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
return file_id.replace(b'_', b'__').replace(b' ', b'_s').replace(b'\x0c', b'_c')
81
def unescape_file_id(file_id):
84
while i < len(file_id):
85
if file_id[i:i+1] != b'_':
86
ret.append(file_id[i])
88
if file_id[i+1:i+2] == b'_':
90
elif file_id[i+1:i+2] == b's':
92
elif file_id[i+1:i+2] == b'c':
93
ret.append(b"\x0c"[0])
95
raise ValueError("unknown escape character %s" %
102
def fix_person_identifier(text):
103
if not b"<" in text and not b">" in text:
107
if text.rindex(b">") < text.rindex(b"<"):
108
raise ValueError(text)
109
username, email = text.split(b"<", 2)[-2:]
110
email = email.split(b">", 1)[0]
111
if username.endswith(b" "):
112
username = username[:-1]
113
return b"%s <%s>" % (username, email)
116
def warn_escaped(commit, num_escaped):
117
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
118
"to regenerate the SHA map.", num_escaped, commit)
121
def warn_unusual_mode(commit, path, mode):
122
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
123
"property. ", mode, path, commit)
126
class BzrGitMapping(foreign.VcsMapping):
127
"""Class that maps between Git and Bazaar semantics."""
130
BZR_FILE_IDS_FILE = None
132
BZR_DUMMY_FILE = None
134
def is_special_file(self, filename):
135
return (filename in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE))
138
super(BzrGitMapping, self).__init__(foreign_vcs_git)
140
def __eq__(self, other):
141
return (type(self) == type(other) and
142
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 lines[-2].startswith("git-svn-id:")):
228
git_svn_id = lines[-2].split(": ", 1)[1]
229
rev.properties[u'git-svn-id'] = git_svn_id
230
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
231
# FIXME: Convert this to converted-from property somehow..
232
return "\n".join(lines[:-2])
234
def _extract_hg_metadata(self, rev, message):
235
(message, renames, branch, extra) = extract_hg_metadata(message)
236
if branch is not None:
237
rev.properties[u'hg:extra:branch'] = branch
238
for name, value in viewitems(extra):
239
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
241
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
242
[(new, old) for (old, new) in viewitems(renames)]))
245
def _extract_bzr_metadata(self, rev, message):
246
(message, metadata) = extract_bzr_metadata(message)
247
return message, metadata
249
def _decode_commit_message(self, rev, message, encoding):
250
return message.decode(encoding), CommitSupplement()
252
def _encode_commit_message(self, rev, message, encoding):
253
return message.encode(encoding)
255
def export_fileid_map(self, fileid_map):
256
"""Export a file id map to a fileid map.
258
:param fileid_map: File id map, mapping paths to file ids
259
:return: A Git blob object (or None if there are no entries)
261
from dulwich.objects import Blob
263
b.set_raw_chunks(serialize_fileid_map(fileid_map))
266
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
268
"""Turn a Bazaar revision in to a Git commit
270
:param tree_sha: Tree sha for the commit
271
:param parent_lookup: Function for looking up the GIT sha equiv of a
273
:param lossy: Whether to store roundtripping information.
274
:param verifiers: Verifiers info
275
:return dulwich.objects.Commit represent the revision:
277
from dulwich.objects import Commit, Tag
279
commit.tree = tree_sha
281
metadata = CommitSupplement()
282
metadata.verifiers = verifiers
286
for p in rev.parent_ids:
288
git_p = parent_lookup(p)
291
if metadata is not None:
292
metadata.explicit_parent_ids = rev.parent_ids
293
if git_p is not None:
295
raise AssertionError("unexpected length for %r" % git_p)
296
parents.append(git_p)
297
commit.parents = parents
299
encoding = rev.properties[u'git-explicit-encoding']
301
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
303
commit.encoding = rev.properties[u'git-explicit-encoding'].encode('ascii')
306
commit.committer = fix_person_identifier(rev.committer.encode(
308
commit.author = fix_person_identifier(
309
rev.get_apparent_authors()[0].encode(encoding))
