fdead31a6323916b640ccda5555560415f830e67
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 repo.lock_read()
261 revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
262 count = 0
263
264 revs = [revid for revid, _, _, _ in revs if not marks.is_marked(revid)]
265
266 for revid in revs:
267
268 rev = repo.get_revision(revid)
269
270 parents = rev.parent_ids
271 time = rev.timestamp
272 tz = rev.timezone
273 committer = rev.committer.encode('utf-8')
274 committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
275 authors = rev.get_apparent_authors()
276 if authors:
277 author = authors[0].encode('utf-8')
278 author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
279 else:
280 author = committer
281 msg = rev.message.encode('utf-8')
282
283 msg += '\n'
284
285 if len(parents) == 0:
286 parent = bzrlib.revision.NULL_REVISION
287 else:
288 parent = parents[0]
289
290 cur_tree = repo.revision_tree(revid)
291 prev = repo.revision_tree(parent)
292 modified, removed = get_filechanges(cur_tree, prev)
293
294 modified_final = export_files(cur_tree, modified)
295
296 if len(parents) == 0:
297 print 'reset %s' % ref
298
299 print "commit %s" % ref
300 print "mark :%d" % (marks.get_mark(revid))
301 print "author %s" % (author)
302 print "committer %s" % (committer)
303 print "data %d" % (len(msg))
304 print msg
305
306 for i, p in enumerate(parents):
307 try:
308 m = rev_to_mark(p)
309 except KeyError:
310 # ghost?
311 continue
312 if i == 0:
313 print "from :%s" % m
314 else:
315 print "merge :%s" % m
316
317 for f in removed:
318 print "D %s" % (f,)
319 for f in modified_final:
320 print "M %s :%u %s" % f
321 print
322
323 count += 1
324 if (count % 100 == 0):
325 print "progress revision %s (%d/%d)" % (revid, count, len(revs))
326 print "#############################################################"
327
328 repo.unlock()
329
330 revid = branch.last_revision()
331
332 # make sure the ref is updated
333 print "reset %s" % ref
334 print "from :%u" % rev_to_mark(revid)
335 print
336
337 marks.set_tip(name, revid)
338
339def export_tag(repo, name):
340 global tags, prefix
341
342 ref = '%s/tags/%s' % (prefix, name)
343 print "reset %s" % ref
344 print "from :%u" % rev_to_mark(tags[name])
345 print
346
347def do_import(parser):
348 global dirname
349
350 branch = parser.repo
351 path = os.path.join(dirname, 'marks-git')
352
353 print "feature done"
354 if os.path.exists(path):
355 print "feature import-marks=%s" % path
356 print "feature export-marks=%s" % path
357 print "feature force"
358 sys.stdout.flush()
359
360 while parser.check('import'):
361 ref = parser[1]
362 if ref.startswith('refs/heads/'):
363 name = ref[len('refs/heads/'):]
364 export_branch(branch, name)
365 if ref.startswith('refs/tags/'):
366 name = ref[len('refs/tags/'):]
367 export_tag(branch, name)
368 parser.next()
369
370 print 'done'
371
372 sys.stdout.flush()
373
374def parse_blob(parser):
375 global blob_marks
376
377 parser.next()
378 mark = parser.get_mark()
379 parser.next()
380 data = parser.get_data()
381 blob_marks[mark] = data
382 parser.next()
383
384class CustomTree():
385
386 def __init__(self, repo, revid, parents, files):
387 global files_cache
388
389 self.updates = {}
390 self.branch = repo
391
392 def copy_tree(revid):
393 files = files_cache[revid] = {}
394 repo.lock_read()
395 tree = repo.repository.revision_tree(revid)
396 try:
397 for path, entry in tree.iter_entries_by_dir():
398 files[path] = [entry.file_id, None]
