Refactor infra git libs and testing.

infra/services/gnumbd/inner_loop.py

 # 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.
 
 """Gnumd (Git NUMber Daemon): Adds metadata to git commits as they land in
 a primary repo.
 
 This is a simple daemon which takes commits pushed to a pending ref, alters
 their message with metadata, and then pushes the altered commits to a parallel
 ref.
 """
 
 import collections
 import logging
 import re
 import sys
 import time
 
 LOGGER = logging.getLogger(__name__)
 
-from infra.services.gnumbd.support import git, data, util
+from infra.libs import git2
+from infra.libs import data_structures
 
 
 DEFAULT_CONFIG_REF = 'refs/pending-config/main'
 DEFAULT_REPO_DIR = 'gnumbd_repos'
 FOOTER_PREFIX = 'Cr-'
 COMMIT_POSITION = FOOTER_PREFIX + 'Commit-Position'
 # takes a Ref and a number
 FMT_COMMIT_POSITION = '{.ref}@{{#{:d}}}'.format
 BRANCHED_FROM = FOOTER_PREFIX + 'Branched-From'
 GIT_SVN_ID = 'git-svn-id'
@@ skipping 15 matching lines @@
         'No {!r} or git-svn-id found for {!r}'.format(COMMIT_POSITION, commit))
 
 
 ################################################################################
 # Commit Manipulation
 ################################################################################
 
 def content_of(commit):
   """Calculates the content of |commit| such that a gnumbd-landed commit and
   the original commit will compare as equals. Returns the content as a
-  data.CommitData object.
+  git2.CommitData object.
 
   This strips out:
     * The parent(s)
     * The committer date
     * footers beginning with 'Cr-'
     * the 'git-svn-id' footer.
 
   Stores a cached copy of the result data on the |commit| instance itself.
   """
   if commit is None:
-    return git.INVALID
+    return git2.INVALID
 
   if not hasattr(commit, '_cr_content'):
     d = commit.data
-    footers = util.thaw(d.footers)
+    footers = data_structures.thaw(d.footers)
     footers[GIT_SVN_ID] = None
     for k in footers.keys():
       if k.startswith(FOOTER_PREFIX):
         footers[k] = None
     commit._cr_content = d.alter(
         parents=(),
-        committer=d.committer.alter(timestamp=data.NULL_TIMESTAMP),
+        committer=d.committer.alter(timestamp=git2.data.NULL_TIMESTAMP),
         footers=footers)
   return commit._cr_content  # pylint: disable=W0212
 
 
 def get_position(commit, _position_re=re.compile('^(.*)@{#(\d*)}$')):
   """Returns (ref, position number) for the given |commit|.
 
   Looks for the Cr-Commit-Position footer. If that's unavailable, it falls back
   to the git-svn-id footer, passing back ref as None.
 
   May raise the MalformedPositionFooter or NoPositionData exceptions.
   """
   f = commit.data.footers
   current_pos = f.get(COMMIT_POSITION)
   if current_pos:
     assert len(current_pos) == 1
     current_pos = current_pos[0]
 
     m = _position_re.match(current_pos)
     if not m:
       raise MalformedPositionFooter(commit, COMMIT_POSITION, current_pos)
-    parent_ref = git.Ref(commit.repo, m.group(1))
+    parent_ref = git2.Ref(commit.repo, m.group(1))
     parent_num = int(m.group(2))
   else:
     # TODO(iannucci): Remove this and rely on a manual initial commit?
     svn_pos = f.get(GIT_SVN_ID)
     if not svn_pos:
       raise NoPositionData(commit)
 
     assert len(svn_pos) == 1
     svn_pos = svn_pos[0]
     parent_ref = None
     try:
       parent_num = int(svn_pos.split()[0].split('@')[1])
     except (IndexError, ValueError):
       raise MalformedPositionFooter(commit, GIT_SVN_ID, svn_pos)
 
   return parent_ref, parent_num
 
 
 def synthesize_commit(commit, new_parent, ref, clock=time):
   """Synthesizes a new Commit given |new_parent| and ref.
 
   The new commit will contain a Cr-Commit-Position footer, and possibly
   Cr-Branched-From footers (if commit is on a branch).
 
   The new commit's committer date will also be updated to 'time.time()', or
   the new parent's date + 1, whichever is higher. This means that within a branch,
   commit timestamps will always increase (at least from the point where this
   daemon went into service).
 
-  @type commit: git.Commit
-  @type new_parent: git.Commit
-  @type ref: git.Ref
+  @type commit: git2.Commit
+  @type new_parent: git2.Commit
+  @type ref: git2.Ref
   @kind clock: implements .time(), used for testing determinisim.
   """
   # TODO(iannucci): See if there are any other footers we want to carry over
   # between new_parent and commit
   footers = collections.OrderedDict()
   parent_ref, parent_num = get_position(new_parent)
   # if parent_ref wasn't encoded, assume that the parent is on the same ref.
   if parent_ref is None:
     parent_ref = ref
 
@@ skipping 11 matching lines @@
   # Gerrit-landed commits.
   for key, value in commit.data.footers.iteritems():
     if key.startswith(FOOTER_PREFIX) or key == GIT_SVN_ID:
       LOGGER.warn('Dropping key on user commit %s: %r -> %r',
                   commit.hsh, key, value)
       footers[key] = None
 
   # Ensure that every commit has a time which is at least 1 second after its
   # parent, and reset the tz to UTC.
   parent_time = new_parent.data.committer.timestamp.secs
-  new_parents = [] if new_parent is git.INVALID else [new_parent.hsh]
+  new_parents = [] if new_parent is git2.INVALID else [new_parent.hsh]
   new_committer = commit.data.committer.alter(
-      timestamp=data.NULL_TIMESTAMP.alter(
+      timestamp=git2.data.NULL_TIMESTAMP.alter(
           secs=max(int(clock.time()), parent_time + 1)))
 
   return commit.alter(
       parents=new_parents,
       committer=new_committer,
       footers=footers,
   )
 
 
 ################################################################################
@@ skipping 22 matching lines @@
 
      v  pending_tag
   A  B  C  D  E  F  <- pending_tip
   A' B' C' D' E' F' <- master
 
   In either case, pending_tag would be advanced, and the method would return
   the commits beteween the tag's proper position and the tip.
 