310
# TODO(jelmer): Don't use this hack.
311
long = getattr(__builtins__, 'long', int)
312
commit.commit_time = long(rev.timestamp)
313
if u'author-timestamp' in rev.properties:
314
commit.author_time = long(rev.properties[u'author-timestamp'])
316
commit.author_time = commit.commit_time
317
commit._commit_timezone_neg_utc = u"commit-timezone-neg-utc" in rev.properties
318
commit.commit_timezone = rev.timezone
319
commit._author_timezone_neg_utc = u"author-timezone-neg-utc" in rev.properties
320
if u'author-timezone' in rev.properties:
321
commit.author_timezone = int(rev.properties[u'author-timezone'])
323
commit.author_timezone = commit.commit_timezone
324
if u'git-gpg-signature' in rev.properties:
325
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode('ascii')
326
commit.message = self._encode_commit_message(rev, rev.message,
328
if not isinstance(commit.message, bytes):
329
raise TypeError(commit.message)
330
if metadata is not None:
332
mapping_registry.parse_revision_id(rev.revision_id)
333
except errors.InvalidRevisionId:
334
metadata.revision_id = rev.revision_id
335
mapping_properties = set(
336
[u'author', u'author-timezone', u'author-timezone-neg-utc',
337
u'commit-timezone-neg-utc', u'git-implicit-encoding',
338
u'git-gpg-signature', u'git-explicit-encoding',
339
u'author-timestamp', u'file-modes'])
340
for k, v in viewitems(rev.properties):
341
if not k in mapping_properties:
342
metadata.properties[k] = v
343
if not lossy and metadata:
344
if self.roundtripping:
345
commit.message = inject_bzr_metadata(commit.message, metadata,
348
raise NoPushSupport(None, None, self, revision_id=rev.revision_id)
349
if not isinstance(commit.message, bytes):
350
raise TypeError(commit.message)
352
propname = u'git-mergetag-0'
353
while propname in rev.properties:
354
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
356
propname = u'git-mergetag-%d' % i
357
if u'git-extra' in rev.properties:
358
commit.extra.extend([l.split(b' ', 1) for l in rev.properties[u'git-extra'].splitlines()])
361
def import_fileid_map(self, blob):
362
"""Convert a git file id map blob.
364
:param blob: Git blob object with fileid map
365
:return: Dictionary mapping paths to file ids
367
return deserialize_fileid_map(blob.data)
369
def import_commit(self, commit, lookup_parent_revid):
370
"""Convert a git commit to a bzr revision.
372
:return: a `breezy.revision.Revision` object, foreign revid and a
376
raise AssertionError("Commit object can't be None")
377
rev = ForeignRevision(commit.id, self,
378
self.revision_id_foreign_to_bzr(commit.id))
379
rev.git_metadata = None
380
def decode_using_encoding(rev, commit, encoding):
381
rev.committer = commit.committer.decode(encoding)
382
if commit.committer != commit.author:
383
rev.properties[u'author'] = commit.author.decode(encoding)
384
rev.message, rev.git_metadata = self._decode_commit_message(
385
rev, commit.message, encoding)
386
if commit.encoding is not None:
387
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode('ascii')
388
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
390
for encoding in ('utf-8', 'latin1'):
392
decode_using_encoding(rev, commit, encoding)
393
except UnicodeDecodeError:
396
if encoding != 'utf-8':
397
rev.properties[u'git-implicit-encoding'] = encoding
399
if commit.commit_time != commit.author_time:
400
rev.properties[u'author-timestamp'] = str(commit.author_time)
401
if commit.commit_timezone != commit.author_timezone:
402
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
403
