.log`, you
1757 would get :file:`app.log`, :file:`app.log.1`, :file:`app.log.2`, up to
1758 :file:`app.log.5`. The file being written to is always :file:`app.log`. When
1759 this file is filled, it is closed and renamed to :file:`app.log.1`, and if files
1760 :file:`app.log.1`, :file:`app.log.2`, etc. exist, then they are renamed to
1761 :file:`app.log.2`, :file:`app.log.3` etc. respectively.
1763 .. versionchanged:: 2.6
1766 .. method:: doRollover()
1768 Does a rollover, as described above.
1771 .. method:: emit(record)
1773 Outputs the record to the file, catering for rollover as described
1776 .. _timed-rotating-file-handler:
1778 TimedRotatingFileHandler
1779 ^^^^^^^^^^^^^^^^^^^^^^^^
1781 The :class:`TimedRotatingFileHandler` class, located in the
1782 :mod:`logging.handlers` module, supports rotation of disk log files at certain
1786 .. class:: TimedRotatingFileHandler(filename [,when [,interval [,backupCount[, encoding[, delay[, utc]]]]]])
1788 Returns a new instance of the :class:`TimedRotatingFileHandler` class. The
1789 specified file is opened and used as the stream for logging. On rotating it also
1790 sets the filename suffix. Rotating happens based on the product of *when* and
1793 You can use the *when* to specify the type of *interval*. The list of possible
1794 values is below. Note that they are not case sensitive.
1796 +----------------+-----------------------+
1797 | Value | Type of interval |
1798 +================+=======================+
1799 | ``'S'`` | Seconds |
1800 +----------------+-----------------------+
1801 | ``'M'`` | Minutes |
1802 +----------------+-----------------------+
1804 +----------------+-----------------------+
1806 +----------------+-----------------------+
1807 | ``'W'`` | Week day (0=Monday) |
1808 +----------------+-----------------------+
1809 | ``'midnight'`` | Roll over at midnight |
1810 +----------------+-----------------------+
1812 The system will save old log files by appending extensions to the filename.
1813 The extensions are date-and-time based, using the strftime format
1814 ``%Y-%m-%d_%H-%M-%S`` or a leading portion thereof, depending on the
1816 If the *utc* argument is true, times in UTC will be used; otherwise
1819 If *backupCount* is nonzero, at most *backupCount* files
1820 will be kept, and if more would be created when rollover occurs, the oldest
1821 one is deleted. The deletion logic uses the interval to determine which
1822 files to delete, so changing the interval may leave old files lying around.
1824 If *delay* is true, then file opening is deferred until the first call to
1827 .. versionchanged:: 2.6
1830 .. method:: doRollover()
1832 Does a rollover, as described above.
1835 .. method:: emit(record)
1837 Outputs the record to the file, catering for rollover as described above.
1845 The :class:`SocketHandler` class, located in the :mod:`logging.handlers` module,
1846 sends logging output to a network socket. The base class uses a TCP socket.
1849 .. class:: SocketHandler(host, port)
1851 Returns a new instance of the :class:`SocketHandler` class intended to
1852 communicate with a remote machine whose address is given by *host* and *port*.
1862 Pickles the record's attribute dictionary and writes it to the socket in
1863 binary format. If there is an error with the socket, silently drops the
1864 packet. If the connection was previously lost, re-establishes the
1865 connection. To unpickle the record at the receiving end into a
1866 :class:`LogRecord`, use the :func:`makeLogRecord` function.
1869 .. method:: handleError()
1871 Handles an error which has occurred during :meth:`emit`. The most likely
1872 cause is a lost connection. Closes the socket so that we can retry on the
1876 .. method:: makeSocket()
1878 This is a factory method which allows subclasses to define the precise
1879 type of socket they want. The default implementation creates a TCP socket
1880 (:const:`socket.SOCK_STREAM`).
1883 .. method:: makePickle(record)
1885 Pickles the record's attribute dictionary in binary format with a length
1886 prefix, and returns it ready for transmission across the socket.
1889 .. method:: send(packet)
1891 Send a pickled string *packet* to the socket. This function allows for
1892 partial sends which can happen when the network is busy.
1895 .. _datagram-handler:
1900 The :class:`DatagramHandler` class, located in the :mod:`logging.handlers`
1901 module, inherits from :class:`SocketHandler` to support sending logging messages
1905 .. class:: DatagramHandler(host, port)
1907 Returns a new instance of the :class:`DatagramHandler` class intended to
1908 communicate with a remote machine whose address is given by *host* and *port*.
