contrib/remote-helpers/git-remote-hg

   1 #!/usr/bin/env python
   2 #
   3 # Copyright (c) 2012 Felipe Contreras
   4 #
   5
   6 # Inspired by Rocco Rutte's hg-fast-export
   7
   8 # Just copy to your ~/bin, or anywhere in your $PATH.
   9 # Then you can clone with:
  10 # git clone hg::/path/to/mercurial/repo/
  11 #
  12 # For remote repositories a local clone is stored in
  13 # "$GIT_DIR/hg/origin/clone/.hg/".
  14
  15 from mercurial import hg, ui, bookmarks, context, encoding, node, error, extensions, discovery, util
  16
  17 import re
  18 import sys
  19 import os
  20 import json
  21 import shutil
  22 import subprocess
  23 import urllib
  24 import atexit
  25 import urlparse, hashlib
  26 import time as ptime
  27
  28 #
  29 # If you want to see Mercurial revisions as Git commit notes:
  30 # git config core.notesRef refs/notes/hg
  31 #
  32 # If you are not in hg-git-compat mode and want to disable the tracking of
  33 # named branches:
  34 # git config --global remote-hg.track-branches false
  35 #
  36 # If you want the equivalent of hg's clone/pull--insecure option:
  37 # git config --global remote-hg.insecure true
  38 #
  39 # If you want to switch to hg-git compatibility mode:
  40 # git config --global remote-hg.hg-git-compat true
  41 #
  42 # git:
  43 # Sensible defaults for git.
  44 # hg bookmarks are exported as git branches, hg branches are prefixed
  45 # with 'branches/', HEAD is a special case.
  46 #
  47 # hg:
  48 # Emulate hg-git.
  49 # Only hg bookmarks are exported as git branches.
  50 # Commits are modified to preserve hg information and allow bidirectionality.
  51 #
  52
  53 NAME_RE = re.compile('^([^<>]+)')
  54 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
  55 EMAIL_RE = re.compile('^([^<>]+[^ \\\t<>])?\\b(?:[ \\t<>]*?)\\b([^ \\t<>]+@[^ \\t<>]+)')
  56 AUTHOR_HG_RE = re.compile('^(.*?) ?<(.*?)(?:>(.+)?)?$')
  57 RAW_AUTHOR_RE = re.compile('^(\w+) (?:(.+)? )?<(.*)> (\d+) ([+-]\d+)')
  58
  59 VERSION = 2
  60
  61 def die(msg, *args):
  62     sys.stderr.write('ERROR: %s\n' % (msg % args))
  63     sys.exit(1)
  64
  65 def warn(msg, *args):
  66     sys.stderr.write('WARNING: %s\n' % (msg % args))
  67
  68 def gitmode(flags):
  69     return 'l' in flags and '120000' or 'x' in flags and '100755' or '100644'
  70
  71 def gittz(tz):
  72     return '%+03d%02d' % (-tz / 3600, -tz % 3600 / 60)
  73
  74 def hgmode(mode):
  75     m = { '100755': 'x', '120000': 'l' }
  76     return m.get(mode, '')
  77
  78 def hghex(n):
  79     return node.hex(n)
  80
  81 def hgbin(n):
  82     return node.bin(n)
  83
  84 def hgref(ref):
  85     return ref.replace('___', ' ')
  86
  87 def gitref(ref):
  88     return ref.replace(' ', '___')
  89
  90 def check_version(*check):
  91     if not hg_version:
  92         return True
  93     return hg_version >= check
  94
  95 def get_config(config):
  96     cmd = ['git', 'config', '--get', config]
  97     process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
  98     output, _ = process.communicate()
  99     return output
 100
 101 def get_config_bool(config, default=False):
 102     value = get_config(config).rstrip('\n')
 103     if value == "true":
 104         return True
 105     elif value == "false":
 106         return False
 107     else:
 108         return default
 109
 110 class Marks:
 111
 112     def __init__(self, path, repo):
 113         self.path = path
 114         self.repo = repo
 115         self.clear()
 116         self.load()
 117
 118         if self.version < VERSION:
 119             if self.version == 1:
 120                 self.upgrade_one()
 121
 122             # upgraded?