if commit._author_timezone_neg_utc:
404
rev.properties[u'author-timezone-neg-utc'] = ""
405
if commit._commit_timezone_neg_utc:
406
rev.properties[u'commit-timezone-neg-utc'] = ""
408
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode('ascii')
410
for i, tag in enumerate(commit.mergetag):
411
rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string()
412
rev.timestamp = commit.commit_time
413
rev.timezone = commit.commit_timezone
414
rev.parent_ids = None
415
if rev.git_metadata is not None:
416
md = rev.git_metadata
417
roundtrip_revid = md.revision_id
418
if md.explicit_parent_ids:
419
rev.parent_ids = md.explicit_parent_ids
420
rev.properties.update(md.properties)
421
verifiers = md.verifiers
423
roundtrip_revid = None
425
if rev.parent_ids is None:
427
for p in commit.parents:
429
parents.append(lookup_parent_revid(p))
431
parents.append(self.revision_id_foreign_to_bzr(p))
432
rev.parent_ids = list(parents)
433
unknown_extra_fields = []
435
for k, v in commit.extra:
436
if k == HG_RENAME_SOURCE:
437
extra_lines.append(k + b' ' + v + b'\n')
439
hgk, hgv = v.split(b':', 1)
440
if hgk not in (HG_EXTRA_AMEND_SOURCE, ):
441
raise UnknownMercurialCommitExtra(commit, hgk)
442
extra_lines.append(k + b' ' + v + b'\n')
444
unknown_extra_fields.append(k)
445
if unknown_extra_fields:
446
raise UnknownCommitExtra(
447
commit, [f.decode('ascii', 'replace') for f in unknown_extra_fields])
449
rev.properties[u'git-extra'] = b''.join(extra_lines)
450
return rev, roundtrip_revid, verifiers
452
def get_fileid_map(self, lookup_object, tree_sha):
453
"""Obtain a fileid map for a particular tree.
455
:param lookup_object: Function for looking up an object
456
:param tree_sha: SHA of the root tree
457
:return: GitFileIdMap instance
460
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
464
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
465
return GitFileIdMap(file_ids, self)
468
class BzrGitMappingv1(BzrGitMapping):
469
revid_prefix = b'git-v1'
473
return self.revid_prefix
476
class BzrGitMappingExperimental(BzrGitMappingv1):
477
revid_prefix = b'git-experimental'
481
BZR_FILE_IDS_FILE = '.bzrfileids'
483
BZR_DUMMY_FILE = '.bzrdummy'
485
def _decode_commit_message(self, rev, message, encoding):
486
message = self._extract_hg_metadata(rev, message)
487
message = self._extract_git_svn_metadata(rev, message)
488
message, metadata = self._extract_bzr_metadata(rev, message)
489
return message.decode(encoding), metadata
491
def _encode_commit_message(self, rev, message, encoding):
492
ret = message.encode(encoding)
493
ret += self._generate_hg_message_tail(rev)
494
ret += self._generate_git_svn_metadata(rev, encoding)
497
def import_commit(self, commit, lookup_parent_revid):
498
rev, roundtrip_revid, verifiers = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
499
rev.properties[u'converted_revision'] = "git %s\n" % commit.id
500
return rev, roundtrip_revid, verifiers
503
class GitMappingRegistry(VcsMappingRegistry):
504
"""Registry with available git mappings."""
506
def revision_id_bzr_to_foreign(self, bzr_revid):
507
if bzr_revid == NULL_REVISION:
508
from dulwich.protocol import ZERO_SHA
509
return ZERO_SHA, None
510
if not bzr_revid.startswith(b"git-"):
511
raise errors.InvalidRevisionId(bzr_revid, None)
512
(mapping_version, git_sha) = bzr_revid.split(b":", 1)
513
mapping = self.get(mapping_version)
514
return mapping.revision_id_bzr_to_foreign(bzr_revid)
516
parse_revision_id = revision_id_bzr_to_foreign
519
mapping_registry = GitMappingRegistry()