1913 Pickles the record's attribute dictionary and writes it to the socket in
1914 binary format. If there is an error with the socket, silently drops the
1915 packet. To unpickle the record at the receiving end into a
1916 :class:`LogRecord`, use the :func:`makeLogRecord` function.
1919 .. method:: makeSocket()
1921 The factory method of :class:`SocketHandler` is here overridden to create
1922 a UDP socket (:const:`socket.SOCK_DGRAM`).
1927 Send a pickled string to a socket.
1935 The :class:`SysLogHandler` class, located in the :mod:`logging.handlers` module,
1936 supports sending logging messages to a remote or local Unix syslog.
1939 .. class:: SysLogHandler([address[, facility]])
1941 Returns a new instance of the :class:`SysLogHandler` class intended to
1942 communicate with a remote Unix machine whose address is given by *address* in
1943 the form of a ``(host, port)`` tuple. If *address* is not specified,
1944 ``('localhost', 514)`` is used. The address is used to open a UDP socket. An
1945 alternative to providing a ``(host, port)`` tuple is providing an address as a
1946 string, for example "/dev/log". In this case, a Unix domain socket is used to
1947 send the message to the syslog. If *facility* is not specified,
1948 :const:`LOG_USER` is used.
1953 Closes the socket to the remote host.
1956 .. method:: emit(record)
1958 The record is formatted, and then sent to the syslog server. If exception
1959 information is present, it is *not* sent to the server.
1962 .. method:: encodePriority(facility, priority)
1964 Encodes the facility and priority into an integer. You can pass in strings
1965 or integers - if strings are passed, internal mapping dictionaries are
1966 used to convert them to integers.
1969 .. _nt-eventlog-handler:
1974 The :class:`NTEventLogHandler` class, located in the :mod:`logging.handlers`
1975 module, supports sending logging messages to a local Windows NT, Windows 2000 or
1976 Windows XP event log. Before you can use it, you need Mark Hammond's Win32
1977 extensions for Python installed.
1980 .. class:: NTEventLogHandler(appname[, dllname[, logtype]])
1982 Returns a new instance of the :class:`NTEventLogHandler` class. The *appname* is
1983 used to define the application name as it appears in the event log. An
1984 appropriate registry entry is created using this name. The *dllname* should give
1985 the fully qualified pathname of a .dll or .exe which contains message
1986 definitions to hold in the log (if not specified, ``'win32service.pyd'`` is used
1987 - this is installed with the Win32 extensions and contains some basic
1988 placeholder message definitions. Note that use of these placeholders will make
1989 your event logs big, as the entire message source is held in the log. If you
1990 want slimmer logs, you have to pass in the name of your own .dll or .exe which
1991 contains the message definitions you want to use in the event log). The
1992 *logtype* is one of ``'Application'``, ``'System'`` or ``'Security'``, and
1993 defaults to ``'Application'``.
1998 At this point, you can remove the application name from the registry as a
1999 source of event log entries. However, if you do this, you will not be able
2000 to see the events as you intended in the Event Log Viewer - it needs to be
2001 able to access the registry to get the .dll name. The current version does
2005 .. method:: emit(record)
2007 Determines the message ID, event category and event type, and then logs
2008 the message in the NT event log.
2011 .. method:: getEventCategory(record)
2013 Returns the event category for the record. Override this if you want to
2014 specify your own categories. This version returns 0.
2017 .. method:: getEventType(record)
2019 Returns the event type for the record. Override this if you want to
2020 specify your own types. This version does a mapping using the handler's
2021 typemap attribute, which is set up in :meth:`__init__` to a dictionary
2022 which contains mappings for :const:`DEBUG`, :const:`INFO`,
2023 :const:`WARNING`, :const:`ERROR` and :const:`CRITICAL`. If you are using
2024 your own levels, you will either need to override this method or place a
2025 suitable dictionary in the handler's *typemap* attribute.
2028 .. method:: getMessageID(record)
2030 Returns the message ID for the record. If you are using your own messages,
2031 you could do this by having the *msg* passed to the logger being an ID
2032 rather than a format string. Then, in here, you could use a dictionary
2033 lookup to get the message ID. This version returns 1, which is the base
2034 message ID in :file:`win32service.pyd`.
2041 The :class:`SMTPHandler` class, located in the :mod:`logging.handlers` module,
2042 supports sending logging messages to an email address via SMTP.
