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 (
54
from .roundtrip import (
61
DEFAULT_FILE_MODE = stat.S_IFREG | 0o644
62
HG_RENAME_SOURCE = b"HG:rename-source"
63
HG_EXTRA = b"HG:extra"
65
# This HG extra is used to indicate the commit that this commit was based on.
66
HG_EXTRA_AMEND_SOURCE = b"amend_source"
68
FILE_ID_PREFIX = b'git:'
71
ROOT_ID = b"TREE_ROOT"
74
class UnknownCommitExtra(errors.BzrError):
75
_fmt = "Unknown extra fields in %(object)r: %(fields)r."
77
def __init__(self, object, fields):
78
errors.BzrError.__init__(self)
80
self.fields = ",".join(fields)
83
class UnknownMercurialCommitExtra(errors.BzrError):
84
_fmt = "Unknown mercurial extra fields in %(object)r: %(fields)r."
86
def __init__(self, object, fields):
87
errors.BzrError.__init__(self)
89
self.fields = b",".join(fields)
92
def escape_file_id(file_id):
93
file_id = file_id.replace(b'_', b'__')
94
file_id = file_id.replace(b' ', b'_s')
95
file_id = file_id.replace(b'\x0c', b'_c')
99
def unescape_file_id(file_id):
102
while i < len(file_id):
103
if file_id[i:i + 1] != b'_':
104
ret.append(file_id[i])
106
if file_id[i + 1:i + 2] == b'_':
108
elif file_id[i + 1:i + 2] == b's':
110
elif file_id[i + 1:i + 2] == b'c':
111
ret.append(b"\x0c"[0])
113
raise ValueError("unknown escape character %s" %
114
file_id[i + 1:i + 2])
120
def fix_person_identifier(text):
121
if b"<" not in text and b">" not in text:
125
if text.rindex(b">") < text.rindex(b"<"):
126
raise ValueError(text)
127
username, email = text.split(b"<", 2)[-2:]
128
email = email.split(b">", 1)[0]
129
if username.endswith(b" "):
130
username = username[:-1]
131
return b"%s <%s>" % (username, email)
134
def warn_escaped(commit, num_escaped):
135
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
136
"to regenerate the SHA map.", num_escaped, commit)
139
def warn_unusual_mode(commit, path, mode):
140
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
141
"property. ", mode, path, commit)
144
class BzrGitMapping(foreign.VcsMapping):
145
"""Class that maps between Git and Bazaar semantics."""
148
BZR_DUMMY_FILE = None
150
def is_special_file(self, filename):
151
return (filename in (self.BZR_DUMMY_FILE, ))
154
super(BzrGitMapping, self).__init__(foreign_vcs_git)
156
def __eq__(self, other):
157
return (type(self) == type(other)
158
and self.revid_prefix == other.revid_prefix)
161
def revision_id_foreign_to_bzr(cls, git_rev_id):
162
"""Convert a git revision id handle to a Bazaar revision id."""
163
from dulwich.protocol import ZERO_SHA
164
if git_rev_id == ZERO_SHA:
166
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
169
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
170
"""Convert a Bazaar revision id to a git revision id handle."""
171
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
172
raise errors.InvalidRevisionId(bzr_rev_id, cls)
173
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
175
def generate_file_id(self, path):
