Andrew's git - gitweb.git/blob

2bae5d034159b7d2c85e82d7bd5269257077a92a
   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
  20bzrlib.initialize()
  21
  22import bzrlib.plugin
  23bzrlib.plugin.load_plugins()
  24
  25import bzrlib.generate_ids
  26
  27import sys
  28import os
  29import json
  30import re
  31import StringIO
  32
  33NAME_RE = re.compile('^([^<>]+)')
  34AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
  35RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
  36
  37def die(msg, *args):
  38    sys.stderr.write('ERROR: %s\n' % (msg % args))
  39    sys.exit(1)
  40
  41def warn(msg, *args):
  42    sys.stderr.write('WARNING: %s\n' % (msg % args))
  43
  44def gittz(tz):
  45    return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
  46
  47class Marks:
  48
  49    def __init__(self, path):
  50        self.path = path
  51        self.tips = {}
  52        self.marks = {}
  53        self.rev_marks = {}
  54        self.last_mark = 0
  55        self.load()
  56
  57    def load(self):
  58        if not os.path.exists(self.path):
  59            return
  60
  61        tmp = json.load(open(self.path))
  62        self.tips = tmp['tips']
  63        self.marks = tmp['marks']
  64        self.last_mark = tmp['last-mark']
  65
  66        for rev, mark in self.marks.iteritems():
  67            self.rev_marks[mark] = rev
  68
  69    def dict(self):
  70        return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
  71
  72    def store(self):
  73        json.dump(self.dict(), open(self.path, 'w'))
  74
  75    def __str__(self):
  76        return str(self.dict())
  77
  78    def from_rev(self, rev):
  79        return self.marks[rev]
  80
  81    def to_rev(self, mark):
  82        return self.rev_marks[mark]
  83
  84    def next_mark(self):
  85        self.last_mark += 1
  86        return self.last_mark
  87
  88    def get_mark(self, rev):
  89        self.last_mark += 1
  90        self.marks[rev] = self.last_mark
  91        return self.last_mark
  92
  93    def is_marked(self, rev):
  94        return self.marks.has_key(rev)
  95
  96    def new_mark(self, rev, mark):
  97        self.marks[rev] = mark
  98        self.rev_marks[mark] = rev
  99        self.last_mark = mark
 100
 101    def get_tip(self, branch):
 102        return self.tips.get(branch, None)
 103
 104    def set_tip(self, branch, tip):
 105        self.tips[branch] = tip
 106
 107class Parser:
 108
 109    def __init__(self, repo):
 110        self.repo = repo
 111        self.line = self.get_line()
 112
 113    def get_line(self):
 114        return sys.stdin.readline().strip()
 115
 116    def __getitem__(self, i):
 117        return self.line.split()[i]
 118
 119    def check(self, word):
 120        return self.line.startswith(word)
 121
 122    def each_block(self, separator):
 123        while self.line != separator:
 124            yield self.line
 125            self.line = self.get_line()
 126
 127    def __iter__(self):
 128        return self.each_block('')
 129
 130    def next(self):
 131        self.line = self.get_line()
 132        if self.line == 'done':
 133            self.line = None
 134
 135    def get_mark(self):
 136        i = self.line.index(':') + 1
 137        return int(self.line[i:])
 138
 139    def get_data(self):
 140        if not self.check('data'):
 141            return None
 142        i = self.line.index(' ') + 1
 143        size = int(self.line[i:])
 144        return sys.stdin.read(size)
 145
 146    def get_author(self):
 147        m = RAW_AUTHOR_RE.match(self.line)
 148        if not m:
 149            return None
 150        _, name, email, date, tz = m.groups()
 151        committer = '%s <%s>' % (name, email)
 152        tz = int(tz)
 153        tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
 154        return (committer, int(date), tz)
 155
 156def rev_to_mark(rev):
 157    global marks
