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 to track (per repo):
17 # % git config remote.origin.bzr-branches 'trunk, devel, test'
19 # Where 'origin' is the name of the repository you want to specify the
26 if hasattr(bzrlib, "initialize"):
30 bzrlib.plugin.load_plugins()
32 import bzrlib.generate_ids
33 import bzrlib.transport
36 import bzrlib.urlutils
44 import atexit, shutil, hashlib, urlparse, subprocess
46 NAME_RE = re.compile('^([^<>]+)')
47 AUTHOR_RE = re.compile('^([^<>]+?)? ?[<>]([^<>]*)(?:$|>)')
48 EMAIL_RE = re.compile(r'([^ \t<>]+@[^ \t<>]+)')
49 RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
52 sys.stderr.write('ERROR: %s\n' % (msg % args))
56 sys.stderr.write('WARNING: %s\n' % (msg % args))
59 return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
61 def get_config(config):
62 cmd = ['git', 'config', '--get', config]
63 process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
64 output, _ = process.communicate()
69 def __init__(self, path):
78 if not os.path.exists(self.path):
81 tmp = json.load(open(self.path))
82 self.tips = tmp['tips']
83 self.marks = tmp['marks']
84 self.last_mark = tmp['last-mark']
86 for rev, mark in self.marks.iteritems():
87 self.rev_marks[mark] = rev
90 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
93 json.dump(self.dict(), open(self.path, 'w'))
96 return str(self.dict())
98 def from_rev(self, rev):
99 return self.marks[rev]
101 def to_rev(self, mark):
102 return str(self.rev_marks[mark])
106 return self.last_mark
108 def get_mark(self, rev):
110 self.marks[rev] = self.last_mark
111 return self.last_mark
113 def is_marked(self, rev):
114 return rev in self.marks
116 def new_mark(self, rev, mark):
117 self.marks[rev] = mark
118 self.rev_marks[mark] = rev
119 self.last_mark = mark
121 def get_tip(self, branch):
123 return str(self.tips[branch])
127 def set_tip(self, branch, tip):
128 self.tips[branch] = tip
132 def __init__(self, repo):
134 self.line = self.get_line()
137 return sys.stdin.readline().strip()
139 def __getitem__(self, i):
140 return self.line.split()[i]
142 def check(self, word):
143 return self.line.startswith(word)
145 def each_block(self, separator):
146 while self.line != separator:
148 self.line = self.get_line()
151 return self.each_block('')
154 self.line = self.get_line()
155 if self.line == 'done':
159 i = self.line.index(':') + 1
160 return int(self.line[i:])
163 if not self.check('data'):
165 i = self.line.index(' ') + 1
166 size = int(self.line[i:])
167 return sys.stdin.read(size)
169 def get_author(self):
170 m = RAW_AUTHOR_RE.match(self.line)
173 _, name, email, date, tz = m.groups()
174 name = name.decode('utf-8')
175 committer = '%s <%s>' % (name, email)
177 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
178 return (committer, int(date), tz)
180 def rev_to_mark(rev):
181 return marks.from_rev(rev)
183 def mark_to_rev(mark):
184 return marks.to_rev(mark)
186 def fixup_user(user):
188 user = user.replace('"', '')
189 m = AUTHOR_RE.match(user)
192 mail = m.group(2).strip()
194 m = EMAIL_RE.match(user)
198 m = NAME_RE.match(user)
200 name = m.group(1).strip()
207 return '%s <%s>' % (name, mail)
209 def get_filechanges(cur, prev):
213 changes = cur.changes_from(prev)
216 return s.encode('utf-8')
218 for path, fid, kind in changes.added:
219 modified[u(path)] = fid
220 for path, fid, kind in changes.removed:
221 removed[u(path)] = None
222 for path, fid, kind, mod, _ in changes.modified:
223 modified[u(path)] = fid
224 for oldpath, newpath, fid, kind, mod, _ in changes.renamed:
225 removed[u(oldpath)] = None
226 if kind == 'directory':
227 lst = cur.list_files(from_dir=newpath, recursive=True)
228 for path, file_class, kind, fid, entry in lst:
229 if kind != 'directory':
230 modified[u(newpath + '/' + path)] = fid
232 modified[u(newpath)] = fid
234 return modified, removed
236 def export_files(tree, files):
238 for path, fid in files.iteritems():
239 kind = tree.kind(fid)
241 h = tree.get_file_sha1(fid)
243 if kind == 'symlink':
244 d = tree.get_symlink_target(fid)
248 if tree.is_executable(fid):
253 # is the blob already exported?
