1
# Copyright (C) 2008-2018 Jelmer Vernooij <jelmer@jelmer.uk>
2
# Copyright (C) 2007 Canonical Ltd
3
# Copyright (C) 2008 John Carr
5
# This program is free software; you can redistribute it and/or modify
6
# it under the terms of the GNU General Public License as published by
7
# the Free Software Foundation; either version 2 of the License, or
8
# (at your option) any later version.
10
# This program is distributed in the hope that it will be useful,
11
# but WITHOUT ANY WARRANTY; without even the implied warranty of
12
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13
# GNU General Public License for more details.
15
# You should have received a copy of the GNU General Public License
16
# along with this program; if not, write to the Free Software
17
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19
"""Converters, etc for going between Bazaar and Git ids."""
21
from __future__ import absolute_import
33
from ..foreign import (
38
from ..revision import (
42
from ..sixish import (
50
UnknownMercurialCommitExtra,
56
from .roundtrip import (
63
DEFAULT_FILE_MODE = stat.S_IFREG | 0o644
64
HG_RENAME_SOURCE = b"HG:rename-source"
65
HG_EXTRA = b"HG:extra"
67
# This HG extra is used to indicate the commit that this commit was based on.
68
HG_EXTRA_AMEND_SOURCE = b"amend_source"
70
FILE_ID_PREFIX = b'git:'
73
ROOT_ID = b"TREE_ROOT"
76
def escape_file_id(file_id):
77
file_id = file_id.replace(b'_', b'__')
78
file_id = file_id.replace(b' ', b'_s')
79
file_id = file_id.replace(b'\x0c', b'_c')
83
def unescape_file_id(file_id):
86
while i < len(file_id):
87
if file_id[i:i + 1] != b'_':
88
ret.append(file_id[i])
90
if file_id[i + 1:i + 2] == b'_':
92
elif file_id[i + 1:i + 2] == b's':
94
elif file_id[i + 1:i + 2] == b'c':
95
ret.append(b"\x0c"[0])
97
raise ValueError("unknown escape character %s" %
104
def fix_person_identifier(text):
105
if b"<" not in text and b">" not in text:
109
if text.rindex(b">") < text.rindex(b"<"):
110
raise ValueError(text)
111
username, email = text.split(b"<", 2)[-2:]
112
email = email.split(b">", 1)[0]
113
if username.endswith(b" "):
114
username = username[:-1]
115
return b"%s <%s>" % (username, email)
118
def warn_escaped(commit, num_escaped):
119
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
120
"to regenerate the SHA map.", num_escaped, commit)
123
def warn_unusual_mode(commit, path, mode):
124
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
125
"property. ", mode, path, commit)
128
class BzrGitMapping(foreign.VcsMapping):
129
"""Class that maps between Git and Bazaar semantics."""
132
BZR_DUMMY_FILE = None
134
def is_special_file(self, filename):
135
return (filename in (self.BZR_DUMMY_FILE, ))
138
super(BzrGitMapping, self).__init__(foreign_vcs_git)
140
def __eq__(self, other):
141
return (type(self) == type(other)
142
and self.revid_prefix == other.revid_prefix)
145
def revision_id_foreign_to_bzr(cls, git_rev_id):
146
"""Convert a git revision id handle to a Bazaar revision id."""
147
from dulwich.protocol import ZERO_SHA
148
if git_rev_id == ZERO_SHA:
150
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
153
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
154
"""Convert a Bazaar revision id to a git revision id handle."""
155
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
156
raise errors.InvalidRevisionId(bzr_rev_id, cls)
157
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
159
def generate_file_id(self, path):
