Python在不同的日志级别记录到多个处理程序？

Question

伙计们,

我正在抓住一个我无法做对的python日志配置.

假设我安装了以下软件包:

mypackage/
   data/mypackage.logging.conf
   module1.py
   module2.py
   scripts/main.py

由于脚本可以交互使用或从crontab运行,因此我有以下要求:

1. 没有打印声明,一切都通过记录;

2. 使用a记录timedRotatingFileHandler,始终处于DEBUG级别;

3. 使用a记录mailinglogger.SummarisingLogger,始终处于INFO级别;

4. 登录到控制台,默认情况下将级别设置为INFO或通过命令行选项覆盖.

问题是,我可以通过命令行更改日志级别,并相应地更改控制台日志级别,但其他处理程序也更改,我不想要...: - /

在日志记录配置文件中,我不确定我是否理解根记录器级别,其他记录器级别和处理程序级别设置之间的优先级.

这是一些示例代码.任何线索将不胜感激:-)

# mypackage/data/mypackage.logging.conf
[loggers]
root,mypackage

[handlers]
keys=consoleHandler,timedRotatingFileHandler,summarisingHandler

[formatters]
keys=simpleFormatter,consoleFormatter,mypackageFormatter

[logger_root]
#level=INFO
handlers=consoleHandler

[logger_mypackage]
#level=INFO
handlers=timedRotatingFileHandler,summarisingHandler
qualname=mypackage

[handler_consoleHandler]
class=StreamHandler
#level=INFO
formatter=consoleFormatter
args=(sys.stdout,)

[handler_timedRotatingFileHandler]
class=logging.handlers.TimedRotatingFileHandler
level=DEBUG
formatter=mypackageFormatter
args=('mypackage.log', 'M', 1, 5)

[handler_summarisingHandler]
class=mailinglogger.SummarisingLogger
level=INFO
formatter=mypackageFormatter
args=('mypackage@someserver.somewhere.com', ('mypackage-alerts@somewhere.com',), 'relay.somewhere.com')

---

#mypackage/scripts/main.py:
import logging
import logging.config
import os
import sys

import mypackage.module1
import mypackage.module2

logging.config.fileConfig('data/mypackage.logging.conf')
log = logging.getLogger(__name__)

if __name__ == '__main__':
    loglevel = 'INFO'
    if len(sys.argv) > 1:
        loglevel = sys.argv[1].upper()

    logging.getLogger('').setLevel(getattr(logging, loglevel))
    # or logging.getLogger('mypackage').setLevel(getattr(logging, loglevel)) ?

    mypackage.module1.do_something()
    mypackage.module2.do_something_else()

---

#mypackage/module1.py:
import logging

log = logging.getLogger(__name__)
log.addHandler(NullHandler())

def do_something():
    log.debug("some debug message from:" + __name__)
    log.info("some info message from:" + __name__)
    log.error("some error message from:" + __name__)

---

#mypackage/module2.py:
import logging

log = logging.getLogger(__name__)
log.addHandler(NullHandler())

def do_something_else():
    log.debug("some debug message from:" + __name__)
    log.info("some info message from:" + __name__)
    log.error("some error message from:" + __name__)

---

更新1

与此同时,我发现了这个答案并成功修改了我的代码:

#mypackage/scripts/main.py:
import logging
import logging.config
import os
import sys
import mailinglogger

import mypackage.module1
import mypackage.module2

def main():
    # get the console log level from the command-line
    loglevel = 'INFO'
    if len(sys.argv) > 1:
        loglevel = sys.argv[1].upper()

    # create formatters
    simple_formatter = logging.Formatter("%(name)s:%(levelname)s: %(message)s")
    detailed_formatter = logging.Formatter("%(asctime)s %(name)s[%(process)d]: %(levelname)s - %(message)s")

    # get a top-level "mypackage" logger,
    # set its log level to DEBUG,
    # BUT PREVENT IT from propagating messages to the root logger
    #
    log = logging.getLogger('mypackage')
    log.setLevel(logging.DEBUG)
    log.propagate = 0

    # create a console handler
    # and set its log level to the command-line option 
    # 
    console_handler = logging.StreamHandler(sys.stdout)
    console_handler.setLevel(getattr(logging, loglevel))
    console_handler.setFormatter(simple_formatter)

    # create a file handler
    # and set its log level to DEBUG
    #
    file_handler = logging.handlers.TimedRotatingFileHandler('mypackage.log', 'M', 1, 5)
    file_handler.setLevel(logging.DEBUG)
    file_handler.setFormatter(detailed_formatter)

    # create a mail handler
    # and set its log level to INFO
    #
    mail_handler = mailinglogger.SummarisingLogger(
        'mypackage@someserver.somewhere.com', ('mypackage-alerts@somewhere.com',), 'relay.somewhere.com')
    mail_handler.setLevel(logging.INFO)
    mail_handler.setFormatter(detailed_formatter)

    # add handlers to the "mypackage" logger
    #
    log.addHandler(console_handler)
    log.addHandler(file_handler)
    log.addHandler(mail_handler)

    # let the modules do their stuff 
    # and log to the "mypackage.module1" and "mypackage.module2" loggers
    #
    mypackage.module1.do_something()
    mypackage.module2.do_something_else()


if __name__ == '__main__':
    main()

现在,我将尝试在logging.config文件中翻译它...

