Roll multiple files from depot_tools into luci.

appengine/third_party_local/depot_tools/git_common.py

Index: appengine/third_party_local/depot_tools/git_common.py
diff --git a/appengine/third_party_local/depot_tools/git_common.py b/appengine/third_party_local/depot_tools/git_common.py
index 1215d9cd41d44385f98583c884188b006b8561f1..a18f59f16474d19e5ed52085dedc70af42f2be8e 100644
--- a/appengine/third_party_local/depot_tools/git_common.py
+++ b/appengine/third_party_local/depot_tools/git_common.py
@@ -1,4 +1,4 @@
-# Copyright 2013 The Chromium Authors. All rights reserved.
+# 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.
 
@@ -16,19 +16,85 @@ IMapIterator.__next__ = IMapIterator.next
 
 
 import binascii
+import collections
 import contextlib
 import functools
 import logging
+import os
+import re
 import signal
 import sys
 import tempfile
+import textwrap
 import threading
 
 import subprocess2
 
+ROOT = os.path.abspath(os.path.dirname(__file__))
 
-GIT_EXE = 'git.bat' if sys.platform.startswith('win') else 'git'
+GIT_EXE = ROOT+'\\git.bat' if sys.platform.startswith('win') else 'git'
+TEST_MODE = False
 
+FREEZE = 'FREEZE'
+FREEZE_SECTIONS = {
+  'indexed': 'soft',
+  'unindexed': 'mixed'
+}
+FREEZE_MATCHER = re.compile(r'%s.(%s)' % (FREEZE, '|'.join(FREEZE_SECTIONS)))
+
+
+# Retry a git operation if git returns a error response with any of these
+# messages. It's all observed 'bad' GoB responses so far.
+#
+# This list is inspired/derived from the one in ChromiumOS's Chromite:
+# <CHROMITE>/lib/git.py::GIT_TRANSIENT_ERRORS
+#
+# It was last imported from '7add3ac29564d98ac35ce426bc295e743e7c0c02'.
+GIT_TRANSIENT_ERRORS = (
+    # crbug.com/285832
+    r'!.*\[remote rejected\].*\(error in hook\)',
+
+    # crbug.com/289932
+    r'!.*\[remote rejected\].*\(failed to lock\)',
+
+    # crbug.com/307156
+    r'!.*\[remote rejected\].*\(error in Gerrit backend\)',
+
+    # crbug.com/285832
+    r'remote error: Internal Server Error',
+
+    # crbug.com/294449
+    r'fatal: Couldn\'t find remote ref ',
+
+    # crbug.com/220543
+    r'git fetch_pack: expected ACK/NAK, got',
+
+    # crbug.com/189455
+    r'protocol error: bad pack header',
+
+    # crbug.com/202807
+    r'The remote end hung up unexpectedly',
+
+    # crbug.com/298189
+    r'TLS packet with unexpected length was received',
+
+    # crbug.com/187444
+    r'RPC failed; result=\d+, HTTP code = \d+',
+
+    # crbug.com/388876
+    r'Connection timed out',
+
+    # crbug.com/430343
+    # TODO(dnj): Resync with Chromite.
+    r'The requested URL returned error: 5\d+',
+)
+
+GIT_TRANSIENT_ERRORS_RE = re.compile('|'.join(GIT_TRANSIENT_ERRORS),
+                                     re.IGNORECASE)
+
+# First version where the for-each-ref command's format string supported the
+# upstream:track token.
+MIN_UPSTREAM_TRACK_GIT_VERSION = (1, 9)
 
 class BadCommitRefException(Exception):
   def __init__(self, refs):
@@ -139,7 +205,7 @@ def ScopedPool(*args, **kwargs):
 
 class ProgressPrinter(object):
   """Threaded single-stat status message printer."""
-  def __init__(self, fmt, enabled=None, stream=sys.stderr, period=0.5):
+  def __init__(self, fmt, enabled=None, fout=sys.stderr, period=0.5):
     """Create a ProgressPrinter.
 
