Implemented filtering by commit message (Fixes #57).

When filtering, "git show" is called in order to pull the commit message
from the git log. While slow, it makes certain that the message itself is
pulled correctly, including any escape characters that might be present.

gitinspector/blame.py

@@ -1,6 +1,6 @@
 # coding: utf-8
 #
-# Copyright © 2012-2014 Ejwa Software. All rights reserved.
+# Copyright © 2012-2015 Ejwa Software. All rights reserved.
 #
 # This file is part of gitinspector.
 #
@@ -82,10 +82,12 @@ class BlameThread(threading.Thread):
 		except KeyError:
 			return
 
-		__blame_lock__.acquire() # Global lock used to protect calls from here...
-
-		if not filtering.set_filtered(author, "author") and not filtering.set_filtered(self.blamechunk_email, "email") and not \
+		if not filtering.set_filtered(author, "author") and not \
+		       filtering.set_filtered(self.blamechunk_email, "email") and not \
 		       filtering.set_filtered(self.blamechunk_revision, "revision"):
+
+			__blame_lock__.acquire() # Global lock used to protect calls from here...
+
 			if self.blames.get((author, self.filename), None) == None:
 				self.blames[(author, self.filename)] = BlameEntry()
 
@@ -96,7 +98,7 @@ class BlameThread(threading.Thread):
 				self.blames[(author, self.filename)].skew += ((self.changes.last_commit_date - self.blamechunk_time).days /
 				                                             (7.0 if self.useweeks else AVG_DAYS_PER_MONTH))
 
-		__blame_lock__.release() # ...to here.
+			__blame_lock__.release() # ...to here.
 
 	def run(self):
 		git_blame_r = subprocess.Popen(self.blame_command, bufsize=1, stdout=subprocess.PIPE).stdout

gitinspector/changes.py

@@ -134,6 +134,7 @@ class ChangesThread(threading.Thread):
 
 		commit = None
 		found_valid_extension = False
+		is_filtered = False
 		commits = []
 
 		__changes_lock__.acquire() # Global lock used to protect calls from here...
@@ -153,11 +154,18 @@ class ChangesThread(threading.Thread):
 					commits.append(commit)
 
 				found_valid_extension = False
+				is_filtered = False
 				commit = Commit(j)
 
-			if FileDiff.is_filediff_line(j) and not filtering.set_filtered(FileDiff.get_filename(j)) and not \
-			   filtering.set_filtered(commit.author, "author") and not filtering.set_filtered(commit.email, "email") and not \
-			   filtering.set_filtered(commit.sha, "revision"):
+				if Commit.is_commit_line(j) and \
+				   (filtering.set_filtered(commit.author, "author") or \
+				   filtering.set_filtered(commit.email, "email") or \
+				   filtering.set_filtered(commit.sha, "revision") or \
+				   filtering.set_filtered(commit.sha, "message")):
+					is_filtered = True
+
+			if FileDiff.is_filediff_line(j) and not \
+			   filtering.set_filtered(FileDiff.get_filename(j)) and not is_filtered:
 				extensions.add_located(FileDiff.get_extension(j))
 
 				if FileDiff.is_valid_extension(j):

gitinspector/filtering.py

@@ -22,10 +22,11 @@ from __future__ import unicode_literals
 from localization import N_
 from outputable import Outputable
 import re
+import subprocess
 import terminal
 import textwrap
 
-__filters__ = {"file": [[], set()], "author": [[], set()], "email": [[], set()], "revision": [[], set()]}
+__filters__ = {"file": [set(), set()], "author": [set(), set()], "email": [set(), set()], "revision": [set(), set()], "message" : [set(), None]}
 
 class InvalidRegExpError(ValueError):
 	def __init__(self, msg):
@@ -38,9 +39,9 @@ def get():
 def __add_one__(string):
 	for i in __filters__:
 		if (i + ":").lower() == string[0:len(i) + 1].lower():
-			__filters__[i][0].append(string[len(i) + 1:])
+			__filters__[i][0].add(string[len(i) + 1:])
 			return
-	__filters__["file"][0].append(string)
+	__filters__["file"][0].add(string)
 
 def add(string):
 	rules = string.split(",")
@@ -49,7 +50,7 @@ def add(string):
 
 def clear():
 	for i in __filters__:
-		__filters__[i][0] = []
+		__filters__[i][0] = set()
 
 def get_filered(filter_type="file"):
 	return __filters__[filter_type][1]
@@ -60,14 +61,32 @@ def has_filtered():
 			return True
 	return False
 
+def __find_commit_message__(sha):
+	git_show_r = subprocess.Popen(filter(None, ["git", "show", "-s", "--pretty=%B", "-w", sha]), bufsize=1,
+	                              stdout=subprocess.PIPE).stdout
+
+	commit_message = git_show_r.read()
+	git_show_r.close()
+
+	commit_message = commit_message.strip().decode("unicode_escape", "ignore")
+	commit_message = commit_message.encode("latin-1", "replace")
+	return commit_message.decode("utf-8", "replace")
+
 def set_filtered(string, filter_type="file"):
 	string = string.strip()
 
 	if len(string) > 0:
 		for i in __filters__[filter_type][0]:
+			search_for = string
+
+			if filter_type == "message":
+				search_for = __find_commit_message__(string)
 			try:
-				if re.search(i, string) != None:
-					__filters__[filter_type][1].add(string)
+				if re.search(i, search_for) != None:
+					if filter_type == "message":
+						__add_one__("revision:" + string)
+					else:
+						__filters__[filter_type][1].add(string)
 					return True
 			except:
 				raise InvalidRegExpError(_("invalid regular expression specified"))
@@ -138,5 +157,5 @@ class Filtering(Outputable):
 			Filtering.__output_xml_section__(_(FILTERING_INFO_TEXT), __filters__["file"][1], "files")
 			Filtering.__output_xml_section__(_(FILTERING_AUTHOR_INFO_TEXT), __filters__["author"][1], "authors")
 			Filtering.__output_xml_section__(_(FILTERING_EMAIL_INFO_TEXT), __filters__["email"][1], "emails")
-			Filtering.__output_xml_section__(_(FILTERING_COMMIT_INFO_TEXT), __filters__["revision"][1].union(), "revisions")
+			Filtering.__output_xml_section__(_(FILTERING_COMMIT_INFO_TEXT), __filters__["revision"][1], "revision")
 			print("\t</filtering>")

gitinspector/help.py

@@ -1,6 +1,6 @@
 # coding: utf-8
 #
-# Copyright © 2012-2014 Ejwa Software. All rights reserved.
+# Copyright © 2012-2015 Ejwa Software. All rights reserved.
 #
 # This file is part of gitinspector.
 #
@@ -61,7 +61,8 @@ Boolean arguments can only be given to long options.
   -w, --weeks[=BOOL]             show all statistical information in weeks
                                    instead of in months
   -x, --exclude=PATTERN          an exclusion pattern describing the file
-                                   paths, revisions, author names or author
+                                   paths, revisions, revisions with certain
+                                   commit messages, author names or author
                                    emails that should be excluded from the
                                    statistics; can be specified multiple times
   -h, --help                     display this help and exit