1#!/usr/bin/env python
2#
3# Copyright (c) 2012 Felipe Contreras
4#
5
6#
7# Just copy to your ~/bin, or anywhere in your $PATH.
8# Then you can clone with:
9# % git clone bzr::/path/to/bzr/repo/or/url
10#
11# For example:
12# % git clone bzr::$HOME/myrepo
13# or
14# % git clone bzr::lp:myrepo
15#
16
17import sys
18
19import bzrlib
20if hasattr(bzrlib, "initialize"):
21 bzrlib.initialize()
22
23import bzrlib.plugin
24bzrlib.plugin.load_plugins()
25
26import bzrlib.generate_ids
27import bzrlib.transport
28import bzrlib.errors
29import bzrlib.ui
30
31import sys
32import os
33import json
34import re
35import StringIO
36import atexit, shutil, hashlib, urlparse, subprocess
37
38NAME_RE = re.compile('^([^<>]+)')
39AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
40RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
41
42def die(msg, *args):
43 sys.stderr.write('ERROR: %s\n' % (msg % args))
44 sys.exit(1)
45
46def warn(msg, *args):
47 sys.stderr.write('WARNING: %s\n' % (msg % args))
48
49def gittz(tz):
50 return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
51
52class Marks:
53
54 def __init__(self, path):
55 self.path = path
56 self.tips = {}
57 self.marks = {}
58 self.rev_marks = {}
59 self.last_mark = 0
60 self.load()
61
62 def load(self):
63 if not os.path.exists(self.path):
64 return
65
66 tmp = json.load(open(self.path))
67 self.tips = tmp['tips']
68 self.marks = tmp['marks']
69 self.last_mark = tmp['last-mark']
70
71 for rev, mark in self.marks.iteritems():
72 self.rev_marks[mark] = rev
73
74 def dict(self):
75 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
76
77 def store(self):
78 json.dump(self.dict(), open(self.path, 'w'))
79
80 def __str__(self):
81 return str(self.dict())
82
83 def from_rev(self, rev):
84 return self.marks[rev]
85
86 def to_rev(self, mark):
87 return self.rev_marks[mark]
88
89 def next_mark(self):
90 self.last_mark += 1
91 return self.last_mark
92
93 def get_mark(self, rev):
94 self.last_mark += 1
95 self.marks[rev] = self.last_mark
96 return self.last_mark
97
98 def is_marked(self, rev):
99 return rev in self.marks
100
101 def new_mark(self, rev, mark):
102 self.marks[rev] = mark
103 self.rev_marks[mark] = rev
104 self.last_mark = mark
105
106 def get_tip(self, branch):
107 return self.tips.get(branch, None)
108
109 def set_tip(self, branch, tip):
110 self.tips[branch] = tip
111
112class Parser:
113
114 def __init__(self, repo):
115 self.repo = repo
116 self.line = self.get_line()
117
118 def get_line(self):
119 return sys.stdin.readline().strip()
120
121 def __getitem__(self, i):
122 return self.line.split()[i]
123
124 def check(self, word):
125 return self.line.startswith(word)
126
127 def each_block(self, separator):
128 while self.line != separator:
129 yield self.line
130 self.line = self.get_line()
131
132 def __iter__(self):
133 return self.each_block('')
134
135 def next(self):
136 self.line = self.get_line()
137 if self.line == 'done':
138 self.line = None
139
140 def get_mark(self):
141 i = self.line.index(':') + 1
142 return int(self.line[i:])
143
144 def get_data(self):
145 if not self.check('data'):
146 return None
147 i = self.line.index(' ') + 1
148 size = int(self.line[i:])
149 return sys.stdin.read(size)
150
151 def get_author(self):
152 m = RAW_AUTHOR_RE.match(self.line)
153 if not m:
154 return None
155 _, name, email, date, tz = m.groups()
156 committer = '%s <%s>' % (name, email)
157 tz = int(tz)
158 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
159 return (committer, int(date), tz)
160
161def rev_to_mark(rev):
162 global marks