---

更新2

这是我找到的最好的日志配置和代码组合.

在mypackage.logging.conf文件中,"mypackage"记录器是:

• 仅使用文件和电子邮件处理程序进行设置;
• 它的传播设置为假;
• 它的级别设置为DEBUG;
• 而文件和电子邮件处理程序分别设置为INFO和DEBUG.

---

#mypackage/data/mypackage.logging.conf
[loggers]
keys=root,mypackage

[handlers]
keys=consoleHandler,timedRotatingFileHandler,summarisingHandler

[formatters]
keys=simpleFormatter,consoleFormatter,mypackageFormatter

[logger_root]
#level=INFO
handlers=consoleHandler

[logger_mypackage]
level=DEBUG
handlers=timedRotatingFileHandler,summarisingHandler
qualname=mypackage
propagate=0

[handler_consoleHandler]
class=StreamHandler
#level=INFO
formatter=consoleFormatter
args=(sys.stdout,)

[handler_timedRotatingFileHandler]
class=logging.handlers.TimedRotatingFileHandler
level=DEBUG
formatter=mypackageFormatter
args=('mypackage.log', 'M', 1, 5)

[handler_summarisingHandler]
class=mailinglogger.SummarisingLogger
level=INFO
formatter=mypackageFormatter
args=('mypackage@someserver.somewhere.com', ('mypackage-alerts@somewhere.com',), 'relay.somewhere.com')

[formatter_consoleFormatter]
format=%(levelname)s: %(message)s
datefmt=

[formatter_mypackageFormatter]
format=%(asctime)s %(name)s[%(process)d]: %(levelname)s - %(message)s
datefmt=

---

在脚本中:

1. 读取日志配置;

2. (重新)创建console_formatter;

3. 使用命令行选项中的日志级别创建控制台处理程序,然后将其添加到"mypackage"记录器中.

---

import logging
import logging.config
import os
import sys

import mypackage.module1
import mypackage.module2

def setup_logging(loglevel):
    #
    # load logging config from file
    #
    logging.config.fileConfig('data/mypackage.logging.conf')

    # (re-)create formatter
    console_formatter = logging.Formatter("%(name)s:%(levelname)s: %(message)s")

    # create a console handler
    # and set its log level to the command-line option 
    # 
    console_handler = logging.StreamHandler(sys.stdout)
    console_handler.setFormatter(console_formatter)
    console_handler.setLevel(getattr(logging, loglevel))

    # add console handler to the pre-configured "mypackage" logger
    #
    logger = logging.getLogger('mypackage')
    logger.addHandler(console_handler)


def main():
    # get the console log level from the command-line
    loglevel = 'INFO'
    if len(sys.argv) > 1:
        loglevel = sys.argv[1].upper()

    # load logging config and setup console handler
    #
    setup_logging(loglevel)

    # log from the script to the "mypackage.scripts.main" logger
    #
    log = logging.getLogger(__name__)
    log.debug("some debug message from:" + __name__)
    log.info("some info message from:" + __name__)
    log.error("some error message from:" + __name__)

    # let the modules do their stuff 
    # and log to the "mypackage.module1" and "mypackage.module2" loggers
    #
    mypackage.module1.do_something()
    mypackage.module2.do_something_else()

if __name__== '__main__':
    main()

---

如果处理程序在从配置文件加载时按名称"可寻址",事情会更简单.

然后,我们可以在配置文件中设置mypackage控制台处理程序,并在代码中更改其日志级别,如下所示:

def setup_logging(loglevel):
    logging.config.fileConfig('data/mypackage.logging.conf')

    logger = logging.getLogger('mypackage')
    console_handler = logger.getHandler('consoleHandler')
    console_handler.setLevel(getattr(logging, loglevel))

也没有必要重新创建格式化程序......

(最后更新:是的,我知道的https://docs.python.org/3/library/logging.config.html#incremental-configuration,但在这种情况下,我坚持使用Python 2.6 ... :-)

Answer 1

更新处理程序的方法：

import logging

from rootmodule.mymodule import mylogger

def update_handler_level(logger,  handler_type, level="INFO"):
    # if not root logger user logger.parent
    for handler in logger.handlers or logger.parent.handlers:
        if isinstance(handler, handler_type):
            print(handler.level)
            handler.setLevel(getattr(logging, level, "INFO"))
            print(handler.level)

mylogger.debug('test')
update_handler_level(mylogger, logging.StreamHandler)
mylogger.debug('test')

我的logging.cfg与你的非常相似，除了记录器名称si在常量模块中设置（可以在不破坏日志记录配置的情况下进行模块重命名）

要从命令行更新，您必须在 opts 值和logging.Handler 子类名称之间有一个映射。