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 (
60
deserialize_fileid_map,
65
DEFAULT_FILE_MODE = stat.S_IFREG | 0o644
66
HG_RENAME_SOURCE = b"HG:rename-source"
67
HG_EXTRA = b"HG:extra"
69
# This HG extra is used to indicate the commit that this commit was based on.
70
HG_EXTRA_AMEND_SOURCE = b"amend_source"
72
FILE_ID_PREFIX = b'git:'
75
ROOT_ID = b"TREE_ROOT"
78
def escape_file_id(file_id):
79
file_id = file_id.replace(b'_', b'__')
80
file_id = file_id.replace(b' ', b'_s')
81
file_id = file_id.replace(b'\x0c', b'_c')
85
def unescape_file_id(file_id):
88
while i < len(file_id):
89
if file_id[i:i + 1] != b'_':
90
ret.append(file_id[i])
92
if file_id[i + 1:i + 2] == b'_':
94
elif file_id[i + 1:i + 2] == b's':
96
elif file_id[i + 1:i + 2] == b'c':
97
ret.append(b"\x0c"[0])
99
raise ValueError("unknown escape character %s" %
100
file_id[i + 1:i + 2])
106
def fix_person_identifier(text):
107
if b"<" not in text and b">" not in text:
111
if text.rindex(b">") < text.rindex(b"<"):
112
raise ValueError(text)
113
username, email = text.split(b"<", 2)[-2:]
114
email = email.split(b">", 1)[0]
115
if username.endswith(b" "):
116
username = username[:-1]
117
return b"%s <%s>" % (username, email)
120
def warn_escaped(commit, num_escaped):
121
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
122
"to regenerate the SHA map.", num_escaped, commit)
125
def warn_unusual_mode(commit, path, mode):
126
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
127
"property. ", mode, path, commit)
130
class BzrGitMapping(foreign.VcsMapping):
131
"""Class that maps between Git and Bazaar semantics."""
134
BZR_FILE_IDS_FILE = None
136
BZR_DUMMY_FILE = None
138
def is_special_file(self, filename):
139
return (filename in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE))
142
super(BzrGitMapping, self).__init__(foreign_vcs_git)
144
def __eq__(self, other):
145
return (type(self) == type(other)
146
and self.revid_prefix == other.revid_prefix)
149
def revision_id_foreign_to_bzr(cls, git_rev_id):
150
"""Convert a git revision id handle to a Bazaar revision id."""
151
from dulwich.protocol import ZERO_SHA
152
if git_rev_id == ZERO_SHA:
154
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
157
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
158
"""Convert a Bazaar revision id to a git revision id handle."""
159
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
160
raise errors.InvalidRevisionId(bzr_rev_id, cls)
161
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
163
def generate_file_id(self, path):
164
# Git paths are just bytestrings
165
# We must just hope they are valid UTF-8..
166
if isinstance(path, text_type):
167
path = path.encode("utf-8")
170
return FILE_ID_PREFIX + escape_file_id(path)
172
def parse_file_id(self, file_id):
173
if file_id == ROOT_ID:
175
if not file_id.startswith(FILE_ID_PREFIX):
177
return unescape_file_id(file_id[len(FILE_ID_PREFIX):]).decode('utf-8')
179
def revid_as_refname(self, revid):
180
if not isinstance(revid, bytes):
181
raise TypeError(revid)
183
revid = revid.decode('utf-8')
184
quoted_revid = urlutils.quote(revid)
185
return b"refs/bzr/" + quoted_revid.encode('utf-8')
187
def import_unusual_file_modes(self, rev, unusual_file_modes):
188
if unusual_file_modes:
189
ret = [(path, unusual_file_modes[path])
190
for path in sorted(unusual_file_modes.keys())]
191
rev.properties[u'file-modes'] = bencode.bencode(ret)
193
def export_unusual_file_modes(self, rev):
195
file_modes = rev.properties[u'file-modes']
199
return dict(bencode.bdecode(file_modes.encode("utf-8")))
201
def _generate_git_svn_metadata(self, rev, encoding):
203
git_svn_id = rev.properties[u"git-svn-id"]
207
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
209
def _generate_hg_message_tail(self, rev):
213
for name in rev.properties:
214
if name == u'hg:extra:branch':
215
branch = rev.properties[u'hg:extra:branch']
216
elif name.startswith(u'hg:extra'):
217
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
218
rev.properties[name])
219
elif name == u'hg:renames':
220
renames = bencode.bdecode(base64.b64decode(
221
rev.properties[u'hg:renames']))