256 final.append((mode, mark, path))
259 d = tree.get_file_text(fid)
260 elif kind == 'directory':
263 die("Unhandled kind '%s' for path '%s'" % (kind, path))
265 mark = marks.next_mark()
269 print "mark :%u" % mark
270 print "data %d" % len(d)
273 final.append((mode, mark, path))
277 def export_branch(repo, name):
278 ref = '%s/heads/%s' % (prefix, name)
279 tip = marks.get_tip(name)
281 branch = get_remote_branch(name)
282 repo = branch.repository
285 revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
287 tip_revno = branch.revision_id_to_revno(tip)
288 last_revno, _ = branch.last_revision_info()
289 total = last_revno - tip_revno
290 except bzrlib.errors.NoSuchRevision:
294 for revid, _, seq, _ in revs:
296 if marks.is_marked(revid):
299 rev = repo.get_revision(revid)
302 parents = rev.parent_ids
305 committer = rev.committer.encode('utf-8')
306 committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
307 authors = rev.get_apparent_authors()
309 author = authors[0].encode('utf-8')
310 author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
313 msg = rev.message.encode('utf-8')
317 if len(parents) == 0:
318 parent = bzrlib.revision.NULL_REVISION
322 cur_tree = repo.revision_tree(revid)
323 prev = repo.revision_tree(parent)
324 modified, removed = get_filechanges(cur_tree, prev)
326 modified_final = export_files(cur_tree, modified)
328 if len(parents) == 0:
329 print 'reset %s' % ref
331 print "commit %s" % ref
332 print "mark :%d" % (marks.get_mark(revid))
333 print "author %s" % (author)
334 print "committer %s" % (committer)
335 print "data %d" % (len(msg))
338 for i, p in enumerate(parents):
347 print "merge :%s" % m
351 for f in modified_final:
352 print "M %s :%u %s" % f
356 # let's skip branch revisions from the progress report
359 progress = (revno - tip_revno)
360 if (progress % 100 == 0):
362 print "progress revision %d '%s' (%d/%d)" % (revno, name, progress, total)
364 print "progress revision %d '%s' (%d)" % (revno, name, progress)
368 revid = branch.last_revision()
370 # make sure the ref is updated
371 print "reset %s" % ref
372 print "from :%u" % rev_to_mark(revid)
375 marks.set_tip(name, revid)
377 def export_tag(repo, name):
378 ref = '%s/tags/%s' % (prefix, name)
379 print "reset %s" % ref
380 print "from :%u" % rev_to_mark(tags[name])
383 def do_import(parser):
385 path = os.path.join(dirname, 'marks-git')
388 if os.path.exists(path):
389 print "feature import-marks=%s" % path
390 print "feature export-marks=%s" % path
391 print "feature force"
394 while parser.check('import'):
396 if ref.startswith('refs/heads/'):
397 name = ref[len('refs/heads/'):]
398 export_branch(repo, name)
399 if ref.startswith('refs/tags/'):
400 name = ref[len('refs/tags/'):]
401 export_tag(repo, name)
408 def parse_blob(parser):
410 mark = parser.get_mark()
412 data = parser.get_data()
413 blob_marks[mark] = data
418 def __init__(self, branch, revid, parents, files):
422 def copy_tree(revid):
423 files = files_cache[revid] = {}
425 tree = branch.repository.revision_tree(revid)
427 for path, entry in tree.iter_entries_by_dir():
428 files[path] = [entry.file_id, None]
433 if len(parents) == 0:
434 self.base_id = bzrlib.revision.NULL_REVISION
437 self.base_id = parents[0]
438 self.base_files = files_cache.get(self.base_id, None)
439 if not self.base_files:
440 self.base_files = copy_tree(self.base_id)
442 self.files = files_cache[revid] = self.base_files.copy()
445 for path, data in self.files.iteritems():
447 self.rev_files[fid] = [path, mark]
449 for path, f in files.iteritems():
450 fid, mark = self.files.get(path, [None, None])
452 fid = bzrlib.generate_ids.gen_file_id(path)
454 self.rev_files[fid] = [path, mark]
455 self.updates[fid] = f
457 def last_revision(self):
460 def iter_changes(self):
463 def get_parent(dirname, basename):
464 parent_fid, mark = self.base_files.get(dirname, [None, None])
467 parent_fid, mark = self.files.get(dirname, [None, None])
472 fid = bzrlib.generate_ids.gen_file_id(path)
473 add_entry(fid, dirname, 'directory')
476 def add_entry(fid, path, kind, mode=None):
477 dirname, basename = os.path.split(path)
478 parent_fid = get_parent(dirname, basename)
483 elif mode == '120000':
