Implement git-drover.

git_drover.py

Index: git_drover.py
diff --git a/git_drover.py b/git_drover.py
new file mode 100755
index 0000000000000000000000000000000000000000..bfd4c83e715966af508b7c57c0f83cddd3c59e59
--- /dev/null
+++ b/git_drover.py
@@ -0,0 +1,506 @@
+#!/usr/bin/env python
+# Copyright (c) 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.
+
+"""Merge/Revert changes to Chromium release branches.
+
+This will use the git clone in the current directory if it matches the commit
+you passed in. Alternately, run this script in an empty directory and it will
+clone the appropriate repo for you (using `git cache` to do the smallest amount
+of network IO possible).
+
+This tool is aware of the following repos:
+"""
+
+import argparse
+import collections
+import multiprocessing
+import os
+import pprint
+import re
+import sys
+import textwrap
+import urllib2
+import urlparse
+
+from multiprocessing.pool import ThreadPool
+
+import git_cache
+import git_common as git
+
+from third_party import fancy_urllib
+
+assert fancy_urllib.can_validate_certs()
+
+CA_CERTS_FILE = os.path.abspath(os.path.join(
+    os.path.dirname(__file__), 'third_party', 'boto', 'cacerts', 'cacerts.txt'
+))
+
+urllib2.install_opener(urllib2.build_opener(
+    fancy_urllib.FancyRedirectHandler(),
+    fancy_urllib.FancyHTTPSHandler()))
+
+
+MISSING = object()
+
+OK_HOST_FMT = '%s.googlesource.com'
+OK_REPOS = {
+  'chrome-internal': ('chrome/src-internal',),
+  'chromium': ('chromium/src', 'chromium/blink',
+               'native_client/src/native_client')
+}
+
+def repo_url(host, repo):
+  assert host in OK_REPOS
+  assert repo in OK_REPOS[host]
+  return 'https://%s/%s.git' % (OK_HOST_FMT % host, repo)
+
+# lambda avoids polluting module with variable names, but still executes at
+# import-time.
+# 'redefining builtin __doc__' pylint: disable=W0622
+__doc__ += (lambda: '\n'.join([
+  '  * %s' % repo_url(host, repo)
+  for host, repos in OK_REPOS.iteritems()
+  for repo in repos
+]))()
+
+
+def die(msg, *args):
+  msg = textwrap.dedent(msg)
+  if args:
+    msg = msg % args
+  print >> sys.stderr, msg
+  sys.exit(1)
+
+
+def retry(fn, args=(), kwargs=None, on=(), but_not=(), upto=3):
+  kwargs = kwargs or {}
+  for attempt in xrange(upto):
+    try:
+      return fn(*args, **kwargs)
+    except but_not:
+      raise
+    except on:
+      if attempt + 1 == upto:
+        raise
+
+
+################################################################################
+
+
+def announce(msg=None, msg_fn=lambda: None):
+  print
+  print
+  print '=' * 80
+  if msg:
+    print textwrap.dedent(msg)
+  msg_fn()
+  print '=' * 80
+  print
+
+
+def confirm(prompt='Is this correct?', abort='No changes have been made.'):
+  while True:
+    v = raw_input('%s (Y/n) ' % prompt)
+    if v == '' or v in 'Yy':
+      break
+    if v in 'Nn':
+      die('Aborting. %s', abort)
+
+
+def summarize_job(correct_url, commits, target_ref, action):
+  def _msg_fn():
+    preposition = 'to' if action == 'merge' else 'from'
+    print "Planning to %s %d change%s %s branch %s of %s." % (
+      action, len(commits), 's' if len(commits) > 1 else '',
+      preposition, target_ref.num, correct_url)
+    print
+    for commit in commits:
+      print git.run('show', '-s', '--format=%H\t%s', commit)