399 finally:
400 repo.unlock()
401 return files
402
403 if len(parents) == 0:
404 self.base_id = bzrlib.revision.NULL_REVISION
405 self.base_files = {}
406 else:
407 self.base_id = parents[0]
408 self.base_files = files_cache.get(self.base_id, None)
409 if not self.base_files:
410 self.base_files = copy_tree(self.base_id)
411
412 self.files = files_cache[revid] = self.base_files.copy()
413 self.rev_files = {}
414
415 for path, data in self.files.iteritems():
416 fid, mark = data
417 self.rev_files[fid] = [path, mark]
418
419 for path, f in files.iteritems():
420 fid, mark = self.files.get(path, [None, None])
421 if not fid:
422 fid = bzrlib.generate_ids.gen_file_id(path)
423 f['path'] = path
424 self.rev_files[fid] = [path, mark]
425 self.updates[fid] = f
426
427 def last_revision(self):
428 return self.base_id
429
430 def iter_changes(self):
431 changes = []
432
433 def get_parent(dirname, basename):
434 parent_fid, mark = self.base_files.get(dirname, [None, None])
435 if parent_fid:
436 return parent_fid
437 parent_fid, mark = self.files.get(dirname, [None, None])
438 if parent_fid:
439 return parent_fid
440 if basename == '':
441 return None
442 fid = bzrlib.generate_ids.gen_file_id(path)
443 add_entry(fid, dirname, 'directory')
444 return fid
445
446 def add_entry(fid, path, kind, mode = None):
447 dirname, basename = os.path.split(path)
448 parent_fid = get_parent(dirname, basename)
449
450 executable = False
451 if mode == '100755':
452 executable = True
453 elif mode == '120000':
454 kind = 'symlink'
455
456 change = (fid,
457 (None, path),
458 True,
459 (False, True),
460 (None, parent_fid),
461 (None, basename),
462 (None, kind),
463 (None, executable))
464 self.files[path] = [change[0], None]
465 changes.append(change)
466
467 def update_entry(fid, path, kind, mode = None):
468 dirname, basename = os.path.split(path)
469 parent_fid = get_parent(dirname, basename)
470
471 executable = False
472 if mode == '100755':
473 executable = True
474 elif mode == '120000':
475 kind = 'symlink'
476
477 change = (fid,
478 (path, path),
479 True,
480 (True, True),
481 (None, parent_fid),
482 (None, basename),
483 (None, kind),
484 (None, executable))
485 self.files[path] = [change[0], None]
486 changes.append(change)
487
488 def remove_entry(fid, path, kind):
489 dirname, basename = os.path.split(path)
490 parent_fid = get_parent(dirname, basename)
491 change = (fid,
492 (path, None),
493 True,
494 (True, False),
495 (parent_fid, None),
496 (None, None),
497 (None, None),
498 (None, None))
499 del self.files[path]
500 changes.append(change)
501
502 for fid, f in self.updates.iteritems():
503 path = f['path']
504
505 if 'deleted' in f:
506 remove_entry(fid, path, 'file')
507 continue
508
509 if path in self.base_files:
510 update_entry(fid, path, 'file', f['mode'])
511 else:
512 add_entry(fid, path, 'file', f['mode'])
513
514 self.files[path][1] = f['mark']
515 self.rev_files[fid][1] = f['mark']
516
517 return changes
518
519 def get_content(self, file_id):
520 path, mark = self.rev_files[file_id]
521 if mark:
522 return blob_marks[mark]
523
524 # last resort
525 tree = self.branch.repository.revision_tree(self.base_id)
526 return tree.get_file_text(file_id)
527
528 def get_file_with_stat(self, file_id, path=None):
529 content = self.get_content(file_id)
530 return (StringIO.StringIO(content), None)
531
532 def get_symlink_target(self, file_id):
533 return self.get_content(file_id)
534
535 def id2path(self, file_id):
536 path, mark = self.rev_files[file_id]
537 return path
538
539def c_style_unescape(string):
540 if string[0] == string[-1] == '"':
541 return string.decode('string-escape')[1:-1]
542 return string
543
544def parse_commit(parser):
545 global marks, blob_marks, parsed_refs
546 global mode
547
548 parents = []
549
550 ref = parser[1]
551 parser.next()
552
553 if ref != 'refs/heads/master':
554 die("bzr doesn't support multiple branches; use 'master'")
555
556 commit_mark = parser.get_mark()
557 parser.next()
558 author = parser.get_author()
559 parser.next()
560 committer = parser.get_author()
561 parser.next()
562 data = parser.get_data()
563 parser.next()
564 if parser.check('from'):
565 parents.append(parser.get_mark())
566 parser.next()
567 while parser.check('merge'):
568 parents.append(parser.get_mark())
569 parser.next()
570
571 # fast-export adds an extra newline
572 if data[-1] == '\n':
573 data = data[:-1]
574
575 files = {}
576
577 for line in parser:
578 if parser.check('M'):
579 t, m, mark_ref, path = line.split(' ', 3)
580 mark = int(mark_ref[1:])
581 f = { 'mode' : m, 'mark' : mark }
582 elif parser.check('D'):
583 t, path = line.split(' ')
584 f = { 'deleted' : True }
585 else:
586 die('Unknown file command: %s' % line)
587 path = c_style_unescape(path).decode('utf-8')
588 files[path] = f
589
590 repo = parser.repo
591
592 committer, date, tz = committer
593 parents = [str(mark_to_rev(p)) for p in parents]
594 revid = bzrlib.generate_ids.gen_revision_id(committer, date)
595 props = {}
596 props['branch-nick'] = repo.nick
597
598 mtree = CustomTree(repo, revid, parents, files)
599 changes = mtree.iter_changes()
600
601 repo.lock_write()
602 try:
603 builder = repo.get_commit_builder(parents, None, date, tz, committer, props, revid)
604 try:
605 list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
606 builder.finish_inventory()
607 builder.commit(data.decode('utf-8', 'replace'))
608 except Exception, e:
609 builder.abort()
610 raise
611 finally:
612 repo.unlock()
613
614 parsed_refs[ref] = revid
615 marks.new_mark(revid, commit_mark)
616
617def parse_reset(parser):
618 global parsed_refs
619
620 ref = parser[1]
621 parser.next()
622
623 if ref != 'refs/heads/master':
624 die("bzr doesn't support multiple branches; use 'master'")
625
626 # ugh
627 if parser.check('commit'):
628 parse_commit(parser)
629 return
630 if not parser.check('from'):
631 return
632 from_mark = parser.get_mark()
633 parser.next()
634
635 parsed_refs[ref] = mark_to_rev(from_mark)
636
637def do_export(parser):
638 global parsed_refs, dirname, peer
639
640 parser.next()
641
642 for line in parser.each_block('done'):
643 if parser.check('blob'):
644 parse_blob(parser)
645 elif parser.check('commit'):
646 parse_commit(parser)
647 elif parser.check('reset'):
648 parse_reset(parser)
649 elif parser.check('tag'):
650 pass
651 elif parser.check('feature'):
652 pass
653 else:
654 die('unhandled export command: %s' % line)
655
656 repo = parser.repo
657
658 for ref, revid in parsed_refs.iteritems():
659 if ref == 'refs/heads/master':
660 repo.generate_revision_history(revid, marks.get_tip('master'))
661 if peer:
662 try:
663 repo.push(peer, stop_revision=revid)
664 except bzrlib.errors.DivergedBranches:
665 print "error %s non-fast forward" % ref
666 continue
667
668 try:
669 wt = repo.bzrdir.open_workingtree()
670 wt.update()
671 except bzrlib.errors.NoWorkingTree:
672 pass
673
674 print "ok %s" % ref
675
676 print
677
678def do_capabilities(parser):
679 global dirname
680
681 print "import"
682 print "export"
683 print "refspec refs/heads/*:%s/heads/*" % prefix
684 print "refspec refs/tags/*:%s/tags/*" % prefix
685
686 path = os.path.join(dirname, 'marks-git')
687
688 if os.path.exists(path):
689 print "*import-marks %s" % path
690 print "*export-marks %s" % path
691
692 print
693
694def ref_is_valid(name):
695 return not True in [c in name for c in '~^: \\']
696
697def do_list(parser):
698 global tags
699 print "? refs/heads/%s" % 'master'
700
701 branch = parser.repo
702 branch.lock_read()
703 for tag, revid in branch.tags.get_tag_dict().items():
704 try:
705 branch.revision_id_to_dotted_revno(revid)
706 except bzrlib.errors.NoSuchRevision:
707 continue
708 if not ref_is_valid(tag):
709 continue
710 print "? refs/tags/%s" % tag
711 tags[tag] = revid
712 branch.unlock()
713 print "@refs/heads/%s HEAD" % 'master'
714 print
715
716def get_repo(url, alias):
717 global dirname, peer
718
719 origin = bzrlib.bzrdir.BzrDir.open(url)
720 branch = origin.open_branch()
721
722 if not isinstance(origin.transport, bzrlib.transport.local.LocalTransport):
723 clone_path = os.path.join(dirname, 'clone')
724 remote_branch = branch
725 if os.path.exists(clone_path):
726 # pull
727 d = bzrlib.bzrdir.BzrDir.open(clone_path)
728 branch = d.open_branch()
729 try:
730 result = branch.pull(remote_branch, [], None, False)
731 except bzrlib.errors.DivergedBranches:
732 # use remote branch for now
733 peer = None
734 return remote_branch
735 else:
736 # clone
737 d = origin.sprout(clone_path, None,
738 hardlink=True, create_tree_if_local=False,
739 source_branch=remote_branch)
740 branch = d.open_branch()
741 branch.bind(remote_branch)
742
743 peer = remote_branch
744 else:
745 peer = None
746
747 return branch
748
749def fix_path(alias, orig_url):
750 url = urlparse.urlparse(orig_url, 'file')
751 if url.scheme != 'file' or os.path.isabs(url.path):
752 return
753 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
754 cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
755 subprocess.call(cmd)
756
757def main(args):
758 global marks, prefix, dirname
759 global tags, filenodes
760 global blob_marks
761 global parsed_refs
762 global files_cache
763 global is_tmp
764
765 alias = args[1]
766 url = args[2]
767
768 tags = {}
769 filenodes = {}
770 blob_marks = {}
771 parsed_refs = {}
772 files_cache = {}
773 marks = None
774
775 if alias[5:] == url:
776 is_tmp = True
777 alias = hashlib.sha1(alias).hexdigest()
778 else:
779 is_tmp = False
780
781 prefix = 'refs/bzr/%s' % alias
782 gitdir = os.environ['GIT_DIR']
783 dirname = os.path.join(gitdir, 'bzr', alias)
784
785 if not is_tmp:
786 fix_path(alias, url)
787
788 if not os.path.exists(dirname):
789 os.makedirs(dirname)
790
791 bzrlib.ui.ui_factory.be_quiet(True)
792
793 repo = get_repo(url, alias)
794
795 marks_path = os.path.join(dirname, 'marks-int')
796 marks = Marks(marks_path)
797
798 parser = Parser(repo)
799 for line in parser:
800 if parser.check('capabilities'):
801 do_capabilities(parser)
802 elif parser.check('list'):
803 do_list(parser)
804 elif parser.check('import'):
805 do_import(parser)
806 elif parser.check('export'):
807 do_export(parser)
808 else:
809 die('unhandled command: %s' % line)
810 sys.stdout.flush()
811
812def bye():
813 if not marks:
814 return
815 if not is_tmp:
816 marks.store()
817 else:
818 shutil.rmtree(dirname)
819
820atexit.register(bye)
821sys.exit(main(sys.argv))