2045 .. class:: SMTPHandler(mailhost, fromaddr, toaddrs, subject[, credentials])
2047 Returns a new instance of the :class:`SMTPHandler` class. The instance is
2048 initialized with the from and to addresses and subject line of the email. The
2049 *toaddrs* should be a list of strings. To specify a non-standard SMTP port, use
2050 the (host, port) tuple format for the *mailhost* argument. If you use a string,
2051 the standard SMTP port is used. If your SMTP server requires authentication, you
2052 can specify a (username, password) tuple for the *credentials* argument.
2054 .. versionchanged:: 2.6
2055 *credentials* was added.
2058 .. method:: emit(record)
2060 Formats the record and sends it to the specified addressees.
2063 .. method:: getSubject(record)
2065 If you want to specify a subject line which is record-dependent, override
2073 The :class:`MemoryHandler` class, located in the :mod:`logging.handlers` module,
2074 supports buffering of logging records in memory, periodically flushing them to a
2075 :dfn:`target` handler. Flushing occurs whenever the buffer is full, or when an
2076 event of a certain severity or greater is seen.
2078 :class:`MemoryHandler` is a subclass of the more general
2079 :class:`BufferingHandler`, which is an abstract class. This buffers logging
2080 records in memory. Whenever each record is added to the buffer, a check is made
2081 by calling :meth:`shouldFlush` to see if the buffer should be flushed. If it
2082 should, then :meth:`flush` is expected to do the needful.
2085 .. class:: BufferingHandler(capacity)
2087 Initializes the handler with a buffer of the specified capacity.
2090 .. method:: emit(record)
2092 Appends the record to the buffer. If :meth:`shouldFlush` returns true,
2093 calls :meth:`flush` to process the buffer.
2098 You can override this to implement custom flushing behavior. This version
2099 just zaps the buffer to empty.
2102 .. method:: shouldFlush(record)
2104 Returns true if the buffer is up to capacity. This method can be
2105 overridden to implement custom flushing strategies.
2108 .. class:: MemoryHandler(capacity[, flushLevel [, target]])
2110 Returns a new instance of the :class:`MemoryHandler` class. The instance is
2111 initialized with a buffer size of *capacity*. If *flushLevel* is not specified,
2112 :const:`ERROR` is used. If no *target* is specified, the target will need to be
2113 set using :meth:`setTarget` before this handler does anything useful.
2118 Calls :meth:`flush`, sets the target to :const:`None` and clears the
2124 For a :class:`MemoryHandler`, flushing means just sending the buffered
2125 records to the target, if there is one. Override if you want different
2129 .. method:: setTarget(target)
2131 Sets the target handler for this handler.
2134 .. method:: shouldFlush(record)
2136 Checks for buffer full or a record at the *flushLevel* or higher.
2144 The :class:`HTTPHandler` class, located in the :mod:`logging.handlers` module,
2145 supports sending logging messages to a Web server, using either ``GET`` or
2149 .. class:: HTTPHandler(host, url[, method])
2151 Returns a new instance of the :class:`HTTPHandler` class. The instance is
2152 initialized with a host address, url and HTTP method. The *host* can be of the
2153 form ``host:port``, should you need to use a specific port number. If no
2154 *method* is specified, ``GET`` is used.
2157 .. method:: emit(record)
2159 Sends the record to the Web server as an URL-encoded dictionary.
2167 .. currentmodule:: logging
2169 :class:`Formatter`\ s have the following attributes and methods. They are
2170 responsible for converting a :class:`LogRecord` to (usually) a string which can
2171 be interpreted by either a human or an external system. The base
2172 :class:`Formatter` allows a formatting string to be specified. If none is
2173 supplied, the default value of ``'%(message)s'`` is used.
2175 A Formatter can be initialized with a format string which makes use of knowledge
2176 of the :class:`LogRecord` attributes - such as the default value mentioned above
2177 making use of the fact that the user's message and arguments are pre-formatted
2178 into a :class:`LogRecord`'s *message* attribute. This format string contains
2179 standard python %-style mapping keys. See section :ref:`string-formatting`
2180 for more information on string formatting.
