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..3235b1e0858e5b70451b35d0f0891b44e2aac600
--- /dev/null
+++ b/git_number.py
@@ -0,0 +1,232 @@
+#!/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.
+

[M-A Ruel, 2013/11/07 20:59:11]

I like a short module docstring that tells what the script does and why, then
use
OptionParser(..., description=sys.modules[__name__].__doc__)

[iannucci, 2013/11/07 21:44:57]

Done. (though I used usage= because description was reflowing the text poorly).

+import binascii
+import collections
+import logging
+import optparse
+import os
+import struct
+import subprocess
+import sys
+import tempfile
+
+import git_common as git
+
+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
+
+# Set this to 'threads' to gather coverage data while testing.
+POOL_KIND = 'procs'
+

[M-A Ruel, 2013/11/07 20:59:11]

one more line

[iannucci, 2013/11/07 21:44:57]

Done.

+@git.memoize_one
+def get_number_tree(prefix_bytes):
+  """Return a dictionary of the blob contents specified by |prefix_bytes|.

[M-A Ruel, 2013/11/07 20:59:11]

Returns

[iannucci, 2013/11/07 21:44:57]

Done.

+  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.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_hash, num = struct.unpack_from(CHUNK_FMT, raw, i * CHUNK_SIZE)
+    ret[commit_hash] = num
+
+  return ret
+
+
+@git.memoize_one
+def get_num(commit_hash):
+  """Takes a hash and returns the generation number for it or None if the
+  commit_hash is unknown."""
+  return get_number_tree(commit_hash[:PREFIX_LEN]).get(commit_hash)
+

[M-A Ruel, 2013/11/07 20:59:11]

space more

[iannucci, 2013/11/07 21:44:57]

Done.

+def clear_caches():
+  get_number_tree.cache.clear()

[M-A Ruel, 2013/11/07 20:59:11]

In general I'm not a fan of global caches, for example this code inhibits
processing multiple git repositories inside a single process. For example let's
say someone in a few years wants to process all the DEPS as pseudo revs and use
this script as a library.

I'm not requesting you to change it but I think it is a significant downside of
the technique used here.

[iannucci, 2013/11/07 21:44:57]

Fair point. It would make sense to refactor this into a class which takes a repo
as it's argument. Maybe for v2?

+  get_num.cache.clear()
+
+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.intern_f(f)
+
+
+def leaf_map_fn((pre, tree)):

[M-A Ruel, 2013/11/07 20:59:11]

Is (()) intended?

[iannucci, 2013/11/07 21:44:57]

Yeah, this function is invoked with a tuple (by multiprocessing). The extra ()
unpacks it automatically.

+  """Converts a prefix and number tree into a git index line."""
+  return '100644 blob %s\t%s\0' % (intern_number_tree(tree),
+                                   git.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(git.run('rev-parse', '--git-dir'), 'number.idx')
+  env = os.environ.copy()
+  env['GIT_INDEX_FILE'] = idx
+
+  progress_message = 'Finalizing: (%%(count)d/%d)' % len(DIRTY_TREES)
+  with git.ProgressPrinter(progress_message) as inc:
+    git.run('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.ScopedPool(kind=POOL_KIND) 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 = git.run('write-tree', env=env)
+    commit_cmd = ['commit-tree', '-m', msg, '-p'] + git.hashes(REF)
+    for t in targets:
+      commit_cmd += ['-p', binascii.hexlify(t)]
+    commit_cmd.append(tree_id)
+    commit_hash = git.run(*commit_cmd)
+    git.run('update-ref', REF, commit_hash)
+  DIRTY_TREES.clear()
+
+
+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)):
+    # This isn't covered because PREFIX_LEN currently == 1
+    if depth > 1:  # pragma: no cover
+      for r in all_prefixes(depth-1):
+        yield x+r
+    else:
+      yield x
+
+
+def load(targets):
+  """Load the generation numbers for targets. Calculates missing numbers if
+  one or more of the targets is past the cached calculations.
+
+  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.tree(REF) is None:
+    empty = git.mktree({})
+    commit_hash = git.run('commit-tree', '-m', 'Initial commit from git-number',
+                        empty)
+    git.run('update-ref', REF, commit_hash)
+
+  with git.ScopedPool(kind=POOL_KIND) as pool:
+    preload_iter = pool.imap_unordered(preload_tree, all_prefixes())
+
+    rev_list = []
+
+    with git.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(binascii.hexlify, targets)
+      for line in git.run(*cmd).splitlines():
+        tokens = map(binascii.unhexlify, line.split())
+        rev_list.append((tokens[0], tokens[1:]))
+        inc()
+
+    for prefix, tree in preload_iter:
+      get_number_tree.cache[prefix] = tree
+
+  with git.ProgressPrinter('Counting: %%(count)d/%d' % len(rev_list)) as inc:
+    for commit_hash, pars in rev_list:
+      num = max(map(get_num, pars)) + 1 if pars else 0
+
+      prefix = commit_hash[:PREFIX_LEN]
+      get_number_tree(prefix)[commit_hash] = num
+      DIRTY_TREES[prefix] += 1
+      get_num.cache[commit_hash] = num
+
+      inc()
+
+
+def git_number(do_reset, do_cache, target_refs):
+  if do_reset:
+    git.run('update-ref', '-d', REF)
+    return
+
+  targets = git.parse_committishes(*target_refs)
+  load(targets)
+  ret = map(get_num, targets)
+  if do_cache:
+    finalize(targets)
+
+  return ret
+
+
+def main():  # pragma: no cover
+  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 opts.verbose == 1:
+    logging.getLogger().setLevel(logging.INFO)
+  elif opts.verbose >= 2:
+    logging.getLogger().setLevel(logging.DEBUG)
+
+  try:
+    ret = git_number(opts.reset, not opts.no_cache, args or ['HEAD'])
+    print '\n'.join(map(str, ret))
+    return 0
+  except KeyboardInterrupt:
+    pass
+
+
+if __name__ == '__main__':  # pragma: no cover
+  sys.exit(main())