222
# TODO: Export other properties as 'bzr:' extras?
223
ret = format_hg_metadata(renames, branch, extra)
224
if not isinstance(ret, bytes):
228
def _extract_git_svn_metadata(self, rev, message):
229
lines = message.split("\n")
230
if not (lines[-1] == "" and len(lines) >= 2 and
231
lines[-2].startswith("git-svn-id:")):
233
git_svn_id = lines[-2].split(": ", 1)[1]
234
rev.properties[u'git-svn-id'] = git_svn_id
235
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
236
# FIXME: Convert this to converted-from property somehow..
237
return "\n".join(lines[:-2])
239
def _extract_hg_metadata(self, rev, message):
240
(message, renames, branch, extra) = extract_hg_metadata(message)
241
if branch is not None:
242
rev.properties[u'hg:extra:branch'] = branch
243
for name, value in viewitems(extra):
244
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
246
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
247
[(new, old) for (old, new) in viewitems(renames)]))
250
def _extract_bzr_metadata(self, rev, message):
251
(message, metadata) = extract_bzr_metadata(message)
252
return message, metadata
254
def _decode_commit_message(self, rev, message, encoding):
255
return message.decode(encoding), CommitSupplement()
257
def _encode_commit_message(self, rev, message, encoding):
258
return message.encode(encoding)
260
def export_fileid_map(self, fileid_map):
261
"""Export a file id map to a fileid map.
263
:param fileid_map: File id map, mapping paths to file ids
264
:return: A Git blob object (or None if there are no entries)
266
from dulwich.objects import Blob
268
b.set_raw_chunks(serialize_fileid_map(fileid_map))
271
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
273
"""Turn a Bazaar revision in to a Git commit
275
:param tree_sha: Tree sha for the commit
276
:param parent_lookup: Function for looking up the GIT sha equiv of a
278
:param lossy: Whether to store roundtripping information.
279
:param verifiers: Verifiers info
280
:return dulwich.objects.Commit represent the revision:
282
from dulwich.objects import Commit, Tag
284
commit.tree = tree_sha
286
metadata = CommitSupplement()
287
metadata.verifiers = verifiers
291
for p in rev.parent_ids:
293
git_p = parent_lookup(p)
296
if metadata is not None:
297
metadata.explicit_parent_ids = rev.parent_ids
298
if git_p is not None:
300
raise AssertionError("unexpected length for %r" % git_p)
301
parents.append(git_p)
302
commit.parents = parents
304
encoding = rev.properties[u'git-explicit-encoding']
306
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
308
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
312
commit.committer = fix_person_identifier(rev.committer.encode(
314
commit.author = fix_person_identifier(
315
rev.get_apparent_authors()[0].encode(encoding))