2182 Currently, the useful mapping keys in a :class:`LogRecord` are:
2184 +-------------------------+-----------------------------------------------+
2185 | Format | Description |
2186 +=========================+===============================================+
2187 | ``%(name)s`` | Name of the logger (logging channel). |
2188 +-------------------------+-----------------------------------------------+
2189 | ``%(levelno)s`` | Numeric logging level for the message |
2190 | | (:const:`DEBUG`, :const:`INFO`, |
2191 | | :const:`WARNING`, :const:`ERROR`, |
2192 | | :const:`CRITICAL`). |
2193 +-------------------------+-----------------------------------------------+
2194 | ``%(levelname)s`` | Text logging level for the message |
2195 | | (``'DEBUG'``, ``'INFO'``, ``'WARNING'``, |
2196 | | ``'ERROR'``, ``'CRITICAL'``). |
2197 +-------------------------+-----------------------------------------------+
2198 | ``%(pathname)s`` | Full pathname of the source file where the |
2199 | | logging call was issued (if available). |
2200 +-------------------------+-----------------------------------------------+
2201 | ``%(filename)s`` | Filename portion of pathname. |
2202 +-------------------------+-----------------------------------------------+
2203 | ``%(module)s`` | Module (name portion of filename). |
2204 +-------------------------+-----------------------------------------------+
2205 | ``%(funcName)s`` | Name of function containing the logging call. |
2206 +-------------------------+-----------------------------------------------+
2207 | ``%(lineno)d`` | Source line number where the logging call was |
2208 | | issued (if available). |
2209 +-------------------------+-----------------------------------------------+
2210 | ``%(created)f`` | Time when the :class:`LogRecord` was created |
2211 | | (as returned by :func:`time.time`). |
2212 +-------------------------+-----------------------------------------------+
2213 | ``%(relativeCreated)d`` | Time in milliseconds when the LogRecord was |
2214 | | created, relative to the time the logging |
2215 | | module was loaded. |
2216 +-------------------------+-----------------------------------------------+
2217 | ``%(asctime)s`` | Human-readable time when the |
2218 | | :class:`LogRecord` was created. By default |
2219 | | this is of the form "2003-07-08 16:49:45,896" |
2220 | | (the numbers after the comma are millisecond |
2221 | | portion of the time). |
2222 +-------------------------+-----------------------------------------------+
2223 | ``%(msecs)d`` | Millisecond portion of the time when the |
2224 | | :class:`LogRecord` was created. |
2225 +-------------------------+-----------------------------------------------+
2226 | ``%(thread)d`` | Thread ID (if available). |
2227 +-------------------------+-----------------------------------------------+
2228 | ``%(threadName)s`` | Thread name (if available). |
2229 +-------------------------+-----------------------------------------------+
2230 | ``%(process)d`` | Process ID (if available). |
2231 +-------------------------+-----------------------------------------------+
2232 | ``%(message)s`` | The logged message, computed as ``msg % |
2234 +-------------------------+-----------------------------------------------+
2236 .. versionchanged:: 2.5
2237 *funcName* was added.
2240 .. class:: Formatter([fmt[, datefmt]])
2242 Returns a new instance of the :class:`Formatter` class. The instance is
2243 initialized with a format string for the message as a whole, as well as a format
2244 string for the date/time portion of a message. If no *fmt* is specified,
2245 ``'%(message)s'`` is used. If no *datefmt* is specified, the ISO8601 date format
2249 .. method:: format(record)
2251 The record's attribute dictionary is used as the operand to a string
2252 formatting operation. Returns the resulting string. Before formatting the
2253 dictionary, a couple of preparatory steps are carried out. The *message*
2254 attribute of the record is computed using *msg* % *args*. If the
2255 formatting string contains ``'(asctime)'``, :meth:`formatTime` is called
2256 to format the event time. If there is exception information, it is
2257 formatted using :meth:`formatException` and appended to the message. Note
2258 that the formatted exception information is cached in attribute
2259 *exc_text*. This is useful because the exception information can be
2260 pickled and sent across the wire, but you should be careful if you have
2261 more than one :class:`Formatter` subclass which customizes the formatting
2262 of exception information. In this case, you will have to clear the cached
2263 value after a formatter has done its formatting, so that the next
2264 formatter to handle the event doesn't use the cached value but
2265 recalculates it afresh.
