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 (
41
from ..sixish import (
49
UnknownMercurialCommitExtra,
55
from .roundtrip import (
59
deserialize_fileid_map,
64
DEFAULT_FILE_MODE = stat.S_IFREG | 0o644
65
HG_RENAME_SOURCE = b"HG:rename-source"
66
HG_EXTRA = b"HG:extra"
68
# This HG extra is used to indicate the commit that this commit was based on.
69
HG_EXTRA_AMEND_SOURCE = b"amend_source"
71
FILE_ID_PREFIX = b'git:'
74
ROOT_ID = b"TREE_ROOT"
77
def escape_file_id(file_id):
78
file_id = file_id.replace(b'_', b'__')
79
file_id = file_id.replace(b' ', b'_s')
80
file_id = file_id.replace(b'\x0c', b'_c')
84
def unescape_file_id(file_id):
87
while i < len(file_id):
88
if file_id[i:i + 1] != b'_':
89
ret.append(file_id[i])
91
if file_id[i + 1:i + 2] == b'_':
93
elif file_id[i + 1:i + 2] == b's':
95
elif file_id[i + 1:i + 2] == b'c':
96
ret.append(b"\x0c"[0])
98
raise ValueError("unknown escape character %s" %
105
def fix_person_identifier(text):
106
if b"<" not in text and b">" not in text:
110
if text.rindex(b">") < text.rindex(b"<"):
111
raise ValueError(text)
112
username, email = text.split(b"<", 2)[-2:]
113
email = email.split(b">", 1)[0]
114
if username.endswith(b" "):
115
username = username[:-1]
116
return b"%s <%s>" % (username, email)
119
def warn_escaped(commit, num_escaped):
120
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
121
"to regenerate the SHA map.", num_escaped, commit)
124
def warn_unusual_mode(commit, path, mode):
125
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
126
"property. ", mode, path, commit)
129
class BzrGitMapping(foreign.VcsMapping):
130
"""Class that maps between Git and Bazaar semantics."""
133
BZR_FILE_IDS_FILE = None
135
BZR_DUMMY_FILE = None
137
def is_special_file(self, filename):
138
return (filename in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE))
141
super(BzrGitMapping, self).__init__(foreign_vcs_git)
143
def __eq__(self, other):
144
return (type(self) == type(other)
145
and self.revid_prefix == other.revid_prefix)
148
def revision_id_foreign_to_bzr(cls, git_rev_id):
149
"""Convert a git revision id handle to a Bazaar revision id."""
150
from dulwich.protocol import ZERO_SHA
151
if git_rev_id == ZERO_SHA:
153
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
156
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
157
"""Convert a Bazaar revision id to a git revision id handle."""
158
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
159
raise errors.InvalidRevisionId(bzr_rev_id, cls)
160
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
162
def generate_file_id(self, path):
163
# Git paths are just bytestrings
164
# We must just hope they are valid UTF-8..
165
if isinstance(path, text_type):
166
path = path.encode("utf-8")
169
return FILE_ID_PREFIX + escape_file_id(path)
171
def parse_file_id(self, file_id):
172
if file_id == ROOT_ID:
174
if not file_id.startswith(FILE_ID_PREFIX):
176
return unescape_file_id(file_id[len(FILE_ID_PREFIX):]).decode('utf-8')
178
def revid_as_refname(self, revid):
179
if not isinstance(revid, bytes):
180
raise TypeError(revid)
182
revid = revid.decode('utf-8')
183
quoted_revid = urlutils.quote(revid)
184
return b"refs/bzr/" + quoted_revid.encode('utf-8')
186
def import_unusual_file_modes(self, rev, unusual_file_modes):
187
if unusual_file_modes:
188
ret = [(path, unusual_file_modes[path])
189
for path in sorted(unusual_file_modes.keys())]
190
rev.properties[u'file-modes'] = bencode.bencode(ret)
192
def export_unusual_file_modes(self, rev):
194
file_modes = rev.properties[u'file-modes']
198
return dict(bencode.bdecode(file_modes.encode("utf-8")))
200
def _generate_git_svn_metadata(self, rev, encoding):
202
git_svn_id = rev.properties[u"git-svn-id"]
206
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
208
def _generate_hg_message_tail(self, rev):
212
for name in rev.properties:
213
if name == u'hg:extra:branch':
214
branch = rev.properties[u'hg:extra:branch']
215
elif name.startswith(u'hg:extra'):
216
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
217
rev.properties[name])
218
elif name == u'hg:renames':
219
renames = bencode.bdecode(base64.b64decode(
220
rev.properties[u'hg:renames']))
