finished

find-orphans.py

@@ -2,6 +2,8 @@
 # encoding: utf-8
 
 # library imports
+import fnmatch
+import glob
 import os
 import over
 import re
@@ -23,8 +25,9 @@ _print = over.core.textui.Output('Find Orphans')
 
 if __name__ == '__main__':
 	main = over.core.app.Main('Find Orphans', None, 'AWARE-Overwatch Joint Software License')
-	main.add_option('filter', 'str', ['/home/*', '/var/*', '*.pyc'], 'Shell globs that should be filtered out.', plural=True)
+	main.add_option('filter', 'str', ['/home/*', '/var/*', '*.pyc', '/lib*/modules/*', '*.pyo'], 'Shell globs that should be filtered out.', plural=True)
 	main.add_option('null-separated', 'bool', False, 'Output each filename separated with a null character. Otherwise use a newline.', short_name='0')
+	main.add_option('output', 'bool', False, 'Output filenames to stdout. Otherwise just counts them.')
 	main.add_help('Description', ['Finds files on Gentoo systems that aren\'t owned by any installed package and lists them.'])
 	main.enable_help('h')
 	main.parse()
@@ -34,17 +37,17 @@ if __name__ == '__main__':
 		main.exit(silent=True)
 	
 	# --------------------------------------------------
-	# Create a file list
+	# Create a list of all files
 	
-	file_list = []
+	all_files = set()
 	threads = [] # {proc, output_buffer}
 	
 	for root in main.targets:
-		proc = over.core.cmd.Command('find', root, '-print0')
+		proc = over.core.cmd.Command('find', root, '-type', 'f', '-print0')
 		proc.run()
-		threads.append({'proc': proc, 'output_buffer': []})
+		threads.append({'proc': proc, 'output_buffer': [], 'root': root})
 	
-	_print('gathering file list using %d processes' %(len(threads)), prefix.start)
+	_print('gathering a list of all files using %d processes' %(len(threads)), prefix.start)
 	
 	while True:
 		at_least_one_lives = False
@@ -57,10 +60,50 @@ if __name__ == '__main__':
 				at_least_one_lives = True
 		
 		if at_least_one_lives:
-			time.sleep(1)
+			time.sleep(.25)
 		else:
 			break
 	
 	_print('post-processing file list')
 	
+	for thread in threads:
+		output = b''.join(thread['output_buffer']).decode('utf-8')
+		all_files.update(os.path.normpath(os.path.join(os.path.abspath(thread['root']), f)) for f in output.split('\x00') if f)
 	
+	_print('found %d files' %(len(all_files)), prefix.done)
+	
+	# --------------------------------------------------
+	# create a list of owned files
+	
+	_print('gathering a list of files owned by installed packages', prefix.start)
+	
+	owned_files = set()
+	
+	for contents_filename in glob.glob('/var/db/pkg/*/*/CONTENTS'):
+		owned_files.update(re.findall('obj (.+) [0-9a-f]+ \d+', open(contents_filename).read()))
+	
+	unowned_files = all_files - owned_files
+	
+	_print('found %d owned files' %(len(owned_files)), prefix.done)
+	_print('there are %d unowned files' %(len(unowned_files)))
+	
+	# --------------------------------------------------
+	# apply filters
+	
+	_print('applying filters', prefix.start)
+	
+	unowned_files_ignored = set()
+	
+	for filter in main.cfg.filter:
+		unowned_files_ignored.update(fnmatch.filter(unowned_files, filter))
+	
+	files_to_output = unowned_files - unowned_files_ignored
+	
+	_print('found §y%d unowned files§/' %(len(files_to_output)), prefix.done)
+	
+	# --------------------------------------------------
+	# output
+	
+	if main.cfg.output:
+		joiner = '\x00' if main.cfg.null_separated else '\n'
+		print(joiner.join(sorted(files_to_output)))