08b0b61ee38eef5e901bfe7ee2b382cf422c9c32
   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# If you want to specify which branches you want track (per repo):
  17# git config remote-bzr.branches 'trunk, devel, test'
  18#
  19
  20import sys
  21
  22import bzrlib
  23if hasattr(bzrlib, "initialize"):
  24    bzrlib.initialize()
  25
  26import bzrlib.plugin
  27bzrlib.plugin.load_plugins()
  28
  29import bzrlib.generate_ids
  30import bzrlib.transport
  31import bzrlib.errors
  32import bzrlib.ui
  33import bzrlib.urlutils
  34import bzrlib.branch
  35
  36import sys
  37import os
  38import json
  39import re
  40import StringIO
  41import atexit, shutil, hashlib, urlparse, subprocess
  42
  43NAME_RE = re.compile('^([^<>]+)')
  44AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
  45EMAIL_RE = re.compile('^([^<>]+[^ \\\t<>])?\\b(?:[ \\t<>]*?)\\b([^ \\t<>]+@[^ \\t<>]+)')
  46RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
  47
  48def die(msg, *args):
  49    sys.stderr.write('ERROR: %s\n' % (msg % args))
  50    sys.exit(1)
  51
  52def warn(msg, *args):
  53    sys.stderr.write('WARNING: %s\n' % (msg % args))
  54
  55def gittz(tz):
  56    return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
  57
  58def get_config(config):
  59    cmd = ['git', 'config', '--get', config]
  60    process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
  61    output, _ = process.communicate()
  62    return output
  63
  64class Marks:
  65
  66    def __init__(self, path):
  67        self.path = path
  68        self.tips = {}
  69        self.marks = {}
  70        self.rev_marks = {}
  71        self.last_mark = 0
  72        self.load()
  73
  74    def load(self):
  75        if not os.path.exists(self.path):
  76            return
  77
  78        tmp = json.load(open(self.path))
  79        self.tips = tmp['tips']
  80        self.marks = tmp['marks']
  81        self.last_mark = tmp['last-mark']
  82
  83        for rev, mark in self.marks.iteritems():
  84            self.rev_marks[mark] = rev
  85
  86    def dict(self):
  87        return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
  88
  89    def store(self):
  90        json.dump(self.dict(), open(self.path, 'w'))
  91
  92    def __str__(self):
  93        return str(self.dict())
  94
  95    def from_rev(self, rev):
  96        return self.marks[rev]
  97
  98    def to_rev(self, mark):
  99        return str(self.rev_marks[mark])
 100
 101    def next_mark(self):
 102        self.last_mark += 1
 103        return self.last_mark
 104
 105    def get_mark(self, rev):
 106        self.last_mark += 1
 107        self.marks[rev] = self.last_mark
 108        return self.last_mark
 109
 110    def is_marked(self, rev):
 111        return rev in self.marks
 112
 113    def new_mark(self, rev, mark):
 114        self.marks[rev] = mark
 115        self.rev_marks[mark] = rev
 116        self.last_mark = mark
 117
 118    def get_tip(self, branch):
 119        try:
 120            return str(self.tips[branch])
 121        except KeyError:
 122            return None
 123
 124    def set_tip(self, branch, tip):
 125        self.tips[branch] = tip
 126
 127class Parser:
 128
 129    def __init__(self, repo):
 130        self.repo = repo
 131        self.line = self.get_line()
 132
 133    def get_line(self):
 134        return sys.stdin.readline().strip()
 135
 136    def __getitem__(self, i):
 137        return self.line.split()[i]
 138
 139    def check(self, word):
 140        return self.line.startswith(word)
 141
 142    def each_block(self, separator):
 143        while self.line != separator:
 144            yield self.line
 145            self.line = self.get_line()
 146
 147    def __iter__(self):
 148        return self.each_block('')
 149
 150    def next(self):
 151        self.line = self.get_line()
 152        if self.line == 'done':
 153            self.line = None
 154
 155    def get_mark(self):
 156        i = self.line.index(':') + 1
 157        return int(self.line[i:])
 158
 159    def get_data(self):
 160        if not self.check('data'):
 161            return None
 162        i = self.line.index(' ') + 1
 163        size = int(self.line[i:])
 164        return sys.stdin.read(size)
 165
 166    def get_author(self):
 167        m = RAW_AUTHOR_RE.match(self.line)
 168        if not m:
 169            return None
 170        _, name, email, date, tz = m.groups()
 171        name = name.decode('utf-8')
 172        committer = '%s <%s>' % (name, email)
 173        tz = int(tz)
 174        tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
