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 $PATH.
   9 # Then you can clone with:
  10 # git clone hg::/path/to/mercurial/repo/
  11
  12 from mercurial import hg, ui, bookmarks, context, util, encoding
  13
  14 import re
  15 import sys
  16 import os
  17 import json
  18 import shutil
  19
  20 NAME_RE = re.compile('^([^<>]+)')
  21 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]+)>$')
  22 RAW_AUTHOR_RE = re.compile('^(\w+) (?:(.+)? )?<(.+)> (\d+) ([+-]\d+)')
  23
  24 def die(msg, *args):
  25     sys.stderr.write('ERROR: %s\n' % (msg % args))
  26     sys.exit(1)
  27
  28 def warn(msg, *args):
  29     sys.stderr.write('WARNING: %s\n' % (msg % args))
  30
  31 def gitmode(flags):
  32     return 'l' in flags and '120000' or 'x' in flags and '100755' or '100644'
  33
  34 def gittz(tz):
  35     return '%+03d%02d' % (-tz / 3600, -tz % 3600 / 60)
  36
  37 def hgmode(mode):
  38     m = { '0100755': 'x', '0120000': 'l' }
  39     return m.get(mode, '')
  40
  41 class Marks:
  42
  43     def __init__(self, path):
  44         self.path = path
  45         self.tips = {}
  46         self.marks = {}
  47         self.rev_marks = {}
  48         self.last_mark = 0
  49
  50         self.load()
  51
  52     def load(self):
  53         if not os.path.exists(self.path):
  54             return
  55
  56         tmp = json.load(open(self.path))
  57
  58         self.tips = tmp['tips']
  59         self.marks = tmp['marks']
  60         self.last_mark = tmp['last-mark']
  61
  62         for rev, mark in self.marks.iteritems():
  63             self.rev_marks[mark] = int(rev)
  64
  65     def dict(self):
  66         return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
  67
  68     def store(self):
  69         json.dump(self.dict(), open(self.path, 'w'))
  70
  71     def __str__(self):
  72         return str(self.dict())
  73
  74     def from_rev(self, rev):
  75         return self.marks[str(rev)]
  76
  77     def to_rev(self, mark):
  78         return self.rev_marks[mark]
  79
  80     def get_mark(self, rev):
  81         self.last_mark += 1
  82         self.marks[str(rev)] = self.last_mark
  83         return self.last_mark
  84
  85     def new_mark(self, rev, mark):
  86         self.marks[str(rev)] = mark
  87         self.rev_marks[mark] = rev
  88         self.last_mark = mark
  89
  90     def is_marked(self, rev):
  91         return self.marks.has_key(str(rev))
  92
  93     def get_tip(self, branch):
  94         return self.tips.get(branch, 0)
  95
  96     def set_tip(self, branch, tip):
  97         self.tips[branch] = tip
  98
  99 class Parser:
 100
 101     def __init__(self, repo):
 102         self.repo = repo
 103         self.line = self.get_line()
 104
 105     def get_line(self):
 106         return sys.stdin.readline().strip()
 107
 108     def __getitem__(self, i):
 109         return self.line.split()[i]
 110
 111     def check(self, word):
 112         return self.line.startswith(word)
 113
 114     def each_block(self, separator):
 115         while self.line != separator:
 116             yield self.line
 117             self.line = self.get_line()
 118
 119     def __iter__(self):
 120         return self.each_block('')
 121
 122     def next(self):
 123         self.line = self.get_line()
 124         if self.line == 'done':
 125             self.line = None
 126
 127     def get_mark(self):
 128         i = self.line.index(':') + 1
 129         return int(self.line[i:])
 130
 131     def get_data(self):
 132         if not self.check('data'):
 133             return None
 134         i = self.line.index(' ') + 1
 135         size = int(self.line[i:])
 136         return sys.stdin.read(size)
 137
 138     def get_author(self):
 139         m = RAW_AUTHOR_RE.match(self.line)
 140         if not m:
 141             return None
 142         _, name, email, date, tz = m.groups()
 143
 144         if email != 'unknown':
 145             if name:
 146                 user = '%s <%s>' % (name, email)
 147             else:
 148                 user = '<%s>' % (email)
 149         else:
 150             user = name
 151
 152         tz = int(tz)
 153         tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
 154         return (user, int(date), -tz)
 155
 156 def export_file(fc):
 157     d = fc.data()
 158     print "M %s inline %s" % (gitmode(fc.flags()), fc.path())
 159     print "data %d" % len(d)
 160     print d
 161
 162 def get_filechanges(repo, ctx, parent):
 163     modified = set()
 164     added = set()
 165     removed = set()
 166
 167     cur = ctx.manifest()
 168     prev = repo[parent].manifest().copy()
 169
 170     for fn in cur:
 171         if fn in prev:
 172             if (cur.flags(fn) != prev.flags(fn) or cur[fn] != prev[fn]):
 173                 modified.add(fn)
 174             del prev[fn]
 175         else:
 176             added.add(fn)
 177     removed |= set(prev.keys())
 178
 179     return added | modified, removed
 180
 181 def fixup_user(user):
 182     user = user.replace('"', '')
 183     name = mail = None
 184     m = AUTHOR_RE.match(user)
 185     if m:
 186         name = m.group(1)
 187         mail = m.group(2).strip()
 188     else:
 189         m = NAME_RE.match(user)
 190         if m:
 191             name = m.group(1).strip()
 192
 193     if not name:
 194         name = 'Unknown'
 195     if not mail:
 196         mail = 'unknown'
 197
 198     return '%s <%s>' % (name, mail)
 199
 200 def get_repo(url, alias):
 201     global dirname, peer
 202
 203     myui = ui.ui()
 204     myui.setconfig('ui', 'interactive', 'off')
 205
 206     if hg.islocal(url):
 207         repo = hg.repository(myui, url)
 208     else:
 209         local_path = os.path.join(dirname, 'clone')
 210         if not os.path.exists(local_path):
 211             peer, dstpeer = hg.clone(myui, {}, url, local_path, update=False, pull=True)
 212             repo = dstpeer.local()
 213         else:
 214             repo = hg.repository(myui, local_path)
 215             peer = hg.peer(myui, {}, url)
 216             repo.pull(peer, heads=None, force=True)
 217
 218     return repo
 219
 220 def rev_to_mark(rev):
 221     global marks
 222     return marks.from_rev(rev)
 223
 224 def mark_to_rev(mark):
 225     global marks
 226     return marks.to_rev(mark)
 227
 228 def export_ref(repo, name, kind, head):
 229     global prefix, marks
 230
 231     ename = '%s/%s' % (kind, name)
 232     tip = marks.get_tip(ename)
 233
 234     # mercurial takes too much time checking this
 235     if tip and tip == head.rev():
 236         # nothing to do
 237         return
 238     revs = repo.revs('%u:%u' % (tip, head))
 239     count = 0
 240
 241     revs = [rev for rev in revs if not marks.is_marked(rev)]
 242
 243     for rev in revs:
 244
 245         c = repo[rev]
 246         (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(c.node())
 247         rev_branch = extra['branch']
 248
 249         author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
 250         if 'committer' in extra:
 251             user, time, tz = extra['committer'].rsplit(' ', 2)
 252             committer = "%s %s %s" % (user, time, gittz(int(tz)))
 253         else:
 254             committer = author
 255
 256         parents = [p for p in repo.changelog.parentrevs(rev) if p >= 0]
 257
 258         if len(parents) == 0:
 259             modified = c.manifest().keys()
 260             removed = []
 261         else:
 262             modified, removed = get_filechanges(repo, c, parents[0])