     Use it as a context manager which produces a simple 'increment' method:
@@ -155,7 +221,7 @@ class ProgressPrinter(object):
         should go.
       enabled (bool) - If this is None, will default to True if
         logging.getLogger() is set to INFO or more verbose.
-      stream (file-like) - The stream to print status messages to.
+      fout (file-like) - The stream to print status messages to.
       period (float) - The time in seconds for the printer thread to wait
         between printing.
     """
@@ -168,7 +234,7 @@ class ProgressPrinter(object):
     self._count = 0
     self._dead = False
     self._dead_cond = threading.Condition()
-    self._stream = stream
+    self._stream = fout
     self._thread = threading.Thread(target=self._run)
     self._period = period
 
@@ -199,6 +265,225 @@ class ProgressPrinter(object):
     del self._thread
 
 
+def once(function):
+  """@Decorates |function| so that it only performs its action once, no matter
+  how many times the decorated |function| is called."""
+  def _inner_gen():
+    yield function()
+    while True:
+      yield
+  return _inner_gen().next
+
+
+## Git functions
+
+
+def branch_config(branch, option, default=None):
+  return config('branch.%s.%s' % (branch, option), default=default)
+
+
+def branch_config_map(option):
+  """Return {branch: <|option| value>} for all branches."""
+  try:
+    reg = re.compile(r'^branch\.(.*)\.%s$' % option)
+    lines = run('config', '--get-regexp', reg.pattern).splitlines()
+    return {reg.match(k).group(1): v for k, v in (l.split() for l in lines)}
+  except subprocess2.CalledProcessError:
+    return {}
+
+
+def branches(*args):
+  NO_BRANCH = ('* (no branch', '* (detached from ')
+
+  key = 'depot-tools.branch-limit'
+  limit = 20
+  try:
+    limit = int(config(key, limit))
+  except ValueError:
+    pass
+
+  raw_branches = run('branch', *args).splitlines()
+
+  num = len(raw_branches)
+  if num > limit:
+    print >> sys.stderr, textwrap.dedent("""\
+    Your git repo has too many branches (%d/%d) for this tool to work well.
+
+    You may adjust this limit by running:
+      git config %s <new_limit>
+    """ % (num, limit, key))
+    sys.exit(1)
+
+  for line in raw_branches:
+    if line.startswith(NO_BRANCH):
+      continue
+    yield line.split()[-1]
+
+
+def config(option, default=None):
+  try:
+    return run('config', '--get', option) or default
+  except subprocess2.CalledProcessError:
+    return default
+
+
+def config_list(option):
+  try:
+    return run('config', '--get-all', option).split()
+  except subprocess2.CalledProcessError:
+    return []
+
+
+def current_branch():
+  try:
+    return run('rev-parse', '--abbrev-ref', 'HEAD')
+  except subprocess2.CalledProcessError:
+    return None
+
+
+def del_branch_config(branch, option, scope='local'):
+  del_config('branch.%s.%s' % (branch, option), scope=scope)
+
+
+def del_config(option, scope='local'):
+  try:
+    run('config', '--' + scope, '--unset', option)
+  except subprocess2.CalledProcessError:
+    pass
+
+
+def freeze():
+  took_action = False
+
+  try:
+    run('commit', '-m', FREEZE + '.indexed')
+    took_action = True
+  except subprocess2.CalledProcessError:
+    pass
+
+  try:
+    run('add', '-A')
+    run('commit', '-m', FREEZE + '.unindexed')
+    took_action = True
+  except subprocess2.CalledProcessError:
+    pass
+
+  if not took_action:
+    return 'Nothing to freeze.'
+
+
+def get_branch_tree():
+  """Get the dictionary of {branch: parent}, compatible with topo_iter.
+
+  Returns a tuple of (skipped, <branch_tree dict>) where skipped is a set of
+  branches without upstream branches defined.
+  """
+  skipped = set()
+  branch_tree = {}
+
+  for branch in branches():
+    parent = upstream(branch)
+    if not parent:
+      skipped.add(branch)
+      continue
+    branch_tree[branch] = parent
+
+  return skipped, branch_tree
+
+
+def get_or_create_merge_base(branch, parent=None):
+  """Finds the configured merge base for branch.
+
+  If parent is supplied, it's used instead of calling upstream(branch).
+  """
+  base = branch_config(branch, 'base')
+  base_upstream = branch_config(branch, 'base-upstream')
+  parent = parent or upstream(branch)
+  if parent is None or branch is None:
+    return None
+  actual_merge_base = run('merge-base', parent, branch)
+
+  if base_upstream != parent:
+    base = None
+    base_upstream = None
+
+  def is_ancestor(a, b):
+    return run_with_retcode('merge-base', '--is-ancestor', a, b) == 0
+
+  if base:
+    if not is_ancestor(base, branch):
+      logging.debug('Found WRONG pre-set merge-base for %s: %s', branch, base)
+      base = None
+    elif is_ancestor(base, actual_merge_base):
+      logging.debug('Found OLD pre-set merge-base for %s: %s', branch, base)
+      base = None
+    else:
+      logging.debug('Found pre-set merge-base for %s: %s', branch, base)
+
+  if not base:
+    base = actual_merge_base
+    manual_merge_base(branch, base, parent)
+
+  return base
+
+
+def hash_multi(*reflike):
+  return run('rev-parse', *reflike).splitlines()
+
+
+def hash_one(reflike, short=False):
+  args = ['rev-parse', reflike]
+  if short:
+    args.insert(1, '--short')
+  return run(*args)
+
+
+def in_rebase():
+  git_dir = run('rev-parse', '--git-dir')
+  return (
+    os.path.exists(os.path.join(git_dir, 'rebase-merge')) or
+    os.path.exists(os.path.join(git_dir, 'rebase-apply')))
+
+
+def intern_f(f, kind='blob'):
+  """Interns a file object into the git object store.
+
+  Args:
+    f (file-like object) - The file-like object to intern
+    kind (git object type) - One of 'blob', 'commit', 'tree', 'tag'.
+
+  Returns the git hash of the interned object (hex encoded).
+  """
+  ret = run('hash-object', '-t', kind, '-w', '--stdin', stdin=f)
+  f.close()
+  return ret
+
+
+def is_dormant(branch):
+  # TODO(iannucci): Do an oldness check?
+  return branch_config(branch, 'dormant', 'false') != 'false'
+
+
+def manual_merge_base(branch, base, parent):
+  set_branch_config(branch, 'base', base)
+  set_branch_config(branch, 'base-upstream', parent)
+
+
+def mktree(treedict):
+  """Makes a git tree object and returns its hash.
+
+  See |tree()| for the values of mode, type, and ref.
+
+  Args:
+    treedict - { name: (mode, type, ref) }
+  """
+  with tempfile.TemporaryFile() as f:
+    for name, (mode, typ, ref) in treedict.iteritems():
+      f.write('%s %s %s\t%s\0' % (mode, typ, ref, name))
+    f.seek(0)
+    return run('mktree', '-z', stdin=f)
+
+
 def parse_commitrefs(*commitrefs):
   """Returns binary encoded commit hashes for one or more commitrefs.
 