2268 .. method:: formatTime(record[, datefmt])
2270 This method should be called from :meth:`format` by a formatter which
2271 wants to make use of a formatted time. This method can be overridden in
2272 formatters to provide for any specific requirement, but the basic behavior
2273 is as follows: if *datefmt* (a string) is specified, it is used with
2274 :func:`time.strftime` to format the creation time of the
2275 record. Otherwise, the ISO8601 format is used. The resulting string is
2279 .. method:: formatException(exc_info)
2281 Formats the specified exception information (a standard exception tuple as
2282 returned by :func:`sys.exc_info`) as a string. This default implementation
2283 just uses :func:`traceback.print_exception`. The resulting string is
2291 Filters can be used by :class:`Handler`\ s and :class:`Logger`\ s for
2292 more sophisticated filtering than is provided by levels. The base filter class
2293 only allows events which are below a certain point in the logger hierarchy. For
2294 example, a filter initialized with "A.B" will allow events logged by loggers
2295 "A.B", "A.B.C", "A.B.C.D", "A.B.D" etc. but not "A.BB", "B.A.B" etc. If
2296 initialized with the empty string, all events are passed.
2299 .. class:: Filter([name])
2301 Returns an instance of the :class:`Filter` class. If *name* is specified, it
2302 names a logger which, together with its children, will have its events allowed
2303 through the filter. If no name is specified, allows every event.
2306 .. method:: filter(record)
2308 Is the specified record to be logged? Returns zero for no, nonzero for
2309 yes. If deemed appropriate, the record may be modified in-place by this
2317 :class:`LogRecord` instances are created every time something is logged. They
2318 contain all the information pertinent to the event being logged. The main
2319 information passed in is in msg and args, which are combined using msg % args to
2320 create the message field of the record. The record also includes information
2321 such as when the record was created, the source line where the logging call was
2322 made, and any exception information to be logged.
2325 .. class:: LogRecord(name, lvl, pathname, lineno, msg, args, exc_info [, func])
2327 Returns an instance of :class:`LogRecord` initialized with interesting
2328 information. The *name* is the logger name; *lvl* is the numeric level;
2329 *pathname* is the absolute pathname of the source file in which the logging
2330 call was made; *lineno* is the line number in that file where the logging
2331 call is found; *msg* is the user-supplied message (a format string); *args*
2332 is the tuple which, together with *msg*, makes up the user message; and
2333 *exc_info* is the exception tuple obtained by calling :func:`sys.exc_info`
2334 (or :const:`None`, if no exception information is available). The *func* is
2335 the name of the function from which the logging call was made. If not
2336 specified, it defaults to ``None``.
2338 .. versionchanged:: 2.5
2342 .. method:: getMessage()
2344 Returns the message for this :class:`LogRecord` instance after merging any
2345 user-supplied arguments with the message.
2349 LoggerAdapter Objects
2350 ---------------------
2352 .. versionadded:: 2.6
2354 :class:`LoggerAdapter` instances are used to conveniently pass contextual
2355 information into logging calls. For a usage example , see the section on
2356 `adding contextual information to your logging output`__.
2360 .. class:: LoggerAdapter(logger, extra)
2362 Returns an instance of :class:`LoggerAdapter` initialized with an
2363 underlying :class:`Logger` instance and a dict-like object.
2365 .. method:: process(msg, kwargs)
2367 Modifies the message and/or keyword arguments passed to a logging call in
2368 order to insert contextual information. This implementation takes the object
2369 passed as *extra* to the constructor and adds it to *kwargs* using key
2370 'extra'. The return value is a (*msg*, *kwargs*) tuple which has the
2371 (possibly modified) versions of the arguments passed in.
2373 In addition to the above, :class:`LoggerAdapter` supports all the logging
2374 methods of :class:`Logger`, i.e. :meth:`debug`, :meth:`info`, :meth:`warning`,
2375 :meth:`error`, :meth:`exception`, :meth:`critical` and :meth:`log`. These
2376 methods have the same signatures as their counterparts in :class:`Logger`, so
2377 you can use the two types of instances interchangeably.
2383 The logging module is intended to be thread-safe without any special work
2384 needing to be done by its clients. It achieves this though using threading
2385 locks; there is one lock to serialize access to the module's shared data, and
2386 each handler also creates a lock to serialize access to its underlying I/O.
2388 If you are implementing asynchronous signal handlers using the :mod:`signal`
2389 module, you may not be able to use logging from within such handlers. This is
2390 because lock implementations in the :mod:`threading` module are not always
2391 re-entrant, and so cannot be invoked from such signal handlers.
2397 .. _logging-config-api:
2399 Configuration functions
2400 ^^^^^^^^^^^^^^^^^^^^^^^
2402 The following functions configure the logging module. They are located in the
2403 :mod:`logging.config` module. Their use is optional --- you can configure the
2404 logging module using these functions or by making calls to the main API (defined
2405 in :mod:`logging` itself) and defining handlers which are declared either in
2406 :mod:`logging` or :mod:`logging.handlers`.
