Andrew's git - gitweb.git/blob - 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
  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
  29
  30import sys
  31import os
  32import json
  33import re
  34import StringIO
  35
  36NAME_RE = re.compile('^([^<>]+)')
  37AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
  38RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
  39
  40def die(msg, *args):
  41    sys.stderr.write('ERROR: %s\n' % (msg % args))
  42    sys.exit(1)
  43
  44def warn(msg, *args):
  45    sys.stderr.write('WARNING: %s\n' % (msg % args))
  46
  47def gittz(tz):
  48    return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
  49
  50class Marks:
  51
  52    def __init__(self, path):
  53        self.path = path
  54        self.tips = {}
  55        self.marks = {}
  56        self.rev_marks = {}
  57        self.last_mark = 0
  58        self.load()
  59
  60    def load(self):
  61        if not os.path.exists(self.path):
  62            return
  63
  64        tmp = json.load(open(self.path))
  65        self.tips = tmp['tips']
  66        self.marks = tmp['marks']
  67        self.last_mark = tmp['last-mark']
  68
  69        for rev, mark in self.marks.iteritems():
  70            self.rev_marks[mark] = rev
  71
  72    def dict(self):
  73        return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
  74
  75    def store(self):
  76        json.dump(self.dict(), open(self.path, 'w'))
  77
  78    def __str__(self):
  79        return str(self.dict())
  80
  81    def from_rev(self, rev):
  82        return self.marks[rev]
  83
  84    def to_rev(self, mark):
  85        return self.rev_marks[mark]
  86
  87    def next_mark(self):
  88        self.last_mark += 1
  89        return self.last_mark
  90
  91    def get_mark(self, rev):
  92        self.last_mark += 1
  93        self.marks[rev] = self.last_mark
  94        return self.last_mark
  95
  96    def is_marked(self, rev):
  97        return self.marks.has_key(rev)
  98
  99    def new_mark(self, rev, mark):
 100        self.marks[rev] = mark
 101        self.rev_marks[mark] = rev
 102        self.last_mark = mark
 103
 104    def get_tip(self, branch):
 105        return self.tips.get(branch, None)
 106
 107    def set_tip(self, branch, tip):
 108        self.tips[branch] = tip
 109
 110class Parser:
 111
 112    def __init__(self, repo):
 113        self.repo = repo
 114        self.line = self.get_line()
 115
 116    def get_line(self):
 117        return sys.stdin.readline().strip()
 118
 119    def __getitem__(self, i):
 120        return self.line.split()[i]
 121
 122    def check(self, word):
 123        return self.line.startswith(word)
 124
 125    def each_block(self, separator):
 126        while self.line != separator:
 127            yield self.line
 128            self.line = self.get_line()
 129
 130    def __iter__(self):
 131        return self.each_block('')
 132
 133    def next(self):
 134        self.line = self.get_line()
 135        if self.line == 'done':
 136            self.line = None
 137
 138    def get_mark(self):
 139        i = self.line.index(':') + 1
 140        return int(self.line[i:])
 141
 142    def get_data(self):
 143        if not self.check('data'):
 144            return None
 145        i = self.line.index(' ') + 1
 146        size = int(self.line[i:])
 147        return sys.stdin.read(size)
 148
 149    def get_author(self):
 150        m = RAW_AUTHOR_RE.match(self.line)
 151        if not m:
 152            return None
 153        _, name, email, date, tz = m.groups()
 154        committer = '%s <%s>' % (name, email)
 155        tz = int(tz)
 156        tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
 157        return (committer, int(date), tz)
 158
 159def rev_to_mark(rev):
 160    global marks
 161    return marks.from_rev(rev)
 162
 163def mark_to_rev(mark):
 164    global marks
 165    return marks.to_rev(mark)
 166
 167def fixup_user(user):
 168    name = mail = None
 169    user = user.replace('"', '')
 170    m = AUTHOR_RE.match(user)
 171    if m:
 172        name = m.group(1)
 173        mail = m.group(2).strip()
 174    else:
 175        m = NAME_RE.match(user)
 176        if m:
 177            name = m.group(1).strip()
 178
 179    return '%s <%s>' % (name, mail)
 180
 181def get_filechanges(cur, prev):
 182    modified = {}
 183    removed = {}
 184
 185    changes = cur.changes_from(prev)
 186
 187    def u(s):
 188        return s.encode('utf-8')
 189
 190    for path, fid, kind in changes.added:
 191        modified[u(path)] = fid
 192    for path, fid, kind in changes.removed:
 193        removed[u(path)] = None
 194    for path, fid, kind, mod, _ in changes.modified:
 195        modified[u(path)] = fid
 196    for oldpath, newpath, fid, kind, mod, _ in changes.renamed:
 197        removed[u(oldpath)] = None
 198        if kind == 'directory':
 199            lst = cur.list_files(from_dir=newpath, recursive=True)
 200            for path, file_class, kind, fid, entry in lst:
 201                if kind != 'directory':
 202                    modified[u(newpath + '/' + path)] = fid
 203        else:
 204            modified[u(newpath)] = fid
 205
 206    return modified, removed
 207
 208def export_files(tree, files):
 209    global marks, filenodes
 210
 211    final = []
 212    for path, fid in files.iteritems():
 213        kind = tree.kind(fid)
 214
 215        h = tree.get_file_sha1(fid)
 216
 217        if kind == 'symlink':
 218            d = tree.get_symlink_target(fid)
 219            mode = '120000'
 220        elif kind == 'file':
 221
 222            if tree.is_executable(fid):
 223                mode = '100755'
 224            else:
 225                mode = '100644'
 226
 227            # is the blog already exported?