@@ -208,45 +493,192 @@ def parse_commitrefs(*commitrefs):
     * 'cool_branch~2'
   """
   try:
-    return map(binascii.unhexlify, hashes(*commitrefs))
+    return map(binascii.unhexlify, hash_multi(*commitrefs))
   except subprocess2.CalledProcessError:
     raise BadCommitRefException(commitrefs)
 
 
+RebaseRet = collections.namedtuple('RebaseRet', 'success stdout stderr')
+
+
+def rebase(parent, start, branch, abort=False):
+  """Rebases |start|..|branch| onto the branch |parent|.
+
+  Args:
+    parent - The new parent ref for the rebased commits.
+    start  - The commit to start from
+    branch - The branch to rebase
+    abort  - If True, will call git-rebase --abort in the event that the rebase
+             doesn't complete successfully.
+
+  Returns a namedtuple with fields:
+    success - a boolean indicating that the rebase command completed
+              successfully.
+    message - if the rebase failed, this contains the stdout of the failed
+              rebase.
+  """
+  try:
+    args = ['--onto', parent, start, branch]
+    if TEST_MODE:
+      args.insert(0, '--committer-date-is-author-date')
+    run('rebase', *args)
+    return RebaseRet(True, '', '')
+  except subprocess2.CalledProcessError as cpe:
+    if abort:
+      run('rebase', '--abort')
+    return RebaseRet(False, cpe.stdout, cpe.stderr)
+
+
+def remove_merge_base(branch):
+  del_branch_config(branch, 'base')
+  del_branch_config(branch, 'base-upstream')
+
+
+def root():
+  return config('depot-tools.upstream', 'origin/master')
+
+
 def run(*cmd, **kwargs):
-  """Runs a git command. Returns stdout as a string.
+  """The same as run_with_stderr, except it only returns stdout."""
+  return run_with_stderr(*cmd, **kwargs)[0]
+
+
+def run_with_retcode(*cmd, **kwargs):
+  """Run a command but only return the status code."""
+  try:
+    run(*cmd, **kwargs)
+    return 0
+  except subprocess2.CalledProcessError as cpe:
+    return cpe.returncode
 