 158    return marks.from_rev(rev)
 159
 160def mark_to_rev(mark):
 161    global marks
 162    return marks.to_rev(mark)
 163
 164def fixup_user(user):
 165    name = mail = None
 166    user = user.replace('"', '')
 167    m = AUTHOR_RE.match(user)
 168    if m:
 169        name = m.group(1)
 170        mail = m.group(2).strip()
 171    else:
 172        m = NAME_RE.match(user)
 173        if m:
 174            name = m.group(1).strip()
 175
 176    return '%s <%s>' % (name, mail)
 177
 178def get_filechanges(cur, prev):
 179    modified = {}
 180    removed = {}
 181
 182    changes = cur.changes_from(prev)
 183
 184    for path, fid, kind in changes.added:
 185        modified[path] = fid
 186    for path, fid, kind in changes.removed:
 187        removed[path] = None
 188    for path, fid, kind, mod, _ in changes.modified:
 189        modified[path] = fid
 190    for oldpath, newpath, fid, kind, mod, _ in changes.renamed:
 191        removed[oldpath] = None
 192        modified[newpath] = fid
 193
 194    return modified, removed
 195
 196def export_files(tree, files):
 197    global marks, filenodes
 198
 199    final = []
 200    for path, fid in files.iteritems():
 201        kind = tree.kind(fid)
 202
 203        h = tree.get_file_sha1(fid)
 204
 205        if kind == 'symlink':
 206            d = tree.get_symlink_target(fid)
 207            mode = '120000'
 208        elif kind == 'file':
 209
 210            if tree.is_executable(fid):
 211                mode = '100755'
 212            else:
 213                mode = '100644'
 214
 215            # is the blog already exported?
 216            if h in filenodes:
 217                mark = filenodes[h]
 218                final.append((mode, mark, path))
 219                continue
 220
 221            d = tree.get_file_text(fid)
 222        elif kind == 'directory':
 223            continue
 224        else:
 225            die("Unhandled kind '%s' for path '%s'" % (kind, path))
 226
 227        mark = marks.next_mark()
 228        filenodes[h] = mark
 229
 230        print "blob"
 231        print "mark :%u" % mark
 232        print "data %d" % len(d)
 233        print d
 234
 235        final.append((mode, mark, path))
 236
 237    return final
 238
 239def export_branch(branch, name):
 240    global prefix, dirname
 241
 242    ref = '%s/heads/%s' % (prefix, name)
 243    tip = marks.get_tip(name)
 244
 245    repo = branch.repository
 246    repo.lock_read()
 247    revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
 248    count = 0
 249
 250    revs = [revid for revid, _, _, _ in revs if not marks.is_marked(revid)]
 251
 252    for revid in revs:
 253
 254        rev = repo.get_revision(revid)
 255
 256        parents = rev.parent_ids
 257        time = rev.timestamp
 258        tz = rev.timezone
 259        committer = rev.committer.encode('utf-8')
 260        committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
 261        author = committer
 262        msg = rev.message.encode('utf-8')
 263
 264        msg += '\n'
 265
 266        if len(parents) == 0:
 267            parent = bzrlib.revision.NULL_REVISION
 268        else:
 269            parent = parents[0]
 270
 271        cur_tree = repo.revision_tree(revid)
 272        prev = repo.revision_tree(parent)
 273        modified, removed = get_filechanges(cur_tree, prev)
 274
 275        modified_final = export_files(cur_tree, modified)
 276
 277        if len(parents) == 0:
 278            print 'reset %s' % ref
 279
 280        print "commit %s" % ref
 281        print "mark :%d" % (marks.get_mark(revid))
 282        print "author %s" % (author)
 283        print "committer %s" % (committer)
 284        print "data %d" % (len(msg))
 285        print msg
 286
 287        for i, p in enumerate(parents):
 288            try:
 289                m = rev_to_mark(p)
 290            except KeyError:
 291                # ghost?