 263
 264         if len(parents) == 0 and rev:
 265             print 'reset %s/%s' % (prefix, ename)
 266
 267         print "commit %s/%s" % (prefix, ename)
 268         print "mark :%d" % (marks.get_mark(rev))
 269         print "author %s" % (author)
 270         print "committer %s" % (committer)
 271         print "data %d" % (len(desc))
 272         print desc
 273
 274         if len(parents) > 0:
 275             print "from :%s" % (rev_to_mark(parents[0]))
 276             if len(parents) > 1:
 277                 print "merge :%s" % (rev_to_mark(parents[1]))
 278
 279         for f in modified:
 280             export_file(c.filectx(f))
 281         for f in removed:
 282             print "D %s" % (f)
 283         print
 284
 285         count += 1
 286         if (count % 100 == 0):
 287             print "progress revision %d '%s' (%d/%d)" % (rev, name, count, len(revs))
 288             print "#############################################################"
 289
 290     # make sure the ref is updated
 291     print "reset %s/%s" % (prefix, ename)
 292     print "from :%u" % rev_to_mark(rev)
 293     print
 294
 295     marks.set_tip(ename, rev)
 296
 297 def export_tag(repo, tag):
 298     export_ref(repo, tag, 'tags', repo[tag])
 299
 300 def export_bookmark(repo, bmark):
 301     head = bmarks[bmark]
 302     export_ref(repo, bmark, 'bookmarks', head)
 303
 304 def export_branch(repo, branch):
 305     tip = get_branch_tip(repo, branch)
 306     head = repo[tip]
 307     export_ref(repo, branch, 'branches', head)
 308
 309 def export_head(repo):
 310     global g_head
 311     export_ref(repo, g_head[0], 'bookmarks', g_head[1])
 312
 313 def do_capabilities(parser):
 314     global prefix, dirname
 315
 316     print "import"
 317     print "export"
 318     print "refspec refs/heads/branches/*:%s/branches/*" % prefix
 319     print "refspec refs/heads/*:%s/bookmarks/*" % prefix
 320     print "refspec refs/tags/*:%s/tags/*" % prefix
 321
 322     path = os.path.join(dirname, 'marks-git')
 323
 324     if os.path.exists(path):
 325         print "*import-marks %s" % path
 326     print "*export-marks %s" % path
 327
 328     print
 329
 330 def get_branch_tip(repo, branch):
 331     global branches
 332
 333     heads = branches.get(branch, None)
 334     if not heads:
 335         return None
 336
 337     # verify there's only one head
 338     if (len(heads) > 1):
 339         warn("Branch '%s' has more than one head, consider merging" % branch)
 340         # older versions of mercurial don't have this
 341         if hasattr(repo, "branchtip"):
 342             return repo.branchtip(branch)
 343
 344     return heads[0]
 345
 346 def list_head(repo, cur):
 347     global g_head
 348
 349     head = bookmarks.readcurrent(repo)
 350     if not head:
 351         return
 352     node = repo[head]
 353     print "@refs/heads/%s HEAD" % head
 354     g_head = (head, node)
 355
 356 def do_list(parser):
 357     global branches, bmarks
 358
 359     repo = parser.repo
 360     for branch in repo.branchmap():
 361         heads = repo.branchheads(branch)
 362         if len(heads):
 363             branches[branch] = heads
 364
 365     for bmark, node in bookmarks.listbookmarks(repo).iteritems():
 366         bmarks[bmark] = repo[node]
 367
 368     cur = repo.dirstate.branch()
 369
 370     list_head(repo, cur)
 371     for branch in branches:
 372         print "? refs/heads/branches/%s" % branch
 373     for bmark in bmarks:
 374         print "? refs/heads/%s" % bmark
 375
 376     for tag, node in repo.tagslist():
 377         if tag == 'tip':
 378             continue
 379         print "? refs/tags/%s" % tag
 380
 381     print
 382
 383 def do_import(parser):
 384     repo = parser.repo
 385