494 self.files[path] = [change[0], None]
495 changes.append(change)
497 def update_entry(fid, path, kind, mode=None):
498 dirname, basename = os.path.split(path)
499 parent_fid = get_parent(dirname, basename)
504 elif mode == '120000':
515 self.files[path] = [change[0], None]
516 changes.append(change)
518 def remove_entry(fid, path, kind):
519 dirname, basename = os.path.split(path)
520 parent_fid = get_parent(dirname, basename)
530 changes.append(change)
532 for fid, f in self.updates.iteritems():
536 remove_entry(fid, path, 'file')
539 if path in self.base_files:
540 update_entry(fid, path, 'file', f['mode'])
542 add_entry(fid, path, 'file', f['mode'])
544 self.files[path][1] = f['mark']
545 self.rev_files[fid][1] = f['mark']
549 def get_content(self, file_id):
550 path, mark = self.rev_files[file_id]
552 return blob_marks[mark]
555 tree = self.branch.repository.revision_tree(self.base_id)
556 return tree.get_file_text(file_id)
558 def get_file_with_stat(self, file_id, path=None):
559 content = self.get_content(file_id)
560 return (StringIO.StringIO(content), None)
562 def get_symlink_target(self, file_id):
563 return self.get_content(file_id)
565 def id2path(self, file_id):
566 path, mark = self.rev_files[file_id]
569 def c_style_unescape(string):
570 if string[0] == string[-1] == '"':
571 return string.decode('string-escape')[1:-1]
574 def parse_commit(parser):
580 if ref.startswith('refs/heads/'):
581 name = ref[len('refs/heads/'):]
582 branch = get_remote_branch(name)
586 commit_mark = parser.get_mark()
588 author = parser.get_author()
590 committer = parser.get_author()
592 data = parser.get_data()
594 if parser.check('from'):
595 parents.append(parser.get_mark())
597 while parser.check('merge'):
598 parents.append(parser.get_mark())
601 # fast-export adds an extra newline
608 if parser.check('M'):
609 t, m, mark_ref, path = line.split(' ', 3)
610 mark = int(mark_ref[1:])
611 f = { 'mode' : m, 'mark' : mark }
612 elif parser.check('D'):
613 t, path = line.split(' ', 1)
614 f = { 'deleted' : True }
616 die('Unknown file command: %s' % line)
617 path = c_style_unescape(path).decode('utf-8')
620 committer, date, tz = committer
621 parents = [mark_to_rev(p) for p in parents]
622 revid = bzrlib.generate_ids.gen_revision_id(committer, date)
624 props['branch-nick'] = branch.nick
626 mtree = CustomTree(branch, revid, parents, files)
627 changes = mtree.iter_changes()
631 builder = branch.get_commit_builder(parents, None, date, tz, committer, props, revid)
633 list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
634 builder.finish_inventory()
635 builder.commit(data.decode('utf-8', 'replace'))
642 parsed_refs[ref] = revid
643 marks.new_mark(revid, commit_mark)
645 def parse_reset(parser):
650 if parser.check('commit'):
653 if not parser.check('from'):
655 from_mark = parser.get_mark()
658 parsed_refs[ref] = mark_to_rev(from_mark)
660 def do_export(parser):
663 for line in parser.each_block('done'):
664 if parser.check('blob'):
666 elif parser.check('commit'):
668 elif parser.check('reset'):
670 elif parser.check('tag'):
672 elif parser.check('feature'):
675 die('unhandled export command: %s' % line)
677 for ref, revid in parsed_refs.iteritems():
678 if ref.startswith('refs/heads/'):
679 name = ref[len('refs/heads/'):]
680 branch = get_remote_branch(name)
681 branch.generate_revision_history(revid, marks.get_tip(name))
684 peer = bzrlib.branch.Branch.open(peers[name],
685 possible_transports=transports)
687 peer.bzrdir.push_branch(branch, revision_id=revid)
688 except bzrlib.errors.DivergedBranches:
689 print "error %s non-fast forward" % ref
693 wt = branch.bzrdir.open_workingtree()
695 except bzrlib.errors.NoWorkingTree:
697 elif ref.startswith('refs/tags/'):
698 # TODO: implement tag push
699 print "error %s pushing tags not supported" % ref
702 # transport-helper/fast-export bugs
709 def do_capabilities(parser):
712 print "refspec refs/heads/*:%s/heads/*" % prefix
713 print "refspec refs/tags/*:%s/tags/*" % prefix
715 path = os.path.join(dirname, 'marks-git')
717 if os.path.exists(path):
718 print "*import-marks %s" % path
719 print "*export-marks %s" % path
723 def ref_is_valid(name):
724 return not True in [c in name for c in '~^: \\']