221
# TODO: Export other properties as 'bzr:' extras?
222
ret = format_hg_metadata(renames, branch, extra)
223
if not isinstance(ret, bytes):
227
def _extract_git_svn_metadata(self, rev, message):
228
lines = message.split("\n")
229
if not (lines[-1] == "" and len(lines) >= 2 and
230
lines[-2].startswith("git-svn-id:")):
232
git_svn_id = lines[-2].split(": ", 1)[1]
233
rev.properties[u'git-svn-id'] = git_svn_id
234
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
235
# FIXME: Convert this to converted-from property somehow..
236
return "\n".join(lines[:-2])
238
def _extract_hg_metadata(self, rev, message):
239
(message, renames, branch, extra) = extract_hg_metadata(message)
240
if branch is not None:
241
rev.properties[u'hg:extra:branch'] = branch
242
for name, value in viewitems(extra):
243
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
245
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
246
[(new, old) for (old, new) in viewitems(renames)]))
249
def _extract_bzr_metadata(self, rev, message):
250
(message, metadata) = extract_bzr_metadata(message)
251
return message, metadata
253
def _decode_commit_message(self, rev, message, encoding):
254
return message.decode(encoding), CommitSupplement()
256
def _encode_commit_message(self, rev, message, encoding):
257
return message.encode(encoding)
259
def export_fileid_map(self, fileid_map):
260
"""Export a file id map to a fileid map.
262
:param fileid_map: File id map, mapping paths to file ids
263
:return: A Git blob object (or None if there are no entries)
265
from dulwich.objects import Blob
267
b.set_raw_chunks(serialize_fileid_map(fileid_map))
270
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
272
"""Turn a Bazaar revision in to a Git commit
274
:param tree_sha: Tree sha for the commit
275
:param parent_lookup: Function for looking up the GIT sha equiv of a
277
:param lossy: Whether to store roundtripping information.
278
:param verifiers: Verifiers info
279
:return dulwich.objects.Commit represent the revision:
281
from dulwich.objects import Commit, Tag
283
commit.tree = tree_sha
285
metadata = CommitSupplement()
286
metadata.verifiers = verifiers
290
for p in rev.parent_ids:
292
git_p = parent_lookup(p)
295
if metadata is not None:
296
metadata.explicit_parent_ids = rev.parent_ids
297
if git_p is not None:
299
raise AssertionError("unexpected length for %r" % git_p)
300
parents.append(git_p)
301
commit.parents = parents
303
encoding = rev.properties[u'git-explicit-encoding']
305
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
307
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
311
commit.committer = fix_person_identifier(rev.committer.encode(
313
commit.author = fix_person_identifier(
314
rev.get_apparent_authors()[0].encode(encoding))
315
# TODO(jelmer): Don't use this hack.
316
long = getattr(__builtins__, 'long', int)
317
commit.commit_time = long(rev.timestamp)
318
if u'author-timestamp' in rev.properties:
319
commit.author_time = long(rev.properties[u'author-timestamp'])
321
commit.author_time = commit.commit_time
322
commit._commit_timezone_neg_utc = (
323
u"commit-timezone-neg-utc" in rev.properties)
324
commit.commit_timezone = rev.timezone
325
commit._author_timezone_neg_utc = (
326
u"author-timezone-neg-utc" in rev.properties)
327
if u'author-timezone' in rev.properties:
328
commit.author_timezone = int(rev.properties[u'author-timezone'])
330
commit.author_timezone = commit.commit_timezone
331
if u'git-gpg-signature' in rev.properties:
332
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
334
commit.message = self._encode_commit_message(rev, rev.message,
336
if not isinstance(commit.message, bytes):
337
raise TypeError(commit.message)
338
if metadata is not None:
340
mapping_registry.parse_revision_id(rev.revision_id)
341
except errors.InvalidRevisionId:
342
metadata.revision_id = rev.revision_id
343
mapping_properties = set(
344
[u'author', u'author-timezone', u'author-timezone-neg-utc',
345
u'commit-timezone-neg-utc', u'git-implicit-encoding',
346
u'git-gpg-signature', u'git-explicit-encoding',
347
u'author-timestamp', u'file-modes'])
348
for k, v in viewitems(rev.properties):
349
if k not in mapping_properties:
350
metadata.properties[k] = v
351
if not lossy and metadata:
352
if self.roundtripping:
353
commit.message = inject_bzr_metadata(commit.message, metadata,
357
None, None, self, revision_id=rev.revision_id)
358
if not isinstance(commit.message, bytes):
359
raise TypeError(commit.message)
361
propname = u'git-mergetag-0'
362
while propname in rev.properties:
363
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
365
propname = u'git-mergetag-%d' % i
366
if u'git-extra' in rev.properties:
369
for l in rev.properties[u'git-extra'].splitlines()])
372
def import_fileid_map(self, blob):
373
"""Convert a git file id map blob.
375
:param blob: Git blob object with fileid map
376
:return: Dictionary mapping paths to file ids
378
return deserialize_fileid_map(blob.data)
380
def import_commit(self, commit, lookup_parent_revid):
381
"""Convert a git commit to a bzr revision.