 175        return (committer, int(date), tz)
 176
 177def rev_to_mark(rev):
 178    global marks
 179    return marks.from_rev(rev)
 180
 181def mark_to_rev(mark):
 182    global marks
 183    return marks.to_rev(mark)
 184
 185def fixup_user(user):
 186    name = mail = None
 187    user = user.replace('"', '')
 188    m = AUTHOR_RE.match(user)
 189    if m:
 190        name = m.group(1)
 191        mail = m.group(2).strip()
 192    else:
 193        m = EMAIL_RE.match(user)
 194        if m:
 195            name = m.group(1)
 196            mail = m.group(2)
 197        else:
 198            m = NAME_RE.match(user)
 199            if m:
 200                name = m.group(1).strip()
 201
 202    if not name:
 203        name = 'unknown'
 204    if not mail:
 205        mail = 'Unknown'
 206
 207    return '%s <%s>' % (name, mail)
 208
 209def get_filechanges(cur, prev):
 210    modified = {}
 211    removed = {}
 212
 213    changes = cur.changes_from(prev)
 214
 215    def u(s):
 216        return s.encode('utf-8')
 217
 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
 231        else:
 232            modified[u(newpath)] = fid
 233
 234    return modified, removed
 235
 236def export_files(tree, files):
 237    global marks, filenodes
 238
 239    final = []
 240    for path, fid in files.iteritems():
 241        kind = tree.kind(fid)
 242
 243        h = tree.get_file_sha1(fid)
 244
 245        if kind == 'symlink':
 246            d = tree.get_symlink_target(fid)
 247            mode = '120000'
 248        elif kind == 'file':
 249
 250            if tree.is_executable(fid):
 251                mode = '100755'
 252            else:
 253                mode = '100644'
 254
 255            # is the blob already exported?
 256            if h in filenodes:
 257                mark = filenodes[h]
 258                final.append((mode, mark, path))
 259                continue
 260
 261            d = tree.get_file_text(fid)
 262        elif kind == 'directory':
 263            continue
 264        else:
 265            die("Unhandled kind '%s' for path '%s'" % (kind, path))
 266
 267        mark = marks.next_mark()
 268        filenodes[h] = mark
 269
 270        print "blob"
 271        print "mark :%u" % mark
 272        print "data %d" % len(d)
 273        print d
 274
 275        final.append((mode, mark, path))
 276
 277    return final
 278
 279def export_branch(repo, name):
 280    global prefix
 281
 282    ref = '%s/heads/%s' % (prefix, name)
 283    tip = marks.get_tip(name)
 284
 285    branch = get_remote_branch(name)
 286    repo = branch.repository
 287
 288    branch.lock_read()
 289    revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
 290    try:
 291        tip_revno = branch.revision_id_to_revno(tip)
 292        last_revno, _ = branch.last_revision_info()
 293        total = last_revno - tip_revno
 294    except bzrlib.errors.NoSuchRevision:
 295        tip_revno = 0
 296        total = 0
 297
 298    for revid, _, seq, _ in revs:
 299
 300        if marks.is_marked(revid):
 301            continue
 302
 303        rev = repo.get_revision(revid)
 304        revno = seq[0]
 305
 306        parents = rev.parent_ids
 307        time = rev.timestamp
 308        tz = rev.timezone
 309        committer = rev.committer.encode('utf-8')
 310        committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
 311        authors = rev.get_apparent_authors()
 312        if authors:
 313            author = authors[0].encode('utf-8')
 314            author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
 315        else:
 316            author = committer
 317        msg = rev.message.encode('utf-8')
 318
 319        msg += '\n'
 320
 321        if len(parents) == 0:
 322            parent = bzrlib.revision.NULL_REVISION
 323        else:
 324            parent = parents[0]
 325
 326        cur_tree = repo.revision_tree(revid)
 327        prev = repo.revision_tree(parent)
 328        modified, removed = get_filechanges(cur_tree, prev)
 329
 330        modified_final = export_files(cur_tree, modified)
 331
 332        if len(parents) == 0:
 333            print 'reset %s' % ref
 334
 335        print "commit %s" % ref
 336        print "mark :%d" % (marks.get_mark(revid))
 337        print "author %s" % (author)
 338        print "committer %s" % (committer)
 339        print "data %d" % (len(msg))
 340        print msg
 341
 342        for i, p in enumerate(parents):
 343            try:
 344                m = rev_to_mark(p)
 345            except KeyError:
 346                # ghost?
 347                continue
 348            if i == 0:
 349                print "from :%s" % m
 350            else:
 351                print "merge :%s" % m
 352
 353        for f in removed:
 354            print "D %s" % (f,)
 355        for f in modified_final:
 356            print "M %s :%u %s" % f
 357        print
 358
 359        if len(seq) > 1:
 360            # let's skip branch revisions from the progress report
 361            continue
 362
 363        progress = (revno - tip_revno)
 364        if (progress % 100 == 0):
 365            if total:
 366                print "progress revision %d '%s' (%d/%d)" % (revno, name, progress, total)
 367            else:
 368                print "progress revision %d '%s' (%d)" % (revno, name, progress)
 369
 370    branch.unlock()
 371
 372    revid = branch.last_revision()
 373
 374    # make sure the ref is updated
 375    print "reset %s" % ref
 376    print "from :%u" % rev_to_mark(revid)
 377    print
 378
 379    marks.set_tip(name, revid)
 380
 381def export_tag(repo, name):
 382    global tags, prefix
 383
 384    ref = '%s/tags/%s' % (prefix, name)
 385    print "reset %s" % ref
 386    print "from :%u" % rev_to_mark(tags[name])
 387    print
 388
 389def do_import(parser):
 390    global dirname
 391
 392    repo = parser.repo
 393    path = os.path.join(dirname, 'marks-git')
 394
 395    print "feature done"
 396    if os.path.exists(path):
 397        print "feature import-marks=%s" % path
 398    print "feature export-marks=%s" % path
 399    print "feature force"
 400    sys.stdout.flush()
 401
 402    while parser.check('import'):
 403        ref = parser[1]
 404        if ref.startswith('refs/heads/'):
 405            name = ref[len('refs/heads/'):]
 406            export_branch(repo, name)
 407        if ref.startswith('refs/tags/'):
 408            name = ref[len('refs/tags/'):]
 409            export_tag(repo, name)
 410        parser.next()
 411
 412    print 'done'
 413
 414    sys.stdout.flush()
 415
 416def parse_blob(parser):
 417    global blob_marks
 418
 419    parser.next()
 420    mark = parser.get_mark()
 421    parser.next()
 422    data = parser.get_data()
 423    blob_marks[mark] = data
 424    parser.next()
 425
 426class CustomTree():
 427
 428    def __init__(self, branch, revid, parents, files):
 429        global files_cache
 430
 431        self.updates = {}
 432        self.branch = branch
 433
 434        def copy_tree(revid):
 435            files = files_cache[revid] = {}
 436            branch.lock_read()
 437            tree = branch.repository.revision_tree(revid)
 438            try:
 439                for path, entry in tree.iter_entries_by_dir():
 440                    files[path] = [entry.file_id, None]
 441            finally:
 442                branch.unlock()
 443            return files
 444
 445        if len(parents) == 0:
 446            self.base_id = bzrlib.revision.NULL_REVISION
 447            self.base_files = {}
 448        else:
 449            self.base_id = parents[0]
 450            self.base_files = files_cache.get(self.base_id, None)
 451            if not self.base_files:
 452                self.base_files = copy_tree(self.base_id)
 453
 454        self.files = files_cache[revid] = self.base_files.copy()
 455        self.rev_files = {}
 456
 457        for path, data in self.files.iteritems():
 458            fid, mark = data
 459            self.rev_files[fid] = [path, mark]
 460
 461        for path, f in files.iteritems():
 462            fid, mark = self.files.get(path, [None, None])
 463            if not fid:
 464                fid = bzrlib.generate_ids.gen_file_id(path)
 465            f['path'] = path
 466            self.rev_files[fid] = [path, mark]
 467            self.updates[fid] = f
 468
 469    def last_revision(self):
 470        return self.base_id
 471
 472    def iter_changes(self):
 473        changes = []
 474
 475        def get_parent(dirname, basename):