160
# Git paths are just bytestrings
161
# We must just hope they are valid UTF-8..
162
if isinstance(path, text_type):
163
path = path.encode("utf-8")
166
return FILE_ID_PREFIX + escape_file_id(path)
168
def parse_file_id(self, file_id):
169
if file_id == ROOT_ID:
171
if not file_id.startswith(FILE_ID_PREFIX):
173
return unescape_file_id(file_id[len(FILE_ID_PREFIX):]).decode('utf-8')
175
def revid_as_refname(self, revid):
176
if not isinstance(revid, bytes):
177
raise TypeError(revid)
179
revid = revid.decode('utf-8')
180
quoted_revid = urlutils.quote(revid)
181
return b"refs/bzr/" + quoted_revid.encode('utf-8')
183
def import_unusual_file_modes(self, rev, unusual_file_modes):
184
if unusual_file_modes:
185
ret = [(path, unusual_file_modes[path])
186
for path in sorted(unusual_file_modes.keys())]
187
rev.properties[u'file-modes'] = bencode.bencode(ret)
189
def export_unusual_file_modes(self, rev):
191
file_modes = rev.properties[u'file-modes']
195
return dict(bencode.bdecode(file_modes.encode("utf-8")))
197
def _generate_git_svn_metadata(self, rev, encoding):
199
git_svn_id = rev.properties[u"git-svn-id"]
203
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
205
def _generate_hg_message_tail(self, rev):
209
for name in rev.properties:
210
if name == u'hg:extra:branch':
211
branch = rev.properties[u'hg:extra:branch']
212
elif name.startswith(u'hg:extra'):
213
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
214
rev.properties[name])
215
elif name == u'hg:renames':
216
renames = bencode.bdecode(base64.b64decode(
217
rev.properties[u'hg:renames']))
218
# TODO: Export other properties as 'bzr:' extras?
219
ret = format_hg_metadata(renames, branch, extra)
220
if not isinstance(ret, bytes):
224
def _extract_git_svn_metadata(self, rev, message):
225
lines = message.split("\n")
226
if not (lines[-1] == "" and len(lines) >= 2 and
227
lines[-2].startswith("git-svn-id:")):
229
git_svn_id = lines[-2].split(": ", 1)[1]
230
rev.properties[u'git-svn-id'] = git_svn_id
231
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
232
# FIXME: Convert this to converted-from property somehow..
233
return "\n".join(lines[:-2])
235
def _extract_hg_metadata(self, rev, message):
236
(message, renames, branch, extra) = extract_hg_metadata(message)
237
if branch is not None:
238
rev.properties[u'hg:extra:branch'] = branch
239
for name, value in viewitems(extra):
240
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
242
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
243
[(new, old) for (old, new) in viewitems(renames)]))
246
def _extract_bzr_metadata(self, rev, message):
247
(message, metadata) = extract_bzr_metadata(message)
248
return message, metadata
250
def _decode_commit_message(self, rev, message, encoding):
251
return message.decode(encoding), CommitSupplement()
253
def _encode_commit_message(self, rev, message, encoding):
254
return message.encode(encoding)
256
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
258
"""Turn a Bazaar revision in to a Git commit
260
:param tree_sha: Tree sha for the commit
261
:param parent_lookup: Function for looking up the GIT sha equiv of a
263
:param lossy: Whether to store roundtripping information.
264
:param verifiers: Verifiers info
265
:return dulwich.objects.Commit represent the revision:
267
from dulwich.objects import Commit, Tag
269
commit.tree = tree_sha
271
metadata = CommitSupplement()
272
metadata.verifiers = verifiers
276
for p in rev.parent_ids:
278
git_p = parent_lookup(p)
281
if metadata is not None:
282
metadata.explicit_parent_ids = rev.parent_ids
283
if git_p is not None:
285
raise AssertionError("unexpected length for %r" % git_p)
286
parents.append(git_p)
287
commit.parents = parents
289
encoding = rev.properties[u'git-explicit-encoding']
291
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
293
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
297
commit.committer = fix_person_identifier(rev.committer.encode(
299
commit.author = fix_person_identifier(
300
rev.get_apparent_authors()[0].encode(encoding))