729 for name in branches:
730 if not master_branch:
732 print "? refs/heads/%s" % name
734 branch = get_remote_branch(master_branch)
736 for tag, revid in branch.tags.get_tag_dict().items():
738 branch.revision_id_to_dotted_revno(revid)
739 except bzrlib.errors.NoSuchRevision:
741 if not ref_is_valid(tag):
743 print "? refs/tags/%s" % tag
747 print "@refs/heads/%s HEAD" % master_branch
750 def clone(path, remote_branch):
752 bdir = bzrlib.bzrdir.BzrDir.create(path, possible_transports=transports)
753 except bzrlib.errors.AlreadyControlDirError:
754 bdir = bzrlib.bzrdir.BzrDir.open(path, possible_transports=transports)
755 repo = bdir.find_repository()
756 repo.fetch(remote_branch.repository)
757 return remote_branch.sprout(bdir, repository=repo)
759 def get_remote_branch(name):
760 remote_branch = bzrlib.branch.Branch.open(branches[name],
761 possible_transports=transports)
762 if isinstance(remote_branch.user_transport, bzrlib.transport.local.LocalTransport):
765 branch_path = os.path.join(dirname, 'clone', name)
768 branch = bzrlib.branch.Branch.open(branch_path,
769 possible_transports=transports)
770 except bzrlib.errors.NotBranchError:
772 branch = clone(branch_path, remote_branch)
776 branch.pull(remote_branch, overwrite=True)
777 except bzrlib.errors.DivergedBranches:
778 # use remote branch for now
783 def find_branches(repo):
784 transport = repo.bzrdir.root_transport
786 for fn in transport.iter_files_recursive():
787 if not fn.endswith('.bzr/branch-format'):
790 name = subdir = fn[:-len('/.bzr/branch-format')]
791 name = name if name != '' else 'master'
792 name = name.replace('/', '+')
795 cur = transport.clone(subdir)
796 branch = bzrlib.branch.Branch.open_from_transport(cur)
797 except bzrlib.errors.NotBranchError:
800 yield name, branch.base
802 def get_repo(url, alias):
803 normal_url = bzrlib.urlutils.normalize_url(url)
804 origin = bzrlib.bzrdir.BzrDir.open(url, possible_transports=transports)
805 is_local = isinstance(origin.transport, bzrlib.transport.local.LocalTransport)
807 shared_path = os.path.join(gitdir, 'bzr')
809 shared_dir = bzrlib.bzrdir.BzrDir.open(shared_path,
810 possible_transports=transports)
811 except bzrlib.errors.NotBranchError:
812 shared_dir = bzrlib.bzrdir.BzrDir.create(shared_path,
813 possible_transports=transports)
815 shared_repo = shared_dir.open_repository()
816 except bzrlib.errors.NoRepositoryPresent:
817 shared_repo = shared_dir.create_repository(shared=True)
820 clone_path = os.path.join(dirname, 'clone')
821 if not os.path.exists(clone_path):
824 # check and remove old organization
826 bdir = bzrlib.bzrdir.BzrDir.open(clone_path,
827 possible_transports=transports)
828 bdir.destroy_repository()
829 except bzrlib.errors.NotBranchError:
831 except bzrlib.errors.NoRepositoryPresent:
834 wanted = get_config('remote.%s.bzr-branches' % alias).rstrip().split(', ')
836 wanted = [e for e in wanted if e]
838 wanted = get_config('remote-bzr.branches').rstrip().split(', ')
840 wanted = [e for e in wanted if e]
844 repo = origin.open_repository()
845 if not repo.user_transport.listable():
846 # this repository is not usable for us
847 raise bzrlib.errors.NoRepositoryPresent(repo.bzrdir)
848 except bzrlib.errors.NoRepositoryPresent:
852 def list_wanted(url, wanted):
854 subdir = name if name != 'master' else ''
855 yield name, bzrlib.urlutils.join(url, subdir)
857 branch_list = list_wanted(url, wanted)
859 branch_list = find_branches(repo)
861 for name, url in branch_list:
868 def fix_path(alias, orig_url):
869 url = urlparse.urlparse(orig_url, 'file')
870 if url.scheme != 'file' or os.path.isabs(url.path):
872 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
873 cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
877 global marks, prefix, gitdir, dirname
878 global tags, filenodes
883 global branches, peers
888 gitdir = os.environ.get('GIT_DIR', None)
891 die('Not enough arguments.')
894 die('GIT_DIR not set')
910 alias = hashlib.sha1(alias).hexdigest()
912 prefix = 'refs/bzr/%s' % alias
913 dirname = os.path.join(gitdir, 'bzr', alias)
918 if not os.path.exists(dirname):
921 if hasattr(bzrlib.ui.ui_factory, 'be_quiet'):
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))