+  announce(msg_fn=_msg_fn)
+
+
+def ensure_working_directory(commits, target_ref):
+  # TODO(iannucci): check all hashes locally after fetching first
+
+  fetch_specs = [
+    '%s:%s' % (target_ref.remote_full_ref, target_ref.remote_full_ref)
+  ] + commits
+
+  if git.check('rev-parse', '--is-inside-work-tree'):
+    actual_url = git.get_remote_url('origin')
+
+    if not actual_url or not is_ok_repo(actual_url):
+      die("""\
+      Inside a git repo, but origin's remote URL doesn't match one of the
+      supported git repos.
+        Current URL: %s""", actual_url)
+
+    s = git.run('status', '--porcelain')
+    if s:
+      die("""\
+      Your current directory is usable for the command you specified, but it
+      appears to be dirty (i.e. there are uncommitted changes). Please commit,
+      freeze, or stash these changes and run this command again.
+
+      %s""", '\n'.join('  '+l for l in s.splitlines()))
+
+    correct_url = get_correct_url(commits, actual_url)
+    if correct_url != actual_url:
+      die("""\
+      The specified commits appear to be from a different repo than the repo
+      in the current directory.
+        Current Repo: %s
+        Expected Repo: %s
+
+      Please re-run this script in an empty working directory and we'll fetch
+      the correct repo.""", actual_url, correct_url)
+
+    m = git_cache.Mirror.from_repo('.')
+    if m:
+      m.populate(bootstrap=True, verbose=True)
+      m.populate(fetch_specs=fetch_specs)
+
+  elif len(os.listdir('.')) == 0:
+    sample_path = '/path/to/cache'
+    if sys.platform.startswith('win'):
+      sample_path = r'X:\path\to\cache'
+    if not git.config('cache.cachepath'):
+      die("""\
+      Automatic drover checkouts require that you configure your global
+      cachepath to make these automatic checkouts as fast as possible. Do this
+      by running:
+        git config --global cache.cachepath "%s"
+
+      We recommend picking a non-network-mounted path with a decent amount of
+      space (at least 4GB).""", sample_path)
+
+    correct_url = get_correct_url(commits)
+
+    m = git_cache.Mirror(correct_url)
+    m.populate(bootstrap=True, verbose=True)
+    m.populate(fetch_specs=fetch_specs)
+    git.run('clone', '-s', '--no-checkout', m.mirror_path, '.')
+    git.run('update-ref', '-d', 'refs/heads/master')
+  else:
+    die('You must either invoke this from a git repo, or from an empty dir.')
+
+  for s in [target_ref.local_full_ref] + commits:
+    git.check('fetch', 'origin', s)
+
+  return correct_url
+
+
+def find_hash_urls(commits, presumed_url=None):
+  """Returns {url -> [commits]}.
+
+  Args:
+    commits - list of 40 char commit hashes
+    presumed_url - the url to the first repo to try. If commits end up not
+                   existing in this repo, find_hash_urls will try all other
+                   known repos.
+  """
+  pool = ThreadPool()
+
+  def process_async_results(asr, results):
+    """Resolves async results from |asr| into |results|.
+
+    Args:
+      asr - {commit -> [multiprocessing.pool.AsyncResult]}
+            The async results are for a url or MISSING.
+      results (in,out) - defaultdict({url -> set(commits)})
+
+    Returns a list of commits which did not have any non-MISSING result.