163 return marks.from_rev(rev)
164
165def mark_to_rev(mark):
166 global marks
167 return marks.to_rev(mark)
168
169def fixup_user(user):
170 name = mail = None
171 user = user.replace('"', '')
172 m = AUTHOR_RE.match(user)
173 if m:
174 name = m.group(1)
175 mail = m.group(2).strip()
176 else:
177 m = NAME_RE.match(user)
178 if m:
179 name = m.group(1).strip()
180
181 return '%s <%s>' % (name, mail)
182
183def get_filechanges(cur, prev):
184 modified = {}
185 removed = {}
186
187 changes = cur.changes_from(prev)
188
189 def u(s):
190 return s.encode('utf-8')
191
192 for path, fid, kind in changes.added:
193 modified[u(path)] = fid
194 for path, fid, kind in changes.removed:
195 removed[u(path)] = None
196 for path, fid, kind, mod, _ in changes.modified:
197 modified[u(path)] = fid
198 for oldpath, newpath, fid, kind, mod, _ in changes.renamed:
199 removed[u(oldpath)] = None
200 if kind == 'directory':
201 lst = cur.list_files(from_dir=newpath, recursive=True)
202 for path, file_class, kind, fid, entry in lst:
203 if kind != 'directory':
204 modified[u(newpath + '/' + path)] = fid
205 else:
206 modified[u(newpath)] = fid
207
208 return modified, removed
209
210def export_files(tree, files):
211 global marks, filenodes
212
213 final = []
214 for path, fid in files.iteritems():
215 kind = tree.kind(fid)
216
217 h = tree.get_file_sha1(fid)
218
219 if kind == 'symlink':
220 d = tree.get_symlink_target(fid)
221 mode = '120000'
222 elif kind == 'file':
223
224 if tree.is_executable(fid):
225 mode = '100755'
226 else:
227 mode = '100644'
228
229 # is the blob already exported?
230 if h in filenodes:
231 mark = filenodes[h]
232 final.append((mode, mark, path))
233 continue
234
235 d = tree.get_file_text(fid)
236 elif kind == 'directory':
237 continue
238 else:
239 die("Unhandled kind '%s' for path '%s'" % (kind, path))
240
241 mark = marks.next_mark()
242 filenodes[h] = mark
243
244 print "blob"
245 print "mark :%u" % mark
246 print "data %d" % len(d)
247 print d
248
249 final.append((mode, mark, path))
250
251 return final
252
253def export_branch(branch, name):
254 global prefix
255
256 ref = '%s/heads/%s' % (prefix, name)
257 tip = marks.get_tip(name)
258
259 repo = branch.repository
260
261 branch.lock_read()
262 revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
263 count = 0
264
265 revs = [revid for revid, _, _, _ in revs if not marks.is_marked(revid)]
266
267 for revid in revs:
268
269 rev = repo.get_revision(revid)
270
271 parents = rev.parent_ids
272 time = rev.timestamp
273 tz = rev.timezone
274 committer = rev.committer.encode('utf-8')
275 committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
276 authors = rev.get_apparent_authors()
277 if authors:
278 author = authors[0].encode('utf-8')
279 author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
280 else:
281 author = committer
282 msg = rev.message.encode('utf-8')
283
284 msg += '\n'
285
286 if len(parents) == 0:
287 parent = bzrlib.revision.NULL_REVISION
288 else:
289 parent = parents[0]
290
291 cur_tree = repo.revision_tree(revid)
292 prev = repo.revision_tree(parent)
293 modified, removed = get_filechanges(cur_tree, prev)
294
295 modified_final = export_files(cur_tree, modified)
296
297 if len(parents) == 0:
298 print 'reset %s' % ref
299
300 print "commit %s" % ref
301 print "mark :%d" % (marks.get_mark(revid))
302 print "author %s" % (author)
303 print "committer %s" % (committer)
304 print "data %d" % (len(msg))
305 print msg
306
307 for i, p in enumerate(parents):
308 try:
309 m = rev_to_mark(p)
310 except KeyError:
311 # ghost?
