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:
118
if text.rindex(b">") < text.rindex(b"<"):
119
raise ValueError(text)
120
username, email = text.split(b"<", 2)[-2:]
121
email = email.split(b">", 1)[0]
122
if username.endswith(b" "):
123
username = username[:-1]
124
return b"%s <%s>" % (username, email)
127
def warn_escaped(commit, num_escaped):
128
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
129
"to regenerate the SHA map.", num_escaped, commit)
132
def warn_unusual_mode(commit, path, mode):
133
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
134
"property. ", mode, path, commit)
137
class BzrGitMapping(foreign.VcsMapping):
138
"""Class that maps between Git and Bazaar semantics."""
141
BZR_DUMMY_FILE = None
143
def is_special_file(self, filename):
144
return (filename in (self.BZR_DUMMY_FILE, ))
147
super(BzrGitMapping, self).__init__(foreign_vcs_git)
149
def __eq__(self, other):
150
return (type(self) == type(other)
151
and self.revid_prefix == other.revid_prefix)
154
def revision_id_foreign_to_bzr(cls, git_rev_id):
155
"""Convert a git revision id handle to a Bazaar revision id."""
156
from dulwich.protocol import ZERO_SHA
157
if git_rev_id == ZERO_SHA:
159
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
162
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
163
"""Convert a Bazaar revision id to a git revision id handle."""
164
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
165
raise errors.InvalidRevisionId(bzr_rev_id, cls)
166
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
168
def generate_file_id(self, path):
169
# Git paths are just bytestrings
170
# We must just hope they are valid UTF-8..
171
if isinstance(path, str):
172
path = path.encode("utf-8")
175
return FILE_ID_PREFIX + escape_file_id(path)
177
def parse_file_id(self, file_id):
178
if file_id == ROOT_ID:
180
if not file_id.startswith(FILE_ID_PREFIX):
182
return unescape_file_id(file_id[len(FILE_ID_PREFIX):]).decode('utf-8')
184
def revid_as_refname(self, revid):
185
if not isinstance(revid, bytes):
186
raise TypeError(revid)
187
revid = revid.decode('utf-8')
188
quoted_revid = urlutils.quote(revid)
189
return b"refs/bzr/" + quoted_revid.encode('utf-8')
191
def import_unusual_file_modes(self, rev, unusual_file_modes):
192
if unusual_file_modes:
193
ret = [(path, unusual_file_modes[path])
194
for path in sorted(unusual_file_modes.keys())]
195
rev.properties[u'file-modes'] = bencode.bencode(ret)
197
def export_unusual_file_modes(self, rev):
199
file_modes = rev.properties[u'file-modes']
203
return dict(bencode.bdecode(file_modes.encode("utf-8")))
205
def _generate_git_svn_metadata(self, rev, encoding):
207
git_svn_id = rev.properties[u"git-svn-id"]
211
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
213
def _generate_hg_message_tail(self, rev):
217
for name in rev.properties:
218
if name == u'hg:extra:branch':
219
branch = rev.properties[u'hg:extra:branch']
220
elif name.startswith(u'hg:extra'):
221
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
222
rev.properties[name])
223
elif name == u'hg:renames':
224
renames = bencode.bdecode(base64.b64decode(
225
rev.properties[u'hg:renames']))
226
# TODO: Export other properties as 'bzr:' extras?
227
ret = format_hg_metadata(renames, branch, extra)
228
if not isinstance(ret, bytes):
232
def _extract_git_svn_metadata(self, rev, message):
233
lines = message.split("\n")
234
if not (lines[-1] == "" and len(lines) >= 2 and
235
lines[-2].startswith("git-svn-id:")):
237
git_svn_id = lines[-2].split(": ", 1)[1]