 228            if h in filenodes:
 229                mark = filenodes[h]
 230                final.append((mode, mark, path))
 231                continue
 232
 233            d = tree.get_file_text(fid)
 234        elif kind == 'directory':
 235            continue
 236        else:
 237            die("Unhandled kind '%s' for path '%s'" % (kind, path))
 238
 239        mark = marks.next_mark()
 240        filenodes[h] = mark
 241
 242        print "blob"
 243        print "mark :%u" % mark
 244        print "data %d" % len(d)
 245        print d
 246
 247        final.append((mode, mark, path))
 248
 249    return final
 250
 251def export_branch(branch, name):
 252    global prefix
 253
 254    ref = '%s/heads/%s' % (prefix, name)
 255    tip = marks.get_tip(name)
 256
 257    repo = branch.repository
 258    repo.lock_read()
 259    revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
 260    count = 0
 261
 262    revs = [revid for revid, _, _, _ in revs if not marks.is_marked(revid)]
 263
 264    for revid in revs:
 265
 266        rev = repo.get_revision(revid)
 267
 268        parents = rev.parent_ids
 269        time = rev.timestamp
 270        tz = rev.timezone
 271        committer = rev.committer.encode('utf-8')
 272        committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
 273        authors = rev.get_apparent_authors()
 274        if authors:
 275            author = authors[0].encode('utf-8')
 276            author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
 277        else:
 278            author = committer
 279        msg = rev.message.encode('utf-8')
 280
 281        msg += '\n'
 282
 283        if len(parents) == 0:
 284            parent = bzrlib.revision.NULL_REVISION
 285        else:
 286            parent = parents[0]
 287
 288        cur_tree = repo.revision_tree(revid)
 289        prev = repo.revision_tree(parent)
 290        modified, removed = get_filechanges(cur_tree, prev)
 291
 292        modified_final = export_files(cur_tree, modified)
 293
 294        if len(parents) == 0:
 295            print 'reset %s' % ref
 296
 297        print "commit %s" % ref
 298        print "mark :%d" % (marks.get_mark(revid))
 299        print "author %s" % (author)
 300        print "committer %s" % (committer)
 301        print "data %d" % (len(msg))
 302        print msg
 303
 304        for i, p in enumerate(parents):
 305            try:
 306                m = rev_to_mark(p)
 307            except KeyError:
 308                # ghost?