 123             if self.version < VERSION:
 124                 self.clear()
 125                 self.version = VERSION
 126
 127     def clear(self):
 128         self.tips = {}
 129         self.marks = {}
 130         self.rev_marks = {}
 131         self.last_mark = 0
 132         self.version = 0
 133         self.last_note = 0
 134
 135     def load(self):
 136         if not os.path.exists(self.path):
 137             return
 138
 139         tmp = json.load(open(self.path))
 140
 141         self.tips = tmp['tips']
 142         self.marks = tmp['marks']
 143         self.last_mark = tmp['last-mark']
 144         self.version = tmp.get('version', 1)
 145         self.last_note = tmp.get('last-note', 0)
 146
 147         for rev, mark in self.marks.iteritems():
 148             self.rev_marks[mark] = rev
 149
 150     def upgrade_one(self):
 151         def get_id(rev):
 152             return hghex(self.repo.changelog.node(int(rev)))
 153         self.tips = dict((name, get_id(rev)) for name, rev in self.tips.iteritems())
 154         self.marks = dict((get_id(rev), mark) for rev, mark in self.marks.iteritems())
 155         self.rev_marks = dict((mark, get_id(rev)) for mark, rev in self.rev_marks.iteritems())
 156         self.version = 2
 157
 158     def dict(self):
 159         return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark, 'version' : self.version, 'last-note' : self.last_note }
 160
 161     def store(self):
 162         json.dump(self.dict(), open(self.path, 'w'))
 163
 164     def __str__(self):
 165         return str(self.dict())
 166
 167     def from_rev(self, rev):
 168         return self.marks[rev]
 169
 170     def to_rev(self, mark):
 171         return str(self.rev_marks[mark])
 172
 173     def next_mark(self):
 174         self.last_mark += 1
 175         return self.last_mark
 176
 177     def get_mark(self, rev):
 178         self.last_mark += 1
 179         self.marks[rev] = self.last_mark
 180         return self.last_mark
 181
 182     def new_mark(self, rev, mark):
 183         self.marks[rev] = mark
 184         self.rev_marks[mark] = rev
 185         self.last_mark = mark
 186
 187     def is_marked(self, rev):
 188         return rev in self.marks
 189
 190     def get_tip(self, branch):
 191         return str(self.tips[branch])
 192
 193     def set_tip(self, branch, tip):
 194         self.tips[branch] = tip
 195
 196 class Parser:
 197
 198     def __init__(self, repo):
 199         self.repo = repo
 200         self.line = self.get_line()
 201
 202     def get_line(self):
 203         return sys.stdin.readline().strip()
 204
 205     def __getitem__(self, i):
 206         return self.line.split()[i]
 207
 208     def check(self, word):
 209         return self.line.startswith(word)
 210
 211     def each_block(self, separator):
 212         while self.line != separator:
 213             yield self.line
 214             self.line = self.get_line()
 215
 216     def __iter__(self):
 217         return self.each_block('')
 218
 219     def next(self):
 220         self.line = self.get_line()
 221         if self.line == 'done':
 222             self.line = None
 223
 224     def get_mark(self):
 225         i = self.line.index(':') + 1
 226         return int(self.line[i:])
 227
 228     def get_data(self):
 229         if not self.check('data'):
 230             return None
 231         i = self.line.index(' ') + 1
 232         size = int(self.line[i:])
 233         return sys.stdin.read(size)
 234
 235     def get_author(self):
 236         ex = None
 237         m = RAW_AUTHOR_RE.match(self.line)
 238         if not m:
 239             return None
 240         _, name, email, date, tz = m.groups()
 241         if name and 'ext:' in name:
 242             m = re.match('^(.+?) ext:\((.+)\)$', name)
 243             if m:
 244                 name = m.group(1)
 245                 ex = urllib.unquote(m.group(2))
 246
 247         if email != bad_mail:
 248             if name:
 249                 user = '%s <%s>' % (name, email)
 250             else:
 251                 user = '<%s>' % (email)
 252         else:
 253             user = name
 254
 255         if ex:
 256             user += ex
 257
 258         tz = int(tz)
 259         tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
 260         return (user, int(date), -tz)
 261
 262 def fix_file_path(path):
 263     if not os.path.isabs(path):
 264         return path
 265     return os.path.relpath(path, '/')
 266
 267 def export_files(files):
 268     final = []
 269     for f in files:
 270         fid = node.hex(f.filenode())
 271
 272         if fid in filenodes:
 273             mark = filenodes[fid]
 274         else:
 275             mark = marks.next_mark()
 276             filenodes[fid] = mark
 277             d = f.data()
 278
 279             print "blob"
 280             print "mark :%u" % mark
 281             print "data %d" % len(d)
 282             print d
 283
 284         path = fix_file_path(f.path())
 285         final.append((gitmode(f.flags()), mark, path))
 286
 287     return final
 288
 289 def get_filechanges(repo, ctx, parent):
 290     modified = set()
 291     added = set()
 292     removed = set()
 293
 294     # load earliest manifest first for caching reasons
