Feat: support log file name pattern (6e4521dd) · Commits · The Tor Project / Onion Services / Onionspray Log Parser

packages/eotk_log_parser/main.py

+14 −1

Original line number	Diff line number	Diff line
		@@ -51,6 +51,9 @@ date_format = "%d/%b/%Y:%H:%M:%S %z"
		# Default extension for the logfiles to search
		default_extension = 'gz'

		# Default pattern
		default_log_file_pattern = '*'

		# Default path where to search logs
		default_path = 'logs'

		@@ -69,6 +72,7 @@ class EotkLogParser():
		ignore_user_agent_pattern=default_ignore_user_agent_pattern,
		pagehit_pattern=default_pagehit_pattern,
		csv_output_file=default_csv_output_file,
		log_file_pattern=default_log_file_pattern,
		):
		"""
		Class initialization.
		@@ -78,7 +82,7 @@ class EotkLogParser():
		raise FileNotFoundError(path)

		self.folder = Path(path)
		self.logfiles = list(self.folder.glob('*/.' + extension))
		self.logfiles = list(self.folder.glob('**/' + log_file_pattern + '.' + extension))
		self.extension = extension
		self.stats = {}
		self.dates = {}
		@@ -102,6 +106,7 @@ class EotkLogParser():
		self.csv_output_file = csv_output_file

		self.log('Folder: ' + str(path))
		self.log('Log file name pattern: ' + str(log_file_pattern))
		self.log('Extension: ' + str(extension))
		self.log('Ignore User-Agent pattern: ' + str(ignore_user_agent_pattern))
		self.log('Page Hit pattern: ' + str(pagehit_pattern))
		@@ -265,6 +270,14 @@ def cmdline_parser():
		default=default_path,
		help="Folder where to search for logs (default: \"%(default)s\")")

		parser.add_argument('-l', '--log_file_pattern',
		default=default_log_file_pattern,
		help="""
		Log file name pattern, without the extension (default: \"%(default)s\")
		Example: '202210 to consider only logs containing 202210 on their file names.
		"""
		)

		parser.add_argument('-e', '--extension',
		default=default_extension,
		help="""