2409 .. function:: fileConfig(fname[, defaults])
2411 Reads the logging configuration from a :mod:`ConfigParser`\-format file named
2412 *fname*. This function can be called several times from an application,
2413 allowing an end user the ability to select from various pre-canned
2414 configurations (if the developer provides a mechanism to present the choices
2415 and load the chosen configuration). Defaults to be passed to the ConfigParser
2416 can be specified in the *defaults* argument.
2419 .. function:: listen([port])
2421 Starts up a socket server on the specified port, and listens for new
2422 configurations. If no port is specified, the module's default
2423 :const:`DEFAULT_LOGGING_CONFIG_PORT` is used. Logging configurations will be
2424 sent as a file suitable for processing by :func:`fileConfig`. Returns a
2425 :class:`Thread` instance on which you can call :meth:`start` to start the
2426 server, and which you can :meth:`join` when appropriate. To stop the server,
2427 call :func:`stopListening`.
2429 To send a configuration to the socket, read in the configuration file and
2430 send it to the socket as a string of bytes preceded by a four-byte length
2431 string packed in binary using ``struct.pack('>L', n)``.
2434 .. function:: stopListening()
2436 Stops the listening server which was created with a call to :func:`listen`.
2437 This is typically called before calling :meth:`join` on the return value from
2441 .. _logging-config-fileformat:
2443 Configuration file format
2444 ^^^^^^^^^^^^^^^^^^^^^^^^^
2446 The configuration file format understood by :func:`fileConfig` is based on
2447 :mod:`ConfigParser` functionality. The file must contain sections called
2448 ``[loggers]``, ``[handlers]`` and ``[formatters]`` which identify by name the
2449 entities of each type which are defined in the file. For each such entity,
2450 there is a separate section which identifies how that entity is configured.
2451 Thus, for a logger named ``log01`` in the ``[loggers]`` section, the relevant
2452 configuration details are held in a section ``[logger_log01]``. Similarly, a
2453 handler called ``hand01`` in the ``[handlers]`` section will have its
2454 configuration held in a section called ``[handler_hand01]``, while a formatter
2455 called ``form01`` in the ``[formatters]`` section will have its configuration
2456 specified in a section called ``[formatter_form01]``. The root logger
2457 configuration must be specified in a section called ``[logger_root]``.
2459 Examples of these sections in the file are given below. ::
2462 keys=root,log02,log03,log04,log05,log06,log07
2465 keys=hand01,hand02,hand03,hand04,hand05,hand06,hand07,hand08,hand09
2468 keys=form01,form02,form03,form04,form05,form06,form07,form08,form09
2470 The root logger must specify a level and a list of handlers. An example of a
2471 root logger section is given below. ::
2477 The ``level`` entry can be one of ``DEBUG, INFO, WARNING, ERROR, CRITICAL`` or
2478 ``NOTSET``. For the root logger only, ``NOTSET`` means that all messages will be
2479 logged. Level values are :func:`eval`\ uated in the context of the ``logging``
2480 package's namespace.
2482 The ``handlers`` entry is a comma-separated list of handler names, which must
2483 appear in the ``[handlers]`` section. These names must appear in the
2484 ``[handlers]`` section and have corresponding sections in the configuration
2487 For loggers other than the root logger, some additional information is required.
2488 This is illustrated by the following example. ::
2494 qualname=compiler.parser
2496 The ``level`` and ``handlers`` entries are interpreted as for the root logger,
2497 except that if a non-root logger's level is specified as ``NOTSET``, the system
2498 consults loggers higher up the hierarchy to determine the effective level of the
2499 logger. The ``propagate`` entry is set to 1 to indicate that messages must
2500 propagate to handlers higher up the logger hierarchy from this logger, or 0 to
2501 indicate that messages are **not** propagated to handlers up the hierarchy. The
2502 ``qualname`` entry is the hierarchical channel name of the logger, that is to
2503 say the name used by the application to get the logger.
2505 Sections which specify handler configuration are exemplified by the following.
2514 The ``class`` entry indicates the handler's class (as determined by :func:`eval`
2515 in the ``logging`` package's namespace). The ``level`` is interpreted as for
2516 loggers, and ``NOTSET`` is taken to mean "log everything".
