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."""
31
from ..foreign import (
36
from ..revision import (
47
from .roundtrip import (
54
DEFAULT_FILE_MODE = stat.S_IFREG | 0o644
55
HG_RENAME_SOURCE = b"HG:rename-source"
56
HG_EXTRA = b"HG:extra"
58
# This HG extra is used to indicate the commit that this commit was based on.
59
HG_EXTRA_AMEND_SOURCE = b"amend_source"
61
FILE_ID_PREFIX = b'git:'
64
ROOT_ID = b"TREE_ROOT"
67
class UnknownCommitExtra(errors.BzrError):
68
_fmt = "Unknown extra fields in %(object)r: %(fields)r."
70
def __init__(self, object, fields):
71
errors.BzrError.__init__(self)
73
self.fields = ",".join(fields)
76
class UnknownMercurialCommitExtra(errors.BzrError):
77
_fmt = "Unknown mercurial extra fields in %(object)r: %(fields)r."
79
def __init__(self, object, fields):
80
errors.BzrError.__init__(self)
82
self.fields = b",".join(fields)
85
def escape_file_id(file_id):
86
file_id = file_id.replace(b'_', b'__')
87
file_id = file_id.replace(b' ', b'_s')
88
file_id = file_id.replace(b'\x0c', b'_c')
92
def unescape_file_id(file_id):
95
while i < len(file_id):
96
if file_id[i:i + 1] != b'_':
97
ret.append(file_id[i])
99
if file_id[i + 1:i + 2] == b'_':
101
elif file_id[i + 1:i + 2] == b's':
103
elif file_id[i + 1:i + 2] == b'c':
104
ret.append(b"\x0c"[0])
106
raise ValueError("unknown escape character %s" %
107
file_id[i + 1:i + 2])
113
def fix_person_identifier(text):
114
if b"<" not in text and b">" not in text:
117
elif b">" not in text:
120
if text.rindex(b">") < text.rindex(b"<"):
121
raise ValueError(text)
122
username, email = text.split(b"<", 2)[-2:]
123
email = email.split(b">", 1)[0]
124
if username.endswith(b" "):
125
username = username[:-1]
126
return b"%s <%s>" % (username, email)
129
def decode_git_path(path):
130
"""Take a git path and decode it."""
132
return path.decode('utf-8')
133
except UnicodeDecodeError:
135
return path.decode('utf-8', 'surrogateescape')
139
def encode_git_path(path):
140
"""Take a regular path and encode it for git."""
142
return path.encode('utf-8')
143
except UnicodeEncodeError:
145
return path.encode('utf-8', 'surrogateescape')
149
def warn_escaped(commit, num_escaped):
150
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
151
"to regenerate the SHA map.", num_escaped, commit)
154
def warn_unusual_mode(commit, path, mode):
155
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
156
"property. ", mode, path, commit)
159
class BzrGitMapping(foreign.VcsMapping):
160
"""Class that maps between Git and Bazaar semantics."""
163
BZR_DUMMY_FILE = None # type: Optional[str]
165
def is_special_file(self, filename):
166
return (filename in (self.BZR_DUMMY_FILE, ))
169
super(BzrGitMapping, self).__init__(foreign_vcs_git)
171
def __eq__(self, other):
172
return (type(self) == type(other)
173
and self.revid_prefix == other.revid_prefix)
176
def revision_id_foreign_to_bzr(cls, git_rev_id):
177
"""Convert a git revision id handle to a Bazaar revision id."""
178
from dulwich.protocol import ZERO_SHA
179
if git_rev_id == ZERO_SHA:
181
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
184
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
185
"""Convert a Bazaar revision id to a git revision id handle."""
186
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
187
raise errors.InvalidRevisionId(bzr_rev_id, cls)
188
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
190
def generate_file_id(self, path):