238
rev.properties[u'git-svn-id'] = git_svn_id
239
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
240
# FIXME: Convert this to converted-from property somehow..
241
return "\n".join(lines[:-2])
243
def _extract_hg_metadata(self, rev, message):
244
(message, renames, branch, extra) = extract_hg_metadata(message)
245
if branch is not None:
246
rev.properties[u'hg:extra:branch'] = branch
247
for name, value in extra.items():
248
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
250
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
251
[(new, old) for (old, new) in renames.items()]))
254
def _extract_bzr_metadata(self, rev, message):
255
(message, metadata) = extract_bzr_metadata(message)
256
return message, metadata
258
def _decode_commit_message(self, rev, message, encoding):
259
return message.decode(encoding), CommitSupplement()
261
def _encode_commit_message(self, rev, message, encoding):
262
return message.encode(encoding)
264
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
266
"""Turn a Bazaar revision in to a Git commit
268
:param tree_sha: Tree sha for the commit
269
:param parent_lookup: Function for looking up the GIT sha equiv of a
271
:param lossy: Whether to store roundtripping information.
272
:param verifiers: Verifiers info
273
:return dulwich.objects.Commit represent the revision:
275
from dulwich.objects import Commit, Tag
277
commit.tree = tree_sha
279
metadata = CommitSupplement()
280
metadata.verifiers = verifiers
284
for p in rev.parent_ids:
286
git_p = parent_lookup(p)
289
if metadata is not None:
290
metadata.explicit_parent_ids = rev.parent_ids
291
if git_p is not None:
293
raise AssertionError("unexpected length for %r" % git_p)
294
parents.append(git_p)
295
commit.parents = parents
297
encoding = rev.properties[u'git-explicit-encoding']
299
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
301
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
305
commit.committer = fix_person_identifier(rev.committer.encode(
307
commit.author = fix_person_identifier(
308
rev.get_apparent_authors()[0].encode(encoding))
309
# TODO(jelmer): Don't use this hack.
310
long = getattr(__builtins__, 'long', int)
311
commit.commit_time = long(rev.timestamp)
312
if u'author-timestamp' in rev.properties:
313
commit.author_time = long(rev.properties[u'author-timestamp'])
315
commit.author_time = commit.commit_time
316
commit._commit_timezone_neg_utc = (
317
u"commit-timezone-neg-utc" in rev.properties)
318
commit.commit_timezone = rev.timezone
319
commit._author_timezone_neg_utc = (
320
u"author-timezone-neg-utc" in rev.properties)
321
if u'author-timezone' in rev.properties:
322
commit.author_timezone = int(rev.properties[u'author-timezone'])
324
commit.author_timezone = commit.commit_timezone
325
if u'git-gpg-signature' in rev.properties:
326
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
328
if u'git-gpg-signature-b64' in rev.properties:
329
commit.gpgsig = base64.b64decode(rev.properties[u'git-gpg-signature-b64'])
330
commit.message = self._encode_commit_message(rev, rev.message,
332
if not isinstance(commit.message, bytes):
333
raise TypeError(commit.message)
334
if metadata is not None:
336
mapping_registry.parse_revision_id(rev.revision_id)
337
except errors.InvalidRevisionId:
338
metadata.revision_id = rev.revision_id
339
mapping_properties = set(
340
[u'author', u'author-timezone', u'author-timezone-neg-utc',
341
u'commit-timezone-neg-utc', u'git-implicit-encoding',
342
u'git-gpg-signature', u'git-gpg-signature-b64',
343
u'git-explicit-encoding',
344
u'author-timestamp', u'file-modes'])
345
for k, v in rev.properties.items():
346
if k not in mapping_properties:
347
metadata.properties[k] = v
348
if not lossy and metadata:
349
if self.roundtripping:
350
commit.message = inject_bzr_metadata(commit.message, metadata,
354
None, None, self, revision_id=rev.revision_id)
355
if not isinstance(commit.message, bytes):
356
raise TypeError(commit.message)
358
propname = u'git-mergetag-0'
359
while propname in rev.properties:
360
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
362
propname = u'git-mergetag-%d' % i
363
if u'git-extra' in rev.properties:
366
for l in rev.properties[u'git-extra'].splitlines()])
369
def get_revision_id(self, commit):
371
encoding = commit.encoding.decode('ascii')
375
message, metadata = self._decode_commit_message(
376
None, commit.message, encoding)
377
except UnicodeDecodeError:
380
if metadata.revision_id:
381
return metadata.revision_id
382
return self.revision_id_foreign_to_bzr(commit.id)
384
def import_commit(self, commit, lookup_parent_revid, strict=True):
385
"""Convert a git commit to a bzr revision.