176
# Git paths are just bytestrings
177
# We must just hope they are valid UTF-8..
178
if isinstance(path, text_type):
179
path = path.encode("utf-8")
182
return FILE_ID_PREFIX + escape_file_id(path)
184
def parse_file_id(self, file_id):
185
if file_id == ROOT_ID:
187
if not file_id.startswith(FILE_ID_PREFIX):
189
return unescape_file_id(file_id[len(FILE_ID_PREFIX):]).decode('utf-8')
191
def revid_as_refname(self, revid):
192
if not isinstance(revid, bytes):
193
raise TypeError(revid)
195
revid = revid.decode('utf-8')
196
quoted_revid = urlutils.quote(revid)
197
return b"refs/bzr/" + quoted_revid.encode('utf-8')
199
def import_unusual_file_modes(self, rev, unusual_file_modes):
200
if unusual_file_modes:
201
ret = [(path, unusual_file_modes[path])
202
for path in sorted(unusual_file_modes.keys())]
203
rev.properties[u'file-modes'] = bencode.bencode(ret)
205
def export_unusual_file_modes(self, rev):
207
file_modes = rev.properties[u'file-modes']
211
return dict(bencode.bdecode(file_modes.encode("utf-8")))
213
def _generate_git_svn_metadata(self, rev, encoding):
215
git_svn_id = rev.properties[u"git-svn-id"]
219
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
221
def _generate_hg_message_tail(self, rev):
225
for name in rev.properties:
226
if name == u'hg:extra:branch':
227
branch = rev.properties[u'hg:extra:branch']
228
elif name.startswith(u'hg:extra'):
229
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
230
rev.properties[name])
231
elif name == u'hg:renames':
232
renames = bencode.bdecode(base64.b64decode(
233
rev.properties[u'hg:renames']))
234
# TODO: Export other properties as 'bzr:' extras?
235
ret = format_hg_metadata(renames, branch, extra)
236
if not isinstance(ret, bytes):
240
def _extract_git_svn_metadata(self, rev, message):
241
lines = message.split("\n")
242
if not (lines[-1] == "" and len(lines) >= 2 and
243
lines[-2].startswith("git-svn-id:")):
245
git_svn_id = lines[-2].split(": ", 1)[1]
246
rev.properties[u'git-svn-id'] = git_svn_id
247
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
248
# FIXME: Convert this to converted-from property somehow..
249
return "\n".join(lines[:-2])
251
def _extract_hg_metadata(self, rev, message):
252
(message, renames, branch, extra) = extract_hg_metadata(message)
253
if branch is not None:
254
rev.properties[u'hg:extra:branch'] = branch
255
for name, value in viewitems(extra):
256
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
258
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
259
[(new, old) for (old, new) in viewitems(renames)]))
262
def _extract_bzr_metadata(self, rev, message):
263
(message, metadata) = extract_bzr_metadata(message)
264
return message, metadata
266
def _decode_commit_message(self, rev, message, encoding):
267
return message.decode(encoding), CommitSupplement()
269
def _encode_commit_message(self, rev, message, encoding):
270
return message.encode(encoding)
272
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
274
"""Turn a Bazaar revision in to a Git commit
276
:param tree_sha: Tree sha for the commit
277
:param parent_lookup: Function for looking up the GIT sha equiv of a
279
:param lossy: Whether to store roundtripping information.
280
:param verifiers: Verifiers info
281
:return dulwich.objects.Commit represent the revision:
283
from dulwich.objects import Commit, Tag
285
commit.tree = tree_sha
287
metadata = CommitSupplement()
288
metadata.verifiers = verifiers
292
for p in rev.parent_ids:
294
git_p = parent_lookup(p)
297
if metadata is not None:
298
metadata.explicit_parent_ids = rev.parent_ids
299
if git_p is not None:
301
raise AssertionError("unexpected length for %r" % git_p)
302
parents.append(git_p)
303
commit.parents = parents
305
encoding = rev.properties[u'git-explicit-encoding']
307
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
309
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
313
commit.committer = fix_person_identifier(rev.committer.encode(
315
commit.author = fix_person_identifier(
316
rev.get_apparent_authors()[0].encode(encoding))