 309                continue
 310            if i == 0:
 311                print "from :%s" % m
 312            else:
 313                print "merge :%s" % m
 314
 315        for f in removed:
 316            print "D %s" % (f,)
 317        for f in modified_final:
 318            print "M %s :%u %s" % f
 319        print
 320
 321        count += 1
 322        if (count % 100 == 0):
 323            print "progress revision %s (%d/%d)" % (revid, count, len(revs))
 324            print "#############################################################"
 325
 326    repo.unlock()
 327
 328    revid = branch.last_revision()
 329
 330    # make sure the ref is updated
 331    print "reset %s" % ref
 332    print "from :%u" % rev_to_mark(revid)
 333    print
 334
 335    marks.set_tip(name, revid)
 336
 337def export_tag(repo, name):
 338    global tags, prefix
 339
 340    ref = '%s/tags/%s' % (prefix, name)
 341    print "reset %s" % ref
 342    print "from :%u" % rev_to_mark(tags[name])
 343    print
 344
 345def 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
 371def 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
 381class 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
 518def c_style_unescape(string):
 519    if string[0] == string[-1] == '"':
 520        return string.decode('string-escape')[1:-1]
 521    return string
 522
 523def 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
 592def 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
 612def 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            if peer:
 637                try:
 638                    repo.push(peer, stop_revision=revid)
 639                except bzrlib.errors.DivergedBranches:
 640                    print "error %s non-fast forward" % ref
 641                    continue
 642            else:
 643                wt = repo.bzrdir.open_workingtree()
 644                wt.update()
 645        print "ok %s" % ref
 646    print
 647
 648def do_capabilities(parser):
 649    global dirname
 650
 651    print "import"
 652    print "export"
 653    print "refspec refs/heads/*:%s/heads/*" % prefix
 654    print "refspec refs/tags/*:%s/tags/*" % prefix
 655
 656    path = os.path.join(dirname, 'marks-git')
 657
 658    if os.path.exists(path):
 659        print "*import-marks %s" % path
 660    print "*export-marks %s" % path
 661
 662    print
 663
 664def ref_is_valid(name):
 665    return not True in [c in name for c in '~^: \\']
 666
 667def do_list(parser):
 668    global tags
 669    print "? refs/heads/%s" % 'master'
 670
 671    branch = parser.repo
 672    branch.lock_read()
 673    for tag, revid in branch.tags.get_tag_dict().items():
 674        try:
 675            branch.revision_id_to_dotted_revno(revid)
 676        except bzrlib.errors.NoSuchRevision:
 677            continue
 678        if not ref_is_valid(tag):
 679            continue
 680        print "? refs/tags/%s" % tag
 681        tags[tag] = revid
 682    branch.unlock()
 683    print "@refs/heads/%s HEAD" % 'master'
 684    print
 685
 686def get_repo(url, alias):
 687    global dirname, peer
 688
 689    origin = bzrlib.bzrdir.BzrDir.open(url)
 690    branch = origin.open_branch()
 691
 692    if not isinstance(origin.transport, bzrlib.transport.local.LocalTransport):
 693        clone_path = os.path.join(dirname, 'clone')
 694        remote_branch = branch
 695        if os.path.exists(clone_path):
 696            # pull
 697            d = bzrlib.bzrdir.BzrDir.open(clone_path)
 698            branch = d.open_branch()
 699            result = branch.pull(remote_branch, [], None, False)
 700        else:
 701            # clone
 702            d = origin.sprout(clone_path, None,
 703                    hardlink=True, create_tree_if_local=False,
 704                    source_branch=remote_branch)
 705            branch = d.open_branch()
 706            branch.bind(remote_branch)
 707
 708        peer = remote_branch
 709    else:
 710        peer = None
 711
 712    return branch
 713
 714def main(args):
 715    global marks, prefix, dirname
 716    global tags, filenodes
 717    global blob_marks
 718    global parsed_refs
 719    global files_cache
 720
 721    alias = args[1]
 722    url = args[2]
 723
 724    prefix = 'refs/bzr/%s' % alias
 725    tags = {}
 726    filenodes = {}
 727    blob_marks = {}
 728    parsed_refs = {}
 729    files_cache = {}
 730
 731    gitdir = os.environ['GIT_DIR']
 732    dirname = os.path.join(gitdir, 'bzr', alias)
 733
 734    if not os.path.exists(dirname):
 735        os.makedirs(dirname)
 736
 737    repo = get_repo(url, alias)
 738
 739    marks_path = os.path.join(dirname, 'marks-int')
 740    marks = Marks(marks_path)
 741
 742    parser = Parser(repo)
 743    for line in parser:
 744        if parser.check('capabilities'):
 745            do_capabilities(parser)
 746        elif parser.check('list'):
 747            do_list(parser)
 748        elif parser.check('import'):
 749            do_import(parser)
 750        elif parser.check('export'):
 751            do_export(parser)
 752        else:
 753            die('unhandled command: %s' % line)
 754        sys.stdout.flush()
 755
 756    marks.store()
 757
 758sys.exit(main(sys.argv))