2518 .. versionchanged:: 2.6
2519 Added support for resolving the handler's class as a dotted module and class
2522 The ``formatter`` entry indicates the key name of the formatter for this
2523 handler. If blank, a default formatter (``logging._defaultFormatter``) is used.
2524 If a name is specified, it must appear in the ``[formatters]`` section and have
2525 a corresponding section in the configuration file.
2527 The ``args`` entry, when :func:`eval`\ uated in the context of the ``logging``
2528 package's namespace, is the list of arguments to the constructor for the handler
2529 class. Refer to the constructors for the relevant handlers, or to the examples
2530 below, to see how typical entries are constructed. ::
2536 args=('python.log', 'w')
2539 class=handlers.SocketHandler
2542 args=('localhost', handlers.DEFAULT_TCP_LOGGING_PORT)
2545 class=handlers.DatagramHandler
2548 args=('localhost', handlers.DEFAULT_UDP_LOGGING_PORT)
2551 class=handlers.SysLogHandler
2554 args=(('localhost', handlers.SYSLOG_UDP_PORT), handlers.SysLogHandler.LOG_USER)
2557 class=handlers.NTEventLogHandler
2560 args=('Python Application', '', 'Application')
2563 class=handlers.SMTPHandler
2566 args=('localhost', 'from@abc', ['user1@abc', 'user2@xyz'], 'Logger Subject')
2569 class=handlers.MemoryHandler
2576 class=handlers.HTTPHandler
2579 args=('localhost:9022', '/log', 'GET')
2581 Sections which specify formatter configuration are typified by the following. ::
2584 format=F1 %(asctime)s %(levelname)s %(message)s
2586 class=logging.Formatter
2588 The ``format`` entry is the overall format string, and the ``datefmt`` entry is
2589 the :func:`strftime`\ -compatible date/time format string. If empty, the
2590 package substitutes ISO8601 format date/times, which is almost equivalent to
2591 specifying the date format string ``"%Y-%m-%d %H:%M:%S"``. The ISO8601 format
2592 also specifies milliseconds, which are appended to the result of using the above
2593 format string, with a comma separator. An example time in ISO8601 format is
2594 ``2003-01-23 00:29:50,411``.
2596 The ``class`` entry is optional. It indicates the name of the formatter's class
2597 (as a dotted module and class name.) This option is useful for instantiating a
2598 :class:`Formatter` subclass. Subclasses of :class:`Formatter` can present
2599 exception tracebacks in an expanded or condensed format.
2602 Configuration server example
2603 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2605 Here is an example of a module using the logging configuration server::
2608 import logging.config
2612 # read initial config file
2613 logging.config.fileConfig("logging.conf")
2615 # create and start listener on port 9999
2616 t = logging.config.listen(9999)
2619 logger = logging.getLogger("simpleExample")
2622 # loop through logging calls to see the difference
2623 # new configurations make, until Ctrl+C is pressed
2625 logger.debug("debug message")
2626 logger.info("info message")
2627 logger.warn("warn message")
2628 logger.error("error message")
2629 logger.critical("critical message")
2631 except KeyboardInterrupt:
2633 logging.config.stopListening()
2636 And here is a script that takes a filename and sends that file to the server,
2637 properly preceded with the binary-encoded length, as the new logging
2640 #!/usr/bin/env python
2641 import socket, sys, struct
2643 data_to_send = open(sys.argv[1], "r").read()
2647 s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
2648 print "connecting..."
2649 s.connect((HOST, PORT))
2650 print "sending config..."
2651 s.send(struct.pack(">L", len(data_to_send)))
2652 s.send(data_to_send)
2660 Multiple handlers and formatters
2661 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2663 Loggers are plain Python objects. The :func:`addHandler` method has no minimum
2664 or maximum quota for the number of handlers you may add. Sometimes it will be
2665 beneficial for an application to log all messages of all severities to a text
2666 file while simultaneously logging errors or above to the console. To set this
2667 up, simply configure the appropriate handlers. The logging calls in the
2668 application code will remain unchanged. Here is a slight modification to the
2669 previous simple module-based configuration example::
2673 logger = logging.getLogger("simple_example")
2674 logger.setLevel(logging.DEBUG)
2675 # create file handler which logs even debug messages
2676 fh = logging.FileHandler("spam.log")
2677 fh.setLevel(logging.DEBUG)
2678 # create console handler with a higher log level
2679 ch = logging.StreamHandler()
2680 ch.setLevel(logging.ERROR)
2681 # create formatter and add it to the handlers
2682 formatter = logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
2683 ch.setFormatter(formatter)
2684 fh.setFormatter(formatter)
2685 # add the handlers to logger
2686 logger.addHandler(ch)
2687 logger.addHandler(fh)
2689 # "application" code
2690 logger.debug("debug message")
2691 logger.info("info message")
2692 logger.warn("warn message")
2693 logger.error("error message")
2694 logger.critical("critical message")
2696 Notice that the "application" code does not care about multiple handlers. All
2697 that changed was the addition and configuration of a new handler named *fh*.