 295     prev = parent.manifest().copy()
 296     cur = ctx.manifest()
 297
 298     for fn in cur:
 299         if fn in prev:
 300             if (cur.flags(fn) != prev.flags(fn) or cur[fn] != prev[fn]):
 301                 modified.add(fn)
 302             del prev[fn]
 303         else:
 304             added.add(fn)
 305     removed |= set(prev.keys())
 306
 307     return added | modified, removed
 308
 309 def fixup_user_git(user):
 310     name = mail = None
 311     user = user.replace('"', '')
 312     m = AUTHOR_RE.match(user)
 313     if m:
 314         name = m.group(1)
 315         mail = m.group(2).strip()
 316     else:
 317         m = EMAIL_RE.match(user)
 318         if m:
 319             name = m.group(1)
 320             mail = m.group(2)
 321         else:
 322             m = NAME_RE.match(user)
 323             if m:
 324                 name = m.group(1).strip()
 325     return (name, mail)
 326
 327 def fixup_user_hg(user):
 328     def sanitize(name):
 329         # stole this from hg-git
 330         return re.sub('[<>\n]', '?', name.lstrip('< ').rstrip('> '))
 331
 332     m = AUTHOR_HG_RE.match(user)
 333     if m:
 334         name = sanitize(m.group(1))
 335         mail = sanitize(m.group(2))
 336         ex = m.group(3)
 337         if ex:
 338             name += ' ext:(' + urllib.quote(ex) + ')'
 339     else:
 340         name = sanitize(user)
 341         if '@' in user:
 342             mail = name
 343         else:
 344             mail = None
 345
 346     return (name, mail)
 347
 348 def fixup_user(user):
 349     if mode == 'git':
 350         name, mail = fixup_user_git(user)
 351     else:
 352         name, mail = fixup_user_hg(user)
 353
 354     if not name:
 355         name = bad_name
 356     if not mail:
 357         mail = bad_mail
 358
 359     return '%s <%s>' % (name, mail)
 360
 361 def updatebookmarks(repo, peer):
 362     remotemarks = peer.listkeys('bookmarks')
 363     localmarks = repo._bookmarks
 364
 365     if not remotemarks:
 366         return
 367
 368     for k, v in remotemarks.iteritems():
 369         localmarks[k] = hgbin(v)
 370
 371     if hasattr(localmarks, 'write'):
 372         localmarks.write()
 373     else:
 374         bookmarks.write(repo)
 375
 376 def get_repo(url, alias):
 377     global peer
 378
 379     myui = ui.ui()
 380     myui.setconfig('ui', 'interactive', 'off')
 381     myui.fout = sys.stderr
 382
 383     if get_config_bool('remote-hg.insecure'):
 384         myui.setconfig('web', 'cacerts', '')
 385
 386     extensions.loadall(myui)
 387
 388     if hg.islocal(url) and not os.environ.get('GIT_REMOTE_HG_TEST_REMOTE'):
 389         repo = hg.repository(myui, url)
 390         if not os.path.exists(dirname):
 391             os.makedirs(dirname)
 392     else:
 393         shared_path = os.path.join(gitdir, 'hg')
 394
 395         # check and upgrade old organization
 396         hg_path = os.path.join(shared_path, '.hg')
 397         if os.path.exists(shared_path) and not os.path.exists(hg_path):
 398             repos = os.listdir(shared_path)
 399             for x in repos:
 400                 local_hg = os.path.join(shared_path, x, 'clone', '.hg')
 401                 if not os.path.exists(local_hg):
 402                     continue
 403                 if not os.path.exists(hg_path):
 404                     shutil.move(local_hg, hg_path)
 405                 shutil.rmtree(os.path.join(shared_path, x, 'clone'))
 406
 407         # setup shared repo (if not there)
 408         try:
 409             hg.peer(myui, {}, shared_path, create=True)
 410         except error.RepoError:
 411             pass
 412
 413         if not os.path.exists(dirname):
 414             os.makedirs(dirname)
 415
 416         local_path = os.path.join(dirname, 'clone')
 417         if not os.path.exists(local_path):
 418             hg.share(myui, shared_path, local_path, update=False)
 419
 420         repo = hg.repository(myui, local_path)
 421         try:
 422             peer = hg.peer(myui, {}, url)
 423         except:
 424             die('Repository error')
 425         repo.pull(peer, heads=None, force=True)
 426
 427         updatebookmarks(repo, peer)
 428
 429     return repo
 430
 431 def rev_to_mark(rev):
 432     return marks.from_rev(rev.hex())
 433
 434 def mark_to_rev(mark):
 435     return marks.to_rev(mark)
 436
 437 def export_ref(repo, name, kind, head):
 438     ename = '%s/%s' % (kind, name)
 439     try:
 440         tip = marks.get_tip(ename)
 441         tip = repo[tip].rev()
 442     except:
 443         tip = 0
 444
 445     revs = xrange(tip, head.rev() + 1)
 446     total = len(revs)
 447
 448     for rev in revs:
 449
 450         c = repo[rev]
 451         node = c.node()
 452
 453         if marks.is_marked(c.hex()):
 454             continue
 455
 456         (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(node)
 457         rev_branch = extra['branch']
 458
 459         author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
 460         if 'committer' in extra:
 461             user, time, tz = extra['committer'].rsplit(' ', 2)
 462             committer = "%s %s %s" % (user, time, gittz(int(tz)))
 463         else:
 464             committer = author
 465