317
# TODO(jelmer): Don't use this hack.
318
long = getattr(__builtins__, 'long', int)
319
commit.commit_time = long(rev.timestamp)
320
if u'author-timestamp' in rev.properties:
321
commit.author_time = long(rev.properties[u'author-timestamp'])
323
commit.author_time = commit.commit_time
324
commit._commit_timezone_neg_utc = (
325
u"commit-timezone-neg-utc" in rev.properties)
326
commit.commit_timezone = rev.timezone
327
commit._author_timezone_neg_utc = (
328
u"author-timezone-neg-utc" in rev.properties)
329
if u'author-timezone' in rev.properties:
330
commit.author_timezone = int(rev.properties[u'author-timezone'])
332
commit.author_timezone = commit.commit_timezone
333
if u'git-gpg-signature' in rev.properties:
334
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
336
if u'git-gpg-signature-b64' in rev.properties:
337
commit.gpgsig = base64.b64decode(rev.properties[u'git-gpg-signature-b64'])
338
commit.message = self._encode_commit_message(rev, rev.message,
340
if not isinstance(commit.message, bytes):
341
raise TypeError(commit.message)
342
if metadata is not None:
344
mapping_registry.parse_revision_id(rev.revision_id)
345
except errors.InvalidRevisionId:
346
metadata.revision_id = rev.revision_id
347
mapping_properties = set(
348
[u'author', u'author-timezone', u'author-timezone-neg-utc',
349
u'commit-timezone-neg-utc', u'git-implicit-encoding',
350
u'git-gpg-signature', u'git-gpg-signature-b64',
351
u'git-explicit-encoding',
352
u'author-timestamp', u'file-modes'])
353
for k, v in viewitems(rev.properties):
354
if k not in mapping_properties:
355
metadata.properties[k] = v
356
if not lossy and metadata:
357
if self.roundtripping:
358
commit.message = inject_bzr_metadata(commit.message, metadata,
362
None, None, self, revision_id=rev.revision_id)
363
if not isinstance(commit.message, bytes):
364
raise TypeError(commit.message)
366
propname = u'git-mergetag-0'
367
while propname in rev.properties:
368
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
370
propname = u'git-mergetag-%d' % i
371
if u'git-extra' in rev.properties:
374
for l in rev.properties[u'git-extra'].splitlines()])
377
def get_revision_id(self, commit):
379
encoding = commit.encoding.decode('ascii')
383
message, metadata = self._decode_commit_message(
384
None, commit.message, encoding)
385
except UnicodeDecodeError:
388
if metadata.revision_id:
389
return metadata.revision_id
390
return self.revision_id_foreign_to_bzr(commit.id)
392
def import_commit(self, commit, lookup_parent_revid, strict=True):
393
"""Convert a git commit to a bzr revision.
395
:return: a `breezy.revision.Revision` object, foreign revid and a
399
raise AssertionError("Commit object can't be None")
400
rev = ForeignRevision(commit.id, self,
401
self.revision_id_foreign_to_bzr(commit.id))
402
rev.git_metadata = None
404
def decode_using_encoding(rev, commit, encoding):
405
rev.committer = commit.committer.decode(encoding)
406
if commit.committer != commit.author:
407
rev.properties[u'author'] = commit.author.decode(encoding)
408
rev.message, rev.git_metadata = self._decode_commit_message(
409
rev, commit.message, encoding)
410
if commit.encoding is not None:
411
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode(
413
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
415
for encoding in ('utf-8', 'latin1'):
417
decode_using_encoding(rev, commit, encoding)
418
except UnicodeDecodeError:
421
if encoding != 'utf-8':
422
rev.properties[u'git-implicit-encoding'] = encoding
424
if commit.commit_time != commit.author_time:
425
rev.properties[u'author-timestamp'] = str(commit.author_time)
426
if commit.commit_timezone != commit.author_timezone:
427
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
428
if commit._author_timezone_neg_utc:
429
rev.properties[u'author-timezone-neg-utc'] = ""
430
if commit._commit_timezone_neg_utc:
431
rev.properties[u'commit-timezone-neg-utc'] = ""
434
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode(
436
except UnicodeDecodeError:
437
rev.properties[u'git-gpg-signature-b64'] = base64.b64encode(
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, 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
def needs_roundtripping(repo, revid):
661
mapping_registry.parse_revision_id(revid)
662
except errors.InvalidRevisionId: