contrib/remote-helpers/git-remote-bzr

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