 466         parents = [repo[p] for p in repo.changelog.parentrevs(rev) if p >= 0]
 467
 468         if len(parents) == 0:
 469             modified = c.manifest().keys()
 470             removed = []
 471         else:
 472             modified, removed = get_filechanges(repo, c, parents[0])
 473
 474         desc += '\n'
 475
 476         if mode == 'hg':
 477             extra_msg = ''
 478
 479             if rev_branch != 'default':
 480                 extra_msg += 'branch : %s\n' % rev_branch
 481
 482             renames = []
 483             for f in c.files():
 484                 if f not in c.manifest():
 485                     continue
 486                 rename = c.filectx(f).renamed()
 487                 if rename:
 488                     renames.append((rename[0], f))
 489
 490             for e in renames:
 491                 extra_msg += "rename : %s => %s\n" % e
 492
 493             for key, value in extra.iteritems():
 494                 if key in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
 495                     continue
 496                 else:
 497                     extra_msg += "extra : %s : %s\n" % (key, urllib.quote(value))
 498
 499             if extra_msg:
 500                 desc += '\n--HG--\n' + extra_msg
 501
 502         if len(parents) == 0 and rev:
 503             print 'reset %s/%s' % (prefix, ename)
 504
 505         modified_final = export_files(c.filectx(f) for f in modified)
 506
 507         print "commit %s/%s" % (prefix, ename)
 508         print "mark :%d" % (marks.get_mark(c.hex()))
 509         print "author %s" % (author)
 510         print "committer %s" % (committer)
 511         print "data %d" % (len(desc))
 512         print desc
 513
 514         if len(parents) > 0:
 515             print "from :%s" % (rev_to_mark(parents[0]))
 516             if len(parents) > 1:
 517                 print "merge :%s" % (rev_to_mark(parents[1]))
 518
 519         for f in removed:
 520             print "D %s" % (fix_file_path(f))
 521         for f in modified_final:
 522             print "M %s :%u %s" % f
 523         print
 524
 525         progress = (rev - tip)
 526         if (progress % 100 == 0):
 527             print "progress revision %d '%s' (%d/%d)" % (rev, name, progress, total)
 528
 529     # make sure the ref is updated
 530     print "reset %s/%s" % (prefix, ename)
 531     print "from :%u" % rev_to_mark(head)
 532     print
 533
 534     pending_revs = set(revs) - notes
 535     if pending_revs:
 536         note_mark = marks.next_mark()
 537         ref = "refs/notes/hg"
 538
 539         print "commit %s" % ref
 540         print "mark :%d" % (note_mark)
 541         print "committer remote-hg <> %s" % (ptime.strftime('%s %z'))
 542         desc = "Notes for %s\n" % (name)
 543         print "data %d" % (len(desc))
 544         print desc
 545         if marks.last_note:
 546             print "from :%u" % marks.last_note
 547
 548         for rev in pending_revs:
 549             notes.add(rev)
 550             c = repo[rev]
 551             print "N inline :%u" % rev_to_mark(c)
 552             msg = c.hex()
 553             print "data %d" % (len(msg))
 554             print msg
 555         print
 556
 557         marks.last_note = note_mark
 558
 559     marks.set_tip(ename, head.hex())
 560
 561 def export_tag(repo, tag):
 562     export_ref(repo, tag, 'tags', repo[hgref(tag)])
 563
 564 def export_bookmark(repo, bmark):
 565     head = bmarks[hgref(bmark)]
 566     export_ref(repo, bmark, 'bookmarks', head)
 567
 568 def export_branch(repo, branch):
 569     tip = get_branch_tip(repo, branch)
 570     head = repo[tip]
 571     export_ref(repo, branch, 'branches', head)
 572
 573 def export_head(repo):
 574     export_ref(repo, g_head[0], 'bookmarks', g_head[1])
 575
 576 def do_capabilities(parser):
 577     print "import"
 578     print "export"
 579     print "refspec refs/heads/branches/*:%s/branches/*" % prefix
 580     print "refspec refs/heads/*:%s/bookmarks/*" % prefix
 581     print "refspec refs/tags/*:%s/tags/*" % prefix
 582
 583     path = os.path.join(dirname, 'marks-git')
 584
 585     if os.path.exists(path):
 586         print "*import-marks %s" % path
 587     print "*export-marks %s" % path
 588     print "option"
 589
 590     print
 591
 592 def branch_tip(branch):
 593     return branches[branch][-1]
 594
 595 def get_branch_tip(repo, branch):
 596     heads = branches.get(hgref(branch), None)
 597     if not heads:
 598         return None
 599
 600     # verify there's only one head
 601     if (len(heads) > 1):
 602         warn("Branch '%s' has more than one head, consider merging" % branch)
 603         return branch_tip(hgref(branch))
 604
 605     return heads[0]
 606
 607 def list_head(repo, cur):
 608     global g_head, fake_bmark
 609
 610     if 'default' not in branches:
 611         # empty repo
 612         return
 613
 614     node = repo[branch_tip('default')]
 615     head = 'master' if not 'master' in bmarks else 'default'
 616     fake_bmark = head
 617     bmarks[head] = node
 618
 619     head = gitref(head)
 620     print "@refs/heads/%s HEAD" % head
 621     g_head = (head, node)
 622
 623 def do_list(parser):
 624     repo = parser.repo
 625     for bmark, node in bookmarks.listbookmarks(repo).iteritems():