-  If logging is DEBUG, we'll print the command before we run it.
+
+def run_stream(*cmd, **kwargs):
+  """Runs a git command. Returns stdout as a PIPE (file-like object).
+
+  stderr is dropped to avoid races if the process outputs to both stdout and
+  stderr.
+  """
+  kwargs.setdefault('stderr', subprocess2.VOID)
+  kwargs.setdefault('stdout', subprocess2.PIPE)
+  cmd = (GIT_EXE, '-c', 'color.ui=never') + cmd
+  proc = subprocess2.Popen(cmd, **kwargs)
+  return proc.stdout
+
+
+def run_with_stderr(*cmd, **kwargs):
+  """Runs a git command.
+
+  Returns (stdout, stderr) as a pair of strings.
 
   kwargs
     autostrip (bool) - Strip the output. Defaults to True.
-  Output string is always strip()'d.
+    indata (str) - Specifies stdin data for the process.
   """
+  kwargs.setdefault('stdin', subprocess2.PIPE)
+  kwargs.setdefault('stdout', subprocess2.PIPE)
+  kwargs.setdefault('stderr', subprocess2.PIPE)
   autostrip = kwargs.pop('autostrip', True)
-  cmd = (GIT_EXE,) + cmd
-  logging.debug('Running %s', ' '.join(repr(tok) for tok in cmd))
-  ret = subprocess2.check_output(cmd, stderr=subprocess2.PIPE, **kwargs)
+  indata = kwargs.pop('indata', None)
+
+  cmd = (GIT_EXE, '-c', 'color.ui=never') + cmd
+  proc = subprocess2.Popen(cmd, **kwargs)
+  ret, err = proc.communicate(indata)
+  retcode = proc.wait()
+  if retcode != 0:
+    raise subprocess2.CalledProcessError(retcode, cmd, os.getcwd(), ret, err)
+
   if autostrip:
     ret = (ret or '').strip()
-  return ret
+    err = (err or '').strip()
 
+  return ret, err
 
-def hashes(*reflike):
-  return run('rev-parse', *reflike).splitlines()
 
+def set_branch_config(branch, option, value, scope='local'):
+  set_config('branch.%s.%s' % (branch, option), value, scope=scope)
 
-def intern_f(f, kind='blob'):
-  """Interns a file object into the git object store.
 
-  Args:
-    f (file-like object) - The file-like object to intern
-    kind (git object type) - One of 'blob', 'commit', 'tree', 'tag'.
+def set_config(option, value, scope='local'):
+  run('config', '--' + scope, option, value)
 