191
# Git paths are just bytestrings
192
# We must just hope they are valid UTF-8..
193
if isinstance(path, str):
194
path = path.encode("utf-8")
197
return FILE_ID_PREFIX + escape_file_id(path)
199
def parse_file_id(self, file_id):
200
if file_id == ROOT_ID:
202
if not file_id.startswith(FILE_ID_PREFIX):
204
return decode_git_path(unescape_file_id(file_id[len(FILE_ID_PREFIX):]))
206
def import_unusual_file_modes(self, rev, unusual_file_modes):
207
if unusual_file_modes:
208
ret = [(path, unusual_file_modes[path])
209
for path in sorted(unusual_file_modes.keys())]
210
rev.properties[u'file-modes'] = bencode.bencode(ret)
212
def export_unusual_file_modes(self, rev):
214
file_modes = rev.properties[u'file-modes']
218
return dict(bencode.bdecode(file_modes.encode("utf-8")))
220
def _generate_git_svn_metadata(self, rev, encoding):
222
git_svn_id = rev.properties[u"git-svn-id"]
226
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
228
def _generate_hg_message_tail(self, rev):
232
for name in rev.properties:
233
if name == u'hg:extra:branch':
234
branch = rev.properties[u'hg:extra:branch']
235
elif name.startswith(u'hg:extra'):
236
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
237
rev.properties[name])
238
elif name == u'hg:renames':
239
renames = bencode.bdecode(base64.b64decode(
240
rev.properties[u'hg:renames']))
241
# TODO: Export other properties as 'bzr:' extras?
242
ret = format_hg_metadata(renames, branch, extra)
243
if not isinstance(ret, bytes):
247
def _extract_git_svn_metadata(self, rev, message):
248
lines = message.split("\n")
249
if not (lines[-1] == "" and len(lines) >= 2 and
250
lines[-2].startswith("git-svn-id:")):
252
git_svn_id = lines[-2].split(": ", 1)[1]
253
rev.properties[u'git-svn-id'] = git_svn_id
254
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
255
# FIXME: Convert this to converted-from property somehow..
256
return "\n".join(lines[:-2])
258
def _extract_hg_metadata(self, rev, message):
259
(message, renames, branch, extra) = extract_hg_metadata(message)
260
if branch is not None:
261
rev.properties[u'hg:extra:branch'] = branch
262
for name, value in extra.items():
263
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
265
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
266
[(new, old) for (old, new) in renames.items()]))
269
def _extract_bzr_metadata(self, rev, message):
270
(message, metadata) = extract_bzr_metadata(message)
271
return message, metadata
273
def _decode_commit_message(self, rev, message, encoding):
274
return message.decode(encoding), CommitSupplement()
276
def _encode_commit_message(self, rev, message, encoding):
277
return message.encode(encoding)
279
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
281
"""Turn a Bazaar revision in to a Git commit
283
:param tree_sha: Tree sha for the commit
284
:param parent_lookup: Function for looking up the GIT sha equiv of a
286
:param lossy: Whether to store roundtripping information.
287
:param verifiers: Verifiers info
288
:return dulwich.objects.Commit represent the revision:
290
from dulwich.objects import Commit, Tag
292
commit.tree = tree_sha
294
metadata = CommitSupplement()
295
metadata.verifiers = verifiers
299
for p in rev.parent_ids:
301
git_p = parent_lookup(p)
304
if metadata is not None:
305
metadata.explicit_parent_ids = rev.parent_ids
306
if git_p is not None:
308
raise AssertionError("unexpected length for %r" % git_p)
309
parents.append(git_p)
310
commit.parents = parents
312
encoding = rev.properties[u'git-explicit-encoding']
314
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
316
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
320
commit.committer = fix_person_identifier(rev.committer.encode(
322
commit.author = fix_person_identifier(
323
rev.get_apparent_authors()[0].encode(encoding))
324
# TODO(jelmer): Don't use this hack.
325
long = getattr(__builtins__, 'long', int)
326
commit.commit_time = long(rev.timestamp)
327
if u'author-timestamp' in rev.properties:
328
commit.author_time = long(rev.properties[u'author-timestamp'])
330
commit.author_time = commit.commit_time
331
commit._commit_timezone_neg_utc = (
332
u"commit-timezone-neg-utc" in rev.properties)
333
commit.commit_timezone = rev.timezone
334
commit._author_timezone_neg_utc = (
335
u"author-timezone-neg-utc" in rev.properties)
336
if u'author-timezone' in rev.properties:
337
commit.author_timezone = int(rev.properties[u'author-timezone'])
339
commit.author_timezone = commit.commit_timezone
340
if u'git-gpg-signature' in rev.properties:
341
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
342
'utf-8', 'surrogateescape')
343
commit.message = self._encode_commit_message(rev, rev.message,
345
if not isinstance(commit.message, bytes):
346
raise TypeError(commit.message)
347
if metadata is not None:
349
mapping_registry.parse_revision_id(rev.revision_id)
350
except errors.InvalidRevisionId:
351
metadata.revision_id = rev.revision_id
352
mapping_properties = set(
353
[u'author', u'author-timezone', u'author-timezone-neg-utc',
354
u'commit-timezone-neg-utc', u'git-implicit-encoding',
355
u'git-gpg-signature', u'git-explicit-encoding',
356
u'author-timestamp', u'file-modes'])
357
for k, v in rev.properties.items():
358
if k not in mapping_properties:
359
metadata.properties[k] = v
360
if not lossy and metadata:
361
if self.roundtripping:
362
commit.message = inject_bzr_metadata(commit.message, metadata,
366
None, None, self, revision_id=rev.revision_id)
367
if not isinstance(commit.message, bytes):
368
raise TypeError(commit.message)
370
propname = u'git-mergetag-0'
371
while propname in rev.properties:
372
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
374
propname = u'git-mergetag-%d' % i
375
if u'git-extra' in rev.properties:
378
for l in rev.properties[u'git-extra'].splitlines()])
381
def get_revision_id(self, commit):
383
encoding = commit.encoding.decode('ascii')
387
message, metadata = self._decode_commit_message(
388
None, commit.message, encoding)
389
except UnicodeDecodeError:
392
if metadata.revision_id:
393
return metadata.revision_id
394
return self.revision_id_foreign_to_bzr(commit.id)
396
def import_commit(self, commit, lookup_parent_revid, strict=True):
397
"""Convert a git commit to a bzr revision.