301
# TODO(jelmer): Don't use this hack.
302
long = getattr(__builtins__, 'long', int)
303
commit.commit_time = long(rev.timestamp)
304
if u'author-timestamp' in rev.properties:
305
commit.author_time = long(rev.properties[u'author-timestamp'])
307
commit.author_time = commit.commit_time
308
commit._commit_timezone_neg_utc = (
309
u"commit-timezone-neg-utc" in rev.properties)
310
commit.commit_timezone = rev.timezone
311
commit._author_timezone_neg_utc = (
312
u"author-timezone-neg-utc" in rev.properties)
313
if u'author-timezone' in rev.properties:
314
commit.author_timezone = int(rev.properties[u'author-timezone'])
316
commit.author_timezone = commit.commit_timezone
317
if u'git-gpg-signature' in rev.properties:
318
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
320
commit.message = self._encode_commit_message(rev, rev.message,
322
if not isinstance(commit.message, bytes):
323
raise TypeError(commit.message)
324
if metadata is not None:
326
mapping_registry.parse_revision_id(rev.revision_id)
327
except errors.InvalidRevisionId:
328
metadata.revision_id = rev.revision_id
329
mapping_properties = set(
330
[u'author', u'author-timezone', u'author-timezone-neg-utc',
331
u'commit-timezone-neg-utc', u'git-implicit-encoding',
332
u'git-gpg-signature', u'git-explicit-encoding',
333
u'author-timestamp', u'file-modes'])
334
for k, v in viewitems(rev.properties):
335
if k not in mapping_properties:
336
metadata.properties[k] = v
337
if not lossy and metadata:
338
if self.roundtripping:
339
commit.message = inject_bzr_metadata(commit.message, metadata,
343
None, None, self, revision_id=rev.revision_id)
344
if not isinstance(commit.message, bytes):
345
raise TypeError(commit.message)
347
propname = u'git-mergetag-0'
348
while propname in rev.properties:
349
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
351
propname = u'git-mergetag-%d' % i
352
if u'git-extra' in rev.properties:
355
for l in rev.properties[u'git-extra'].splitlines()])
358
def get_revision_id(self, commit):
360
encoding = commit.encoding.decode('ascii')
364
message, metadata = self._decode_commit_message(
365
None, commit.message, encoding)
366
except UnicodeDecodeError:
369
if metadata.revision_id:
370
return metadata.revision_id
371
return self.revision_id_foreign_to_bzr(commit.id)
373
def import_commit(self, commit, lookup_parent_revid):
374
"""Convert a git commit to a bzr revision.
376
:return: a `breezy.revision.Revision` object, foreign revid and a
380
raise AssertionError("Commit object can't be None")
381
rev = ForeignRevision(commit.id, self,
382
self.revision_id_foreign_to_bzr(commit.id))
383
rev.git_metadata = None
385
def decode_using_encoding(rev, commit, encoding):
386
rev.committer = commit.committer.decode(encoding)
387
if commit.committer != commit.author:
388
rev.properties[u'author'] = commit.author.decode(encoding)
389
rev.message, rev.git_metadata = self._decode_commit_message(
390
rev, commit.message, encoding)
391
if commit.encoding is not None:
392
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode(
394
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
396
for encoding in ('utf-8', 'latin1'):
398
decode_using_encoding(rev, commit, encoding)
399
except UnicodeDecodeError:
402
if encoding != 'utf-8':
403
rev.properties[u'git-implicit-encoding'] = encoding
405
if commit.commit_time != commit.author_time:
406
rev.properties[u'author-timestamp'] = str(commit.author_time)
407
if commit.commit_timezone != commit.author_timezone:
408
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
409
if commit._author_timezone_neg_utc:
410
rev.properties[u'author-timezone-neg-utc'] = ""
411
if commit._commit_timezone_neg_utc:
412
rev.properties[u'commit-timezone-neg-utc'] = ""
414
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode(
417
for i, tag in enumerate(commit.mergetag):
418
rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string()
419
rev.timestamp = commit.commit_time
420
rev.timezone = commit.commit_timezone
421
rev.parent_ids = None
422
if rev.git_metadata is not None:
423
md = rev.git_metadata
424
roundtrip_revid = md.revision_id
425
if md.explicit_parent_ids:
426
rev.parent_ids = md.explicit_parent_ids
427
rev.properties.update(md.properties)
428
verifiers = md.verifiers
430
roundtrip_revid = None
432
if rev.parent_ids is None:
434
for p in commit.parents:
436
parents.append(lookup_parent_revid(p))
438
parents.append(self.revision_id_foreign_to_bzr(p))
439
rev.parent_ids = list(parents)
440
unknown_extra_fields = []
442
for k, v in commit.extra:
443
if k == HG_RENAME_SOURCE:
444
extra_lines.append(k + b' ' + v + b'\n')
446
hgk, hgv = v.split(b':', 1)
447
if hgk not in (HG_EXTRA_AMEND_SOURCE, ):
448
raise UnknownMercurialCommitExtra(commit, [hgk])
449
extra_lines.append(k + b' ' + v + b'\n')
451
unknown_extra_fields.append(k)
452
if unknown_extra_fields:
453
raise UnknownCommitExtra(
455
[f.decode('ascii', 'replace') for f in unknown_extra_fields])
457
rev.properties[u'git-extra'] = b''.join(extra_lines)
458
return rev, roundtrip_revid, verifiers
461
class BzrGitMappingv1(BzrGitMapping):
462
revid_prefix = b'git-v1'
466
return self.revid_prefix
469
class BzrGitMappingExperimental(BzrGitMappingv1):
470
revid_prefix = b'git-experimental'
472
roundtripping = False
474
BZR_DUMMY_FILE = '.bzrdummy'
476
def _decode_commit_message(self, rev, message, encoding):
479
message = self._extract_hg_metadata(rev, message)
480
message = self._extract_git_svn_metadata(rev, message)
481
message, metadata = self._extract_bzr_metadata(rev, message)
482
return message.decode(encoding), metadata
484
def _encode_commit_message(self, rev, message, encoding):
485
ret = message.encode(encoding)
486
ret += self._generate_hg_message_tail(rev)
487
ret += self._generate_git_svn_metadata(rev, encoding)
490
def import_commit(self, commit, lookup_parent_revid):
491
rev, roundtrip_revid, verifiers = super(
492
BzrGitMappingExperimental, self).import_commit(
493
commit, lookup_parent_revid)
494
rev.properties[u'converted_revision'] = "git %s\n" % commit.id
495
return rev, roundtrip_revid, verifiers
498
class GitMappingRegistry(VcsMappingRegistry):
499
"""Registry with available git mappings."""
