3 # Copyright (c) 2012 Felipe Contreras
7 # Just copy to your ~/bin, or anywhere in your $PATH.
8 # Then you can clone with:
9 # % git clone bzr::/path/to/bzr/repo/or/url
12 # % git clone bzr::$HOME/myrepo
14 # % git clone bzr::lp:myrepo
16 # If you want to specify which branches you want track (per repo):
17 # git config remote-bzr.branches 'trunk, devel, test'
23 if hasattr(bzrlib, "initialize"):
27 bzrlib.plugin.load_plugins()
29 import bzrlib.generate_ids
30 import bzrlib.transport
33 import bzrlib.urlutils
40 import atexit, shutil, hashlib, urlparse, subprocess
42 NAME_RE = re.compile('^([^<>]+)')
43 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
44 EMAIL_RE = re.compile('^([^<>]+[^ \\\t<>])?\\b(?:[ \\t<>]*?)\\b([^ \\t<>]+@[^ \\t<>]+)')
45 RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
48 sys.stderr.write('ERROR: %s\n' % (msg % args))
52 sys.stderr.write('WARNING: %s\n' % (msg % args))
55 return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
57 def get_config(config):
58 cmd = ['git', 'config', '--get', config]
59 process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
60 output, _ = process.communicate()
65 def __init__(self, path):
74 if not os.path.exists(self.path):
77 tmp = json.load(open(self.path))
78 self.tips = tmp['tips']
79 self.marks = tmp['marks']
80 self.last_mark = tmp['last-mark']
82 for rev, mark in self.marks.iteritems():
83 self.rev_marks[mark] = rev
86 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
89 json.dump(self.dict(), open(self.path, 'w'))
92 return str(self.dict())
94 def from_rev(self, rev):
95 return self.marks[rev]
97 def to_rev(self, mark):
98 return str(self.rev_marks[mark])
102 return self.last_mark
104 def get_mark(self, rev):
106 self.marks[rev] = self.last_mark
107 return self.last_mark
109 def is_marked(self, rev):
110 return rev in self.marks
112 def new_mark(self, rev, mark):
113 self.marks[rev] = mark
114 self.rev_marks[mark] = rev
115 self.last_mark = mark
117 def get_tip(self, branch):
118 return self.tips.get(branch, None)
120 def set_tip(self, branch, tip):
121 self.tips[branch] = tip
125 def __init__(self, repo):
127 self.line = self.get_line()
130 return sys.stdin.readline().strip()
132 def __getitem__(self, i):
133 return self.line.split()[i]
135 def check(self, word):
136 return self.line.startswith(word)
138 def each_block(self, separator):
139 while self.line != separator:
141 self.line = self.get_line()
144 return self.each_block('')
147 self.line = self.get_line()
148 if self.line == 'done':
152 i = self.line.index(':') + 1
153 return int(self.line[i:])
156 if not self.check('data'):
158 i = self.line.index(' ') + 1
159 size = int(self.line[i:])
160 return sys.stdin.read(size)
162 def get_author(self):
163 m = RAW_AUTHOR_RE.match(self.line)
166 _, name, email, date, tz = m.groups()
167 committer = '%s <%s>' % (name, email)
169 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
170 return (committer, int(date), tz)
172 def rev_to_mark(rev):
174 return marks.from_rev(rev)
176 def mark_to_rev(mark):
178 return marks.to_rev(mark)
180 def fixup_user(user):
182 user = user.replace('"', '')
183 m = AUTHOR_RE.match(user)
186 mail = m.group(2).strip()
188 m = EMAIL_RE.match(user)
193 m = NAME_RE.match(user)
195 name = m.group(1).strip()
202 return '%s <%s>' % (name, mail)
204 def get_filechanges(cur, prev):
208 changes = cur.changes_from(prev)
211 return s.encode('utf-8')
213 for path, fid, kind in changes.added:
214 modified[u(path)] = fid
215 for path, fid, kind in changes.removed:
216 removed[u(path)] = None
217 for path, fid, kind, mod, _ in changes.modified:
218 modified[u(path)] = fid
219 for oldpath, newpath, fid, kind, mod, _ in changes.renamed:
220 removed[u(oldpath)] = None
221 if kind == 'directory':
222 lst = cur.list_files(from_dir=newpath, recursive=True)
223 for path, file_class, kind, fid, entry in lst:
224 if kind != 'directory':
225 modified[u(newpath + '/' + path)] = fid
227 modified[u(newpath)] = fid
229 return modified, removed
231 def export_files(tree, files):
232 global marks, filenodes
235 for path, fid in files.iteritems():
236 kind = tree.kind(fid)
238 h = tree.get_file_sha1(fid)
240 if kind == 'symlink':
241 d = tree.get_symlink_target(fid)
245 if tree.is_executable(fid):
250 # is the blob already exported?