 626         bmarks[bmark] = repo[node]
 627
 628     cur = repo.dirstate.branch()
 629     orig = peer if peer else repo
 630
 631     for branch, heads in orig.branchmap().iteritems():
 632         # only open heads
 633         heads = [h for h in heads if 'close' not in repo.changelog.read(h)[5]]
 634         if heads:
 635             branches[branch] = heads
 636
 637     list_head(repo, cur)
 638
 639     if track_branches:
 640         for branch in branches:
 641             print "? refs/heads/branches/%s" % gitref(branch)
 642
 643     for bmark in bmarks:
 644         if  bmarks[bmark].hex() == '0000000000000000000000000000000000000000':
 645             warn("Ignoring invalid bookmark '%s'", bmark)
 646         else:
 647             print "? refs/heads/%s" % gitref(bmark)
 648
 649     for tag, node in repo.tagslist():
 650         if tag == 'tip':
 651             continue
 652         print "? refs/tags/%s" % gitref(tag)
 653
 654     print
 655
 656 def do_import(parser):
 657     repo = parser.repo
 658
 659     path = os.path.join(dirname, 'marks-git')
 660
 661     print "feature done"
 662     if os.path.exists(path):
 663         print "feature import-marks=%s" % path
 664     print "feature export-marks=%s" % path
 665     print "feature force"
 666     sys.stdout.flush()
 667
 668     tmp = encoding.encoding
 669     encoding.encoding = 'utf-8'
 670
 671     # lets get all the import lines
 672     while parser.check('import'):
 673         ref = parser[1]
 674
 675         if (ref == 'HEAD'):
 676             export_head(repo)
 677         elif ref.startswith('refs/heads/branches/'):
 678             branch = ref[len('refs/heads/branches/'):]
 679             export_branch(repo, branch)
 680         elif ref.startswith('refs/heads/'):
 681             bmark = ref[len('refs/heads/'):]
 682             export_bookmark(repo, bmark)
 683         elif ref.startswith('refs/tags/'):
 684             tag = ref[len('refs/tags/'):]
 685             export_tag(repo, tag)
 686
 687         parser.next()
 688
 689     encoding.encoding = tmp
 690
 691     print 'done'
 692
 693 def parse_blob(parser):
 694     parser.next()
 695     mark = parser.get_mark()
 696     parser.next()
 697     data = parser.get_data()
 698     blob_marks[mark] = data
 699     parser.next()
 700
 701 def get_merge_files(repo, p1, p2, files):
 702     for e in repo[p1].files():
 703         if e not in files:
 704             if e not in repo[p1].manifest():
 705                 continue
 706             f = { 'ctx' : repo[p1][e] }
 707             files[e] = f
 708
 709 def c_style_unescape(string):
 710     if string[0] == string[-1] == '"':
 711         return string.decode('string-escape')[1:-1]
 712     return string
 713
 714 def parse_commit(parser):
 715     from_mark = merge_mark = None
 716
 717     ref = parser[1]
 718     parser.next()
 719
 720     commit_mark = parser.get_mark()
 721     parser.next()
 722     author = parser.get_author()
 723     parser.next()
 724     committer = parser.get_author()
 725     parser.next()
 726     data = parser.get_data()
 727     parser.next()
 728     if parser.check('from'):
 729         from_mark = parser.get_mark()
 730         parser.next()
 731     if parser.check('merge'):
 732         merge_mark = parser.get_mark()
 733         parser.next()
 734         if parser.check('merge'):
 735             die('octopus merges are not supported yet')
 736
 737     # fast-export adds an extra newline
 738     if data[-1] == '\n':
 739         data = data[:-1]
 740
 741     files = {}
 742
 743     for line in parser:
 744         if parser.check('M'):
 745             t, m, mark_ref, path = line.split(' ', 3)
 746             mark = int(mark_ref[1:])
 747             f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
 748         elif parser.check('D'):
 749             t, path = line.split(' ', 1)
 750             f = { 'deleted' : True }
 751         else:
 752             die('Unknown file command: %s' % line)
 753         path = c_style_unescape(path)
 754         files[path] = f
 755
 756     # only export the commits if we are on an internal proxy repo
 757     if dry_run and not peer:
 758         parsed_refs[ref] = None
 759         return
 760
 761     def getfilectx(repo, memctx, f):
 762         of = files[f]
 763         if 'deleted' in of:
 764             raise IOError
 765         if 'ctx' in of:
 766             return of['ctx']
 767         is_exec = of['mode'] == 'x'
 768         is_link = of['mode'] == 'l'
 769         rename = of.get('rename', None)
 770         return context.memfilectx(f, of['data'],
 771                 is_link, is_exec, rename)
 772
 773     repo = parser.repo
 774
 775     user, date, tz = author
 776     extra = {}
 777
 778     if committer != author:
 779         extra['committer'] = "%s %u %u" % committer
 780
 781     if from_mark:
 782         p1 = mark_to_rev(from_mark)
 783     else:
 784         p1 = '0' * 40
 785
 786     if merge_mark:
 787         p2 = mark_to_rev(merge_mark)
 788     else:
 789         p2 = '0' * 40
 790
 791     #
 792     # If files changed from any of the parents, hg wants to know, but in git if
 793     # nothing changed from the first parent, nothing changed.