383
:return: a `breezy.revision.Revision` object, foreign revid and a
387
raise AssertionError("Commit object can't be None")
388
rev = ForeignRevision(commit.id, self,
389
self.revision_id_foreign_to_bzr(commit.id))
390
rev.git_metadata = None
392
def decode_using_encoding(rev, commit, encoding):
393
rev.committer = commit.committer.decode(encoding)
394
if commit.committer != commit.author:
395
rev.properties[u'author'] = commit.author.decode(encoding)
396
rev.message, rev.git_metadata = self._decode_commit_message(
397
rev, commit.message, encoding)
398
if commit.encoding is not None:
399
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode(
401
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
403
for encoding in ('utf-8', 'latin1'):
405
decode_using_encoding(rev, commit, encoding)
406
except UnicodeDecodeError:
409
if encoding != 'utf-8':
410
rev.properties[u'git-implicit-encoding'] = encoding
412
if commit.commit_time != commit.author_time:
413
rev.properties[u'author-timestamp'] = str(commit.author_time)
414
if commit.commit_timezone != commit.author_timezone:
415
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
416
if commit._author_timezone_neg_utc:
417
rev.properties[u'author-timezone-neg-utc'] = ""
418
if commit._commit_timezone_neg_utc:
419
rev.properties[u'commit-timezone-neg-utc'] = ""
421
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode(
424
for i, tag in enumerate(commit.mergetag):
425
rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string()
426
rev.timestamp = commit.commit_time
427
rev.timezone = commit.commit_timezone
428
rev.parent_ids = None
429
if rev.git_metadata is not None:
430
md = rev.git_metadata
431
roundtrip_revid = md.revision_id
432
if md.explicit_parent_ids:
433
rev.parent_ids = md.explicit_parent_ids
434
rev.properties.update(md.properties)
435
verifiers = md.verifiers
437
roundtrip_revid = None
439
if rev.parent_ids is None:
441
for p in commit.parents:
443
parents.append(lookup_parent_revid(p))
445
parents.append(self.revision_id_foreign_to_bzr(p))
446
rev.parent_ids = list(parents)
447
unknown_extra_fields = []
449
for k, v in commit.extra:
450
if k == HG_RENAME_SOURCE:
451
extra_lines.append(k + b' ' + v + b'\n')
453
hgk, hgv = v.split(b':', 1)
454
if hgk not in (HG_EXTRA_AMEND_SOURCE, ):
455
raise UnknownMercurialCommitExtra(commit, [hgk])
456
extra_lines.append(k + b' ' + v + b'\n')
458
unknown_extra_fields.append(k)
459
if unknown_extra_fields:
460
raise UnknownCommitExtra(
462
[f.decode('ascii', 'replace') for f in unknown_extra_fields])
464
rev.properties[u'git-extra'] = b''.join(extra_lines)
465
return rev, roundtrip_revid, verifiers
467
def get_fileid_map(self, lookup_object, tree_sha):
468
"""Obtain a fileid map for a particular tree.
470
:param lookup_object: Function for looking up an object
471
:param tree_sha: SHA of the root tree
472
:return: GitFileIdMap instance
475
file_id_map_sha = lookup_object(
476
tree_sha)[self.BZR_FILE_IDS_FILE][1]
480
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
481
return GitFileIdMap(file_ids, self)
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'
497
BZR_FILE_IDS_FILE = '.bzrfileids'
499
BZR_DUMMY_FILE = '.bzrdummy'
501
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):
514
rev, roundtrip_revid, verifiers = super(
515
BzrGitMappingExperimental, self).import_commit(
516
commit, lookup_parent_revid)
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, text_type):
585
symlink_target = symlink_target.encode('utf-8')
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
class GitFileIdMap(object):
661
def __init__(self, file_ids, mapping):
662
self.file_ids = file_ids
664
self.mapping = mapping
666
def set_file_id(self, path, file_id):
667
if type(path) is not str:
668
raise TypeError(path)
669
if not isinstance(file_id, bytes):
670
raise TypeError(file_id)
671
self.file_ids[path] = file_id
673
def lookup_file_id(self, path):
674
if not isinstance(path, text_type):
675
raise TypeError(path)
677
file_id = self.file_ids[path]
679
file_id = self.mapping.generate_file_id(path)
680
if not isinstance(file_id, bytes):
681
raise TypeError(file_id)
684
def lookup_path(self, file_id):
685
if self.paths is None:
687
for k, v in viewitems(self.file_ids):
690
path = self.paths[file_id]
692
return self.mapping.parse_file_id(file_id)
694
if not isinstance(path, text_type):
695
raise TypeError(path)
699
return self.__class__(dict(self.file_ids), self.mapping)
702
def needs_roundtripping(repo, revid):
704
mapping_registry.parse_revision_id(revid)
705
except errors.InvalidRevisionId:
added
removed
breezy/git/mapping.py
