[Findit] Add local git parsers.

appengine/findit/common/repo_util.py

Index: appengine/findit/common/repo_util.py
diff --git a/appengine/findit/common/repo_util.py b/appengine/findit/common/repo_util.py
new file mode 100644
index 0000000000000000000000000000000000000000..78662f91565b4b483345d3dd20abc927235bf625
--- /dev/null
+++ b/appengine/findit/common/repo_util.py
@@ -0,0 +1,85 @@
+# Copyright 2016 The Chromium Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+import json
+import re
+import urllib2
+
+CODE_REVIEW_URL_PATTERN = re.compile(
+    '^(?:Review URL|Review-Url): (.*\d+).*$', re.IGNORECASE)
+COMMIT_POSITION_PATTERN = re.compile(
+    '^Cr-Commit-Position: refs/heads/master@{#(\d+)}$', re.IGNORECASE)
+REVERTED_REVISION_PATTERN = re.compile(
+    '^> Committed: https://.+/([0-9a-fA-F]{40})$', re.IGNORECASE)
+
+
+def ExtractCommitPositionAndCodeReviewUrl(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:]

[lijeffrey, 2016/10/20 23:31:32]

move this to a constant

[Sharu Jiang, 2016/10/21 01:07:42]

Cannot think of a good name except a extremely long one:

COMMIT_POSITION_CODE_REVIEW_URL_PARSING_START_LINE

[wrengr, 2016/10/28 18:15:36]

START_OF_CR_COMMIT_POSITION (or expand "CR" if preferred)

[Sharu Jiang, 2016/10/28 21:05:36]

Sounds good :)

+  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(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 GetRevertedRevision(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 GetRepoToCloneUrlDict(host_url):
+  # Gerrit prepends )]}' to json-formatted response.
+  content = urllib2.urlopen('%s?format=json' % host_url).read()
+  prefix = ')]}\'\n'
+  repo_infos = json.loads(content[len(prefix):])
+
+  repo_to_clone_url = {}
+  for repo, repo_info in repo_infos.iteritems():
+    repo_to_clone_url[repo] = repo_info['clone_url']
+
+  return repo_to_clone_url