 794     #
 795     if merge_mark:
 796         get_merge_files(repo, p1, p2, files)
 797
 798     # Check if the ref is supposed to be a named branch
 799     if ref.startswith('refs/heads/branches/'):
 800         branch = ref[len('refs/heads/branches/'):]
 801         extra['branch'] = hgref(branch)
 802
 803     if mode == 'hg':
 804         i = data.find('\n--HG--\n')
 805         if i >= 0:
 806             tmp = data[i + len('\n--HG--\n'):].strip()
 807             for k, v in [e.split(' : ', 1) for e in tmp.split('\n')]:
 808                 if k == 'rename':
 809                     old, new = v.split(' => ', 1)
 810                     files[new]['rename'] = old
 811                 elif k == 'branch':
 812                     extra[k] = v
 813                 elif k == 'extra':
 814                     ek, ev = v.split(' : ', 1)
 815                     extra[ek] = urllib.unquote(ev)
 816             data = data[:i]
 817
 818     ctx = context.memctx(repo, (p1, p2), data,
 819             files.keys(), getfilectx,
 820             user, (date, tz), extra)
 821
 822     tmp = encoding.encoding
 823     encoding.encoding = 'utf-8'
 824
 825     node = hghex(repo.commitctx(ctx))
 826
 827     encoding.encoding = tmp
 828
 829     parsed_refs[ref] = node
 830     marks.new_mark(node, commit_mark)
 831
 832 def parse_reset(parser):
 833     ref = parser[1]
 834     parser.next()
 835     # ugh
 836     if parser.check('commit'):
 837         parse_commit(parser)
 838         return
 839     if not parser.check('from'):
 840         return
 841     from_mark = parser.get_mark()
 842     parser.next()
 843
 844     try:
 845         rev = mark_to_rev(from_mark)
 846     except KeyError:
 847         rev = None
 848     parsed_refs[ref] = rev
 849
 850 def parse_tag(parser):
 851     name = parser[1]
 852     parser.next()
 853     from_mark = parser.get_mark()
 854     parser.next()
 855     tagger = parser.get_author()
 856     parser.next()
 857     data = parser.get_data()
 858     parser.next()
 859
 860     parsed_tags[name] = (tagger, data)
 861
 862 def write_tag(repo, tag, node, msg, author):
 863     branch = repo[node].branch()
 864     tip = branch_tip(branch)
 865     tip = repo[tip]
 866
 867     def getfilectx(repo, memctx, f):
 868         try:
 869             fctx = tip.filectx(f)
 870             data = fctx.data()
 871         except error.ManifestLookupError:
 872             data = ""
 873         content = data + "%s %s\n" % (node, tag)
 874         return context.memfilectx(f, content, False, False, None)
 875
 876     p1 = tip.hex()
 877     p2 = '0' * 40
 878     if author:
 879         user, date, tz = author
 880         date_tz = (date, tz)
 881     else:
 882         cmd = ['git', 'var', 'GIT_COMMITTER_IDENT']
 883         process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
 884         output, _ = process.communicate()
 885         m = re.match('^.* <.*>', output)
 886         if m:
 887             user = m.group(0)
 888         else:
 889             user = repo.ui.username()
 890         date_tz = None
 891
 892     ctx = context.memctx(repo, (p1, p2), msg,
 893             ['.hgtags'], getfilectx,
 894             user, date_tz, {'branch' : branch})
 895
 896     tmp = encoding.encoding
 897     encoding.encoding = 'utf-8'
 898
 899     tagnode = repo.commitctx(ctx)
 900
 901     encoding.encoding = tmp
 902
 903     return (tagnode, branch)
 904
 905 def checkheads_bmark(repo, ref, ctx):
 906     bmark = ref[len('refs/heads/'):]
 907     if not bmark in bmarks:
 908         # new bmark
 909         return True
 910
 911     ctx_old = bmarks[bmark]
 912     ctx_new = ctx
 913     if not repo.changelog.descendant(ctx_old.rev(), ctx_new.rev()):
 914         if force_push:
 915             print "ok %s forced update" % ref
 916         else:
 917             print "error %s non-fast forward" % ref
 918             return False
 919
 920     return True
 921
 922 def checkheads(repo, remote, p_revs):
 923
 924     remotemap = remote.branchmap()
 925     if not remotemap:
 926         # empty repo
 927         return True
 928
 929     new = {}
 930     ret = True
 931
 932     for node, ref in p_revs.iteritems():
 933         ctx = repo[node]
 934         branch = ctx.branch()
 935         if not branch in remotemap:
 936             # new branch
 937             continue
 938         if not ref.startswith('refs/heads/branches'):
 939             if ref.startswith('refs/heads/'):
 940                 if not checkheads_bmark(repo, ref, ctx):
 941                     ret = False
 942
 943             # only check branches
 944             continue
 945         new.setdefault(branch, []).append(ctx.rev())
 946
 947     for branch, heads in new.iteritems():
 948         old = [repo.changelog.rev(x) for x in remotemap[branch]]
 949         for rev in heads:
 950             if check_version(2, 3):
 951                 ancestors = repo.changelog.ancestors([rev], stoprev=min(old))
 952             else:
 953                 ancestors = repo.changelog.ancestors(rev)
 954             found = False
 955
 956             for x in old:
 957                 if x in ancestors:
 958                     found = True
 959                     break
 960
 961             if found:
 962                 continue
 963
 964             node = repo.changelog.node(rev)
 965             ref = p_revs[node]
 966             if force_push:
 967                 print "ok %s forced update" % ref
 968             else:
 969                 print "error %s non-fast forward" % ref
 970                 ret = False
 971
 972     return ret
 973
 974 def push_unsafe(repo, remote, parsed_refs, p_revs):
 975
 976     force = force_push
 977
 978     fci = discovery.findcommonincoming
 979     commoninc = fci(repo, remote, force=force)
 980     common, _, remoteheads = commoninc
 981
 982     if not checkheads(repo, remote, p_revs):
 983         return None
 984
 985     cg = repo.getbundle('push', heads=list(p_revs), common=common)
 986
 987     unbundle = remote.capable('unbundle')
 988     if unbundle:
 989         if force:
 990             remoteheads = ['force']
 991         return remote.unbundle(cg, remoteheads, 'push')
 992     else:
 993         return remote.addchangegroup(cg, 'push', repo.url())
 994
 995 def push(repo, remote, parsed_refs, p_revs):
 996     if hasattr(remote, 'canpush') and not remote.canpush():
 997         print "error cannot push"
 998
 999     if not p_revs:
1000         # nothing to push
1001         return
1002
1003     lock = None
1004     unbundle = remote.capable('unbundle')
1005     if not unbundle:
1006         lock = remote.lock()
1007     try:
1008         ret = push_unsafe(repo, remote, parsed_refs, p_revs)
1009     finally:
1010         if lock is not None:
1011             lock.release()
1012
1013     return ret
1014
1015 def check_tip(ref, kind, name, heads):
1016     try:
1017         ename = '%s/%s' % (kind, name)
1018         tip = marks.get_tip(ename)
1019     except KeyError:
1020         return True
1021     else:
1022         return tip in heads
1023
1024 def do_export(parser):
1025     p_bmarks = []
1026     p_revs = {}
1027
1028     parser.next()
1029
1030     for line in parser.each_block('done'):
1031         if parser.check('blob'):
1032             parse_blob(parser)
1033         elif parser.check('commit'):
1034             parse_commit(parser)
1035         elif parser.check('reset'):
1036             parse_reset(parser)
1037         elif parser.check('tag'):
1038             parse_tag(parser)
1039         elif parser.check('feature'):
1040             pass
1041         else:
1042             die('unhandled export command: %s' % line)
1043
1044     need_fetch = False
1045
1046     for ref, node in parsed_refs.iteritems():
1047         bnode = hgbin(node) if node else None
1048         if ref.startswith('refs/heads/branches'):
1049             branch = ref[len('refs/heads/branches/'):]
1050             if branch in branches and bnode in branches[branch]:
1051                 # up to date
1052                 continue
1053
1054             if peer:
1055                 remotemap = peer.branchmap()
1056                 if remotemap and branch in remotemap:
1057                     heads = [hghex(e) for e in remotemap[branch]]
1058                     if not check_tip(ref, 'branches', branch, heads):
1059                         print "error %s fetch first" % ref
1060                         need_fetch = True
1061                         continue
1062
1063             p_revs[bnode] = ref
1064             print "ok %s" % ref
1065         elif ref.startswith('refs/heads/'):
1066             bmark = ref[len('refs/heads/'):]
1067             new = node
1068             old = bmarks[bmark].hex() if bmark in bmarks else ''
1069
1070             if old == new:
1071                 continue
1072
1073             print "ok %s" % ref
1074             if bmark != fake_bmark and \
1075                     not (bmark == 'master' and bmark not in parser.repo._bookmarks):
1076                 p_bmarks.append((ref, bmark, old, new))
1077
1078             if peer:
1079                 remote_old = peer.listkeys('bookmarks').get(bmark)
1080                 if remote_old:
1081                     if not check_tip(ref, 'bookmarks', bmark, remote_old):
1082                         print "error %s fetch first" % ref
1083                         need_fetch = True
1084                         continue
1085
1086             p_revs[bnode] = ref
1087         elif ref.startswith('refs/tags/'):
1088             if dry_run:
1089                 print "ok %s" % ref
1090                 continue
1091             tag = ref[len('refs/tags/'):]
1092             tag = hgref(tag)
1093             author, msg = parsed_tags.get(tag, (None, None))
1094             if mode == 'git':
1095                 if not msg:
1096                     msg = 'Added tag %s for changeset %s' % (tag, node[:12])
1097                 tagnode, branch = write_tag(parser.repo, tag, node, msg, author)
1098                 p_revs[tagnode] = 'refs/heads/branches/' + gitref(branch)
1099             else:
1100                 fp = parser.repo.opener('localtags', 'a')
1101                 fp.write('%s %s\n' % (node, tag))
1102                 fp.close()
1103             p_revs[bnode] = ref
1104             print "ok %s" % ref
1105         else:
1106             # transport-helper/fast-export bugs
1107             continue
1108
1109     if need_fetch:
1110         print
1111         return
1112
1113     if dry_run:
1114         if peer and not force_push:
1115             checkheads(parser.repo, peer, p_revs)
1116         print
1117         return
1118
1119     if peer:
1120         if not push(parser.repo, peer, parsed_refs, p_revs):
1121             # do not update bookmarks
1122             print
1123             return
1124
1125         # update remote bookmarks
1126         remote_bmarks = peer.listkeys('bookmarks')
1127         for ref, bmark, old, new in p_bmarks:
1128             if force_push:
1129                 old = remote_bmarks.get(bmark, '')
1130             if not peer.pushkey('bookmarks', bmark, old, new):
1131                 print "error %s" % ref
1132     else:
1133         # update local bookmarks
1134         for ref, bmark, old, new in p_bmarks:
1135             if not bookmarks.pushbookmark(parser.repo, bmark, old, new):
1136                 print "error %s" % ref
1137
1138     print
1139
1140 def do_option(parser):
1141     global dry_run, force_push
1142     _, key, value = parser.line.split(' ')
1143     if key == 'dry-run':
1144         dry_run = (value == 'true')
1145         print 'ok'
1146     elif key == 'force':
1147         force_push = (value == 'true')
1148         print 'ok'
1149     else:
1150         print 'unsupported'
1151
1152 def fix_path(alias, repo, orig_url):
1153     url = urlparse.urlparse(orig_url, 'file')
1154     if url.scheme != 'file' or os.path.isabs(os.path.expanduser(url.path)):
1155         return
1156     abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
1157     cmd = ['git', 'config', 'remote.%s.url' % alias, "hg::%s" % abs_url]
1158     subprocess.call(cmd)
1159
1160 def main(args):
1161     global prefix, gitdir, dirname, branches, bmarks
1162     global marks, blob_marks, parsed_refs
1163     global peer, mode, bad_mail, bad_name
1164     global track_branches, force_push, is_tmp
1165     global parsed_tags
1166     global filenodes
1167     global fake_bmark, hg_version
1168     global dry_run
1169     global notes, alias
1170
1171     alias = args[1]
1172     url = args[2]
1173     peer = None
1174
1175     hg_git_compat = get_config_bool('remote-hg.hg-git-compat')
1176     track_branches = get_config_bool('remote-hg.track-branches', True)
1177     force_push = False
1178
1179     if hg_git_compat:
1180         mode = 'hg'
1181         bad_mail = 'none@none'
1182         bad_name = ''
1183     else:
1184         mode = 'git'
1185         bad_mail = 'unknown'
1186         bad_name = 'Unknown'
1187
1188     if alias[4:] == url:
1189         is_tmp = True
1190         alias = hashlib.sha1(alias).hexdigest()
1191     else:
1192         is_tmp = False
1193
1194     gitdir = os.environ['GIT_DIR']
1195     dirname = os.path.join(gitdir, 'hg', alias)
1196     branches = {}
1197     bmarks = {}
1198     blob_marks = {}
1199     parsed_refs = {}
1200     marks = None
1201     parsed_tags = {}
1202     filenodes = {}
1203     fake_bmark = None
1204     try:
1205         hg_version = tuple(int(e) for e in util.version().split('.'))
1206     except:
1207         hg_version = None
1208     dry_run = False
1209     notes = set()
1210
1211     repo = get_repo(url, alias)
1212     prefix = 'refs/hg/%s' % alias
1213
1214     if not is_tmp:
1215         fix_path(alias, peer or repo, url)
1216
1217     marks_path = os.path.join(dirname, 'marks-hg')
1218     marks = Marks(marks_path, repo)
1219
1220     if sys.platform == 'win32':
1221         import msvcrt
1222         msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
1223
1224     parser = Parser(repo)
1225     for line in parser:
1226         if parser.check('capabilities'):
1227             do_capabilities(parser)
1228         elif parser.check('list'):
1229             do_list(parser)
1230         elif parser.check('import'):
1231             do_import(parser)
1232         elif parser.check('export'):
1233             do_export(parser)
1234         elif parser.check('option'):
1235             do_option(parser)
1236         else:
1237             die('unhandled command: %s' % line)
1238         sys.stdout.flush()
1239
1240 def bye():
1241     if not marks:
1242         return
1243     if not is_tmp:
1244         marks.store()
1245     else:
1246         shutil.rmtree(dirname)
1247
1248 atexit.register(bye)
1249 sys.exit(main(sys.argv))