253 final.append((mode, mark, path))
256 d = tree.get_file_text(fid)
257 elif kind == 'directory':
260 die("Unhandled kind '%s' for path '%s'" % (kind, path))
262 mark = marks.next_mark()
266 print "mark :%u" % mark
267 print "data %d" % len(d)
270 final.append((mode, mark, path))
274 def export_branch(repo, name):
277 ref = '%s/heads/%s' % (prefix, name)
278 tip = marks.get_tip(name)
280 branch = bzrlib.branch.Branch.open(branches[name])
281 repo = branch.repository
284 revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
286 tip_revno = branch.revision_id_to_revno(tip)
287 last_revno, _ = branch.last_revision_info()
288 total = last_revno - tip_revno
289 except bzrlib.errors.NoSuchRevision:
293 for revid, _, seq, _ in revs:
295 if marks.is_marked(revid):
298 rev = repo.get_revision(revid)
301 parents = rev.parent_ids
304 committer = rev.committer.encode('utf-8')
305 committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
306 authors = rev.get_apparent_authors()
308 author = authors[0].encode('utf-8')
309 author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
312 msg = rev.message.encode('utf-8')
316 if len(parents) == 0:
317 parent = bzrlib.revision.NULL_REVISION
321 cur_tree = repo.revision_tree(revid)
322 prev = repo.revision_tree(parent)
323 modified, removed = get_filechanges(cur_tree, prev)
325 modified_final = export_files(cur_tree, modified)
327 if len(parents) == 0:
328 print 'reset %s' % ref
330 print "commit %s" % ref
331 print "mark :%d" % (marks.get_mark(revid))
332 print "author %s" % (author)
333 print "committer %s" % (committer)
334 print "data %d" % (len(msg))
337 for i, p in enumerate(parents):
346 print "merge :%s" % m
350 for f in modified_final:
351 print "M %s :%u %s" % f
355 # let's skip branch revisions from the progress report
358 progress = (revno - tip_revno)
359 if (progress % 100 == 0):
361 print "progress revision %d '%s' (%d/%d)" % (revno, name, progress, total)
363 print "progress revision %d '%s' (%d)" % (revno, name, progress)
367 revid = branch.last_revision()
369 # make sure the ref is updated
370 print "reset %s" % ref
371 print "from :%u" % rev_to_mark(revid)
374 marks.set_tip(name, revid)
376 def export_tag(repo, name):
379 ref = '%s/tags/%s' % (prefix, name)
380 print "reset %s" % ref
381 print "from :%u" % rev_to_mark(tags[name])
384 def do_import(parser):
388 path = os.path.join(dirname, 'marks-git')
391 if os.path.exists(path):
392 print "feature import-marks=%s" % path
393 print "feature export-marks=%s" % path
394 print "feature force"
397 while parser.check('import'):
399 if ref.startswith('refs/heads/'):
400 name = ref[len('refs/heads/'):]
401 export_branch(repo, name)
402 if ref.startswith('refs/tags/'):
403 name = ref[len('refs/tags/'):]
404 export_tag(repo, name)
411 def parse_blob(parser):
415 mark = parser.get_mark()
417 data = parser.get_data()
418 blob_marks[mark] = data
423 def __init__(self, branch, revid, parents, files):
429 def copy_tree(revid):
430 files = files_cache[revid] = {}
432 tree = branch.repository.revision_tree(revid)
434 for path, entry in tree.iter_entries_by_dir():
435 files[path] = [entry.file_id, None]
440 if len(parents) == 0:
441 self.base_id = bzrlib.revision.NULL_REVISION
444 self.base_id = parents[0]
445 self.base_files = files_cache.get(self.base_id, None)
446 if not self.base_files:
447 self.base_files = copy_tree(self.base_id)
449 self.files = files_cache[revid] = self.base_files.copy()
452 for path, data in self.files.iteritems():
454 self.rev_files[fid] = [path, mark]
456 for path, f in files.iteritems():
457 fid, mark = self.files.get(path, [None, None])
459 fid = bzrlib.generate_ids.gen_file_id(path)
461 self.rev_files[fid] = [path, mark]
462 self.updates[fid] = f
464 def last_revision(self):
467 def iter_changes(self):
470 def get_parent(dirname, basename):
471 parent_fid, mark = self.base_files.get(dirname, [None, None])
474 parent_fid, mark = self.files.get(dirname, [None, None])
479 fid = bzrlib.generate_ids.gen_file_id(path)
480 add_entry(fid, dirname, 'directory')
483 def add_entry(fid, path, kind, mode = None):
484 dirname, basename = os.path.split(path)