312 continue
313 if i == 0:
314 print "from :%s" % m
315 else:
316 print "merge :%s" % m
317
318 for f in removed:
319 print "D %s" % (f,)
320 for f in modified_final:
321 print "M %s :%u %s" % f
322 print
323
324 count += 1
325 if (count % 100 == 0):
326 print "progress revision %s (%d/%d)" % (revid, count, len(revs))
327 print "#############################################################"
328
329 branch.unlock()
330
331 revid = branch.last_revision()
332
333 # make sure the ref is updated
334 print "reset %s" % ref
335 print "from :%u" % rev_to_mark(revid)
336 print
337
338 marks.set_tip(name, revid)
339
340def export_tag(repo, name):
341 global tags, prefix
342
343 ref = '%s/tags/%s' % (prefix, name)
344 print "reset %s" % ref
345 print "from :%u" % rev_to_mark(tags[name])
346 print
347
348def do_import(parser):
349 global dirname
350
351 branch = parser.repo
352 path = os.path.join(dirname, 'marks-git')
353
354 print "feature done"
355 if os.path.exists(path):
356 print "feature import-marks=%s" % path
357 print "feature export-marks=%s" % path
358 print "feature force"
359 sys.stdout.flush()
360
361 while parser.check('import'):
362 ref = parser[1]
363 if ref.startswith('refs/heads/'):
364 name = ref[len('refs/heads/'):]
365 export_branch(branch, name)
366 if ref.startswith('refs/tags/'):
367 name = ref[len('refs/tags/'):]
368 export_tag(branch, name)
369 parser.next()
370
371 print 'done'
372
373 sys.stdout.flush()
374
375def parse_blob(parser):
376 global blob_marks
377
378 parser.next()
379 mark = parser.get_mark()
380 parser.next()
381 data = parser.get_data()
382 blob_marks[mark] = data
383 parser.next()
384
385class CustomTree():
386
387 def __init__(self, branch, revid, parents, files):
388 global files_cache
389
390 self.updates = {}
391 self.branch = branch
392
393 def copy_tree(revid):
394 files = files_cache[revid] = {}
395 branch.lock_read()
396 tree = branch.repository.revision_tree(revid)
397 try:
398 for path, entry in tree.iter_entries_by_dir():
399 files[path] = [entry.file_id, None]
400 finally:
401 branch.unlock()
402 return files
403
404 if len(parents) == 0:
405 self.base_id = bzrlib.revision.NULL_REVISION
406 self.base_files = {}
407 else:
408 self.base_id = parents[0]
409 self.base_files = files_cache.get(self.base_id, None)
410 if not self.base_files:
411 self.base_files = copy_tree(self.base_id)
412
413 self.files = files_cache[revid] = self.base_files.copy()
414 self.rev_files = {}
415
416 for path, data in self.files.iteritems():
417 fid, mark = data
418 self.rev_files[fid] = [path, mark]
419
420 for path, f in files.iteritems():
421 fid, mark = self.files.get(path, [None, None])
422 if not fid:
423 fid = bzrlib.generate_ids.gen_file_id(path)
424 f['path'] = path
425 self.rev_files[fid] = [path, mark]
426 self.updates[fid] = f
427
428 def last_revision(self):
429 return self.base_id
430
431 def iter_changes(self):
432 changes = []
433
434 def get_parent(dirname, basename):
435 parent_fid, mark = self.base_files.get(dirname, [None, None])
436 if parent_fid:
437 return parent_fid
438 parent_fid, mark = self.files.get(dirname, [None, None])
439 if parent_fid:
440 return parent_fid
441 if basename == '':
442 return None
443 fid = bzrlib.generate_ids.gen_file_id(path)
444 add_entry(fid, dirname, 'directory')
445 return fid
446
447 def add_entry(fid, path, kind, mode = None):
448 dirname, basename = os.path.split(path)
449 parent_fid = get_parent(dirname, basename)
450
451 executable = False
452 if mode == '100755':
453 executable = True
454 elif mode == '120000':
455 kind = 'symlink'
456
457 change = (fid,
458 (None, path),
459 True,
460 (False, True),
461 (None, parent_fid),
462 (None, basename),
463 (None, kind),
464 (None, executable))
465 self.files[path] = [change[0], None]
466 changes.append(change)
467
468 def update_entry(fid, path, kind, mode = None):
469 dirname, basename = os.path.split(path)
470 parent_fid = get_parent(dirname, basename)
471
472 executable = False
473 if mode == '100755':
474 executable = True
475 elif mode == '120000':
476 kind = 'symlink'
477
478 change = (fid,
479 (path, path),
480 True,
481 (True, True),
482 (None, parent_fid),