 292                continue
 293            if i == 0:
 294                print "from :%s" % m
 295            else:
 296                print "merge :%s" % m
 297
 298        for f in modified_final:
 299            print "M %s :%u %s" % f
 300        for f in removed:
 301            print "D %s" % (f)
 302        print
 303
 304        count += 1
 305        if (count % 100 == 0):
 306            print "progress revision %s (%d/%d)" % (revid, count, len(revs))
 307            print "#############################################################"
 308
 309    repo.unlock()
 310
 311    revid = branch.last_revision()
 312
 313    # make sure the ref is updated
 314    print "reset %s" % ref
 315    print "from :%u" % rev_to_mark(revid)
 316    print
 317
 318    marks.set_tip(name, revid)
 319
 320def export_tag(repo, name):
 321    global tags
 322    try:
 323        print "reset refs/tags/%s" % name
 324        print "from :%u" % rev_to_mark(tags[name])
 325        print
 326    except KeyError:
 327        warn("TODO: fetch tag '%s'" % name)
 328
 329def do_import(parser):
 330    global dirname
 331
 332    branch = parser.repo
 333    path = os.path.join(dirname, 'marks-git')
 334
 335    print "feature done"
 336    if os.path.exists(path):
 337        print "feature import-marks=%s" % path
 338    print "feature export-marks=%s" % path
 339    sys.stdout.flush()
 340
 341    while parser.check('import'):
 342        ref = parser[1]
 343        if ref.startswith('refs/heads/'):
 344            name = ref[len('refs/heads/'):]
 345            export_branch(branch, name)
 346        if ref.startswith('refs/tags/'):
 347            name = ref[len('refs/tags/'):]
 348            export_tag(branch, name)
 349        parser.next()
 350
 351    print 'done'
 352
 353    sys.stdout.flush()
 354
 355def parse_blob(parser):
 356    global blob_marks
 357
 358    parser.next()
 359    mark = parser.get_mark()
 360    parser.next()
 361    data = parser.get_data()
 362    blob_marks[mark] = data
 363    parser.next()
 364
 365class CustomTree():
 366
 367    def __init__(self, repo, revid, parents, files):
 368        global files_cache
 369
 370        self.repo = repo
 371        self.revid = revid
 372        self.parents = parents
 373        self.updates = files
 374
 375        def copy_tree(revid):
 376            files = files_cache[revid] = {}
 377            tree = repo.repository.revision_tree(revid)
 378            repo.lock_read()
 379            try:
 380                for path, entry in tree.iter_entries_by_dir():
 381                    files[path] = entry.file_id
 382            finally:
 383                repo.unlock()
 384            return files
 385
 386        if len(parents) == 0:
 387            self.base_id = bzrlib.revision.NULL_REVISION
 388            self.base_files = {}
 389        else:
 390            self.base_id = parents[0]
 391            self.base_files = files_cache.get(self.base_id, None)
 392            if not self.base_files:
 393                self.base_files = copy_tree(self.base_id)
 394
 395        self.files = files_cache[revid] = self.base_files.copy()
 396
 397    def last_revision(self):
 398        return self.base_id
 399
 400    def iter_changes(self):
 401        changes = []
 402
 403        def get_parent(dirname, basename):
 404            parent_fid = self.base_files.get(dirname, None)
 405            if parent_fid:
 406                return parent_fid
 407            parent_fid = self.files.get(dirname, None)
 408            if parent_fid:
 409                return parent_fid
 410            if basename == '':
 411                return None
 412            d = add_entry(dirname, 'directory')
 413            return d[0]
 414
 415        def add_entry(path, kind):
 416            dirname, basename = os.path.split(path)
 417            parent_fid = get_parent(dirname, basename)
 418            fid = bzrlib.generate_ids.gen_file_id(path)
 419            change = (fid,
 420                    (None, path),
 421                    True,
 422                    (False, True),
 423                    (None, parent_fid),