[agable, 2014/04/28 21:38:22]

"...which had only MISSING results."

+    """
+    try:
+      lost_commits = []
+      passes = 0
+      while asr and passes <= 10:
+        new_asr = {}
+        for commit, attempts in asr.iteritems():
+          new_attempts = []
+          for attempt in attempts:
+            try:
+              attempt = attempt.get(.5)
+              if attempt is not MISSING:
+                results[attempt].add(commit)
+                break
+            except multiprocessing.TimeoutError:
+              new_attempts.append(attempt)
+          else:
+            if new_attempts:
+              new_asr[commit] = new_attempts
+            else:
+              lost_commits.append(commit)
+        asr = new_asr
+        passes += 1
+      lost_commits += asr.keys()
+      return lost_commits
+    except Exception:
+      import traceback
+      traceback.print_exc()
+
+  def exists(url, commit):
+    """Queries the repo at |url| for |commit|.
+
+    Returns MISSING or the repo |url|
+    """
+    query_url = '%s/+/%s?format=JSON' % (url, commit)
+    return MISSING if GET(query_url) is MISSING else url
+
+  def go_fish(commit, except_for=()):
+    """Given a |commit|, search for it in all repos simultaneously, except for
+    repos indicated by |except_for|.
+
+    Returns the repo url for the commit or None.
+    """
+    async_results = {commit: set()}
+    for host, repos in OK_REPOS.iteritems():
+      for repo in repos:
+        url = repo_url(host, repo)
+        if url in except_for:
+          continue
+        async_results[commit].add(
+            pool.apply_async(exists, args=(url, commit)))
+
+    results = collections.defaultdict(set)
+    lost = process_async_results(async_results, results)
+    if not lost:
+      return results.popitem()[0]
+
+  # map of url -> set(commits)
+  results = collections.defaultdict(set)
+
+  # Try to find one hash which matches some repo
+  while commits and not presumed_url:
+    presumed_url = go_fish(commits[0])
+    results[presumed_url].add(commits[0])
+    commits = commits[1:]
+
+  # map of commit -> attempts
+  async_results = collections.defaultdict(list)
+  for commit in commits:
+    async_results[commit].append(
+        pool.apply_async(exists, args=(presumed_url, commit)))
+
+  lost = process_async_results(async_results, results)
+
+  if lost:
+    fishing_pool = ThreadPool()
+    async_results = collections.defaultdict(list)
+    for commit in lost:
+      async_results[commit].append(
+        fishing_pool.apply_async(go_fish, (commit,),
+                                 {'except_for': presumed_url})
+      )
+    lost = process_async_results(async_results, results)
+    if lost:
+      results[None].update(lost)
+
+  return {(k or 'UNKNOWN'): list(v) for k, v in results.iteritems()}
+
+
+def GET(url, **kwargs):
+  try:
+    kwargs.setdefault('timeout', 5)
+    request = fancy_urllib.FancyRequest(url)
+    request.set_ssl_info(ca_certs=CA_CERTS_FILE)
+    return retry(urllib2.urlopen, [request], kwargs,
+                 on=urllib2.URLError, but_not=urllib2.HTTPError, upto=3)
+  except urllib2.HTTPError as e:
+    if e.getcode() / 100 == 4:
+      return MISSING
+    raise
+
+
+def get_correct_url(commits, presumed_url=None):
+  unverified = commits
+  if presumed_url:
+    unverified = [c for c in unverified if not git.verify_commit(c)]
+    if not unverified:
+      return presumed_url
+    git.cached_fetch(unverified)
+    unverified = [c for c in unverified if not git.verify_commit(c)]
+    if not unverified:
+      return presumed_url
+
+  url_hashes = find_hash_urls(unverified, presumed_url)
+  if None in url_hashes:
+    die("""\
+    Could not determine what repo the following commits originate from:
+    %r""", url_hashes[None])
+
+  if len(url_hashes) > 1:
+    die("""\
+    Ambiguous commits specified. You supplied multiple commits, but they
+    appear to be from more than one repo?
+    %s""", pprint.pformat(dict(url_hashes)))
+
+  return url_hashes.popitem()[0]
+
+
+def is_ok_repo(url):
+  parsed = urlparse.urlsplit(url)
+
+  if parsed.scheme == 'https':
+    host = parsed.netloc[:-len(OK_HOST_FMT % '')]
+  elif parsed.scheme == 'sso':
+    host = parsed.netloc
+  else:
+    return False
+
+  if host not in OK_REPOS:
+    return False
+
+  path = parsed.path.strip('/')
+  if path.endswith('.git'):
+    path = path[:-4]
+
+  return path in OK_REPOS[host]
+
+
+class NumberedBranch(collections.namedtuple('NumberedBranch', 'num')):
+  # pylint: disable=W0232
+  @property
+  def remote_full_ref(self):
+    return 'refs/branch-heads/%d' % self.num
+
+  @property
+  def local_full_ref(self):
+    return 'refs/origin/branch-heads/%d' % self.num
+
+
+def parse_opts():
+  epilog = textwrap.dedent("""\
+  REF in the above may take the form of:
+    DDDD - a numbered branch (i.e. refs/branch-heads/DDDD)
+  """)
+
+  commit_re = re.compile('^[0-9a-fA-F]{40}$')
+  def commit_type(s):
+    if not commit_re.match(s):
+      raise argparse.ArgumentTypeError("%r is not a valid commit hash" % s)
+    return s
+
+  def ref_type(s):
+    if not s:
+      raise argparse.ArgumentTypeError("Empty ref: %r" % s)
+    if not s.isdigit():

[agable, 2014/04/28 21:38:22]