 476            parent_fid, mark = self.base_files.get(dirname, [None, None])
 477            if parent_fid:
 478                return parent_fid
 479            parent_fid, mark = self.files.get(dirname, [None, None])
 480            if parent_fid:
 481                return parent_fid
 482            if basename == '':
 483                return None
 484            fid = bzrlib.generate_ids.gen_file_id(path)
 485            add_entry(fid, dirname, 'directory')
 486            return fid
 487
 488        def add_entry(fid, path, kind, mode = None):
 489            dirname, basename = os.path.split(path)
 490            parent_fid = get_parent(dirname, basename)
 491
 492            executable = False
 493            if mode == '100755':
 494                executable = True
 495            elif mode == '120000':
 496                kind = 'symlink'
 497
 498            change = (fid,
 499                    (None, path),
 500                    True,
 501                    (False, True),
 502                    (None, parent_fid),
 503                    (None, basename),
 504                    (None, kind),
 505                    (None, executable))
 506            self.files[path] = [change[0], None]
 507            changes.append(change)
 508
 509        def update_entry(fid, path, kind, mode = None):
 510            dirname, basename = os.path.split(path)
 511            parent_fid = get_parent(dirname, basename)
 512
 513            executable = False
 514            if mode == '100755':
 515                executable = True
 516            elif mode == '120000':
 517                kind = 'symlink'
 518
 519            change = (fid,
 520                    (path, path),
 521                    True,
 522                    (True, True),
 523                    (None, parent_fid),
 524                    (None, basename),
 525                    (None, kind),
 526                    (None, executable))
 527            self.files[path] = [change[0], None]
 528            changes.append(change)
 529
 530        def remove_entry(fid, path, kind):
 531            dirname, basename = os.path.split(path)
 532            parent_fid = get_parent(dirname, basename)
 533            change = (fid,
 534                    (path, None),
 535                    True,
 536                    (True, False),
 537                    (parent_fid, None),
 538                    (None, None),
 539                    (None, None),
 540                    (None, None))
 541            del self.files[path]
 542            changes.append(change)
 543
 544        for fid, f in self.updates.iteritems():
 545            path = f['path']
 546
 547            if 'deleted' in f:
 548                remove_entry(fid, path, 'file')
 549                continue
 550
 551            if path in self.base_files:
 552                update_entry(fid, path, 'file', f['mode'])
 553            else:
 554                add_entry(fid, path, 'file', f['mode'])
 555
 556            self.files[path][1] = f['mark']
 557            self.rev_files[fid][1] = f['mark']
 558
 559        return changes
 560
 561    def get_content(self, file_id):
 562        path, mark = self.rev_files[file_id]
 563        if mark:
 564            return blob_marks[mark]
 565
 566        # last resort
 567        tree = self.branch.repository.revision_tree(self.base_id)
 568        return tree.get_file_text(file_id)
 569
 570    def get_file_with_stat(self, file_id, path=None):
 571        content = self.get_content(file_id)
 572        return (StringIO.StringIO(content), None)
 573
 574    def get_symlink_target(self, file_id):
 575        return self.get_content(file_id)
 576
 577    def id2path(self, file_id):
 578        path, mark = self.rev_files[file_id]
 579        return path
 580
 581def c_style_unescape(string):
 582    if string[0] == string[-1] == '"':
 583        return string.decode('string-escape')[1:-1]
 584    return string
 585