 424                    (None, basename),
 425                    (None, kind),
 426                    (None, False))
 427            self.files[path] = change[0]
 428            changes.append(change)
 429            return change
 430
 431        def update_entry(path, kind):
 432            dirname, basename = os.path.split(path)
 433            fid = self.base_files[path]
 434            parent_fid = get_parent(dirname, basename)
 435            change = (fid,
 436                    (path, path),
 437                    True,
 438                    (True, True),
 439                    (None, parent_fid),
 440                    (None, basename),
 441                    (None, kind),
 442                    (None, False))
 443            self.files[path] = change[0]
 444            changes.append(change)
 445            return change
 446
 447        def remove_entry(path, kind):
 448            dirname, basename = os.path.split(path)
 449            fid = self.base_files[path]
 450            parent_fid = get_parent(dirname, basename)
 451            change = (fid,
 452                    (path, None),
 453                    True,
 454                    (True, False),
 455                    (parent_fid, None),
 456                    (None, None),
 457                    (None, None),
 458                    (None, None))
 459            del self.files[path]
 460            changes.append(change)
 461            return change
 462
 463        for path, f in self.updates.iteritems():
 464            if 'deleted' in f:
 465                remove_entry(path, 'file')
 466            elif path in self.base_files:
 467                update_entry(path, 'file')
 468            else:
 469                add_entry(path, 'file')
 470
 471        return changes
 472
 473    def get_file_with_stat(self, file_id, path=None):
 474        return (StringIO.StringIO(self.updates[path]['data']), None)
 475
 476def parse_commit(parser):
 477    global marks, blob_marks, bmarks, parsed_refs
 478    global mode
 479
 480    parents = []
 481
 482    ref = parser[1]
 483    parser.next()
 484
 485    if ref != 'refs/heads/master':
 486        die("bzr doesn't support multiple branches; use 'master'")
 487
 488    commit_mark = parser.get_mark()
 489    parser.next()
 490    author = parser.get_author()
 491    parser.next()
 492    committer = parser.get_author()
 493    parser.next()
 494    data = parser.get_data()
 495    parser.next()
 496    if parser.check('from'):
 497        parents.append(parser.get_mark())
 498        parser.next()
 499    while parser.check('merge'):
 500        parents.append(parser.get_mark())
 501        parser.next()
 502
 503    files = {}
 504
 505    for line in parser:
 506        if parser.check('M'):
 507            t, m, mark_ref, path = line.split(' ', 3)
 508            mark = int(mark_ref[1:])
 509            f = { 'mode' : m, 'data' : blob_marks[mark] }
 510        elif parser.check('D'):
 511            t, path = line.split(' ')
 512            f = { 'deleted' : True }
 513        else:
 514            die('Unknown file command: %s' % line)
 515        files[path] = f
 516
 517    repo = parser.repo
 518
 519    committer, date, tz = committer
 520    parents = [str(mark_to_rev(p)) for p in parents]
 521    revid = bzrlib.generate_ids.gen_revision_id(committer, date)
 522    props = {}
 523    props['branch-nick'] = repo.nick
 524
 525    mtree = CustomTree(repo, revid, parents, files)
 526    changes = mtree.iter_changes()
 527
 528    repo.lock_write()
 529    try:
 530        builder = repo.get_commit_builder(parents, None, date, tz, committer, props, revid, False)
 531        try:
 532            list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
 533            builder.finish_inventory()
 534            builder.commit(data.decode('utf-8', 'replace'))
 535        except Exception, e:
 536            builder.abort()
 537            raise
 538    finally:
 539        repo.unlock()
 540
 541    parsed_refs[ref] = revid
 542    marks.new_mark(revid, commit_mark)
 543
 544def parse_reset(parser):
 545    global parsed_refs
 546
