svn-fast-export.py

   1 #!/usr/bin/python
   2 #
   3 # svn-fast-export.py
   4 # ----------
   5 #  Walk through each revision of a local Subversion repository and export it
   6 #  in a stream that git-fast-import can consume.
   7 #
   8 # Author: Chris Lee <clee@kde.org>
   9 # License: MIT <http://www.opensource.org/licenses/mit-license.php>
  10
  11 trunk_path = '/trunk/'
  12 branches_path = '/branches/'
  13 tags_path = '/tags/'
  14
  15 first_rev = 1
  16 final_rev = 0
  17
  18 import gc, sys, os.path
  19 from optparse import OptionParser
  20 from time import sleep, mktime, localtime, strftime, strptime
  21 from svn.fs import svn_fs_dir_entries, svn_fs_file_length, svn_fs_file_contents, svn_fs_is_dir, svn_fs_revision_root, svn_fs_youngest_rev, svn_fs_revision_proplist, svn_fs_revision_prop, svn_fs_paths_changed
  22 from svn.core import svn_pool_create, svn_pool_clear, svn_pool_destroy, svn_stream_read, svn_stream_for_stdout, svn_stream_copy, svn_stream_close, run_app
  23 from svn.repos import svn_repos_open, svn_repos_fs
  24
  25 ct_short = ['M', 'A', 'D', 'R', 'X']
  26
  27 def dump_file_blob(root, full_path, pool):
  28     stream_length = svn_fs_file_length(root, full_path, pool)
  29     stream = svn_fs_file_contents(root, full_path, pool)
  30     sys.stdout.write("data %s\n" % stream_length)
  31     sys.stdout.flush()
  32     ostream = svn_stream_for_stdout(pool)
  33     svn_stream_copy(stream, ostream, pool)
  34     svn_stream_close(ostream)
  35     sys.stdout.write("\n")
  36
  37
  38 def export_revision(rev, repo, fs, pool):
  39     sys.stderr.write("Exporting revision %s... " % rev)
  40
  41     revpool = svn_pool_create(pool)
  42     svn_pool_clear(revpool)
  43
  44     # Open a root object representing the youngest (HEAD) revision.
  45     root = svn_fs_revision_root(fs, rev, revpool)
  46
  47     # And the list of what changed in this revision.
  48     changes = svn_fs_paths_changed(root, revpool)
  49
  50     i = 1
  51     marks = {}
  52     file_changes = []
  53
  54     for path, change_type in changes.iteritems():
  55         c_t = ct_short[change_type.change_kind]
  56         if svn_fs_is_dir(root, path, revpool):
  57             continue
  58
  59         if not path.startswith(trunk_path):
  60             # We don't handle branches. Or tags. Yet.
  61             pass
  62         else:
  63             if c_t == 'D':
  64                 file_changes.append("D %s" % path.replace(trunk_path, ''))
  65             else:
  66                 marks[i] = path.replace(trunk_path, '')
  67                 file_changes.append("M 644 :%s %s" % (i, marks[i]))
  68                 sys.stdout.write("blob\nmark :%s\n" % i)
  69                 dump_file_blob(root, path, revpool)
  70                 i += 1
  71
  72     # Get the commit author and message
  73     props = svn_fs_revision_proplist(fs, rev, revpool)
  74
  75     # Do the recursive crawl.
  76     if props.has_key('svn:author'):
  77         author = "%s <%s@localhost>" % (props['svn:author'], props['svn:author'])
  78     else:
  79         author = 'nobody <nobody@localhost>'
  80
  81     if len(file_changes) == 0:
  82         svn_pool_destroy(revpool)
  83         sys.stderr.write("skipping.\n")
  84         return
  85
  86     svndate = props['svn:date'][0:-8]
  87     commit_time = mktime(strptime(svndate, '%Y-%m-%dT%H:%M:%S'))
  88     sys.stdout.write("commit refs/heads/master\n")
  89     sys.stdout.write("committer %s %s -0000\n" % (author, int(commit_time)))
  90     sys.stdout.write("data %s\n" % len(props['svn:log']))
  91     sys.stdout.write(props['svn:log'])
  92     sys.stdout.write("\n")
  93     sys.stdout.write('\n'.join(file_changes))
  94     sys.stdout.write("\n\n")
  95
  96     svn_pool_destroy(revpool)
  97
  98     sys.stderr.write("done!\n")
  99
 100     #if rev % 1000 == 0:
 101     #    sys.stderr.write("gc: %s objects\n" % len(gc.get_objects()))
 102     #    sleep(5)
 103
 104
 105 def crawl_revisions(pool, repos_path):
 106     """Open the repository at REPOS_PATH, and recursively crawl all its
 107     revisions."""
 108     global final_rev
 109
 110     # Open the repository at REPOS_PATH, and get a reference to its
 111     # versioning filesystem.
 112     repos_obj = svn_repos_open(repos_path, pool)
 113     fs_obj = svn_repos_fs(repos_obj)
 114
 115     # Query the current youngest revision.
 116     youngest_rev = svn_fs_youngest_rev(fs_obj, pool)
 117
 118
 119     first_rev = 1
 120     if final_rev == 0:
 121         final_rev = youngest_rev
 122     for rev in xrange(first_rev, final_rev + 1):
 123         export_revision(rev, repos_obj, fs_obj, pool)
 124
 125
 126 if __name__ == '__main__':
 127     usage = '%prog [options] REPOS_PATH'
 128     parser = OptionParser()
 129     parser.set_usage(usage)
 130     parser.add_option('-f', '--final-rev', help='Final revision to import',
 131                       dest='final_rev', metavar='FINAL_REV', type='int')
 132     parser.add_option('-t', '--trunk-path', help='Path in repo to /trunk',
 133                       dest='trunk_path', metavar='TRUNK_PATH')
 134     parser.add_option('-b', '--branches-path', help='Path in repo to /branches',
 135                       dest='branches_path', metavar='BRANCHES_PATH')
 136     parser.add_option('-T', '--tags-path', help='Path in repo to /tags',
 137                       dest='tags_path', metavar='TAGS_PATH')
 138     (options, args) = parser.parse_args()
 139
 140     if options.trunk_path != None:
 141         trunk_path = options.trunk_path
 142     if options.branches_path != None:
 143         branches_path = options.branches_path
 144     if options.tags_path != None:
 145         tags_path = options.tags_path
 146     if options.final_rev != None:
 147         final_rev = options.final_rev
 148
 149     if len(args) != 1:
 150         parser.print_help()
 151         sys.exit(2)
 152
 153     # Canonicalize (enough for Subversion, at least) the repository path.
 154     repos_path = os.path.normpath(args[0])
 155     if repos_path == '.':
 156         repos_path = ''
 157
 158     # Call the app-wrapper, which takes care of APR initialization/shutdown
 159     # and the creation and cleanup of our top-level memory pool.
 160     run_app(crawl_revisions, repos_path)