387
:return: a `breezy.revision.Revision` object, foreign revid and a
391
raise AssertionError("Commit object can't be None")
392
rev = ForeignRevision(commit.id, self,
393
self.revision_id_foreign_to_bzr(commit.id))
394
rev.git_metadata = None
396
def decode_using_encoding(rev, commit, encoding):
397
rev.committer = commit.committer.decode(encoding)
398
if commit.committer != commit.author:
399
rev.properties[u'author'] = commit.author.decode(encoding)
400
rev.message, rev.git_metadata = self._decode_commit_message(
401
rev, commit.message, encoding)
402
if commit.encoding is not None:
403
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode(
405
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
407
for encoding in ('utf-8', 'latin1'):
409
decode_using_encoding(rev, commit, encoding)
410
except UnicodeDecodeError:
413
if encoding != 'utf-8':
414
rev.properties[u'git-implicit-encoding'] = encoding
416
if commit.commit_time != commit.author_time:
417
rev.properties[u'author-timestamp'] = str(commit.author_time)
418
if commit.commit_timezone != commit.author_timezone:
419
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
420
if commit._author_timezone_neg_utc:
421
rev.properties[u'author-timezone-neg-utc'] = ""
422
if commit._commit_timezone_neg_utc:
423
rev.properties[u'commit-timezone-neg-utc'] = ""
426
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode(
428
except UnicodeDecodeError:
429
rev.properties[u'git-gpg-signature-b64'] = base64.b64encode(
432
for i, tag in enumerate(commit.mergetag):
433
rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string()
434
rev.timestamp = commit.commit_time
435
rev.timezone = commit.commit_timezone
436
rev.parent_ids = None
437
if rev.git_metadata is not None:
438
md = rev.git_metadata
439
roundtrip_revid = md.revision_id
440
if md.explicit_parent_ids:
441
rev.parent_ids = md.explicit_parent_ids
442
rev.properties.update(md.properties)
443
verifiers = md.verifiers
445
roundtrip_revid = None
447
if rev.parent_ids is None:
449
for p in commit.parents:
451
parents.append(lookup_parent_revid(p))
453
parents.append(self.revision_id_foreign_to_bzr(p))
454
rev.parent_ids = list(parents)
455
unknown_extra_fields = []
457
for k, v in commit.extra:
458
if k == HG_RENAME_SOURCE:
459
extra_lines.append(k + b' ' + v + b'\n')
461
hgk, hgv = v.split(b':', 1)
462
if hgk not in (HG_EXTRA_AMEND_SOURCE, ) and strict:
463
raise UnknownMercurialCommitExtra(commit, [hgk])
464
extra_lines.append(k + b' ' + v + b'\n')
466
unknown_extra_fields.append(k)
467
if unknown_extra_fields and strict:
468
raise UnknownCommitExtra(
470
[f.decode('ascii', 'replace') for f in unknown_extra_fields])
472
rev.properties[u'git-extra'] = b''.join(extra_lines)
473
return rev, roundtrip_revid, verifiers
476
class BzrGitMappingv1(BzrGitMapping):
477
revid_prefix = b'git-v1'
481
return self.revid_prefix
484
class BzrGitMappingExperimental(BzrGitMappingv1):
485
revid_prefix = b'git-experimental'
487
roundtripping = False
489
BZR_DUMMY_FILE = '.bzrdummy'
491
def _decode_commit_message(self, rev, message, encoding):
494
message = self._extract_hg_metadata(rev, message)
495
message = self._extract_git_svn_metadata(rev, message)
496
message, metadata = self._extract_bzr_metadata(rev, message)
497
return message.decode(encoding), metadata
499
def _encode_commit_message(self, rev, message, encoding):
500
ret = message.encode(encoding)
501
ret += self._generate_hg_message_tail(rev)
502
ret += self._generate_git_svn_metadata(rev, encoding)
505
def import_commit(self, commit, lookup_parent_revid, strict=True):
506
rev, roundtrip_revid, verifiers = super(
507
BzrGitMappingExperimental, self).import_commit(
508
commit, lookup_parent_revid, strict)
509
rev.properties[u'converted_revision'] = "git %s\n" % commit.id
510
return rev, roundtrip_revid, verifiers
513
class GitMappingRegistry(VcsMappingRegistry):
514
"""Registry with available git mappings."""