316
# TODO(jelmer): Don't use this hack.
317
long = getattr(__builtins__, 'long', int)
318
commit.commit_time = long(rev.timestamp)
319
if u'author-timestamp' in rev.properties:
320
commit.author_time = long(rev.properties[u'author-timestamp'])
322
commit.author_time = commit.commit_time
323
commit._commit_timezone_neg_utc = (
324
u"commit-timezone-neg-utc" in rev.properties)
325
commit.commit_timezone = rev.timezone
326
commit._author_timezone_neg_utc = (
327
u"author-timezone-neg-utc" in rev.properties)
328
if u'author-timezone' in rev.properties:
329
commit.author_timezone = int(rev.properties[u'author-timezone'])
331
commit.author_timezone = commit.commit_timezone
332
if u'git-gpg-signature' in rev.properties:
333
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
335
commit.message = self._encode_commit_message(rev, rev.message,
337
if not isinstance(commit.message, bytes):
338
raise TypeError(commit.message)
339
if metadata is not None:
341
mapping_registry.parse_revision_id(rev.revision_id)
342
except errors.InvalidRevisionId:
343
metadata.revision_id = rev.revision_id
344
mapping_properties = set(
345
[u'author', u'author-timezone', u'author-timezone-neg-utc',
346
u'commit-timezone-neg-utc', u'git-implicit-encoding',
347
u'git-gpg-signature', u'git-explicit-encoding',
348
u'author-timestamp', u'file-modes'])
349
for k, v in viewitems(rev.properties):
350
if k not in mapping_properties:
351
metadata.properties[k] = v
352
if not lossy and metadata:
353
if self.roundtripping:
354
commit.message = inject_bzr_metadata(commit.message, metadata,
358
None, None, self, revision_id=rev.revision_id)
359
if not isinstance(commit.message, bytes):
360
raise TypeError(commit.message)
362
propname = u'git-mergetag-0'
363
while propname in rev.properties:
364
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
366
propname = u'git-mergetag-%d' % i
367
if u'git-extra' in rev.properties:
370
for l in rev.properties[u'git-extra'].splitlines()])
373
def import_fileid_map(self, blob):
374
"""Convert a git file id map blob.
376
:param blob: Git blob object with fileid map
377
:return: Dictionary mapping paths to file ids
379
return deserialize_fileid_map(blob.data)
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):
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(
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, ):
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:
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
483
def get_fileid_map(self, lookup_object, tree_sha):
484
"""Obtain a fileid map for a particular tree.
486
:param lookup_object: Function for looking up an object
487
:param tree_sha: SHA of the root tree
488
:return: GitFileIdMap instance
491
file_id_map_sha = lookup_object(
492
tree_sha)[self.BZR_FILE_IDS_FILE][1]
496
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
497
return GitFileIdMap(file_ids, self)
500
class BzrGitMappingv1(BzrGitMapping):
501
revid_prefix = b'git-v1'
505
return self.revid_prefix
508
class BzrGitMappingExperimental(BzrGitMappingv1):
509
revid_prefix = b'git-experimental'
513
BZR_FILE_IDS_FILE = '.bzrfileids'
515
BZR_DUMMY_FILE = '.bzrdummy'
517
def _decode_commit_message(self, rev, message, encoding):
520
message = self._extract_hg_metadata(rev, message)
521
message = self._extract_git_svn_metadata(rev, message)
522
message, metadata = self._extract_bzr_metadata(rev, message)
523
return message.decode(encoding), metadata
525
def _encode_commit_message(self, rev, message, encoding):
526
ret = message.encode(encoding)
527
ret += self._generate_hg_message_tail(rev)
528
ret += self._generate_git_svn_metadata(rev, encoding)
531
def import_commit(self, commit, lookup_parent_revid):
532
rev, roundtrip_revid, verifiers = super(
533
BzrGitMappingExperimental, self).import_commit(
534
commit, lookup_parent_revid)
535
rev.properties[u'converted_revision'] = "git %s\n" % commit.id
536
return rev, roundtrip_revid, verifiers
539
class GitMappingRegistry(VcsMappingRegistry):
540
"""Registry with available git mappings."""
