The zen of logging and YAML

Python 2.7 and 3.2 come with a new way to configure the Python logging services with a standard dict. Is there a better way to let an user provide a dict than YAML ?

ConfigParser ? Uh ! Not sure. In addition, YAML provides OOTB lists, aliases, and even arbitrary Python objects.

Just having got the zen of complex logging and YAML, I write this self explanatory recipe for my memory and anyone who want to configure a complex logging with only some lines of Python. Any comment, question, improvement suggestion is welcome.

# -*- coding: utf-8
"""\
A simple demo of logging configuration with YAML (Python 2.7)
=============================================================
Requires PyYAML -> "easy_install PyYAML"
See the recipes for configuring logging with dicts and YAML
- http://docs.python.org/2.7/howto/logging-cookbook.html
- http://stackoverflow.com/questions/10519392/python2-7-logging-configuration-with-yaml
"""
import logging
import logging.config
import yaml
import StringIO
# Should be a config file in real app
YAML_CONF = """\
%YAML 1.2
---
# Config for my application
# --------------------------
myapp:
foo: bar
bar: [1, 2]
# Config for logging
# ------------------
# See http://docs.python.org/2.7/library/logging.config.html#configuration-dictionary-schema
logging:
version: 1
disable_existing_loggers: true
# Configuring the default (root) logger is highly recommended
root:
level: !!python/name:logging.NOTSET
handlers: [console]
loggers:
# Logging from my application
myapp.lib:
level: !!python/name:logging.WARN
handlers: [logfile]
qualname: myapp.lib
propagate: false
myapp.cli:
level: !!python/name:logging.WARN
handlers: [console]
qualname: myapp.cli
propagate: false
# Controlling logging of 3rd party libs
sqlalchemy.engine:
level: !!python/name:logging.WARN
handlers: [logfile]
qualname: sqlalchemy.engine
propagate: false
sqlalchemy.pool:
level: !!python/name:logging.WARN
handlers: [logfile]
qualname: sqlalchemy.pool
propagate: false
handlers:
logfile:
class: logging.FileHandler
filename: sample.log
formatter: simpleFormatter
level: !!python/name:logging.NOTSET
console:
class: logging.StreamHandler
stream: ext://sys.stdout
formatter: simpleFormatter
level: !!python/name:logging.NOTSET
formatters:
simpleFormatter:
class: !!python/name:logging.Formatter
format: '%(name)s %(asctime)s %(levelname)s %(message)s'
datefmt: '%d/%m/%Y %H:%M:%S'
"""
# Loading config. Of course this is in another file in the real life
global_config = yaml.load(StringIO.StringIO(YAML_CONF))
# Configuring logging with the subset of the dict
#
logging.config.dictConfig(global_config['logging'])
# Using explicitely the root logger always logs to the console
logging.info("This is an info of the root logger")
# The unconfigured loggers are captured by the root logger (-> console)
unconfigured_logger = logging.getLogger('unconfigured')
unconfigured_logger.info("This is an info from an unknown / unconfigured source")
# Logging from myapp.cli
myapp_cli_logger = logging.getLogger('myapp.cli')
myapp_cli_logger.info("This is an info from myapp.cli") # Not recorded
myapp_cli_logger.warning("This is a warning from myapp.cli") # -> console
# Logging from myapp.lib
myapp_lib_logger = logging.getLogger('myapp.lib')
myapp_lib_logger.info("This is an info from myapp.lib") # Not recorded
myapp_lib_logger.warning("This is a warning from myapp.lib") # -> sample.log
# Controlling logs from 3rd party libs
sqla_logger = logging.getLogger('sqlalchemy.engine')
sqla_logger.info("This is an info from SQLAlchemy") # Not recorded
sqla_logger.warning("This is a warning from SQLAlchemy") # -> sample.log
print "Now look at the file 'sample.log'"
view raw logging_yaml_config.py hosted with ❤ by GitHub

Comments !

links

social