-  Returns the git hash of the interned object (hex encoded).
+
+def squash_current_branch(header=None, merge_base=None):
+  header = header or 'git squash commit.'
+  merge_base = merge_base or get_or_create_merge_base(current_branch())
+  log_msg = header + '\n'
+  if log_msg:
+    log_msg += '\n'
+  log_msg += run('log', '--reverse', '--format=%H%n%B', '%s..HEAD' % merge_base)
+  run('reset', '--soft', merge_base)
+  run('commit', '-a', '-F', '-', indata=log_msg)
+
+
+def tags(*args):
+  return run('tag', *args).splitlines()
+
+
+def thaw():
+  took_action = False
+  for sha in (s.strip() for s in run_stream('rev-list', 'HEAD').xreadlines()):
+    msg = run('show', '--format=%f%b', '-s', 'HEAD')
+    match = FREEZE_MATCHER.match(msg)
+    if not match:
+      if not took_action:
+        return 'Nothing to thaw.'
+      break
+
+    run('reset', '--' + FREEZE_SECTIONS[match.group(1)], sha)
+    took_action = True
+
+
+def topo_iter(branch_tree, top_down=True):
+  """Generates (branch, parent) in topographical order for a branch tree.
+
+  Given a tree:
+
+            A1
+        B1      B2
+      C1  C2    C3
+                D1
+
+  branch_tree would look like: {
+    'D1': 'C3',
+    'C3': 'B2',
+    'B2': 'A1',
+    'C1': 'B1',
+    'C2': 'B1',
+    'B1': 'A1',
+  }
+
+  It is OK to have multiple 'root' nodes in your graph.
+
+  if top_down is True, items are yielded from A->D. Otherwise they're yielded
+  from D->A. Within a layer the branches will be yielded in sorted order.
   """
-  ret = run('hash-object', '-t', kind, '-w', '--stdin', stdin=f)
-  f.close()
-  return ret
+  branch_tree = branch_tree.copy()
+
+  # TODO(iannucci): There is probably a more efficient way to do these.
+  if top_down:
+    while branch_tree:
+      this_pass = [(b, p) for b, p in branch_tree.iteritems()
+                   if p not in branch_tree]
+      assert this_pass, "Branch tree has cycles: %r" % branch_tree
+      for branch, parent in sorted(this_pass):
+        yield branch, parent
+        del branch_tree[branch]
+  else:
+    parent_to_branches = collections.defaultdict(set)
+    for branch, parent in branch_tree.iteritems():
+      parent_to_branches[parent].add(branch)
+
+    while branch_tree:
+      this_pass = [(b, p) for b, p in branch_tree.iteritems()
+                   if not parent_to_branches[b]]
+      assert this_pass, "Branch tree has cycles: %r" % branch_tree
+      for branch, parent in sorted(this_pass):
+        yield branch, parent
+        parent_to_branches[parent].discard(branch)
+        del branch_tree[branch]
 
 
 def tree(treeref, recurse=False):
@@ -286,16 +718,54 @@ def tree(treeref, recurse=False):
   return ret
 
 
-def mktree(treedict):
-  """Makes a git tree object and returns its hash.
+def upstream(branch):
+  try:
+    return run('rev-parse', '--abbrev-ref', '--symbolic-full-name',
+               branch+'@{upstream}')
+  except subprocess2.CalledProcessError:
+    return None
 
-  See |tree()| for the values of mode, type, and ref.
 
-  Args:
-    treedict - { name: (mode, type, ref) }
-  """
-  with tempfile.TemporaryFile() as f:
-    for name, (mode, typ, ref) in treedict.iteritems():
-      f.write('%s %s %s\t%s\0' % (mode, typ, ref, name))
-    f.seek(0)
-    return run('mktree', '-z', stdin=f)
+def get_git_version():
+  """Returns a tuple that contains the numeric components of the current git
+  version."""
+  version_string = run('--version')
+  version_match = re.search(r'(\d+.)+(\d+)', version_string)
+  version = version_match.group() if version_match else ''
+
+  return tuple(int(x) for x in version.split('.'))
+
+
+def get_branches_info(include_tracking_status):
+  format_string = (
+      '--format=%(refname:short):%(objectname:short):%(upstream:short):')
+
+  # This is not covered by the depot_tools CQ which only has git version 1.8.
+  if (include_tracking_status and
+      get_git_version() >= MIN_UPSTREAM_TRACK_GIT_VERSION):  # pragma: no cover
+    format_string += '%(upstream:track)'
+
+  info_map = {}
+  data = run('for-each-ref', format_string, 'refs/heads')
+  BranchesInfo = collections.namedtuple(
+      'BranchesInfo', 'hash upstream ahead behind')
+  for line in data.splitlines():
+    (branch, branch_hash, upstream_branch, tracking_status) = line.split(':')
+
+    ahead_match = re.search(r'ahead (\d+)', tracking_status)
+    ahead = int(ahead_match.group(1)) if ahead_match else None
+
+    behind_match = re.search(r'behind (\d+)', tracking_status)
+    behind = int(behind_match.group(1)) if behind_match else None
+
+    info_map[branch] = BranchesInfo(
+        hash=branch_hash, upstream=upstream_branch, ahead=ahead, behind=behind)
+
+  # Set None for upstreams which are not branches (e.g empty upstream, remotes
+  # and deleted upstream branches).
+  missing_upstreams = {}
+  for info in info_map.values():
+    if info.upstream not in info_map and info.upstream not in missing_upstreams:
+      missing_upstreams[info.upstream] = None
+
+  return dict(info_map.items() + missing_upstreams.items())