485 parent_fid = get_parent(dirname, basename)
490 elif mode == '120000':
501 self.files[path] = [change[0], None]
502 changes.append(change)
504 def update_entry(fid, path, kind, mode = None):
505 dirname, basename = os.path.split(path)
506 parent_fid = get_parent(dirname, basename)
511 elif mode == '120000':
522 self.files[path] = [change[0], None]
523 changes.append(change)
525 def remove_entry(fid, path, kind):
526 dirname, basename = os.path.split(path)
527 parent_fid = get_parent(dirname, basename)
537 changes.append(change)
539 for fid, f in self.updates.iteritems():
543 remove_entry(fid, path, 'file')
546 if path in self.base_files:
547 update_entry(fid, path, 'file', f['mode'])
549 add_entry(fid, path, 'file', f['mode'])
551 self.files[path][1] = f['mark']
552 self.rev_files[fid][1] = f['mark']
556 def get_content(self, file_id):
557 path, mark = self.rev_files[file_id]
559 return blob_marks[mark]
562 tree = self.branch.repository.revision_tree(self.base_id)
563 return tree.get_file_text(file_id)
565 def get_file_with_stat(self, file_id, path=None):
566 content = self.get_content(file_id)
567 return (StringIO.StringIO(content), None)
569 def get_symlink_target(self, file_id):
570 return self.get_content(file_id)
572 def id2path(self, file_id):
573 path, mark = self.rev_files[file_id]
576 def c_style_unescape(string):
577 if string[0] == string[-1] == '"':
578 return string.decode('string-escape')[1:-1]
581 def parse_commit(parser):
582 global marks, blob_marks, parsed_refs
590 if ref.startswith('refs/heads/'):
591 name = ref[len('refs/heads/'):]
592 branch = bzrlib.branch.Branch.open(branches[name])
596 commit_mark = parser.get_mark()
598 author = parser.get_author()
600 committer = parser.get_author()
602 data = parser.get_data()
604 if parser.check('from'):
605 parents.append(parser.get_mark())
607 while parser.check('merge'):
608 parents.append(parser.get_mark())
611 # fast-export adds an extra newline
618 if parser.check('M'):
619 t, m, mark_ref, path = line.split(' ', 3)
620 mark = int(mark_ref[1:])
621 f = { 'mode' : m, 'mark' : mark }
622 elif parser.check('D'):
623 t, path = line.split(' ')
624 f = { 'deleted' : True }
626 die('Unknown file command: %s' % line)
627 path = c_style_unescape(path).decode('utf-8')
630 committer, date, tz = committer
631 parents = [mark_to_rev(p) for p in parents]
632 revid = bzrlib.generate_ids.gen_revision_id(committer, date)
634 props['branch-nick'] = branch.nick
636 mtree = CustomTree(branch, revid, parents, files)
637 changes = mtree.iter_changes()
641 builder = branch.get_commit_builder(parents, None, date, tz, committer, props, revid)
643 list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
644 builder.finish_inventory()
645 builder.commit(data.decode('utf-8', 'replace'))
652 parsed_refs[ref] = revid
653 marks.new_mark(revid, commit_mark)
655 def parse_reset(parser):
662 if parser.check('commit'):
665 if not parser.check('from'):
667 from_mark = parser.get_mark()
670 parsed_refs[ref] = mark_to_rev(from_mark)
672 def do_export(parser):
673 global parsed_refs, dirname
677 for line in parser.each_block('done'):
678 if parser.check('blob'):
680 elif parser.check('commit'):
682 elif parser.check('reset'):
684 elif parser.check('tag'):
686 elif parser.check('feature'):
689 die('unhandled export command: %s' % line)
691 for ref, revid in parsed_refs.iteritems():
692 if ref.startswith('refs/heads/'):
693 name = ref[len('refs/heads/'):]
694 branch = bzrlib.branch.Branch.open(branches[name])
695 branch.generate_revision_history(revid, marks.get_tip(name))
698 peer = bzrlib.branch.Branch.open(peers[name])
700 peer.bzrdir.push_branch(branch, revision_id=revid)
701 except bzrlib.errors.DivergedBranches:
702 print "error %s non-fast forward" % ref
706 wt = branch.bzrdir.open_workingtree()
708 except bzrlib.errors.NoWorkingTree:
710 elif ref.startswith('refs/tags/'):
711 # TODO: implement tag push
712 print "error %s pushing tags not supported" % ref
715 # transport-helper/fast-export bugs
722 def do_capabilities(parser):
727 print "refspec refs/heads/*:%s/heads/*" % prefix
728 print "refspec refs/tags/*:%s/tags/*" % prefix
