From 242896ba37f409042856112261a0e4c4cb5f9657 Mon Sep 17 00:00:00 2001
From: Martinez <overwatch@covalent.cz>
Date: Thu, 12 Feb 2015 13:15:51 +0100
Subject: [PATCH] app config, parallel filelist generation

---
 .gitignore      |  3 +++
 find-orphans.py | 66 +++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 69 insertions(+)
 create mode 100644 .gitignore
 create mode 100755 find-orphans.py

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..6a5929e
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,3 @@
+*.pyc
+__pycache__
+over
diff --git a/find-orphans.py b/find-orphans.py
new file mode 100755
index 0000000..93c6d0a
--- /dev/null
+++ b/find-orphans.py
@@ -0,0 +1,66 @@
+#! /bin/env python3
+# encoding: utf-8
+
+# library imports
+import os
+import over
+import re
+import sys
+import time
+
+# --------------------------------------------------
+
+prefix = over.core.textui.prefix
+_print = over.core.textui.Output('Find Orphans')
+
+# --------------------------------------------------
+# Functions
+
+# --------------------------------------------------
+# Classes
+
+# --------------------------------------------------
+
+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('null-separated', 'bool', False, 'Output each filename separated with a null character. Otherwise use a newline.', short_name='0')
+	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()
+	
+	if not main.targets:
+		_print('specify at least one directory to work on, e.g. \'/\'', prefix.fail)
+		main.exit(silent=True)
+	
+	# --------------------------------------------------
+	# Create a file list
+	
+	file_list = []
+	threads = [] # {proc, output_buffer}
+	
+	for root in main.targets:
+		proc = over.core.cmd.Command('find', root, '-print0')
+		proc.run()
+		threads.append({'proc': proc, 'output_buffer': []})
+	
+	_print('gathering file list using %d processes' %(len(threads)), prefix.start)
+	
+	while True:
+		at_least_one_lives = False
+		
+		for thread in threads:
+			tmp = thread['proc'].get_output()
+			
+			if not tmp is None:
+				thread['output_buffer'].append(tmp)
+				at_least_one_lives = True
+		
+		if at_least_one_lives:
+			time.sleep(1)
+		else:
+			break
+	
+	_print('post-processing file list')
+	
+