[Findit] Factoring the gitiles (etc) stuff out into its own directory

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
deleted file mode 100644
index 9d49da023fb1b42dfb78e7dc4f2db334005ddb61..0000000000000000000000000000000000000000
--- a/appengine/findit/common/git_repository.py
+++ /dev/null
@@ -1,291 +0,0 @@
-# Copyright 2014 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 base64
-from datetime import datetime
-from datetime import timedelta
-import json
-import re
-
-from common import diff
-from common.blame import Blame
-from common.blame import Region
-from common.cache_decorator import Cached
-from common.cache_decorator import CompressedMemCacher
-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
-
-
-class GitRepository(Repository):
-  """Represents a git repository on https://chromium.googlesource.com."""
-
-  def __init__(self, repo_url=None, http_client=None):
-    super(GitRepository, self).__init__()
-    if repo_url and repo_url.endswith('/'):
-      self._repo_url = repo_url[:-1]
-    else:
-      self._repo_url = repo_url
-
-    self._http_client = http_client
-
-  @property
-  def repo_url(self):
-    return self._repo_url
-
-  @repo_url.setter
-  def repo_url(self, repo_url):
-    self._repo_url = repo_url
-
-  @property
-  def http_client(self):
-    return self._http_client
-
-  @property
-  def identifier(self):
-    return self.repo_url
-
-  @Cached(namespace='Gitiles-json-view', expire_time=CACHE_EXPIRE_TIME_SECONDS,
-          cacher=CompressedMemCacher())
-  def _SendRequestForJsonResponse(self, url, params=None):
-    if params is None:  # pragma: no cover
-      params = {}
-    params['format'] = 'json'
-
-    # Gerrit prepends )]}' to json-formatted response.
-    prefix = ')]}\'\n'
-
-    status_code, content = self.http_client.Get(url, params)
-    if status_code != 200:
-      return None
-    elif not content or not content.startswith(prefix):
-      raise Exception('Response does not begin with %s' % prefix)
-
-    return json.loads(content[len(prefix):])
-
-  @Cached(namespace='Gitiles-text-view', expire_time=CACHE_EXPIRE_TIME_SECONDS)
-  def _SendRequestForTextResponse(self, url):
-    status_code, content = self.http_client.Get(url, {'format': 'text'})
-    if status_code != 200:
-      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):
-      # Need to handle timezone conversion.
-      naive_datetime_str, _, offset_str = datetime_string.rpartition(' ')
-      naive_datetime = datetime.strptime(naive_datetime_str,
-                                         date_format)
-      hour_offset = int(offset_str[-4:-2])
-      minute_offset = int(offset_str[-2:])
-      if(offset_str[0]) == '-':
-        hour_offset = -hour_offset
-        minute_offset = -minute_offset
-
-      time_delta = timedelta(hours=hour_offset, minutes=minute_offset)
-
-      utc_datetime = naive_datetime - time_delta
-      return utc_datetime
-
-    return datetime.strptime(datetime_string, date_format)
-
-  def _DownloadChangeLogData(self, revision):
-    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']))
-
-    touched_files = []
-    for file_diff in data['tree_diff']:
-      change_type = file_diff['type'].lower()
-      if not diff.IsKnownChangeType(change_type):
-        raise Exception('Unknown change type "%s"' % change_type)
-      touched_files.append(
-          FileChangeInfo(
-              change_type, file_diff['old_path'], file_diff['new_path']))
-
-    author_time = self._GetDateTimeFromString(data['author']['time'])
-    committer_time = self._GetDateTimeFromString(data['committer']['time'])
-    reverted_revision = self.GetRevertedRevision(data['message'])
-    url = '%s/+/%s' % (self.repo_url, data['commit'])
-
-    return ChangeLog(
-        data['author']['name'], self._NormalizeEmail(data['author']['email']),
-        author_time,
-        data['committer']['name'],
-        self._NormalizeEmail(data['committer']['email']),
-        committer_time, data['commit'], commit_position,
-        data['message'], touched_files, url, code_review_url,
-        reverted_revision)
-
-  def GetChangeLog(self, revision):
-    """Returns the change log of the given revision."""
-    _, data = self._DownloadChangeLogData(revision)
-    if not data:
-      return None
-
-    return self._ParseChangeLogFromLogData(data)
-
-  def GetCommitsBetweenRevisions(self, start_revision, end_revision, n=1000):
-    """Gets a list of commit hashes between start_revision and end_revision.
-
-    Args:
-      start_revision: The oldest revision in the range.
-      end_revision: The latest revision in the range.
-      n: The maximum number of revisions to request at a time.
-
-    Returns:
-      A list of commit hashes made since start_revision through and including
-      end_revision in order from most-recent to least-recent. This includes
-      end_revision, but not start_revision.
-    """
-    params = {'n': n}
-    next_end_revision = end_revision
-    commits = []
-
-    while next_end_revision:
-      url = '%s/+log/%s..%s' % (
-          self.repo_url, start_revision, next_end_revision)
-      data = self._SendRequestForJsonResponse(url, params)
-
-      if not data:
-        break
-
-      for log in data.get('log', []):
-        commit = log.get('commit')
-        if commit:
-          commits.append(commit)
-
-      next_end_revision = data.get('next')
-
-    return commits
-
-  def GetChangeDiff(self, revision):
-    """Returns the raw diff of the given revision."""
-    url = '%s/+/%s%%5E%%21/' % (self.repo_url, revision)
-    return self._SendRequestForTextResponse(url)
-
-  def GetBlame(self, path, revision):
-    """Returns blame of the file at ``path`` of the given revision."""
-    url = '%s/+blame/%s/%s' % (self.repo_url, revision, path)
-
-    data = self._SendRequestForJsonResponse(url)
-    if not data:
-      return None
-
-    blame = Blame(revision, path)
-    for region in data['regions']:
-      author_time = self._GetDateTimeFromString(
-          region['author']['time'], '%Y-%m-%d %H:%M:%S')
-
-      blame.AddRegion(
-          Region(region['start'], region['count'], region['commit'],
-                 region['author']['name'],
-                 self._NormalizeEmail(region['author']['email']), author_time))
-
-    return blame
-
-  def GetSource(self, path, revision):
-    """Returns source code of the file at ``path`` of the given revision."""
-    url = '%s/+/%s/%s' % (self.repo_url, revision, path)
-    return self._SendRequestForTextResponse(url)
-
-  def GetChangeLogs(self, start_revision, end_revision, n=1000):
-    """Gets a list of ChangeLogs in revision range by batch.
-
-    Args:
-      start_revision (str): The oldest revision in the range.
-      end_revision (str): The latest revision in the range.
-      n (int): The maximum number of revisions to request at a time (default
-        to 1000).
-
-    Returns:
-      A list of changelogs in (start_revision, end_revision].
-    """
-    next_end_revision = end_revision
-    changelogs = []
-
-    while next_end_revision:
-      url = '%s/+log/%s..%s' % (self.repo_url,
-                                start_revision, next_end_revision)
-      data = self._SendRequestForJsonResponse(url, params={'n': str(n),
-                                                           'name-status': '1'})
-
-      for log in data['log']:
-        changelogs.append(self._ParseChangeLogFromLogData(log))
-
-      if 'next' in data:
-        next_end_revision = data['next']
-      else:
-        next_end_revision = None
-
-    return changelogs