 547    ref = parser[1]
 548    parser.next()
 549
 550    if ref != 'refs/heads/master':
 551        die("bzr doesn't support multiple branches; use 'master'")
 552
 553    # ugh
 554    if parser.check('commit'):
 555        parse_commit(parser)
 556        return
 557    if not parser.check('from'):
 558        return
 559    from_mark = parser.get_mark()
 560    parser.next()
 561
 562    parsed_refs[ref] = mark_to_rev(from_mark)
 563
 564def do_export(parser):
 565    global parsed_refs, dirname, peer
 566
 567    parser.next()
 568
 569    for line in parser.each_block('done'):
 570        if parser.check('blob'):
 571            parse_blob(parser)
 572        elif parser.check('commit'):
 573            parse_commit(parser)
 574        elif parser.check('reset'):
 575            parse_reset(parser)
 576        elif parser.check('tag'):
 577            pass
 578        elif parser.check('feature'):
 579            pass
 580        else:
 581            die('unhandled export command: %s' % line)
 582
 583    repo = parser.repo
 584
 585    for ref, revid in parsed_refs.iteritems():
 586        if ref == 'refs/heads/master':
 587            repo.generate_revision_history(revid, marks.get_tip('master'))
 588            revno, revid = repo.last_revision_info()
 589            peer.import_last_revision_info_and_tags(repo, revno, revid)
 590            wt = peer.bzrdir.open_workingtree()
 591            wt.update()
 592        print "ok %s" % ref
 593    print
 594
 595def do_capabilities(parser):
 596    global dirname
 597
 598    print "import"
 599    print "export"
 600    print "refspec refs/heads/*:%s/heads/*" % prefix
 601
 602    path = os.path.join(dirname, 'marks-git')
 603
 604    if os.path.exists(path):
 605        print "*import-marks %s" % path
 606    print "*export-marks %s" % path
 607
 608    print
 609
 610def do_list(parser):
 611    global tags
 612    print "? refs/heads/%s" % 'master'
 613    for tag, revid in parser.repo.tags.get_tag_dict().items():
 614        print "? refs/tags/%s" % tag
 615        tags[tag] = revid
 616    print "@refs/heads/%s HEAD" % 'master'
 617    print
 618
 619def get_repo(url, alias):
 620    global dirname, peer
 621
 622    clone_path = os.path.join(dirname, 'clone')
 623    origin = bzrlib.controldir.ControlDir.open(url)
 624    remote_branch = origin.open_branch()
 625
 626    if os.path.exists(clone_path):
 627        # pull
 628        d = bzrlib.controldir.ControlDir.open(clone_path)
 629        branch = d.open_branch()
 630        result = branch.pull(remote_branch, [], None, False)
 631    else:
 632        # clone
 633        d = origin.sprout(clone_path, None,
 634                hardlink=True, create_tree_if_local=False,
 635                source_branch=remote_branch)
 636        branch = d.open_branch()
 637        branch.bind(remote_branch)
 638
 639    peer = remote_branch
 640
 641    return branch
 642
 643def main(args):
 644    global marks, prefix, dirname
 645    global tags, filenodes
 646    global blob_marks
 647    global parsed_refs
 648    global files_cache
 649
 650    alias = args[1]
 651    url = args[2]
 652
 653    prefix = 'refs/bzr/%s' % alias
 654    tags = {}
 655    filenodes = {}
 656    blob_marks = {}
 657    parsed_refs = {}
 658    files_cache = {}
 659
 660    gitdir = os.environ['GIT_DIR']
 661    dirname = os.path.join(gitdir, 'bzr', alias)
 662
 663    if not os.path.exists(dirname):
 664        os.makedirs(dirname)
 665
 666    repo = get_repo(url, alias)
 667
 668    marks_path = os.path.join(dirname, 'marks-int')
 669    marks = Marks(marks_path)
 670
 671    parser = Parser(repo)
 672    for line in parser:
 673        if parser.check('capabilities'):
 674            do_capabilities(parser)
 675        elif parser.check('list'):
 676            do_list(parser)
 677        elif parser.check('import'):
 678            do_import(parser)
 679        elif parser.check('export'):
 680            do_export(parser)
 681        else:
 682            die('unhandled command: %s' % line)
 683        sys.stdout.flush()
 684
 685    marks.store()
 686
 687sys.exit(main(sys.argv))