added support for blacklists for authors and paths

Author: Monnoroch

gitstats

@@ -47,7 +47,9 @@ conf = {
 	'linear_linestats': 1,
 	'project_name': '',
 	'processes': 8,
-	'start_date': ''
+	'start_date': '',
+	'excluded_authors': [],
+	'excluded_prefixes': []
 }
 
 def getpipeoutput(cmds, quiet = False):
@@ -322,6 +324,8 @@ class GitDataCollector(DataCollector):
 				parts = re.split('\s+', line, 2)
 				commits = int(parts[1])
 				author = parts[2]
+				if author in conf["excluded_authors"]:
+					continue
 				self.tags[tag]['commits'] += commits
 				self.tags[tag]['authors'][author] = commits
 
@@ -338,6 +342,8 @@ class GitDataCollector(DataCollector):
 			timezone = parts[3]
 			author, mail = parts[4].split('<', 1)
 			author = author.rstrip()
+			if author in conf["excluded_authors"]:
+				continue
 			mail = mail.rstrip('>')
 			domain = '?'
 			if mail.find('@') != -1:
@@ -434,14 +440,16 @@ class GitDataCollector(DataCollector):
 			self.commits_by_timezone[timezone] = self.commits_by_timezone.get(timezone, 0) + 1
 
 		# outputs "<stamp> <files>" for each revision
-		revlines = getpipeoutput(['git rev-list --pretty=format:"%%at %%T" %s' % getlogrange('HEAD'), 'grep -v ^commit']).strip().split('\n')
+		revlines = getpipeoutput(['git rev-list --pretty=format:"%%at %%T %%an" %s' % getlogrange('HEAD'), 'grep -v ^commit']).strip().split('\n')
 		lines = []
 		revs_to_read = []
 		time_rev_count = []
 		#Look up rev in cache and take info from cache if found
 		#If not append rev to list of rev to read from repo
 		for revline in revlines:
-			time, rev = revline.split(' ')
+			time, rev, author = revline.split(' ')
+			if author in conf["excluded_authors"]:
+				continue
 			#if cache empty then add time and rev to list of new rev's
 			#otherwise try to read needed info from cache
 			if 'files_in_tree' not in self.cache.keys():
@@ -489,6 +497,14 @@ class GitDataCollector(DataCollector):
 			blob_id = parts[2]
 			size = int(parts[3])
 			fullpath = parts[4]
+			exclude = False
+			for path in conf["excluded_prefixes"]:
+				if fullpath.startswith(path):
+					exclude = True
+					break
+				
+			if exclude:
+				continue
 
 			self.total_size += size
 			self.total_files += 1
@@ -540,6 +556,7 @@ class GitDataCollector(DataCollector):
 		lines.reverse()
 		files = 0; inserted = 0; deleted = 0; total_lines = 0
 		author = None
+		last_line = ""
 		for line in lines:
 			if len(line) == 0:
 				continue
@@ -550,35 +567,36 @@ class GitDataCollector(DataCollector):
 				if pos != -1:
 					try:
 						(stamp, author) = (int(line[:pos]), line[pos+1:])
-						self.changes_by_date[stamp] = { 'files': files, 'ins': inserted, 'del': deleted, 'lines': total_lines }
-
-						date = datetime.datetime.fromtimestamp(stamp)
-						yymm = date.strftime('%Y-%m')
-						self.lines_added_by_month[yymm] = self.lines_added_by_month.get(yymm, 0) + inserted
-						self.lines_removed_by_month[yymm] = self.lines_removed_by_month.get(yymm, 0) + deleted
-
-						yy = date.year
-						self.lines_added_by_year[yy] = self.lines_added_by_year.get(yy,0) + inserted
-						self.lines_removed_by_year[yy] = self.lines_removed_by_year.get(yy, 0) + deleted
-
-						files, inserted, deleted = 0, 0, 0
+						if author not in conf["excluded_authors"]:
+							self.changes_by_date[stamp] = { 'files': files, 'ins': inserted, 'del': deleted, 'lines': total_lines }
+
+							date = datetime.datetime.fromtimestamp(stamp)
+							yymm = date.strftime('%Y-%m')
+							self.lines_added_by_month[yymm] = self.lines_added_by_month.get(yymm, 0) + inserted
+							self.lines_removed_by_month[yymm] = self.lines_removed_by_month.get(yymm, 0) + deleted
+
+							yy = date.year
+							self.lines_added_by_year[yy] = self.lines_added_by_year.get(yy,0) + inserted
+							self.lines_removed_by_year[yy] = self.lines_removed_by_year.get(yy, 0) + deleted
+
+							files, inserted, deleted = 0, 0, 0
+							
+							numbers = getstatsummarycounts(last_line)
+							if len(numbers) == 3:
+								(files, inserted, deleted) = map(lambda el : int(el), numbers)
+								total_lines += inserted
+								total_lines -= deleted
+								self.total_lines_added += inserted
+								self.total_lines_removed += deleted
+							else:
+								print 'Warning: failed to handle line "%s"' % line
+								(files, inserted, deleted) = (0, 0, 0)
 					except ValueError:
 						print 'Warning: unexpected line "%s"' % line
 				else:
 					print 'Warning: unexpected line "%s"' % line
 			else:
-				numbers = getstatsummarycounts(line)
-
-				if len(numbers) == 3:
-					(files, inserted, deleted) = map(lambda el : int(el), numbers)
-					total_lines += inserted
-					total_lines -= deleted
-					self.total_lines_added += inserted
-					self.total_lines_removed += deleted
-
-				else:
-					print 'Warning: failed to handle line "%s"' % line
-					(files, inserted, deleted) = (0, 0, 0)
+				last_line = line
 				#self.changes_by_date[stamp] = { 'files': files, 'ins': inserted, 'del': deleted }
 		self.total_lines += total_lines
 