399
:return: a `breezy.revision.Revision` object, foreign revid and a
403
raise AssertionError("Commit object can't be None")
404
rev = ForeignRevision(commit.id, self,
405
self.revision_id_foreign_to_bzr(commit.id))
406
rev.git_metadata = None
408
def decode_using_encoding(rev, commit, encoding):
409
rev.committer = commit.committer.decode(encoding)
410
if commit.committer != commit.author:
411
rev.properties[u'author'] = commit.author.decode(encoding)
412
rev.message, rev.git_metadata = self._decode_commit_message(
413
rev, commit.message, encoding)
414
if commit.encoding is not None:
415
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode(
417
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
419
for encoding in ('utf-8', 'latin1'):
421
decode_using_encoding(rev, commit, encoding)
422
except UnicodeDecodeError:
425
if encoding != 'utf-8':
426
rev.properties[u'git-implicit-encoding'] = encoding
428
if commit.commit_time != commit.author_time:
429
rev.properties[u'author-timestamp'] = str(commit.author_time)
430
if commit.commit_timezone != commit.author_timezone:
431
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
432
if commit._author_timezone_neg_utc:
433
rev.properties[u'author-timezone-neg-utc'] = ""
434
if commit._commit_timezone_neg_utc:
435
rev.properties[u'commit-timezone-neg-utc'] = ""
437
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode(
438
'utf-8', 'surrogateescape')
440
for i, tag in enumerate(commit.mergetag):
441
rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string()
442
rev.timestamp = commit.commit_time
443
rev.timezone = commit.commit_timezone
444
rev.parent_ids = None
445
if rev.git_metadata is not None:
446
md = rev.git_metadata
447
roundtrip_revid = md.revision_id
448
if md.explicit_parent_ids:
449
rev.parent_ids = md.explicit_parent_ids
450
rev.properties.update(md.properties)
451
verifiers = md.verifiers
453
roundtrip_revid = None
455
if rev.parent_ids is None:
457
for p in commit.parents:
459
parents.append(lookup_parent_revid(p))
461
parents.append(self.revision_id_foreign_to_bzr(p))
462
rev.parent_ids = list(parents)
463
unknown_extra_fields = []
465
for k, v in commit.extra:
466
if k == HG_RENAME_SOURCE:
467
extra_lines.append(k + b' ' + v + b'\n')
469
hgk, hgv = v.split(b':', 1)
470
if hgk not in (HG_EXTRA_AMEND_SOURCE, ) and strict:
471
raise UnknownMercurialCommitExtra(commit, [hgk])
472
extra_lines.append(k + b' ' + v + b'\n')
474
unknown_extra_fields.append(k)
475
if unknown_extra_fields and strict:
476
raise UnknownCommitExtra(
478
[f.decode('ascii', 'replace') for f in unknown_extra_fields])
480
rev.properties[u'git-extra'] = b''.join(extra_lines)
481
return rev, roundtrip_revid, verifiers
484
class BzrGitMappingv1(BzrGitMapping):
485
revid_prefix = b'git-v1'
489
return self.revid_prefix
492
class BzrGitMappingExperimental(BzrGitMappingv1):
493
revid_prefix = b'git-experimental'
495
roundtripping = False
497
BZR_DUMMY_FILE = '.bzrdummy'
499
def _decode_commit_message(self, rev, message, encoding):
502
message = self._extract_hg_metadata(rev, message)
503
message = self._extract_git_svn_metadata(rev, message)
504
message, metadata = self._extract_bzr_metadata(rev, message)
505
return message.decode(encoding), metadata
507
def _encode_commit_message(self, rev, message, encoding):
508
ret = message.encode(encoding)
509
ret += self._generate_hg_message_tail(rev)
510
ret += self._generate_git_svn_metadata(rev, encoding)
513
def import_commit(self, commit, lookup_parent_revid, strict=True):
514
rev, roundtrip_revid, verifiers = super(
515
BzrGitMappingExperimental, self).import_commit(
516
commit, lookup_parent_revid, strict)
517
rev.properties[u'converted_revision'] = "git %s\n" % commit.id
518
return rev, roundtrip_revid, verifiers
521
class GitMappingRegistry(VcsMappingRegistry):
522
"""Registry with available git mappings."""