Doesn't handle 1780_21

+      raise argparse.ArgumentTypeError("Invalid ref: %r" % s)
+    return NumberedBranch(int(s))
+
+  parser = argparse.ArgumentParser(
+      description=__doc__, epilog=epilog,
+      formatter_class=argparse.RawDescriptionHelpFormatter
+  )
+
+  parser.add_argument('commit', nargs=1, metavar='HASH',
+                      type=commit_type, help='commit hash to revert/merge')
+
+  parser.add_argument('--prep_only', action='store_true', default=False,
+                      help=(
+                        'Prep and upload the CL (without sending mail) but '
+                        'don\'t push.'))
+
+  parser.add_argument('--bug', metavar='NUM', action='append', dest='bugs',
+                      help='optional bug number(s)')
+
+  grp = parser.add_mutually_exclusive_group(required=True)
+  grp.add_argument('--merge_to', metavar='REF', type=ref_type,
+                   help='branch to merge to')
+  grp.add_argument('--revert_from', metavar='REF', type=ref_type,
+                   help='branch ref to revert from')
+  opts = parser.parse_args()
+
+  # TODO(iannucci): Support multiple commits
+  opts.commits = opts.commit
+  del opts.commit
+
+  if opts.merge_to:
+    opts.action = 'merge'
+    opts.ref = opts.merge_to
+  elif opts.revert_from:
+    opts.action = 'revert'
+    opts.ref = opts.revert_from
+  else:
+    parser.error("?confusion? must specify either revert_from or merge_to")
+
+  del opts.merge_to
+  del opts.revert_from
+
+  return opts
+
+
+def main():
+  opts = parse_opts()
+
+  announce('Preparing working directory')
+
+  correct_url = ensure_working_directory(opts.commits, opts.ref)
+  summarize_job(correct_url, opts.commits, opts.ref, opts.action)
+  confirm()
+
+  announce('Checking out branches to %s changes' % opts.action)
+
+  git.run('fetch', 'origin',
+          '%s:%s' % (opts.ref.remote_full_ref, opts.ref.local_full_ref))
+  git.check('update-ref', '-d', 'refs/heads/__drover_base')
+  git.run('checkout', '-b', '__drover_base', opts.ref.local_full_ref,
+          stdout=None, stderr=None)
+  git.run('config', 'branch.__drover_base.remote', 'origin')
+  git.run('config', 'branch.__drover_base.merge', opts.ref.remote_full_ref)
+  git.check('update-ref', '-d', 'refs/heads/__drover_change')
+  git.run('checkout', '-t', '__drover_base', '-b', '__drover_change',
+          stdout=None, stderr=None)
+
+  announce('Performing %s' % opts.action)
+
+  # TODO(iannucci): support --signoff ?
+  authors = []
+  for commit in opts.commits:
+    success = False
+    if opts.action == 'merge':
+      success = git.check('cherry-pick', '-x', commit, verbose=True,
+                          stdout=None, stderr=None)
+    else:  # revert
+      success = git.check('revert', '--no-edit', commit, verbose=True,
+                          stdout=None, stderr=None)
+    if not success:
+      die("Aborting. Failed to %s.", opts.action)
+
+    email = git.run('show', '--format=%ae', '-s')
+    # git-svn email addresses take the form of:
+    #   user@domain.com@<svn id>
+    authors.append('@'.join(email.split('@', 2)[:2]))
+
+  announce('Success! Uploading codereview...')
+
+  if opts.prep_only:
+    print "Prep only mode, uploading CL but not sending mail."
+    mail = []
+  else:
+    mail = ['--send-mail', '--reviewers=' + ','.join(authors)]
+
+  args = [
+    '-c', 'gitcl.remotebranch=__drover_base',
+    '-c', 'branch.__drover_change.base-url=%s' % correct_url,
+    'cl', 'upload', '--bypass-hooks'
+  ] + mail
+
+  # TODO(iannucci): option to not bypass hooks?
+  git.check(*args, stdout=None, stderr=None, stdin=None)
+
+  if opts.prep_only:
+    announce('Issue created. To push to the branch, run `git cl push`')
+  else:
+    announce('About to push! This will make the commit live.')
+    confirm(abort=('Issue has been created, '
+                   'but change was not pushed to the repo.'))
+    git.run('cl', 'push', '-f', '--bypass-hooks')
+
+  return 0
+
+
+if __name__ == '__main__':
+  sys.exit(main())