 586def parse_commit(parser):
 587    global marks, blob_marks, parsed_refs
 588    global mode
 589
 590    parents = []
 591
 592    ref = parser[1]
 593    parser.next()
 594
 595    if ref.startswith('refs/heads/'):
 596        name = ref[len('refs/heads/'):]
 597        branch = get_remote_branch(name)
 598    else:
 599        die('unknown ref')
 600
 601    commit_mark = parser.get_mark()
 602    parser.next()
 603    author = parser.get_author()
 604    parser.next()
 605    committer = parser.get_author()
 606    parser.next()
 607    data = parser.get_data()
 608    parser.next()
 609    if parser.check('from'):
 610        parents.append(parser.get_mark())
 611        parser.next()
 612    while parser.check('merge'):
 613        parents.append(parser.get_mark())
 614        parser.next()
 615
 616    # fast-export adds an extra newline
 617    if data[-1] == '\n':
 618        data = data[:-1]
 619
 620    files = {}
 621
 622    for line in parser:
 623        if parser.check('M'):
 624            t, m, mark_ref, path = line.split(' ', 3)
 625            mark = int(mark_ref[1:])
 626            f = { 'mode' : m, 'mark' : mark }
 627        elif parser.check('D'):
 628            t, path = line.split(' ', 1)
 629            f = { 'deleted' : True }
 630        else:
 631            die('Unknown file command: %s' % line)
 632        path = c_style_unescape(path).decode('utf-8')
 633        files[path] = f
 634
 635    committer, date, tz = committer
 636    parents = [mark_to_rev(p) for p in parents]
 637    revid = bzrlib.generate_ids.gen_revision_id(committer, date)
 638    props = {}
 639    props['branch-nick'] = branch.nick
 640
 641    mtree = CustomTree(branch, revid, parents, files)
 642    changes = mtree.iter_changes()
 643
 644    branch.lock_write()
 645    try:
 646        builder = branch.get_commit_builder(parents, None, date, tz, committer, props, revid)
 647        try:
 648            list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
 649            builder.finish_inventory()
 650            builder.commit(data.decode('utf-8', 'replace'))
 651        except Exception, e:
 652            builder.abort()
 653            raise
 654    finally:
 655        branch.unlock()
 656
 657    parsed_refs[ref] = revid
 658    marks.new_mark(revid, commit_mark)
 659
 660def parse_reset(parser):
 661    global parsed_refs
 662
 663    ref = parser[1]
 664    parser.next()
 665
 666    # ugh
 667    if parser.check('commit'):
 668        parse_commit(parser)
 669        return
 670    if not parser.check('from'):
 671        return
 672    from_mark = parser.get_mark()
 673    parser.next()
 674
 675    parsed_refs[ref] = mark_to_rev(from_mark)
 676
 677def do_export(parser):
 678    global parsed_refs, dirname
 679
 680    parser.next()
 681
 682    for line in parser.each_block('done'):
 683        if parser.check('blob'):
 684            parse_blob(parser)
 685        elif parser.check('commit'):
 686            parse_commit(parser)
 687        elif parser.check('reset'):
 688            parse_reset(parser)
 689        elif parser.check('tag'):
 690            pass
 691        elif parser.check('feature'):
 692            pass
 693        else:
 694            die('unhandled export command: %s' % line)
 695
 696    for ref, revid in parsed_refs.iteritems():
 697        if ref.startswith('refs/heads/'):
 698            name = ref[len('refs/heads/'):]
 699            branch = get_remote_branch(name)
 700            branch.generate_revision_history(revid, marks.get_tip(name))
 701
 702            if name in peers:
 703                peer = bzrlib.branch.Branch.open(peers[name])
 704                try:
 705                    peer.bzrdir.push_branch(branch, revision_id=revid)
 706                except bzrlib.errors.DivergedBranches:
 707                    print "error %s non-fast forward" % ref
 708                    continue
 709
 710            try:
