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