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:
124
elif b">" not in text:
127
if text.rindex(b">") < text.rindex(b"<"):
128
raise ValueError(text)
129
username, email = text.split(b"<", 2)[-2:]
130
email = email.split(b">", 1)[0]
131
if username.endswith(b" "):
132
username = username[:-1]
133
return b"%s <%s>" % (username, email)
136
def decode_git_path(path):
137
"""Take a git path and decode it."""
139
return path.decode('utf-8')
140
except UnicodeDecodeError:
142
return path.decode('utf-8', 'surrogateescape')
146
def encode_git_path(path):
147
"""Take a regular path and encode it for git."""
149
return path.encode('utf-8')
150
except UnicodeEncodeError:
152
return path.encode('utf-8', 'surrogateescape')
156
def warn_escaped(commit, num_escaped):
157
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
158
"to regenerate the SHA map.", num_escaped, commit)
161
def warn_unusual_mode(commit, path, mode):
162
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
163
"property. ", mode, path, commit)
166
class BzrGitMapping(foreign.VcsMapping):
167
"""Class that maps between Git and Bazaar semantics."""
170
BZR_DUMMY_FILE = None # type: Optional[str]
172
def is_special_file(self, filename):
173
return (filename in (self.BZR_DUMMY_FILE, ))
176
super(BzrGitMapping, self).__init__(foreign_vcs_git)
178
def __eq__(self, other):
179
return (type(self) == type(other)
180
and self.revid_prefix == other.revid_prefix)
183
def revision_id_foreign_to_bzr(cls, git_rev_id):
184
"""Convert a git revision id handle to a Bazaar revision id."""
185
from dulwich.protocol import ZERO_SHA
186
if git_rev_id == ZERO_SHA:
188
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
191
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
192
"""Convert a Bazaar revision id to a git revision id handle."""
193
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
194
raise errors.InvalidRevisionId(bzr_rev_id, cls)
195
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
197
def generate_file_id(self, path):
198
# Git paths are just bytestrings
199
# We must just hope they are valid UTF-8..
200
if isinstance(path, text_type):
201
path = encode_git_path(path)
204
return FILE_ID_PREFIX + escape_file_id(path)
206
def parse_file_id(self, file_id):
207
if file_id == ROOT_ID:
209
if not file_id.startswith(FILE_ID_PREFIX):
211
return decode_git_path(unescape_file_id(file_id[len(FILE_ID_PREFIX):]))
213
def import_unusual_file_modes(self, rev, unusual_file_modes):
214
if unusual_file_modes:
215
ret = [(path, unusual_file_modes[path])
216
for path in sorted(unusual_file_modes.keys())]
217
rev.properties[u'file-modes'] = bencode.bencode(ret)
219
def export_unusual_file_modes(self, rev):
221
file_modes = rev.properties[u'file-modes']
225
return dict(bencode.bdecode(file_modes.encode("utf-8")))
227
def _generate_git_svn_metadata(self, rev, encoding):
229
git_svn_id = rev.properties[u"git-svn-id"]
233
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
235
def _generate_hg_message_tail(self, rev):
239
for name in rev.properties:
240
if name == u'hg:extra:branch':
241
branch = rev.properties[u'hg:extra:branch']
242
elif name.startswith(u'hg:extra'):
243
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
244
rev.properties[name])
245
elif name == u'hg:renames':
246
renames = bencode.bdecode(base64.b64decode(
247
rev.properties[u'hg:renames']))
248
# TODO: Export other properties as 'bzr:' extras?
249
ret = format_hg_metadata(renames, branch, extra)
250
if not isinstance(ret, bytes):
254
def _extract_git_svn_metadata(self, rev, message):
255
lines = message.split("\n")
256
if not (lines[-1] == "" and len(lines) >= 2 and
257
lines[-2].startswith("git-svn-id:")):
259
git_svn_id = lines[-2].split(": ", 1)[1]
260
rev.properties[u'git-svn-id'] = git_svn_id
261
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
262
# FIXME: Convert this to converted-from property somehow..
263
return "\n".join(lines[:-2])
265
def _extract_hg_metadata(self, rev, message):
266
(message, renames, branch, extra) = extract_hg_metadata(message)
267
if branch is not None:
268
rev.properties[u'hg:extra:branch'] = branch
269
for name, value in viewitems(extra):
270
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
272
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
273
[(new, old) for (old, new) in viewitems(renames)]))
276
def _extract_bzr_metadata(self, rev, message):
277
(message, metadata) = extract_bzr_metadata(message)
278
return message, metadata
280
def _decode_commit_message(self, rev, message, encoding):
281
return message.decode(encoding), CommitSupplement()
283
def _encode_commit_message(self, rev, message, encoding):
284
return message.encode(encoding)
286
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
288
"""Turn a Bazaar revision in to a Git commit
290
:param tree_sha: Tree sha for the commit
291
:param parent_lookup: Function for looking up the GIT sha equiv of a
293
:param lossy: Whether to store roundtripping information.
