contrib/git-move-tags.py

   1 #!/usr/bin/python
   2
   3 """Remove redundant tag fixup commits from a cvs2svn-converted git repository.
   4
   5 Process each tag in a git repository. If the tagged commit is
   6 tree-wise identical with another tagged commit, the tag is moved to
   7 point at the other commit (i.e., tags pointing at identical content
   8 will all point at a single tag fixup commit).
   9
  10 Furthermore, if one of the parents of the tag fixup commit is
  11 identical to the tag fixup commit itself, then the tag is moved to the
  12 parent.
  13
  14 The script is meant to be run against a repository converted by
  15 cvs2svn, since cvs2svn creates empty commits for some tags.
  16
  17 """
  18
  19 from subprocess import Popen, PIPE, call
  20
  21 # Cache trees we have already seen, and that are suitable targets for
  22 # moved tags
  23 tree_cache = {} # tree SHA1 -> commit SHA1
  24
  25 # Cache parent commit -> parent tree mapping
  26 parent_cache = {} # commit SHA1 -> tree SHA1
  27
  28 def resolve_commit(commit):
  29     """Return the tree object associated with the given commit."""
  30
  31     get_tree_cmd = ["git", "rev-parse", commit + "^{tree}"]
  32     tree = Popen(get_tree_cmd, stdout = PIPE).communicate()[0].strip()
  33     return tree
  34
  35 def move_tag(tag, from_commit, to_commit):
  36     """Move the given tag to the given commit."""
  37
  38     print "Moving tag %s from %s to %s..." % (tag, from_commit, to_commit),
  39     retcode = call(["git", "tag", "-f", tag, to_commit])
  40     if retcode == 0:
  41         print "done"
  42     else:
  43         print "FAILED"
  44
  45 def try_to_move_tag(tag, commit, tree, parents):
  46     """Try to move the given tag to a separate commit (with identical tree)."""
  47
  48     if tree in tree_cache:
  49         # We have already found a suitable commit for this tree
  50         move_tag(tag, commit, tree_cache[tree])
  51         return
  52
  53     # Try to move this tag to one of its commit's parents
  54     for p in parents:
  55         if p not in parent_cache:
  56             # Not in cache
  57             parent_cache[p] = resolve_commit(p)
  58         p_tree = parent_cache[p]
  59         if tree == p_tree:
  60             # We can move tag to parent p
  61             move_tag(tag, commit, p)
  62             commit = p
  63             break
  64
  65     # Register the resulting commit object in the tree_cache
  66     assert tree not in tree_cache # Sanity check
  67     tree_cache[tree] = commit
  68
  69 # Command for retrieving tags and associated metadata
  70 # See 'git for-each-ref' manual page for --format details
  71 get_tag_info_cmd = [
  72     "git",
  73     "for-each-ref",
  74     "--format=%(refname)%00%(objecttype)%00%(subject)%00"
  75               "%(objectname)%00%(tree)%00%(parent)%00"
  76               "%(*objectname)%00%(*tree)%00%(*parent)",
  77     "refs/tags",
  78 ]
  79
  80 get_tag_info = Popen(get_tag_info_cmd, stdout = PIPE)
  81
  82 while True: # While get_tag_info process is still running
  83     for line in get_tag_info.stdout:
  84         line = line.strip()
  85         (tag, objtype, subject,
  86          commit, tree, parents,
  87          commit_alt, tree_alt, parents_alt) = line.split(chr(0))
  88         if objtype == "tag":
  89             commit = commit_alt
  90             tree = tree_alt
  91             parents = parents_alt
  92         elif objtype != "commit":
  93             continue
  94
  95         if subject.startswith("This commit was manufactured by cvs2svn") \
  96                or not subject:
  97             # We shall try to move this tag, if possible
  98             parent_list = []
  99             if parents:
 100                 parent_list = parents.split(" ")
 101             for p in parent_list:
 102                 assert len(p) == 40
 103             assert tag.startswith("refs/tags/")
 104             try_to_move_tag(tag[10:], commit, tree, parent_list)
 105         else:
 106             # We shall not move this tag, but it is a possible target
 107             # for other tags that we _do_ want to move
 108             tree_cache.setdefault(tree, commit)
 109
 110     if get_tag_info.poll() is not None:
 111         # Break if no longer running:
 112         break
 113
 114 assert get_tag_info.returncode == 0
 115