524
def revision_id_bzr_to_foreign(self, bzr_revid):
525
if bzr_revid == NULL_REVISION:
526
from dulwich.protocol import ZERO_SHA
527
return ZERO_SHA, None
528
if not bzr_revid.startswith(b"git-"):
529
raise errors.InvalidRevisionId(bzr_revid, None)
530
(mapping_version, git_sha) = bzr_revid.split(b":", 1)
531
mapping = self.get(mapping_version)
532
return mapping.revision_id_bzr_to_foreign(bzr_revid)
534
parse_revision_id = revision_id_bzr_to_foreign
537
mapping_registry = GitMappingRegistry()
538
mapping_registry.register_lazy(b'git-v1', __name__,
540
mapping_registry.register_lazy(b'git-experimental',
541
__name__, "BzrGitMappingExperimental")
542
# Uncomment the next line to enable the experimental bzr-git mappings.
543
# This will make sure all bzr metadata is pushed into git, allowing for
544
# full roundtripping later.
545
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
546
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
547
# mapping_registry.set_default('git-experimental')
548
mapping_registry.set_default(b'git-v1')
551
class ForeignGit(ForeignVcs):
552
"""The Git Stupid Content Tracker"""
555
def branch_format(self):
556
from .branch import LocalGitBranchFormat
557
return LocalGitBranchFormat()
560
def repository_format(self):
561
from .repository import GitRepositoryFormat
562
return GitRepositoryFormat()
565
super(ForeignGit, self).__init__(mapping_registry)
566
self.abbreviation = "git"
569
def serialize_foreign_revid(self, foreign_revid):
573
def show_foreign_revid(cls, foreign_revid):
574
return {"git commit": foreign_revid.decode('utf-8')}
577
foreign_vcs_git = ForeignGit()
578
default_mapping = mapping_registry.get_default()()
581
def symlink_to_blob(symlink_target):
582
from dulwich.objects import Blob
584
if isinstance(symlink_target, str):
585
symlink_target = encode_git_path(symlink_target)
586
blob.data = symlink_target
590
def mode_is_executable(mode):
591
"""Check if mode should be considered executable."""
592
return bool(mode & 0o111)
596
"""Determine the Bazaar inventory kind based on Unix file mode."""
599
entry_kind = (mode & 0o700000) / 0o100000
602
elif entry_kind == 1:
603
file_kind = (mode & 0o70000) / 0o10000
609
return 'tree-reference'
611
raise AssertionError(
612
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
614
raise AssertionError(
615
"Unknown kind, perms=%r." % (mode,))
618
def object_mode(kind, executable):
619
if kind == 'directory':
621
elif kind == 'symlink':
627
mode = stat.S_IFREG | 0o644
631
elif kind == 'tree-reference':
632
from dulwich.objects import S_IFGITLINK
638
def entry_mode(entry):
639
"""Determine the git file mode for an inventory entry."""
640
return object_mode(entry.kind, getattr(entry, 'executable', False))
643
def extract_unusual_modes(rev):
645
foreign_revid, mapping = mapping_registry.parse_revision_id(
647
except errors.InvalidRevisionId:
650
return mapping.export_unusual_file_modes(rev)
653
def parse_git_svn_id(text):
654
(head, uuid) = text.rsplit(" ", 1)
655
(full_url, rev) = head.rsplit("@", 1)
656
return (full_url, int(rev), uuid)
659
def needs_roundtripping(repo, revid):
661
mapping_registry.parse_revision_id(revid)
662
except errors.InvalidRevisionId:
added
removed
breezy/git/mapping.py
