contrib/git-move-refs.py

   1 #!/usr/bin/python
   2
   3 """Remove redundant fixup commits from a cvs2svn-converted git repository.
   4
   5 Process each head ref and/or tag in a git repository. If the
   6 associated commit is tree-wise identical with another commit, the head
   7 or tag is moved to point at the other commit (i.e., refs pointing at
   8 identical content will all point at a single fixup commit).
   9
  10 Furthermore, if one of the parents of the fixup commit is identical to
  11 the fixup commit itself, then the head or tag is moved to the parent.
  12
  13 The script is meant to be run against a repository converted by
  14 cvs2svn, since cvs2svn creates empty commits for some tags and head
  15 refs (branches).
  16
  17 """
  18
  19 usage = 'USAGE: %prog [options]'
  20
  21 import sys
  22 import optparse
  23 from subprocess import Popen, PIPE, call
  24
  25
  26 # Cache trees we have already seen, and that are suitable targets for
  27 # moved refs
  28 tree_cache = {} # tree SHA1 -> commit SHA1
  29
  30 # Cache parent commit -> parent tree mapping
  31 parent_cache = {} # commit SHA1 -> tree SHA1
  32
  33
  34 def resolve_commit(commit):
  35     """Return the tree object associated with the given commit."""
  36
  37     get_tree_cmd = ["git", "rev-parse", commit + "^{tree}"]
  38     tree = Popen(get_tree_cmd, stdout = PIPE).communicate()[0].strip()
  39     return tree
  40
  41
  42 def move_ref(ref, from_commit, to_commit, ref_type):
  43     """Move the given head to the given commit.
  44     ref_type is either "tags" or "heads"
  45     """
  46     if from_commit != to_commit:
  47         print "Moving ref %s from %s to %s..." % (ref, from_commit, to_commit),
  48         if ref_type == "tags":
  49             command = "tag"
  50         else:
  51             command = "branch"
  52         retcode = call(["git", command, "-f", ref, to_commit])
  53         if retcode == 0:
  54             print "done"
  55         else:
  56             print "FAILED"
  57
  58
  59 def try_to_move_ref(ref, commit, tree, parents, ref_type):
  60     """Try to move the given ref to a separate commit (with identical tree)."""
  61
  62     if tree in tree_cache:
  63         # We have already found a suitable commit for this tree
  64         move_ref(ref, commit, tree_cache[tree], ref_type)
  65         return
  66
  67     # Try to move this ref to one of its commit's parents
  68     for p in parents:
  69         if p not in parent_cache:
  70             # Not in cache
  71             parent_cache[p] = resolve_commit(p)
  72         p_tree = parent_cache[p]
  73         if tree == p_tree:
  74             # We can move ref to parent p
  75             move_ref(ref, commit, p, ref_type)
  76             commit = p
  77             break
  78
  79     # Register the resulting commit object in the tree_cache
  80     assert tree not in tree_cache # Sanity check
  81     tree_cache[tree] = commit
  82
  83
  84 def process_refs(ref_type):
  85     tree_cache.clear()
  86     parent_cache.clear()
  87
  88     # Command for retrieving refs and associated metadata
  89     # See 'git for-each-ref' manual page for --format details
  90     get_ref_info_cmd = [
  91         "git",
  92         "for-each-ref",
  93         "--format=%(refname)%00%(objecttype)%00%(subject)%00"
  94                   "%(objectname)%00%(tree)%00%(parent)%00"
  95                   "%(*objectname)%00%(*tree)%00%(*parent)",
  96         "refs/%s" % (ref_type,),
  97     ]
  98
  99     get_ref_info = Popen(get_ref_info_cmd, stdout = PIPE)
 100
 101     while True: # While get_ref_info process is still running
 102         for line in get_ref_info.stdout:
 103             line = line.strip()
 104             (ref, objtype, subject,
 105              commit, tree, parents,
 106              commit_alt, tree_alt, parents_alt) = line.split(chr(0))
 107             if objtype == "tag":
 108                 commit = commit_alt
 109                 tree = tree_alt
 110                 parents = parents_alt
 111             elif objtype != "commit":
 112                 continue
 113
 114             if subject.startswith("This commit was manufactured by cvs2svn") \
 115                    or not subject:
 116                 # We shall try to move this ref, if possible
 117                 parent_list = []
 118                 if parents:
 119                     parent_list = parents.split(" ")
 120                 for p in parent_list:
 121                     assert len(p) == 40
 122                 ref_prefix = "refs/%s/" % (ref_type,)
 123                 assert ref.startswith(ref_prefix)
 124                 try_to_move_ref(
 125                     ref[len(ref_prefix):], commit, tree, parent_list, ref_type
 126                     )
 127             else:
 128                 # We shall not move this ref, but it is a possible target
 129                 # for other refs that we _do_ want to move
 130                 tree_cache.setdefault(tree, commit)
 131
 132         if get_ref_info.poll() is not None:
 133             # Break if no longer running:
 134             break
 135
 136     assert get_ref_info.returncode == 0
 137
 138
 139 def main(args):
 140     parser = optparse.OptionParser(usage=usage, description=__doc__)
 141     parser.add_option(
 142         '--tags', '-t',
 143         action='store_true', default=False,
 144         help='process tags',
 145         )
 146     parser.add_option(
 147         '--branches', '-b',
 148         action='store_true', default=False,
 149         help='process branches',
 150         )
 151
 152     (options, args) = parser.parse_args(args=args)
 153
 154     if args:
 155         parser.error('Unexpected command-line arguments')
 156
 157     if not (options.tags or options.branches):
 158         # By default, process tags but not branches:
 159         options.tags = True
 160
 161     if options.tags:
 162         process_refs("tags")
 163
 164     if options.branches:
 165         process_refs("branches")
 166
 167
 168 main(sys.argv[1:])
 169
 170