483 (None, basename),
484 (None, kind),
485 (None, executable))
486 self.files[path] = [change[0], None]
487 changes.append(change)
488
489 def remove_entry(fid, path, kind):
490 dirname, basename = os.path.split(path)
491 parent_fid = get_parent(dirname, basename)
492 change = (fid,
493 (path, None),
494 True,
495 (True, False),
496 (parent_fid, None),
497 (None, None),
498 (None, None),
499 (None, None))
500 del self.files[path]
501 changes.append(change)
502
503 for fid, f in self.updates.iteritems():
504 path = f['path']
505
506 if 'deleted' in f:
507 remove_entry(fid, path, 'file')
508 continue
509
510 if path in self.base_files:
511 update_entry(fid, path, 'file', f['mode'])
512 else:
513 add_entry(fid, path, 'file', f['mode'])
514
515 self.files[path][1] = f['mark']
516 self.rev_files[fid][1] = f['mark']
517
518 return changes
519
520 def get_content(self, file_id):
521 path, mark = self.rev_files[file_id]
522 if mark:
523 return blob_marks[mark]
524
525 # last resort
526 tree = self.branch.repository.revision_tree(self.base_id)
527 return tree.get_file_text(file_id)
528
529 def get_file_with_stat(self, file_id, path=None):
530 content = self.get_content(file_id)
531 return (StringIO.StringIO(content), None)
532
533 def get_symlink_target(self, file_id):
534 return self.get_content(file_id)
535
536 def id2path(self, file_id):
537 path, mark = self.rev_files[file_id]
538 return path
539
540def c_style_unescape(string):
541 if string[0] == string[-1] == '"':
542 return string.decode('string-escape')[1:-1]
543 return string
544
545def parse_commit(parser):
546 global marks, blob_marks, parsed_refs
547 global mode
548
549 parents = []
550
551 ref = parser[1]
552 parser.next()
553
554 if ref != 'refs/heads/master':
555 die("bzr doesn't support multiple branches; use 'master'")
556
557 commit_mark = parser.get_mark()
558 parser.next()
559 author = parser.get_author()
560 parser.next()
561 committer = parser.get_author()
562 parser.next()
563 data = parser.get_data()
564 parser.next()
565 if parser.check('from'):
566 parents.append(parser.get_mark())
567 parser.next()
568 while parser.check('merge'):
569 parents.append(parser.get_mark())
570 parser.next()
571
572 # fast-export adds an extra newline
573 if data[-1] == '\n':
574 data = data[:-1]
575
576 files = {}
577
578 for line in parser:
579 if parser.check('M'):
580 t, m, mark_ref, path = line.split(' ', 3)
581 mark = int(mark_ref[1:])
582 f = { 'mode' : m, 'mark' : mark }
583 elif parser.check('D'):
584 t, path = line.split(' ')
585 f = { 'deleted' : True }
586 else:
587 die('Unknown file command: %s' % line)
588 path = c_style_unescape(path).decode('utf-8')
589 files[path] = f
590
591 branch = parser.repo
592
593 committer, date, tz = committer
594 parents = [str(mark_to_rev(p)) for p in parents]
595 revid = bzrlib.generate_ids.gen_revision_id(committer, date)
596 props = {}
597 props['branch-nick'] = branch.nick
598
599 mtree = CustomTree(branch, revid, parents, files)
600 changes = mtree.iter_changes()
601
602 branch.lock_write()
603 try:
604 builder = branch.get_commit_builder(parents, None, date, tz, committer, props, revid)
605 try:
606 list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
607 builder.finish_inventory()
608 builder.commit(data.decode('utf-8', 'replace'))
609 except Exception, e:
610 builder.abort()
611 raise
612 finally:
613 branch.unlock()
614
615 parsed_refs[ref] = revid
616 marks.new_mark(revid, commit_mark)
617
618def parse_reset(parser):
619 global parsed_refs
620
621 ref = parser[1]
622 parser.next()
623
624 if ref != 'refs/heads/master':
625 die("bzr doesn't support multiple branches; use 'master'")
626
627 # ugh
628 if parser.check('commit'):
629 parse_commit(parser)
630 return
631 if not parser.check('from'):
632 return
633 from_mark = parser.get_mark()
634 parser.next()
635
636 parsed_refs[ref] = mark_to_rev(from_mark)
637
638def do_export(parser):
639 global parsed_refs, dirname, peer
640
641 parser.next()
642
643 for line in parser.each_block('done'):
644 if parser.check('blob'):
645 parse_blob(parser)
646 elif parser.check('commit'):
647 parse_commit(parser)
648 elif parser.check('reset'):
