path: root/exclude.patch

--- lb.py~	2008-03-04 19:14:37.000000000 -0500
+++ lb.py	2008-05-13 22:05:54.000000000 -0400
@@ -6,7 +6,7 @@
 of backed-up trees, and intelligently handles renames, moves, and
 duplicate files without additional storage or transfer.
 
-Transfer occurs over standard i/o locally or remotely between a client and
+Transfer occurs over standard I/O locally or remotely between a client and
 server instance of this script. Remote backups rely on the secure remote
 shell program ssh.
 
@@ -28,18 +28,21 @@
 
 Options:
 
-    --verify          Run rsync with --dry-run to cross-verify
-    --numeric-ids     Keep uid/gid values instead of mapping; requires root
-    --minutes <mins>  Only run for <mins> minutes. Incremental backup.
-    --showfiles       Don't backup, only list relative path files needing
-                      backup
-    --catalogonly     Update catalog only
-    --filelist <- or file> Specify filelist. Files relative to srcdir.
-    --lock            Ensure only one backup to a given dest will run at a time
-    --verbose         Show what is happening
-    --ssh-i <file>    Select id file to use for authentication (ssh -i)
-    --ssh-C           Use ssh compression (ssh -C)
-    --ssh-p <port>    ssh port on remote host (ssh -p)
+  --catalogonly             Update catalog only
+  --exclude     <file>      Exclude paths matching the regexp rules in file
+  --filelist    <- or file> Specify filelist. Files relative to srcdir.
+  --lock                    Ensure only one backup to a given dest will run at
+                            a time
+  --minutes     <mins>      Only run for <mins> minutes. Incremental backup
+  --numeric-ids             Keep uid/gid values instead of mapping;
+                            requires root
+  --showfiles               Don't backup, only list relative path files needing
+                            backup
+  --ssh-C                   Use ssh compression (ssh -C)
+  --ssh-i       <file>      Select id file to use for authentication (ssh -i)
+  --ssh-p       <port>      ssh port on remote host (ssh -p)
+  --verbose                 Show what is happening
+  --verify                  Run rsync with --dry-run to cross-verify
 
 Comments:
 
@@ -69,19 +72,19 @@
 
 Example 1:
 
-python lb.py pictures pictures-backup
+lb pictures pictures-backup
 
 Makes a new backup of pictures in pictures-backup.
 
 Example 2:
 
-python lb.py pictures me@fluffy:~/pictures-backup
+lb pictures me@fluffy:~/pictures-backup
 
 Backs up on remote machine fluffy instead of locally.
 
 Example 3:
 
-python lb.py --minutes 240 pictures me@remote:~/pictures-backup
+lb --minutes 240 pictures me@remote:~/pictures-backup
 
 Same as above except for 240 minutes only. This is useful if backing up over
 the internet only during specific times (at night for example). Does what it
@@ -89,22 +92,22 @@
 hardlinked to the catalog.
 
 Example 4:
-python lb.py --showfiles pictures pictures-backup | \
-python lb.py --filelist - pictures pictures-backup
+lb --showfiles pictures pictures-backup | \
+lb --filelist - pictures pictures-backup
 
 Same as example #1.
 
 Example 5:
 
 1)
-python lb.py --showfiles pictures me@remote:~/pictures-backup | \
-python lb.py --filelist - pictures me@laptop:~/pictures-transfer
+lb --showfiles pictures me@remote:~/pictures-backup | \
+lb --filelist - pictures me@laptop:~/pictures-transfer
 
 2)
-python lb.py --catalogonly pictures-transfer me@remote:~/pictures-backup
+lb --catalogonly pictures-transfer me@remote:~/pictures-backup
 
 3)
-python lb.py pictures me@remote:~/pictures-backup
+lb pictures me@remote:~/pictures-backup
 
 If the difference between pictures and pictures-backup (for example) is too
 large for internet backup, the steps above can be used. Step 1 transfers only
@@ -118,11 +121,11 @@
 since only the catalog is being updated (however it would be a speedup).
  
 History:
-
+  
 v 0.8 12/23/2006 scottlu
-  - allow backups to occur while files are changing
-  - minor --verify command bug
-  - added --verbose logging to tree building
+  - Allow backups to occur while files are changing
+  - Minor --verify command bug
+  - Added --verbose logging to tree building
 
 v 0.7 09/02/2006 scottlu
   - Ignore pipe, socket, and device file types
@@ -159,14 +162,14 @@
   - Added backup stat query methods
   - Changed log file format
   - Added viewlb.cgi, a web interface for viewing backups
-  - added gzip compression of filemap
-  - added --numeric-ids
+  - Added gzip compression of filemap
+  - Added --numeric-ids
 
 v 0.2 8/28/2004 scottlu
-  - filemap format change
-  - added --rmempty
-  - added --verify to run rsync in verify mode
-  - added uid/gid mapping by default unless --numeric-ids is specified
+  - Filemap format change
+  - Added --rmempty
+  - Added --verify to run rsync in verify mode
+  - Added uid/gid mapping by default unless --numeric-ids is specified
 
 v 0.1 8/19/2004 scottlu
   - Fully working backup, hardlinking between trees
@@ -747,55 +750,68 @@
     if stat.S_ISSOCK(mode):
         return False
     return True
+    
+def is_excluded(filepath, exclude_rules):
+	exclude_filepath = False
+	if exclude_rules:
+		for rule in exclude_rules:
+			if rule.search(filepath):
+				exclude_filepath = True
+				break
+	return exclude_filepath
 
-def build_filelist_from_tree(treepath):
+def build_filelist_from_tree(treepath, exclude_rules):
     class ListBuilder:
 	def __init__(self, basepath):
 	    self.lenbase = len('%s%s' % (basepath, os.sep))
 
         def callback(self, arg, dirpath, filelist):
             for file in filelist:
-		# Sometimes a stat may fail, like if there are broken
-		# symlinks in the file system
-		try:
-		    # Collect stat values instead of stat objects. It's 6
-		    # times smaller (measured) and mutuable
-		    # (for uid/gid mapping at the dest)
-		    filepath = join(dirpath, file)
-		    s = os.stat(filepath)
-		    if not is_mode_ok(s.st_mode):
-			continue
-		    arg.append((filepath[self.lenbase:], [s.st_mode, s.st_size, s.st_mtime, s.st_uid, s.st_gid]))
-		except:
-		    pass
+				# Sometimes a stat may fail, like if there are broken
+				# symlinks in the file system
+				try:
+					# Collect stat values instead of stat objects. It's 6
+					# times smaller (measured) and mutuable
+					# (for uid/gid mapping at the dest)
+					filepath = join(dirpath, file)
+					if is_excluded(filepath, exclude_rules):
+						continue							
+					s = os.stat(filepath)
+					if not is_mode_ok(s.st_mode):
+						continue
+					arg.append((filepath[self.lenbase:], [s.st_mode, s.st_size, s.st_mtime, s.st_uid, s.st_gid]))
+				except:
+					pass
 
     treepath_abs = os.path.abspath(treepath)
     filelist = []
     os.path.walk(treepath_abs, ListBuilder(treepath_abs).callback, filelist)
     return filelist
 
-def build_filelist_from_file(treepath, file):
+def build_filelist_from_file(treepath, file, exclude_rules):
     filelist = []
     for line in file.readlines():
 	filepath_rel = line.rstrip('\n')
+	if is_excluded(filepath_rel, exclude_rules):
+		continue
 	s = os.stat(join(treepath, filepath_rel))
 	if not is_mode_ok(s.st_mode):
 	    continue
 	filelist.append((filepath_rel, [s.st_mode, s.st_size, s.st_mtime, s.st_uid, s.st_gid]))
     return filelist
 
-def build_filelist(treepath):
+def build_filelist(treepath, exclude_rules):
     verbose_log('building filelist...')
     for n in xrange(len(sys.argv)):
 	if sys.argv[n] == '--filelist':
 	    if sys.argv[n + 1] == '-':
-		return build_filelist_from_file(treepath, sys.stdin)
+		return build_filelist_from_file(treepath, sys.stdin, exclude_rules)
 	    else:
 		file = open(sys.argv[n + 1])
-		filelist = build_filelist_from_file(treepath, file)
+		filelist = build_filelist_from_file(treepath, file, exclude_rules)
 		file.close()
 		return filelist
-    return build_filelist_from_tree(treepath)
+    return build_filelist_from_tree(treepath, exclude_rules)
 
 def build_uidgidmap(filelist):
     """Build a map of uid's to names and gid's to names
@@ -973,7 +989,7 @@
     if not treepath_last:
 	verbose_log('tree not equal: no last tree!')
 	return False
-    filelist_old = build_filelist_from_tree(treepath_last)
+    filelist_old = build_filelist_from_tree(treepath_last, None)
     if len(filelist) != len(filelist_old):
 	verbose_log('tree not equal: filelists different sizes!')
 	return False
@@ -1006,9 +1022,29 @@
     if is_source:
 	# Sending side
 	# Create filelist, calc name map, send both
-
+	
+	exclude_rules_file_path = None
+	exclude_rules = None
+	for n in xrange(len(sys.argv)):
+		if sys.argv[n] == '--exclude':
+			exclude_rules_file_path = sys.argv[n + 1]
+			break
+	if exclude_rules_file_path:
+		if  os.path.isfile(exclude_rules_file_path):
+			exclude_rules_file = open(exclude_rules_file_path, "r")
+			try:
+				exclude_rules = []
+				for rule in exclude_rules_file:
+					exclude_rules.append(re.compile(rule.strip(), re.UNICODE))
+			except IOError, e:
+				error('an error occured while reading the excludes file')
+			finally:
+				exclude_rules_file.close()
+		else:
+			error('excludes file "' + exclude_rules_file_path + '" not found')
+		
 	srcpath = os.path.abspath(os.path.expanduser(src['path']))
-	filelist = build_filelist(srcpath)
+	filelist = build_filelist(srcpath, exclude_rules)
 	send_object(filelist)
 	idname_map = build_uidgidmap(filelist)
 	send_object(idname_map)