730 path = os.path.join(dirname, 'marks-git')
732 if os.path.exists(path):
733 print "*import-marks %s" % path
734 print "*export-marks %s" % path
738 def ref_is_valid(name):
739 return not True in [c in name for c in '~^: \\']
746 for name in branches:
747 if not master_branch:
749 print "? refs/heads/%s" % name
751 branch = bzrlib.branch.Branch.open(branches[master_branch])
753 for tag, revid in branch.tags.get_tag_dict().items():
755 branch.revision_id_to_dotted_revno(revid)
756 except bzrlib.errors.NoSuchRevision:
758 if not ref_is_valid(tag):
760 print "? refs/tags/%s" % tag
764 print "@refs/heads/%s HEAD" % master_branch
767 def get_remote_branch(origin, remote_branch, name):
768 global dirname, peers
770 branch_path = os.path.join(dirname, 'clone', name)
771 if os.path.exists(branch_path):
773 d = bzrlib.bzrdir.BzrDir.open(branch_path)
774 branch = d.open_branch()
776 branch.pull(remote_branch, [], None, False)
777 except bzrlib.errors.DivergedBranches:
778 # use remote branch for now
782 d = origin.sprout(branch_path, None,
783 hardlink=True, create_tree_if_local=False,
784 force_new_repo=False,
785 source_branch=remote_branch)
786 branch = d.open_branch()
790 def find_branches(repo, wanted):
791 transport = repo.user_transport
793 for fn in transport.iter_files_recursive():
794 if not fn.endswith('.bzr/branch-format'):
797 name = subdir = fn[:-len('/.bzr/branch-format')]
798 name = name if name != '' else 'master'
799 name = name.replace('/', '+')
801 if wanted and not name in wanted:
805 cur = transport.clone(subdir)
806 branch = bzrlib.branch.Branch.open_from_transport(cur)
807 except bzrlib.errors.NotBranchError:
812 def get_repo(url, alias):
813 global dirname, peer, branches
815 normal_url = bzrlib.urlutils.normalize_url(url)
816 origin = bzrlib.bzrdir.BzrDir.open(url)
817 is_local = isinstance(origin.transport, bzrlib.transport.local.LocalTransport)
819 shared_path = os.path.join(gitdir, 'bzr')
821 shared_dir = bzrlib.bzrdir.BzrDir.open(shared_path)
822 except bzrlib.errors.NotBranchError:
823 shared_dir = bzrlib.bzrdir.BzrDir.create(shared_path)
825 shared_repo = shared_dir.open_repository()
826 except bzrlib.errors.NoRepositoryPresent:
827 shared_repo = shared_dir.create_repository(shared=True)
830 clone_path = os.path.join(dirname, 'clone')
831 if not os.path.exists(clone_path):
834 # check and remove old organization
836 bdir = bzrlib.bzrdir.BzrDir.open(clone_path)
837 bdir.destroy_repository()
838 except bzrlib.errors.NotBranchError:
842 repo = origin.open_repository()
843 except bzrlib.errors.NoRepositoryPresent:
847 remote_branch = origin.open_branch()
850 peers[name] = remote_branch.base
851 branch = get_remote_branch(origin, remote_branch, name)
853 branch = remote_branch
855 branches[name] = branch.base
857 return branch.repository
861 wanted = get_config('remote-bzr.branches').rstrip().split(', ')
863 wanted = [e for e in wanted if e]
865 for name, remote_branch in find_branches(repo, wanted):
868 peers[name] = remote_branch.base
869 branch = get_remote_branch(origin, remote_branch, name)
871 branch = remote_branch
873 branches[name] = branch.base
877 def fix_path(alias, orig_url):
878 url = urlparse.urlparse(orig_url, 'file')
879 if url.scheme != 'file' or os.path.isabs(url.path):
881 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
882 cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
886 global marks, prefix, gitdir, dirname
887 global tags, filenodes
892 global branches, peers
908 alias = hashlib.sha1(alias).hexdigest()
912 prefix = 'refs/bzr/%s' % alias
913 gitdir = os.environ['GIT_DIR']
914 dirname = os.path.join(gitdir, 'bzr', alias)
919 if not os.path.exists(dirname):
922 bzrlib.ui.ui_factory.be_quiet(True)
924 repo = get_repo(url, alias)
926 marks_path = os.path.join(dirname, 'marks-int')
927 marks = Marks(marks_path)
929 parser = Parser(repo)
931 if parser.check('capabilities'):
932 do_capabilities(parser)
933 elif parser.check('list'):
935 elif parser.check('import'):
937 elif parser.check('export'):
940 die('unhandled command: %s' % line)
949 shutil.rmtree(dirname)
952 sys.exit(main(sys.argv))