[Findit] Add local git parsers.

appengine/findit/common/git_repository.py

Index: appengine/findit/common/git_repository.py
diff --git a/appengine/findit/common/git_repository.py b/appengine/findit/common/git_repository.py
index b4f5845455a0574c36f11207e39ceb575a4d9ca6..5078dbf9068b8de55fcbdacf113625ad7ef2eaee 100644
--- a/appengine/findit/common/git_repository.py
+++ b/appengine/findit/common/git_repository.py
@@ -10,6 +10,7 @@ import re
 
 from common import diff
 from common import http_client_appengine
+from common import repo_util
 from common.blame import Blame
 from common.blame import Region
 from common.cache_decorator import Cached
@@ -18,13 +19,6 @@ from common.change_log import ChangeLog
 from common.change_log import FileChangeInfo
 from common.repository import Repository
 
-
-COMMIT_POSITION_PATTERN = re.compile(
-    '^Cr-Commit-Position: refs/heads/master@{#(\d+)}$', re.IGNORECASE)
-CODE_REVIEW_URL_PATTERN = re.compile(
-    '^(?:Review URL|Review-Url): (.*\d+).*$', re.IGNORECASE)
-REVERTED_REVISION_PATTERN = re.compile(
-    '^> Committed: https://.+/([0-9a-fA-F]{40})$', re.IGNORECASE)
 TIMEZONE_PATTERN = re.compile('[-+]\d{4}$')
 CACHE_EXPIRE_TIME_SECONDS = 24 * 60 * 60
 
@@ -69,49 +63,6 @@ class GitRepository(Repository):
       return None
     return base64.b64decode(content)
 
-  def ExtractCommitPositionAndCodeReviewUrl(self, message):
-    """Returns the commit position and code review url in the commit message.
-
-    A "commit position" is something similar to SVN version ids; i.e.,
-    numeric identifiers which are issued in sequential order. The reason
-    we care about them is that they're easier for humans to read than
-    the hashes that Git uses internally for identifying commits. We
-    should never actually use them for *identifying* commits; they're
-    only for pretty printing to humans.
-
-    Returns:
-      (commit_position, code_review_url)
-    """
-    if not message:
-      return (None, None)
-
-    commit_position = None
-    code_review_url = None
-
-    # Commit position and code review url are in the last 5 lines.
-    lines = message.strip().split('\n')[-5:]
-    lines.reverse()
-
-    for line in lines:
-      if commit_position is None:
-        match = COMMIT_POSITION_PATTERN.match(line)
-        if match:
-          commit_position = int(match.group(1))
-
-      if code_review_url is None:
-        match = CODE_REVIEW_URL_PATTERN.match(line)
-        if match:
-          code_review_url = match.group(1)
-    return (commit_position, code_review_url)
-
-  def _NormalizeEmail(self, email):
-    """Normalizes the email from git repo.
-
-    Some email is like: test@chromium.org@bbb929c8-8fbe-4397-9dbb-9b2b20218538.
-    """
-    parts = email.split('@')
-    return '@'.join(parts[0:2])
-
   def _GetDateTimeFromString(self, datetime_string,
                              date_format='%a %b %d %H:%M:%S %Y'):
     if TIMEZONE_PATTERN.findall(datetime_string):
@@ -136,21 +87,9 @@ class GitRepository(Repository):
     url = '%s/+/%s' % (self.repo_url, revision)
     return url, self._SendRequestForJsonResponse(url)
 
-  def GetRevertedRevision(self, message):
-    """Parse message to get the reverted revision if there is one."""
-    lines = message.strip().splitlines()
-    if not lines[0].lower().startswith('revert'):
-      return None
-
-    for line in reversed(lines):  # pragma: no cover
-      # TODO: Handle cases where no reverted_revision in reverting message.
-      reverted_revision_match = REVERTED_REVISION_PATTERN.match(line)
-      if reverted_revision_match:
-        return reverted_revision_match.group(1)
-
   def _ParseChangeLogFromLogData(self, data):
     commit_position, code_review_url = (
-        self.ExtractCommitPositionAndCodeReviewUrl(data['message']))
+        repo_util.ExtractCommitPositionAndCodeReviewUrl(data['message']))
 
     touched_files = []
     for file_diff in data['tree_diff']:
@@ -163,14 +102,15 @@ class GitRepository(Repository):
 
     author_time = self._GetDateTimeFromString(data['author']['time'])
     committer_time = self._GetDateTimeFromString(data['committer']['time'])
-    reverted_revision = self.GetRevertedRevision(data['message'])
+    reverted_revision = repo_util.GetRevertedRevision(data['message'])
     url = '%s/+/%s' % (self.repo_url, data['commit'])
 
     return ChangeLog(
-        data['author']['name'], self._NormalizeEmail(data['author']['email']),
+        data['author']['name'],
+        repo_util.NormalizeEmail(data['author']['email']),
         author_time,
         data['committer']['name'],
-        self._NormalizeEmail(data['committer']['email']),
+        repo_util.NormalizeEmail(data['committer']['email']),
         committer_time, data['commit'], commit_position,
         data['message'], touched_files, url, code_review_url,
         reverted_revision)
@@ -238,7 +178,8 @@ class GitRepository(Repository):
       blame.AddRegion(
           Region(region['start'], region['count'], region['commit'],
                  region['author']['name'],
-                 self._NormalizeEmail(region['author']['email']), author_time))
+                 repo_util.NormalizeEmail(region['author']['email']),
+                 author_time))
 
     return blame