2699 The ability to create new handlers with higher- or lower-severity filters can be
2700 very helpful when writing and testing an application. Instead of using many
2701 ``print`` statements for debugging, use ``logger.debug``: Unlike the print
2702 statements, which you will have to delete or comment out later, the logger.debug
2703 statements can remain intact in the source code and remain dormant until you
2704 need them again. At that time, the only change that needs to happen is to
2705 modify the severity level of the logger and/or handler to debug.
2708 Using logging in multiple modules
2709 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2711 It was mentioned above that multiple calls to
2712 ``logging.getLogger('someLogger')`` return a reference to the same logger
2713 object. This is true not only within the same module, but also across modules
2714 as long as it is in the same Python interpreter process. It is true for
2715 references to the same object; additionally, application code can define and
2716 configure a parent logger in one module and create (but not configure) a child
2717 logger in a separate module, and all logger calls to the child will pass up to
2718 the parent. Here is a main module::
2721 import auxiliary_module
2723 # create logger with "spam_application"
2724 logger = logging.getLogger("spam_application")
2725 logger.setLevel(logging.DEBUG)
2726 # create file handler which logs even debug messages
2727 fh = logging.FileHandler("spam.log")
2728 fh.setLevel(logging.DEBUG)
2729 # create console handler with a higher log level
2730 ch = logging.StreamHandler()
2731 ch.setLevel(logging.ERROR)
2732 # create formatter and add it to the handlers
2733 formatter = logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
2734 fh.setFormatter(formatter)
2735 ch.setFormatter(formatter)
2736 # add the handlers to the logger
2737 logger.addHandler(fh)
2738 logger.addHandler(ch)
2740 logger.info("creating an instance of auxiliary_module.Auxiliary")
2741 a = auxiliary_module.Auxiliary()
2742 logger.info("created an instance of auxiliary_module.Auxiliary")
2743 logger.info("calling auxiliary_module.Auxiliary.do_something")
2745 logger.info("finished auxiliary_module.Auxiliary.do_something")
2746 logger.info("calling auxiliary_module.some_function()")
2747 auxiliary_module.some_function()
2748 logger.info("done with auxiliary_module.some_function()")
2750 Here is the auxiliary module::
2755 module_logger = logging.getLogger("spam_application.auxiliary")
2759 self.logger = logging.getLogger("spam_application.auxiliary.Auxiliary")
2760 self.logger.info("creating an instance of Auxiliary")
2761 def do_something(self):
2762 self.logger.info("doing something")
2764 self.logger.info("done doing something")
2766 def some_function():
2767 module_logger.info("received a call to \"some_function\"")
2769 The output looks like this::
2771 2005-03-23 23:47:11,663 - spam_application - INFO -
2772 creating an instance of auxiliary_module.Auxiliary
2773 2005-03-23 23:47:11,665 - spam_application.auxiliary.Auxiliary - INFO -
2774 creating an instance of Auxiliary
2775 2005-03-23 23:47:11,665 - spam_application - INFO -
2776 created an instance of auxiliary_module.Auxiliary
2777 2005-03-23 23:47:11,668 - spam_application - INFO -
2778 calling auxiliary_module.Auxiliary.do_something
2779 2005-03-23 23:47:11,668 - spam_application.auxiliary.Auxiliary - INFO -
2781 2005-03-23 23:47:11,669 - spam_application.auxiliary.Auxiliary - INFO -
2782 done doing something
2783 2005-03-23 23:47:11,670 - spam_application - INFO -
2784 finished auxiliary_module.Auxiliary.do_something
2785 2005-03-23 23:47:11,671 - spam_application - INFO -
2786 calling auxiliary_module.some_function()
2787 2005-03-23 23:47:11,672 - spam_application.auxiliary - INFO -
2788 received a call to "some_function"
2789 2005-03-23 23:47:11,673 - spam_application - INFO -
2790 done with auxiliary_module.some_function()