 711                wt = branch.bzrdir.open_workingtree()
 712                wt.update()
 713            except bzrlib.errors.NoWorkingTree:
 714                pass
 715        elif ref.startswith('refs/tags/'):
 716            # TODO: implement tag push
 717            print "error %s pushing tags not supported" % ref
 718            continue
 719        else:
 720            # transport-helper/fast-export bugs
 721            continue
 722
 723        print "ok %s" % ref
 724
 725    print
 726
 727def do_capabilities(parser):
 728    global dirname
 729
 730    print "import"
 731    print "export"
 732    print "refspec refs/heads/*:%s/heads/*" % prefix
 733    print "refspec refs/tags/*:%s/tags/*" % prefix
 734
 735    path = os.path.join(dirname, 'marks-git')
 736
 737    if os.path.exists(path):
 738        print "*import-marks %s" % path
 739    print "*export-marks %s" % path
 740
 741    print
 742
 743def ref_is_valid(name):
 744    return not True in [c in name for c in '~^: \\']
 745
 746def do_list(parser):
 747    global tags
 748
 749    master_branch = None
 750
 751    for name in branches:
 752        if not master_branch:
 753            master_branch = name
 754        print "? refs/heads/%s" % name
 755
 756    branch = get_remote_branch(master_branch)
 757    branch.lock_read()
 758    for tag, revid in branch.tags.get_tag_dict().items():
 759        try:
 760            branch.revision_id_to_dotted_revno(revid)
 761        except bzrlib.errors.NoSuchRevision:
 762            continue
 763        if not ref_is_valid(tag):
 764            continue
 765        print "? refs/tags/%s" % tag
 766        tags[tag] = revid
 767    branch.unlock()
 768
 769    print "@refs/heads/%s HEAD" % master_branch
 770    print
 771
 772def clone(path, remote_branch):
 773    try:
 774        bdir = bzrlib.bzrdir.BzrDir.create(path)
 775    except bzrlib.errors.AlreadyControlDirError:
 776        bdir = bzrlib.bzrdir.BzrDir.open(path)
 777    repo = bdir.find_repository()
 778    repo.fetch(remote_branch.repository)
 779    return remote_branch.sprout(bdir, repository=repo)
 780
 781def get_remote_branch(name):
 782    global dirname, branches
 783
 784    remote_branch = bzrlib.branch.Branch.open(branches[name])
 785    if isinstance(remote_branch.user_transport, bzrlib.transport.local.LocalTransport):
 786        return remote_branch
 787
 788    branch_path = os.path.join(dirname, 'clone', name)
 789
 790    try:
 791        branch = bzrlib.branch.Branch.open(branch_path)
 792    except bzrlib.errors.NotBranchError:
 793        # clone
 794        branch = clone(branch_path, remote_branch)
 795    else:
 796        # pull
 797        try:
 798            branch.pull(remote_branch, overwrite=True)
 799        except bzrlib.errors.DivergedBranches:
 800            # use remote branch for now
 801            return remote_branch
 802
 803    return branch
 804
 805def find_branches(repo):
 806    transport = repo.bzrdir.root_transport
 807
 808    for fn in transport.iter_files_recursive():
 809        if not fn.endswith('.bzr/branch-format'):
 810            continue
 811
 812        name = subdir = fn[:-len('/.bzr/branch-format')]
 813        name = name if name != '' else 'master'
 814        name = name.replace('/', '+')
 815
 816        try:
 817            cur = transport.clone(subdir)
 818            branch = bzrlib.branch.Branch.open_from_transport(cur)
 819        except bzrlib.errors.NotBranchError:
 820            continue
 821        else:
 822            yield name, branch.base
 823
 824def get_repo(url, alias):
 825    global dirname, peer, branches
 826
 827    normal_url = bzrlib.urlutils.normalize_url(url)
 828    origin = bzrlib.bzrdir.BzrDir.open(url)
 829    is_local = isinstance(origin.transport, bzrlib.transport.local.LocalTransport)
 830
 831    shared_path = os.path.join(gitdir, 'bzr')
 832    try:
 833        shared_dir = bzrlib.bzrdir.BzrDir.open(shared_path)
 834    except bzrlib.errors.NotBranchError:
