Configurable logging

[janosh]

run_locally allows toggling a python std lib logger

jobflow/src/jobflow/managers/local.py

Line 18 in 947072a

log: bool = True,

whose settings are hard-coded in initialize_logger. would be great if run_locally accepted a list of custom handlers. we run DFT in containers on cloud infra and would like to send our logs to ElasticSearch which currently seems to require a lot of hackery

jobflow/src/jobflow/utils/log.py

Lines 5 to 28 in 947072a

	def initialize_logger(level: int = logging.INFO) -> logging.Logger:
	"""Initialize the default logger.
	Parameters
	----------
	level
	The log level.
	Returns
	-------
	Logger
	A logging instance with customized formatter and handlers.
	"""
	import sys
	log = logging.getLogger("jobflow")
	log.setLevel(level)
	log.handlers = [] # reset logging handlers if they already exist
	fmt = logging.Formatter("%(asctime)s %(levelname)s %(message)s")
	screen_handler = logging.StreamHandler(stream=sys.stdout)
	screen_handler.setFormatter(fmt)
	log.addHandler(screen_handler)