501
def revision_id_bzr_to_foreign(self, bzr_revid):
502
if bzr_revid == NULL_REVISION:
503
from dulwich.protocol import ZERO_SHA
504
return ZERO_SHA, None
505
if not bzr_revid.startswith(b"git-"):
506
raise errors.InvalidRevisionId(bzr_revid, None)
507
(mapping_version, git_sha) = bzr_revid.split(b":", 1)
508
mapping = self.get(mapping_version)
509
return mapping.revision_id_bzr_to_foreign(bzr_revid)
511
parse_revision_id = revision_id_bzr_to_foreign
514
mapping_registry = GitMappingRegistry()
515
mapping_registry.register_lazy(b'git-v1', __name__,
517
mapping_registry.register_lazy(b'git-experimental',
518
__name__, "BzrGitMappingExperimental")
519
# Uncomment the next line to enable the experimental bzr-git mappings.
520
# This will make sure all bzr metadata is pushed into git, allowing for
521
# full roundtripping later.
522
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
523
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
524
# mapping_registry.set_default('git-experimental')
525
mapping_registry.set_default(b'git-v1')
528
class ForeignGit(ForeignVcs):
529
"""The Git Stupid Content Tracker"""
532
def branch_format(self):
533
from .branch import LocalGitBranchFormat
534
return LocalGitBranchFormat()
537
def repository_format(self):
538
from .repository import GitRepositoryFormat
539
return GitRepositoryFormat()
542
super(ForeignGit, self).__init__(mapping_registry)
543
self.abbreviation = "git"
546
def serialize_foreign_revid(self, foreign_revid):
550
def show_foreign_revid(cls, foreign_revid):
551
return {"git commit": foreign_revid.decode('utf-8')}
554
foreign_vcs_git = ForeignGit()
555
default_mapping = mapping_registry.get_default()()
558
def symlink_to_blob(symlink_target):
559
from dulwich.objects import Blob
561
if isinstance(symlink_target, text_type):
562
symlink_target = symlink_target.encode('utf-8')
563
blob.data = symlink_target
567
def mode_is_executable(mode):
568
"""Check if mode should be considered executable."""
569
return bool(mode & 0o111)
573
"""Determine the Bazaar inventory kind based on Unix file mode."""
576
entry_kind = (mode & 0o700000) / 0o100000
579
elif entry_kind == 1:
580
file_kind = (mode & 0o70000) / 0o10000
586
return 'tree-reference'
588
raise AssertionError(
589
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
591
raise AssertionError(
592
"Unknown kind, perms=%r." % (mode,))
595
def object_mode(kind, executable):
596
if kind == 'directory':
598
elif kind == 'symlink':
604
mode = stat.S_IFREG | 0o644
608
elif kind == 'tree-reference':
609
from dulwich.objects import S_IFGITLINK
615
def entry_mode(entry):
616
"""Determine the git file mode for an inventory entry."""
617
return object_mode(entry.kind, getattr(entry, 'executable', False))
620
def extract_unusual_modes(rev):
622
foreign_revid, mapping = mapping_registry.parse_revision_id(
624
except errors.InvalidRevisionId:
627
return mapping.export_unusual_file_modes(rev)
630
def parse_git_svn_id(text):
631
(head, uuid) = text.rsplit(" ", 1)
632
(full_url, rev) = head.rsplit("@", 1)
633
return (full_url, int(rev), uuid)
636
def needs_roundtripping(repo, revid):
638
mapping_registry.parse_revision_id(revid)
639
except errors.InvalidRevisionId: