Add git-number script to calculate generation numbers for commits.

git_number.py

Index: git_number.py
diff --git a/git_number.py b/git_number.py
new file mode 100755
index 0000000000000000000000000000000000000000..198e31b45f828a7f943bb046bafc4f2eb6b13153
--- /dev/null
+++ b/git_number.py
@@ -0,0 +1,217 @@
+#!/usr/bin/env python
+# Copyright (c) 2013 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 collections
+import optparse
+import os
+import struct
+import subprocess
+import sys
+import tempfile
+
+import git_common
+from git_common import run_git, ProgressPrinter
+
+
+CHUNK_FMT = '!20sL'
+CHUNK_SIZE = struct.calcsize(CHUNK_FMT)
+DIRTY_TREES = collections.defaultdict(int)
+REF = 'refs/number/commits'
+
+# Number of bytes to use for the prefix on our internal number structure.
+# 0 is slow to deserialize. 2 creates way too much bookeeping overhead (would
+# need to reimplement cache data structures to be a bit more sophisticated than
+# dicts. 1 seems to be just right.
+PREFIX_LEN = 1
+
+
+@git_common.memoize_one
+def get_number_tree(prefix_bytes):
+  """Return a dictionary of the blob contents specified by |prefix_bytes|.
+  This is in the form of {<full binary ref>: <gen num> ...}
+
+  >>> get_number_tree('\x83\xb4')
+  {'\x83\xb4\xe3\xe4W\xf9J*\x8f/c\x16\xecD\xd1\x04\x8b\xa9qz': 169, ...}
+  """
+  ret = {}
+  ref = '%s:%s' % (REF, git_common.pathlify(prefix_bytes))
+
+  p = subprocess.Popen(['git', 'cat-file', 'blob', ref],
+                        stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+  raw = buffer(p.communicate()[0])
+  for i in xrange(len(raw) / CHUNK_SIZE):
+    commit_id, num = struct.unpack_from(CHUNK_FMT, raw, i * CHUNK_SIZE)
+    ret[commit_id] = num
+
+  return ret
+
+
+@git_common.memoize_one
+def get_num(commit_id):
+  """Takes a hash and returns the generation number for it or None if the
+  commit_id is unknown."""
+  return get_number_tree(commit_id[:PREFIX_LEN]).get(commit_id)
+
+
+def intern_number_tree(tree):
+  """Transforms a number tree (in the form returned by |get_number_tree|) into
+  a git blob.
+
+  Returns the git blob id as hex-encoded string.
+
+  >>> d = {'\x83\xb4\xe3\xe4W\xf9J*\x8f/c\x16\xecD\xd1\x04\x8b\xa9qz': 169}
+  >>> intern_number_tree(d)
+  'c552317aa95ca8c3f6aae3357a4be299fbcb25ce'
+  """
+  with tempfile.TemporaryFile() as f:
+    for k, v in sorted(tree.iteritems()):
+      f.write(struct.pack(CHUNK_FMT, k, v))
+    f.seek(0)
+    return git_common.git_intern_f(f)
+
+
+def leaf_map_fn((pre, tree)):
+  """Converts a prefix and number tree into a git index line."""
+  return '100644 blob %s\t%s\0' % (intern_number_tree(tree),
+                                   git_common.pathlify(pre))
+
+
+def finalize(targets):
+  """Saves all cache data to the git repository.
+
+  After calculating the generation number for |targets|, call finalize() to
+  save all the work to the git repository.
+
+  This in particular saves the trees referred to by DIRTY_TREES.
+  """
+  if not DIRTY_TREES:
+    return
+
+  msg = 'git-number Added %s numbers' % sum(DIRTY_TREES.itervalues())
+
+  idx = os.path.join(run_git('rev-parse', '--git-dir'), 'number.idx')
+  env = os.environ.copy()
+  env['GIT_INDEX_FILE'] = idx
+
+  with ProgressPrinter('Finalizing: (%%(count)d/%d)' % len(DIRTY_TREES)) as inc:
+    run_git('read-tree', REF, env=env)
+
+    prefixes_trees = ((p, get_number_tree(p)) for p in sorted(DIRTY_TREES))
+    updater = subprocess.Popen(['git', 'update-index', '-z', '--index-info'],
+                               stdin=subprocess.PIPE, env=env)
+
+    with git_common.ScopedPool() as leaf_pool:
+      for item in leaf_pool.imap(leaf_map_fn, prefixes_trees):
+        updater.stdin.write(item)
+        inc()
+
+    updater.stdin.close()
+    updater.wait()
+
+    tree_id = run_git('write-tree', env=env)
+    commit_cmd = ['commit-tree', '-m', msg, '-p', git_common.git_hash(REF)]
+    for t in targets:
+      commit_cmd += ['-p', git_common.hexlify(t)]
+    commit_cmd.append(tree_id)
+    commit_id = run_git(*commit_cmd)
+    run_git('update-ref', REF, commit_id)
+
+
+def preload_tree(prefix):
+  """Returns the prefix and parsed tree object for the specified prefix."""
+  return prefix, get_number_tree(prefix)
+
+
+def all_prefixes(depth=PREFIX_LEN):
+  for x in (chr(i) for i in xrange(255)):
+    if depth > 1:
+      for r in all_prefixes(depth-1):
+        yield x+r
+    else:
+      yield x
+
+
+def load(targets):
+  """Load/calculate the generation numbers for targets.

