Added git footers tool to parse conventional metadata from git commits

git_footers.py

Index: git_footers.py
diff --git a/git_footers.py b/git_footers.py
new file mode 100755
index 0000000000000000000000000000000000000000..ad35fbeff1d65809358457dd54cb46720d9edac2
--- /dev/null
+++ b/git_footers.py
@@ -0,0 +1,132 @@
+#!/usr/bin/env python
+# 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 argparse
+import re
+import sys
+
+from collections import defaultdict
+
+import git_common as git
+
+FOOTER_PATTERN = re.compile(r'^\s*([\w-]+): (.*)$')
+CHROME_COMMIT_POSITION_PATTERN = re.compile(r'^([\w/-]+)@{#(\d+)}$')
+GIT_SVN_ID_PATTERN = re.compile('^([^\s@]+)@(\d+)')
+
+def normalize_name(header):
+  return '-'.join([ word.title() for word in header.strip().split('-') ])
+
+
+def parse_footer(line):
+  match = FOOTER_PATTERN.match(line)
+  if match:
+    return (match.group(1), match.group(2))
+  else:
+    return None
+
+
+def parse_footers(message):
+  """Parses a git commit message into a multimap of footers."""
+  footer_lines = []
+  for line in reversed(message.splitlines()):
+    if line == '' or line.isspace():
+      break
+    footer_lines.append(line)
+
+  footers = map(parse_footer, footer_lines)
+  if not all(footers):
+    return defaultdict(list)
+
+  footer_map = defaultdict(list)
+  for (k, v) in footers:
+    footer_map[normalize_name(k)].append(v.strip())
+
+  return footer_map
+
+
+def get_unique(footers, key):
+  key = normalize_name(key)
+  values = footers[key]
+  assert len(values) <= 1, 'Multiple %s footers' % key
+  if values:
+    return values[0]
+  else:
+    return None
+
+
+def get_position(footers):
+  """Get the chrome commit position from a footer multimap using a heuristic.
+
+  Returns:
+    A tuple of the branch and the position on that branch. For example,
+
+    Cr-Commit-Position: refs/heads/master@{#292272}
+
+    would give the return value ('refs/heads/master', 292272).  The position
+    number can be None if that doesn't make sense (e.g. a branch).

[iannucci, 2014/09/03 00:21:44]

mention that this falls back to git-svn-id

also... branches have position numbers

[luqui, 2014/09/03 02:28:48]

Done.

+  """
+
+  position = get_unique(footers, 'Cr-Commit-Position')
+  if position:
+    match = CHROME_COMMIT_POSITION_PATTERN.match(position)
+    assert match, 'Invalid Cr-Commit-Position value: %s' % position
+    return (match.group(1), match.group(2))
+
+  svn_commit = get_unique(footers, 'Git-Svn-Id')

[iannucci, 2014/09/03 00:21:44]

note that it's always git-svn-id, though you're lower()ing, so I guess it
doesn't matter?

[luqui, 2014/09/03 02:28:49]

Made it git-svn-id for greppability.

+  if svn_commit:
+    match = GIT_SVN_ID_PATTERN.match(svn_commit)
+    assert match, 'Invalid Git-Svn-Id value: %s' % svn_commit
+    if re.match('.*/chrome/trunk/src$', match.group(1)):
+      return ('refs/heads/master', match.group(2))
+    branch_match = re.match('.*/chrome/branches/([\w/-]+)/src$', match.group(1))
+    if branch_match:
+      return ('refs/branch-heads/%s' % branch_match.group(1), None)
+
+  raise ValueError('Unable to infer commit position from footers')
+
+
+def main(args):
+  parser = argparse.ArgumentParser(
+    formatter_class=argparse.ArgumentDefaultsHelpFormatter
+  )
+  parser.add_argument('ref')
+
+  g = parser.add_mutually_exclusive_group()
+  g.add_argument('--key', metavar='KEY',
+                 help='Get all values for the given footer name, one per '
+                 'line (case insensitive)')
+  g.add_argument('--position', action='store_true')
+  g.add_argument('--position-ref', action='store_true')
+  g.add_argument('--position-num', action='store_true')

[iannucci, 2014/09/03 00:21:44]

use an exclusive group for all of these

[luqui, 2014/09/03 02:28:48]

Uh... they already are?

+
+
+  opts = parser.parse_args(args)
+
+  message = git.run('log', '-1', '--format=%B', opts.ref)
+  footers = parse_footers(message)
+
+  if opts.key:
+    for v in footers.get(normalize_name(opts.key), []):
+      print v
+  elif opts.position:
+    pos = get_position(footers)
+    if pos[1] is not None:
+      print '%s@{#%s}' % (pos[0], pos[1])
+    else:
+      print pos[0]

[iannucci, 2014/09/03 00:21:44]

I would actually print it as

  refs/branch-heads/12345@{#0}

[luqui, 2014/09/03 02:28:48]

Printing as refs/branch-heads/12345@{#?} 

+  elif opts.position_ref:
+    print get_position(footers)[0]
+  elif opts.position_num:
+    pos = get_position(footers)
+    assert pos[1], 'No valid position for commit'
+    print pos[1]
+  else:
+    for k in footers.keys():
+      for v in footers[k]:
+        print '%s: %s' % (k, v)
+
+
+if __name__ == '__main__':
+  sys.exit(main(sys.argv[1:]))