649 parse_reset(parser)
650 elif parser.check('tag'):
651 pass
652 elif parser.check('feature'):
653 pass
654 else:
655 die('unhandled export command: %s' % line)
656
657 branch = parser.repo
658
659 for ref, revid in parsed_refs.iteritems():
660 if ref == 'refs/heads/master':
661 branch.generate_revision_history(revid, marks.get_tip('master'))
662 if peer:
663 try:
664 branch.push(peer, stop_revision=revid)
665 except bzrlib.errors.DivergedBranches:
666 print "error %s non-fast forward" % ref
667 continue
668
669 try:
670 wt = branch.bzrdir.open_workingtree()
671 wt.update()
672 except bzrlib.errors.NoWorkingTree:
673 pass
674
675 print "ok %s" % ref
676
677 print
678
679def do_capabilities(parser):
680 global dirname
681
682 print "import"
683 print "export"
684 print "refspec refs/heads/*:%s/heads/*" % prefix
685 print "refspec refs/tags/*:%s/tags/*" % prefix
686
687 path = os.path.join(dirname, 'marks-git')
688
689 if os.path.exists(path):
690 print "*import-marks %s" % path
691 print "*export-marks %s" % path
692
693 print
694
695def ref_is_valid(name):
696 return not True in [c in name for c in '~^: \\']
697
698def do_list(parser):
699 global tags
700 print "? refs/heads/%s" % 'master'
701
702 branch = parser.repo
703 branch.lock_read()
704 for tag, revid in branch.tags.get_tag_dict().items():
705 try:
706 branch.revision_id_to_dotted_revno(revid)
707 except bzrlib.errors.NoSuchRevision:
708 continue
709 if not ref_is_valid(tag):
710 continue
711 print "? refs/tags/%s" % tag
712 tags[tag] = revid
713 branch.unlock()
714 print "@refs/heads/%s HEAD" % 'master'
715 print
716
717def get_repo(url, alias):
718 global dirname, peer
719
720 origin = bzrlib.bzrdir.BzrDir.open(url)
721 branch = origin.open_branch()
722
723 if not isinstance(origin.transport, bzrlib.transport.local.LocalTransport):
724 clone_path = os.path.join(dirname, 'clone')
725 remote_branch = branch
726 if os.path.exists(clone_path):
727 # pull
728 d = bzrlib.bzrdir.BzrDir.open(clone_path)
729 branch = d.open_branch()
730 try:
731 result = branch.pull(remote_branch, [], None, False)
732 except bzrlib.errors.DivergedBranches:
733 # use remote branch for now
734 peer = None
735 return remote_branch
736 else:
737 # clone
738 d = origin.sprout(clone_path, None,
739 hardlink=True, create_tree_if_local=False,
740 source_branch=remote_branch)
741 branch = d.open_branch()
742 branch.bind(remote_branch)
743
744 peer = remote_branch
745 else:
746 peer = None
747
748 return branch
749
750def fix_path(alias, orig_url):
751 url = urlparse.urlparse(orig_url, 'file')
752 if url.scheme != 'file' or os.path.isabs(url.path):
753 return
754 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
755 cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
756 subprocess.call(cmd)
757
758def main(args):
759 global marks, prefix, dirname
760 global tags, filenodes
761 global blob_marks
762 global parsed_refs
763 global files_cache
764 global is_tmp
765
766 alias = args[1]
767 url = args[2]
768
769 tags = {}
770 filenodes = {}
771 blob_marks = {}
772 parsed_refs = {}
773 files_cache = {}
774 marks = None
775
776 if alias[5:] == url:
777 is_tmp = True
778 alias = hashlib.sha1(alias).hexdigest()
779 else:
780 is_tmp = False
781
782 prefix = 'refs/bzr/%s' % alias
783 gitdir = os.environ['GIT_DIR']
784 dirname = os.path.join(gitdir, 'bzr', alias)
785
786 if not is_tmp:
787 fix_path(alias, url)
788
789 if not os.path.exists(dirname):
790 os.makedirs(dirname)
791
792 bzrlib.ui.ui_factory.be_quiet(True)
793
794 repo = get_repo(url, alias)
795
796 marks_path = os.path.join(dirname, 'marks-int')
797 marks = Marks(marks_path)
798
799 parser = Parser(repo)
800 for line in parser:
801 if parser.check('capabilities'):
802 do_capabilities(parser)
803 elif parser.check('list'):
804 do_list(parser)
805 elif parser.check('import'):
806 do_import(parser)
807 elif parser.check('export'):
808 do_export(parser)
809 else:
810 die('unhandled command: %s' % line)
811 sys.stdout.flush()
812
813def bye():
814 if not marks:
815 return
816 if not is_tmp:
817 marks.store()
818 else:
819 shutil.rmtree(dirname)
820
821atexit.register(bye)
822sys.exit(main(sys.argv))