 835        shared_dir = bzrlib.bzrdir.BzrDir.create(shared_path)
 836    try:
 837        shared_repo = shared_dir.open_repository()
 838    except bzrlib.errors.NoRepositoryPresent:
 839        shared_repo = shared_dir.create_repository(shared=True)
 840
 841    if not is_local:
 842        clone_path = os.path.join(dirname, 'clone')
 843        if not os.path.exists(clone_path):
 844            os.mkdir(clone_path)
 845        else:
 846            # check and remove old organization
 847            try:
 848                bdir = bzrlib.bzrdir.BzrDir.open(clone_path)
 849                bdir.destroy_repository()
 850            except bzrlib.errors.NotBranchError:
 851                pass
 852            except bzrlib.errors.NoRepositoryPresent:
 853                pass
 854
 855    wanted = get_config('remote-bzr.branches').rstrip().split(', ')
 856    # stupid python
 857    wanted = [e for e in wanted if e]
 858
 859    if not wanted:
 860        try:
 861            repo = origin.open_repository()
 862            if not repo.user_transport.listable():
 863                # this repository is not usable for us
 864                raise bzrlib.errors.NoRepositoryPresent(repo.bzrdir)
 865        except bzrlib.errors.NoRepositoryPresent:
 866            wanted = ['master']
 867
 868    if wanted:
 869        def list_wanted(url, wanted):
 870            for name in wanted:
 871                subdir = name if name != 'master' else ''
 872                yield name, bzrlib.urlutils.join(url, subdir)
 873
 874        branch_list = list_wanted(url, wanted)
 875    else:
 876        branch_list = find_branches(repo)
 877
 878    for name, url in branch_list:
 879        if not is_local:
 880            peers[name] = url
 881        branches[name] = url
 882
 883    return origin
 884
 885def fix_path(alias, orig_url):
 886    url = urlparse.urlparse(orig_url, 'file')
 887    if url.scheme != 'file' or os.path.isabs(url.path):
 888        return
 889    abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
 890    cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
 891    subprocess.call(cmd)
 892
 893def main(args):
 894    global marks, prefix, gitdir, dirname
 895    global tags, filenodes
 896    global blob_marks
 897    global parsed_refs
 898    global files_cache
 899    global is_tmp
 900    global branches, peers
 901
 902    alias = args[1]
 903    url = args[2]
 904
 905    tags = {}
 906    filenodes = {}
 907    blob_marks = {}
 908    parsed_refs = {}
 909    files_cache = {}
 910    marks = None
 911    branches = {}
 912    peers = {}
 913
 914    if alias[5:] == url:
 915        is_tmp = True
 916        alias = hashlib.sha1(alias).hexdigest()
 917    else:
 918        is_tmp = False
 919
 920    prefix = 'refs/bzr/%s' % alias
 921    gitdir = os.environ['GIT_DIR']
 922    dirname = os.path.join(gitdir, 'bzr', alias)
 923
 924    if not is_tmp:
 925        fix_path(alias, url)
 926
 927    if not os.path.exists(dirname):
 928        os.makedirs(dirname)
 929
 930    if hasattr(bzrlib.ui.ui_factory, 'be_quiet'):
 931        bzrlib.ui.ui_factory.be_quiet(True)
 932
 933    repo = get_repo(url, alias)
 934
 935    marks_path = os.path.join(dirname, 'marks-int')
 936    marks = Marks(marks_path)
 937
 938    parser = Parser(repo)
 939    for line in parser:
 940        if parser.check('capabilities'):
 941            do_capabilities(parser)
 942        elif parser.check('list'):
 943            do_list(parser)
 944        elif parser.check('import'):
 945            do_import(parser)
 946        elif parser.check('export'):
 947            do_export(parser)
 948        else:
 949            die('unhandled command: %s' % line)
 950        sys.stdout.flush()
 951
 952def bye():
 953    if not marks:
 954        return
 955    if not is_tmp:
 956        marks.store()
 957    else:
 958        shutil.rmtree(dirname)
 959
 960atexit.register(bye)
 961sys.exit(main(sys.argv))