 386     path = os.path.join(dirname, 'marks-git')
 387
 388     print "feature done"
 389     if os.path.exists(path):
 390         print "feature import-marks=%s" % path
 391     print "feature export-marks=%s" % path
 392     sys.stdout.flush()
 393
 394     tmp = encoding.encoding
 395     encoding.encoding = 'utf-8'
 396
 397     # lets get all the import lines
 398     while parser.check('import'):
 399         ref = parser[1]
 400
 401         if (ref == 'HEAD'):
 402             export_head(repo)
 403         elif ref.startswith('refs/heads/branches/'):
 404             branch = ref[len('refs/heads/branches/'):]
 405             export_branch(repo, branch)
 406         elif ref.startswith('refs/heads/'):
 407             bmark = ref[len('refs/heads/'):]
 408             export_bookmark(repo, bmark)
 409         elif ref.startswith('refs/tags/'):
 410             tag = ref[len('refs/tags/'):]
 411             export_tag(repo, tag)
 412
 413         parser.next()
 414
 415     encoding.encoding = tmp
 416
 417     print 'done'
 418
 419 def parse_blob(parser):
 420     global blob_marks
 421
 422     parser.next()
 423     mark = parser.get_mark()
 424     parser.next()
 425     data = parser.get_data()
 426     blob_marks[mark] = data
 427     parser.next()
 428     return
 429
 430 def get_merge_files(repo, p1, p2, files):
 431     for e in repo[p1].files():
 432         if e not in files:
 433             if e not in repo[p1].manifest():
 434                 continue
 435             f = { 'ctx' : repo[p1][e] }
 436             files[e] = f
 437
 438 def parse_commit(parser):
 439     global marks, blob_marks, bmarks, parsed_refs
 440
 441     from_mark = merge_mark = None
 442
 443     ref = parser[1]
 444     parser.next()
 445
 446     commit_mark = parser.get_mark()
 447     parser.next()
 448     author = parser.get_author()
 449     parser.next()
 450     committer = parser.get_author()
 451     parser.next()
 452     data = parser.get_data()
 453     parser.next()
 454     if parser.check('from'):
 455         from_mark = parser.get_mark()
 456         parser.next()
 457     if parser.check('merge'):
 458         merge_mark = parser.get_mark()
 459         parser.next()
 460         if parser.check('merge'):
 461             die('octopus merges are not supported yet')
 462
 463     files = {}
 464
 465     for line in parser:
 466         if parser.check('M'):
 467             t, m, mark_ref, path = line.split(' ')
 468             mark = int(mark_ref[1:])
 469             f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
 470         elif parser.check('D'):
 471             t, path = line.split(' ')
 472             f = { 'deleted' : True }
 473         else:
 474             die('Unknown file command: %s' % line)
 475         files[path] = f
 476
 477     def getfilectx(repo, memctx, f):
 478         of = files[f]
 479         if 'deleted' in of:
 480             raise IOError
 481         if 'ctx' in of:
 482             return of['ctx']
 483         is_exec = of['mode'] == 'x'
 484         is_link = of['mode'] == 'l'
 485         return context.memfilectx(f, of['data'], is_link, is_exec, None)
 486
 487     repo = parser.repo
 488
 489     user, date, tz = author
 490     extra = {}
 491
 492     if committer != author:
 493         extra['committer'] = "%s %u %u" % committer
 494
 495     if from_mark:
 496         p1 = repo.changelog.node(mark_to_rev(from_mark))
 497     else:
 498         p1 = '\0' * 20
 499
 500     if merge_mark:
 501         p2 = repo.changelog.node(mark_to_rev(merge_mark))
 502     else:
 503         p2 = '\0' * 20
 504
 505     #
 506     # If files changed from any of the parents, hg wants to know, but in git if
 507     # nothing changed from the first parent, nothing changed.