542
def revision_id_bzr_to_foreign(self, bzr_revid):
543
if bzr_revid == NULL_REVISION:
544
from dulwich.protocol import ZERO_SHA
545
return ZERO_SHA, None
546
if not bzr_revid.startswith(b"git-"):
547
raise errors.InvalidRevisionId(bzr_revid, None)
548
(mapping_version, git_sha) = bzr_revid.split(b":", 1)
549
mapping = self.get(mapping_version)
550
return mapping.revision_id_bzr_to_foreign(bzr_revid)
552
parse_revision_id = revision_id_bzr_to_foreign
555
mapping_registry = GitMappingRegistry()
556
mapping_registry.register_lazy(b'git-v1', __name__,
558
mapping_registry.register_lazy(b'git-experimental',
559
__name__, "BzrGitMappingExperimental")
560
# Uncomment the next line to enable the experimental bzr-git mappings.
561
# This will make sure all bzr metadata is pushed into git, allowing for
562
# full roundtripping later.
563
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
564
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
565
# mapping_registry.set_default('git-experimental')
566
mapping_registry.set_default(b'git-v1')
569
class ForeignGit(ForeignVcs):
570
"""The Git Stupid Content Tracker"""
573
def branch_format(self):
574
from .branch import LocalGitBranchFormat
575
return LocalGitBranchFormat()
578
def repository_format(self):
579
from .repository import GitRepositoryFormat
580
return GitRepositoryFormat()
583
super(ForeignGit, self).__init__(mapping_registry)
584
self.abbreviation = "git"
587
def serialize_foreign_revid(self, foreign_revid):
591
def show_foreign_revid(cls, foreign_revid):
592
return {"git commit": foreign_revid.decode('utf-8')}
595
foreign_vcs_git = ForeignGit()
596
default_mapping = mapping_registry.get_default()()
599
def symlink_to_blob(symlink_target):
600
from dulwich.objects import Blob
602
if isinstance(symlink_target, text_type):
603
symlink_target = symlink_target.encode('utf-8')
604
blob.data = symlink_target
608
def mode_is_executable(mode):
609
"""Check if mode should be considered executable."""
610
return bool(mode & 0o111)
614
"""Determine the Bazaar inventory kind based on Unix file mode."""
617
entry_kind = (mode & 0o700000) / 0o100000
620
elif entry_kind == 1:
621
file_kind = (mode & 0o70000) / 0o10000
627
return 'tree-reference'
629
raise AssertionError(
630
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
632
raise AssertionError(
633
"Unknown kind, perms=%r." % (mode,))
636
def object_mode(kind, executable):
637
if kind == 'directory':
639
elif kind == 'symlink':
645
mode = stat.S_IFREG | 0o644
649
elif kind == 'tree-reference':
650
from dulwich.objects import S_IFGITLINK
656
def entry_mode(entry):
657
"""Determine the git file mode for an inventory entry."""
658
return object_mode(entry.kind, getattr(entry, 'executable', False))
661
def extract_unusual_modes(rev):
663
foreign_revid, mapping = mapping_registry.parse_revision_id(
665
except errors.InvalidRevisionId:
668
return mapping.export_unusual_file_modes(rev)
671
def parse_git_svn_id(text):
672
(head, uuid) = text.rsplit(" ", 1)
673
(full_url, rev) = head.rsplit("@", 1)
674
return (full_url, int(rev), uuid)
677
class GitFileIdMap(object):
679
def __init__(self, file_ids, mapping):
680
self.file_ids = file_ids
682
self.mapping = mapping
684
def set_file_id(self, path, file_id):
685
if type(path) is not str:
686
raise TypeError(path)
687
if not isinstance(file_id, bytes):
688
raise TypeError(file_id)
689
self.file_ids[path] = file_id
691
def lookup_file_id(self, path):
692
if not isinstance(path, text_type):
693
raise TypeError(path)
695
file_id = self.file_ids[path]
697
file_id = self.mapping.generate_file_id(path)
698
if not isinstance(file_id, bytes):
699
raise TypeError(file_id)
702
def lookup_path(self, file_id):
703
if self.paths is None:
705
for k, v in viewitems(self.file_ids):
708
path = self.paths[file_id]
710
return self.mapping.parse_file_id(file_id)
712
if not isinstance(path, text_type):
713
raise TypeError(path)
717
return self.__class__(dict(self.file_ids), self.mapping)
720
def needs_roundtripping(repo, revid):
722
mapping_registry.parse_revision_id(revid)
723
except errors.InvalidRevisionId:
added
removed
breezy/git/mapping.py