516
def revision_id_bzr_to_foreign(self, bzr_revid):
517
if bzr_revid == NULL_REVISION:
518
from dulwich.protocol import ZERO_SHA
519
return ZERO_SHA, None
520
if not bzr_revid.startswith(b"git-"):
521
raise errors.InvalidRevisionId(bzr_revid, None)
522
(mapping_version, git_sha) = bzr_revid.split(b":", 1)
523
mapping = self.get(mapping_version)
524
return mapping.revision_id_bzr_to_foreign(bzr_revid)
526
parse_revision_id = revision_id_bzr_to_foreign
529
mapping_registry = GitMappingRegistry()
530
mapping_registry.register_lazy(b'git-v1', __name__,
532
mapping_registry.register_lazy(b'git-experimental',
533
__name__, "BzrGitMappingExperimental")
534
# Uncomment the next line to enable the experimental bzr-git mappings.
535
# This will make sure all bzr metadata is pushed into git, allowing for
536
# full roundtripping later.
537
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
538
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
539
# mapping_registry.set_default('git-experimental')
540
mapping_registry.set_default(b'git-v1')
543
class ForeignGit(ForeignVcs):
544
"""The Git Stupid Content Tracker"""
547
def branch_format(self):
548
from .branch import LocalGitBranchFormat
549
return LocalGitBranchFormat()
552
def repository_format(self):
553
from .repository import GitRepositoryFormat
554
return GitRepositoryFormat()
557
super(ForeignGit, self).__init__(mapping_registry)
558
self.abbreviation = "git"
561
def serialize_foreign_revid(self, foreign_revid):
565
def show_foreign_revid(cls, foreign_revid):
566
return {"git commit": foreign_revid.decode('utf-8')}
569
foreign_vcs_git = ForeignGit()
570
default_mapping = mapping_registry.get_default()()
573
def symlink_to_blob(symlink_target):
574
from dulwich.objects import Blob
576
if isinstance(symlink_target, str):
577
symlink_target = symlink_target.encode('utf-8')
578
blob.data = symlink_target
582
def mode_is_executable(mode):
583
"""Check if mode should be considered executable."""
584
return bool(mode & 0o111)
588
"""Determine the Bazaar inventory kind based on Unix file mode."""
591
entry_kind = (mode & 0o700000) / 0o100000
594
elif entry_kind == 1:
595
file_kind = (mode & 0o70000) / 0o10000
601
return 'tree-reference'
603
raise AssertionError(
604
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
606
raise AssertionError(
607
"Unknown kind, perms=%r." % (mode,))
610
def object_mode(kind, executable):
611
if kind == 'directory':
613
elif kind == 'symlink':
619
mode = stat.S_IFREG | 0o644
623
elif kind == 'tree-reference':
624
from dulwich.objects import S_IFGITLINK
630
def entry_mode(entry):
631
"""Determine the git file mode for an inventory entry."""
632
return object_mode(entry.kind, getattr(entry, 'executable', False))
635
def extract_unusual_modes(rev):
637
foreign_revid, mapping = mapping_registry.parse_revision_id(
639
except errors.InvalidRevisionId:
642
return mapping.export_unusual_file_modes(rev)
645
def parse_git_svn_id(text):
646
(head, uuid) = text.rsplit(" ", 1)
647
(full_url, rev) = head.rsplit("@", 1)
648
return (full_url, int(rev), uuid)
651
def needs_roundtripping(repo, revid):
653
mapping_registry.parse_revision_id(revid)
654
except errors.InvalidRevisionId:
added
removed
breezy/git/mapping.py