@@ -606,21 +624,22 @@ class GitDataCollector(DataCollector):
 					try:
 						oldstamp = stamp
 						(stamp, author) = (int(line[:pos]), line[pos+1:])
-						if oldstamp > stamp:
-							# clock skew, keep old timestamp to avoid having ugly graph
-							stamp = oldstamp
-						if author not in self.authors:
-							self.authors[author] = { 'lines_added' : 0, 'lines_removed' : 0, 'commits' : 0}
-						self.authors[author]['commits'] = self.authors[author].get('commits', 0) + 1
-						self.authors[author]['lines_added'] = self.authors[author].get('lines_added', 0) + inserted
-						self.authors[author]['lines_removed'] = self.authors[author].get('lines_removed', 0) + deleted
-						if stamp not in self.changes_by_date_by_author:
-							self.changes_by_date_by_author[stamp] = {}
-						if author not in self.changes_by_date_by_author[stamp]:
-							self.changes_by_date_by_author[stamp][author] = {}
-						self.changes_by_date_by_author[stamp][author]['lines_added'] = self.authors[author]['lines_added']
-						self.changes_by_date_by_author[stamp][author]['commits'] = self.authors[author]['commits']
-						files, inserted, deleted = 0, 0, 0
+						if author not in conf["excluded_authors"]:
+							if oldstamp > stamp:
+								# clock skew, keep old timestamp to avoid having ugly graph
+								stamp = oldstamp
+							if author not in self.authors:
+								self.authors[author] = { 'lines_added' : 0, 'lines_removed' : 0, 'commits' : 0}
+							self.authors[author]['commits'] = self.authors[author].get('commits', 0) + 1
+							self.authors[author]['lines_added'] = self.authors[author].get('lines_added', 0) + inserted
+							self.authors[author]['lines_removed'] = self.authors[author].get('lines_removed', 0) + deleted
+							if stamp not in self.changes_by_date_by_author:
+								self.changes_by_date_by_author[stamp] = {}
+							if author not in self.changes_by_date_by_author[stamp]:
+								self.changes_by_date_by_author[stamp][author] = {}
+							self.changes_by_date_by_author[stamp][author]['lines_added'] = self.authors[author]['lines_added']
+							self.changes_by_date_by_author[stamp][author]['commits'] = self.authors[author]['commits']
+							files, inserted, deleted = 0, 0, 0
 					except ValueError:
 						print 'Warning: unexpected line "%s"' % line
 				else:
@@ -644,6 +663,8 @@ class GitDataCollector(DataCollector):
 
 		for name in self.authors.keys():
 			a = self.authors[name]
+			#if a is None:
+			#	continue
 			a['commits_frac'] = (100 * float(a['commits'])) / self.getTotalCommits()
 			date_first = datetime.datetime.fromtimestamp(a['first_commit_stamp'])
 			date_last = datetime.datetime.fromtimestamp(a['last_commit_stamp'])
@@ -1489,3 +1510,5 @@ if __name__=='__main__':
 	g = GitStats()
 	g.run(sys.argv[1:])
 
+
+mon@open-freelancer:~/dev/go/src/openfreelancers$