[M-A Ruel, 2013/10/24 13:23:03]

Loads and then calculates

[iannucci, 2013/10/25 00:52:41]

Clarified this docstring.

+
+  Args:
+    targets - An iterable of binary-encoded full git commit id hashes.
+  """
+  if all(get_num(t) is not None for t in targets):
+    return
+
+  if git_common.git_tree(REF) is None:
+    empty = git_common.git_mktree({})
+    commit_id = run_git('commit-tree', '-m', 'Initial commit from git-number',
+                        empty)
+    run_git('update-ref', REF, commit_id)
+
+  with git_common.ScopedPool() as pool:
+    preload_iter = pool.imap_unordered(preload_tree, all_prefixes())
+
+    rev_list = []
+
+    with ProgressPrinter('Loading commits: %(count)d') as inc:
+      # Curiously, buffering the list into memory seems to be the fastest
+      # approach in python (as opposed to iterating over the lines in the
+      # stdout as they're produced). GIL strikes again :/
+      cmd = [
+        'rev-list', '--topo-order', '--parents', '--reverse', '^' + REF
+      ] + map(git_common.hexlify, targets)
+      for line in run_git(*cmd).splitlines():
+        tokens = map(git_common.unhexlify, line.split())
+        rev_list.append((tokens[0], tokens[1:]))
+        inc()
+
+    for prefix, tree in preload_iter:
+      get_number_tree.cache[prefix] = tree
+
+  with ProgressPrinter('Counting: %%(count)d/%d' % len(rev_list)) as inc:
+    for commit_id, pars in rev_list:
+      num = max(map(get_num, pars)) + 1 if pars else 0
+
+      prefix = commit_id[:PREFIX_LEN]
+      get_number_tree(prefix)[commit_id] = num
+      DIRTY_TREES[prefix] += 1
+      get_num.cache[commit_id] = num
+
+      inc()
+
+
+def main():
+  try:
+    parser = optparse.OptionParser(
+      usage='usage: %prog [options] [<committish>]\n\n'
+            '<committish> defaults to HEAD')
+    parser.add_option('--no-cache', action='store_true',
+                      help='Do not actually cache anything we calculate.')
+    parser.add_option('--reset', action='store_true',
+                      help='Reset the generation number cache and quit.')
+    parser.add_option('-v', '--verbose', action='count', default=0,
+                      help='Be verbose. Use more times for more verbosity.')
+    opts, args = parser.parse_args()
+
+    if not args:
+      args = ['HEAD']
+
+    git_common.VERBOSE_LEVEL = opts.verbose
+
+    if opts.reset:

[M-A Ruel, 2013/10/24 13:23:03]

Personally, I'd start the try: here, or preferably create a function here where
you call the worker function. I don't mind much.

[iannucci, 2013/10/25 00:52:41]

Cleaned this up as you suggest. Much better now :)

+      run_git('update-ref', '-d', REF)
+      return 0
+
+    targets = git_common.parse_committishes(*args)
+    load(targets)
+    for t in targets:
+      print get_num(t)
+    if not opts.no_cache:
+      finalize(targets)
+
+    return 0
+  except KeyboardInterrupt:
+    pass
+

[M-A Ruel, 2013/10/24 13:23:03]

2 lines

[iannucci, 2013/10/25 00:52:41]

Done.

+if __name__ == '__main__':
+  sys.exit(main())