520
mapping_registry.register_lazy(b'git-v1', __name__,
522
mapping_registry.register_lazy(b'git-experimental',
523
__name__, "BzrGitMappingExperimental")
524
# Uncomment the next line to enable the experimental bzr-git mappings.
525
# This will make sure all bzr metadata is pushed into git, allowing for
526
# full roundtripping later.
527
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
528
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
529
#mapping_registry.set_default('git-experimental')
530
mapping_registry.set_default(b'git-v1')
533
class ForeignGit(ForeignVcs):
534
"""The Git Stupid Content Tracker"""
537
def branch_format(self):
538
from .branch import LocalGitBranchFormat
539
return LocalGitBranchFormat()
542
def repository_format(self):
543
from .repository import GitRepositoryFormat
544
return GitRepositoryFormat()
547
super(ForeignGit, self).__init__(mapping_registry)
548
self.abbreviation = "git"
551
def serialize_foreign_revid(self, foreign_revid):
555
def show_foreign_revid(cls, foreign_revid):
556
return { "git commit": foreign_revid.decode('utf-8') }
559
foreign_vcs_git = ForeignGit()
560
default_mapping = mapping_registry.get_default()()
563
def symlink_to_blob(symlink_target):
564
from dulwich.objects import Blob
566
if isinstance(symlink_target, text_type):
567
symlink_target = symlink_target.encode('utf-8')
568
blob.data = symlink_target
572
def mode_is_executable(mode):
573
"""Check if mode should be considered executable."""
574
return bool(mode & 0o111)
578
"""Determine the Bazaar inventory kind based on Unix file mode."""
581
entry_kind = (mode & 0o700000) / 0o100000
584
elif entry_kind == 1:
585
file_kind = (mode & 0o70000) / 0o10000
591
return 'tree-reference'
593
raise AssertionError(
594
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
596
raise AssertionError(
597
"Unknown kind, perms=%r." % (mode,))
600
def object_mode(kind, executable):
601
if kind == 'directory':
603
elif kind == 'symlink':
609
mode = stat.S_IFREG | 0o644
613
elif kind == 'tree-reference':
614
from dulwich.objects import S_IFGITLINK
620
def entry_mode(entry):
621
"""Determine the git file mode for an inventory entry."""
622
return object_mode(entry.kind, getattr(entry, 'executable', False))
625
def extract_unusual_modes(rev):
627
foreign_revid, mapping = mapping_registry.parse_revision_id(
629
except errors.InvalidRevisionId:
632
return mapping.export_unusual_file_modes(rev)
635
def parse_git_svn_id(text):
636
(head, uuid) = text.rsplit(" ", 1)
637
(full_url, rev) = head.rsplit("@", 1)
638
return (full_url, int(rev), uuid)
641
class GitFileIdMap(object):
643
def __init__(self, file_ids, mapping):
644
self.file_ids = file_ids
646
self.mapping = mapping
648
def all_file_ids(self):
649
return self.file_ids.values()
651
def set_file_id(self, path, file_id):
652
if type(path) is not str:
653
raise TypeError(path)
654
if not isinstance(file_id, bytes):
655
raise TypeError(file_id)
656
self.file_ids[path] = file_id
658
def lookup_file_id(self, path):
659
if not isinstance(path, text_type):
660
raise TypeError(path)
662
file_id = self.file_ids[path]
664
file_id = self.mapping.generate_file_id(path)
665
if not isinstance(file_id, bytes):
666
raise TypeError(file_id)
669
def lookup_path(self, file_id):
670
if self.paths is None:
672
for k, v in viewitems(self.file_ids):
675
path = self.paths[file_id]
677
return self.mapping.parse_file_id(file_id)
679
if not isinstance(path, text_type):
680
raise TypeError(path)
684
return self.__class__(dict(self.file_ids), self.mapping)
687
def needs_roundtripping(repo, revid):
689
mapping_registry.parse_revision_id(revid)
690
except errors.InvalidRevisionId:
added
removed
breezy/git/mapping.py
