Recently, I was made a service which can provide a simple way to get best model. so, i spent lot of time to read source code of auto-sklearn, auto-skl ...
Recently, I was made a service which can provide a simple way to get best model. so, i spent lot of time to read source code of auto-sklearn, auto-sklearn is an automated machine learning toolkit and a drop-in replacement for a scikit-learn estimator.
when I write my logging module I found the logging module of auto-sklearn which used yaml file as config, the config file:
1 --- 2 version: 1 3 disable_existing_loggers: False 4 formatters: 5 simple: 6 format: '[%(asctime)s: %(levelname)s] %(message)s' 7 8 handlers: 9 console: 10 class: logging.StreamHandler 11 level: WARNING 12 formatter: simple 13 stream: ext://sys.stdout 14 15 file_handler: 16 class: logging.FileHandler 17 level: DEBUG 18 formatter: simple 19 filename: autosklearn.log 20 21 root: 22 level: ERROR 23 handlers: [console, file_handler] 24 25 loggers: 26 server: 27 level: INFO 28 handlers: [file_handler] 29 propagate: no
the caller can define loggers、handlers、formatters here, after that, caller need write setup code and get logger code.
1 # -*- encoding: utf-8 -*- 2 import logging 3 import logging.config 4 import os 5 import yaml 6 7 8 def setup_logger(output_file=None, logging_config=None): 9 # logging_config must be a dictionary object specifying the configuration 10 if not os.path.exists(os.path.dirname(output_file)): 11 os.makedirs(os.path.dirname(output_file)) 12 if logging_config is not None: 13 if output_file is not None: 14 logging_config['handlers']['file_handler']['filename'] = output_file 15 logging.config.dictConfig(logging_config) 16 else: 17 with open(os.path.join(os.path.dirname(__file__), 'logging.yaml'), 18 'r') as fh: 19 logging_config = yaml.safe_load(fh) 20 if output_file is not None: 21 logging_config['handlers']['file_handler']['filename'] = output_file 22 logging.config.dictConfig(logging_config) 23 24 25 def _create_logger(name): 26 return logging.getLogger(name) 27 28 29 def get_logger(name): 30 logger = PickableLoggerAdapter(name) 31 return logger 32 33 34 class PickableLoggerAdapter(object): 35 36 def __init__(self, name): 37 self.name = name 38 self.logger = _create_logger(name) 39 40 def __getstate__(self): 41 """ 42 Method is called when pickle dumps an object. 43 44 Returns 45 ------- 46 Dictionary, representing the object state to be pickled. Ignores 47 the self.logger field and only returns the logger name. 48 """ 49 return {'name': self.name} 50 51 def __setstate__(self, state): 52 """ 53 Method is called when pickle loads an object. Retrieves the name and 54 creates a logger. 55 56 Parameters 57 ---------- 58 state - dictionary, containing the logger name. 59 60 """ 61 self.name = state['name'] 62 self.logger = _create_logger(self.name) 63 64 def debug(self, msg, *args, **kwargs): 65 self.logger.debug(msg, *args, **kwargs) 66 67 def info(self, msg, *args, **kwargs): 68 self.logger.info(msg, *args, **kwargs) 69 70 def warning(self, msg, *args, **kwargs): 71 self.logger.warning(msg, *args, **kwargs) 72 73 def error(self, msg, *args, **kwargs): 74 self.logger.error(msg, *args, **kwargs) 75 76 def exception(self, msg, *args, **kwargs): 77 self.logger.exception(msg, *args, **kwargs) 78 79 def critical(self, msg, *args, **kwargs): 80 self.logger.critical(msg, *args, **kwargs) 81 82 def log(self, level, msg, *args, **kwargs): 83 self.logger.log(level, msg, *args, **kwargs) 84 85 def isEnabledFor(self, level): 86 return self.logger.isEnabledFor(level)
get_logger return a logger object, setup_logger setup logger handler which define output where.
when you use it,like this
1 from .util import get_logger, setup_logger 2 3 4 APP_DIR = os.path.dirname(os.path.abspath(__file__)) 5 setup_logger(output_file=os.path.join(APP_DIR, 'logs', 'server.log')) 6 LOG = get_logger("server")
Define a global variable LOG for other modules to call.
all done.