 508     #
 509     if merge_mark:
 510         get_merge_files(repo, p1, p2, files)
 511
 512     ctx = context.memctx(repo, (p1, p2), data,
 513             files.keys(), getfilectx,
 514             user, (date, tz), extra)
 515
 516     tmp = encoding.encoding
 517     encoding.encoding = 'utf-8'
 518
 519     node = repo.commitctx(ctx)
 520
 521     encoding.encoding = tmp
 522
 523     rev = repo[node].rev()
 524
 525     parsed_refs[ref] = node
 526
 527     marks.new_mark(rev, commit_mark)
 528
 529 def parse_reset(parser):
 530     ref = parser[1]
 531     parser.next()
 532     # ugh
 533     if parser.check('commit'):
 534         parse_commit(parser)
 535         return
 536     if not parser.check('from'):
 537         return
 538     from_mark = parser.get_mark()
 539     parser.next()
 540
 541     node = parser.repo.changelog.node(mark_to_rev(from_mark))
 542     parsed_refs[ref] = node
 543
 544 def parse_tag(parser):
 545     name = parser[1]
 546     parser.next()
 547     from_mark = parser.get_mark()
 548     parser.next()
 549     tagger = parser.get_author()
 550     parser.next()
 551     data = parser.get_data()
 552     parser.next()
 553
 554     # nothing to do
 555
 556 def do_export(parser):
 557     global parsed_refs, bmarks, peer
 558
 559     parser.next()
 560
 561     for line in parser.each_block('done'):
 562         if parser.check('blob'):
 563             parse_blob(parser)
 564         elif parser.check('commit'):
 565             parse_commit(parser)
 566         elif parser.check('reset'):
 567             parse_reset(parser)
 568         elif parser.check('tag'):
 569             parse_tag(parser)
 570         elif parser.check('feature'):
 571             pass
 572         else:
 573             die('unhandled export command: %s' % line)
 574
 575     for ref, node in parsed_refs.iteritems():
 576         if ref.startswith('refs/heads/branches'):
 577             pass
 578         elif ref.startswith('refs/heads/'):
 579             bmark = ref[len('refs/heads/'):]
 580             if bmark in bmarks:
 581                 old = bmarks[bmark].hex()
 582             else:
 583                 old = ''
 584             if not bookmarks.pushbookmark(parser.repo, bmark, old, node):
 585                 continue
 586         elif ref.startswith('refs/tags/'):
 587             tag = ref[len('refs/tags/'):]
 588             parser.repo.tag([tag], node, None, True, None, {})
 589         print "ok %s" % ref
 590
 591     print
 592
 593     if peer:
 594         parser.repo.push(peer, force=False)
 595
 596 def main(args):
 597     global prefix, dirname, branches, bmarks
 598     global marks, blob_marks, parsed_refs
 599     global peer
 600
 601     alias = args[1]
 602     url = args[2]
 603     peer = None
 604
 605     if alias[4:] == url:
 606         is_tmp = True
 607         alias = util.sha1(alias).hexdigest()
 608     else:
 609         is_tmp = False
 610
 611     gitdir = os.environ['GIT_DIR']
 612     dirname = os.path.join(gitdir, 'hg', alias)
 613     branches = {}
 614     bmarks = {}
 615     blob_marks = {}
 616     parsed_refs = {}
 617
 618     repo = get_repo(url, alias)
 619     prefix = 'refs/hg/%s' % alias
 620
 621     if not os.path.exists(dirname):
 622         os.makedirs(dirname)
 623
 624     marks_path = os.path.join(dirname, 'marks-hg')
 625     marks = Marks(marks_path)
 626
 627     parser = Parser(repo)
 628     for line in parser:
 629         if parser.check('capabilities'):
 630             do_capabilities(parser)
 631         elif parser.check('list'):
 632             do_list(parser)
 633         elif parser.check('import'):
 634             do_import(parser)
 635         elif parser.check('export'):
 636             do_export(parser)
 637         else:
 638             die('unhandled command: %s' % line)
 639         sys.stdout.flush()
 640
 641     if not is_tmp:
 642         marks.store()
 643     else:
 644         shutil.rmtree(dirname)
 645
 646 sys.exit(main(sys.argv))