   Other discrepancies are errors and this method will return an empty list.
 
-  @type pending_tag: git.Ref
-  @type pending_tip: git.Ref
-  @type real_ref: git.Ref
-  @returns [git.Commit]
+  @type pending_tag: git2.Ref
+  @type pending_tip: git2.Ref
+  @type real_ref: git2.Ref
+  @returns [git2.Commit]
   """
   assert pending_tag.commit != pending_tip.commit
   i = 0
   new_commits = list(pending_tag.to(pending_tip))
   if not new_commits:
     LOGGER.error('%r doesn\'t match %r, but there are no new_commits?',
                  pending_tag.ref, pending_tip.ref)
     return []
 
   for commit in new_commits:
     parent = commit.parent
-    if parent is git.INVALID:
+    if parent is git2.INVALID:
       LOGGER.error('Cannot process pending merge commit %r', commit)
       return []
 
     if content_of(parent) == content_of(real_ref.commit):
       break
 
     LOGGER.warn('Skipping already-processed commit on real_ref %r: %r',
                 real_ref, commit.hsh)
     i += 1
 
@@ skipping 28 matching lines @@
         v  pending_tag
   A  B  C  D  E  F  <- pending_tip
   A' B' C' <- master
 
   This function will produce:
 
                  v  pending_tag
   A  B  C  D  E  F  <- pending_tip
   A' B' C' D' E' F' <- master
 
-  @type real_ref: git.Ref
-  @type pending_tag: git.Ref
-  @type new_commits: [git.Commit]
+  @type real_ref: git2.Ref
+  @type pending_tag: git2.Ref
+  @type new_commits: [git2.Commit]
   @kind clock: implements .time(), used for testing determinisim.
   """
   # TODO(iannucci): use push --force-with-lease to reset pending to the real
   # ref?
   # TODO(iannucci): The ACL rejection message for the real ref should point
   # users to the pending ref.
   assert content_of(pending_tag.commit) == content_of(real_ref.commit)
   real_parent = real_ref.commit
   for commit in new_commits:
     assert content_of(commit.parent) == content_of(real_parent)
@@ skipping 13 matching lines @@
   """Execute a single pass over a fetched Repo.
 
   Will call |process_ref| for every branch indicated by the enabled_refglobs
   config option.
   """
   pending_tag_prefix = cref['pending_tag_prefix']
   pending_ref_prefix = cref['pending_ref_prefix']
   enabled_refglobs = cref['enabled_refglobs']
 
   def join(prefix, ref):
-    return git.Ref(repo, '/'.join((prefix, ref.ref[len('refs/'):])))
+    return git2.Ref(repo, '/'.join((prefix, ref.ref[len('refs/'):])))
 
   for refglob in enabled_refglobs:
-    glob = join(pending_ref_prefix, git.Ref(repo, refglob))
+    glob = join(pending_ref_prefix, git2.Ref(repo, refglob))
     for pending_tip in repo.refglob(glob.ref):
       # TODO(iannucci): each real_ref could have its own thread.
       try:
-        real_ref = git.Ref(repo, pending_tip.ref.replace(
+        real_ref = git2.Ref(repo, pending_tip.ref.replace(
             pending_ref_prefix, 'refs'))
 
-        if real_ref.commit is git.INVALID:
+        if real_ref.commit is git2.INVALID:
           LOGGER.error('Missing real ref %r', real_ref)
           continue
 
         LOGGER.info('Processing %r', real_ref)
         pending_tag = join(pending_tag_prefix, real_ref)
 
-        if pending_tag.commit is git.INVALID:
+        if pending_tag.commit is git2.INVALID:
           LOGGER.error('Missing pending tag %r for %r', pending_tag, real_ref)
           continue
 
         if pending_tag.commit != pending_tip.commit:
           new_commits = get_new_commits(real_ref, pending_tag, pending_tip)
           if new_commits:
             process_ref(real_ref, pending_tag, new_commits, clock)
         else:
           if content_of(pending_tag.commit) != content_of(real_ref.commit):
             LOGGER.error('%r and %r match, but %r\'s content doesn\'t match!',
                          pending_tag, pending_tip, real_ref)
           else:
             LOGGER.info('%r is up to date', real_ref)
       except (NoPositionData, MalformedPositionFooter) as e:
         LOGGER.error('%s %s', e.__class__.__name__, e)
       except Exception:  # pragma: no cover
         LOGGER.exception('Uncaught exception while processing %r', real_ref)
 
 
 def inner_loop(repo, cref, clock=time):
   LOGGER.debug('fetching %r', repo)
   repo.run('fetch', stdout=sys.stdout, stderr=sys.stderr)
   cref.evaluate()
   process_repo(repo, cref, clock)