294
:param verifiers: Verifiers info
295
:return dulwich.objects.Commit represent the revision:
297
from dulwich.objects import Commit, Tag
299
commit.tree = tree_sha
301
metadata = CommitSupplement()
302
metadata.verifiers = verifiers
306
for p in rev.parent_ids:
308
git_p = parent_lookup(p)
311
if metadata is not None:
312
metadata.explicit_parent_ids = rev.parent_ids
313
if git_p is not None:
315
raise AssertionError("unexpected length for %r" % git_p)
316
parents.append(git_p)
317
commit.parents = parents
319
encoding = rev.properties[u'git-explicit-encoding']
321
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
323
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
327
commit.committer = fix_person_identifier(rev.committer.encode(
329
commit.author = fix_person_identifier(
330
rev.get_apparent_authors()[0].encode(encoding))
331
# TODO(jelmer): Don't use this hack.
332
long = getattr(__builtins__, 'long', int)
333
commit.commit_time = long(rev.timestamp)
334
if u'author-timestamp' in rev.properties:
335
commit.author_time = long(rev.properties[u'author-timestamp'])
337
commit.author_time = commit.commit_time
338
commit._commit_timezone_neg_utc = (
339
u"commit-timezone-neg-utc" in rev.properties)
340
commit.commit_timezone = rev.timezone
341
commit._author_timezone_neg_utc = (
342
u"author-timezone-neg-utc" in rev.properties)
343
if u'author-timezone' in rev.properties:
344
commit.author_timezone = int(rev.properties[u'author-timezone'])
346
commit.author_timezone = commit.commit_timezone
347
if u'git-gpg-signature' in rev.properties:
348
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
349
'utf-8', 'surrogateescape')
350
commit.message = self._encode_commit_message(rev, rev.message,
352
if not isinstance(commit.message, bytes):
353
raise TypeError(commit.message)
354
if metadata is not None:
356
mapping_registry.parse_revision_id(rev.revision_id)
357
except errors.InvalidRevisionId:
358
metadata.revision_id = rev.revision_id
359
mapping_properties = set(
360
[u'author', u'author-timezone', u'author-timezone-neg-utc',
361
u'commit-timezone-neg-utc', u'git-implicit-encoding',
362
u'git-gpg-signature', u'git-explicit-encoding',
363
u'author-timestamp', u'file-modes'])
364
for k, v in viewitems(rev.properties):
365
if k not in mapping_properties:
366
metadata.properties[k] = v
367
if not lossy and metadata:
368
if self.roundtripping:
369
commit.message = inject_bzr_metadata(commit.message, metadata,
373
None, None, self, revision_id=rev.revision_id)
374
if not isinstance(commit.message, bytes):
375
raise TypeError(commit.message)
377
propname = u'git-mergetag-0'
378
while propname in rev.properties:
379
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
381
propname = u'git-mergetag-%d' % i
382
if u'git-extra' in rev.properties:
385
for l in rev.properties[u'git-extra'].splitlines()])
388
def get_revision_id(self, commit):
390
encoding = commit.encoding.decode('ascii')
394
message, metadata = self._decode_commit_message(
395
None, commit.message, encoding)
396
except UnicodeDecodeError:
399
if metadata.revision_id:
400
return metadata.revision_id
401
return self.revision_id_foreign_to_bzr(commit.id)
403
def import_commit(self, commit, lookup_parent_revid, strict=True):
404
"""Convert a git commit to a bzr revision.
406
:return: a `breezy.revision.Revision` object, foreign revid and a
410
raise AssertionError("Commit object can't be None")
411
rev = ForeignRevision(commit.id, self,
412
self.revision_id_foreign_to_bzr(commit.id))
413
rev.git_metadata = None
415
def decode_using_encoding(rev, commit, encoding):
416
rev.committer = commit.committer.decode(encoding)
417
if commit.committer != commit.author:
418
rev.properties[u'author'] = commit.author.decode(encoding)
419
rev.message, rev.git_metadata = self._decode_commit_message(
420
rev, commit.message, encoding)
421
if commit.encoding is not None:
422
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode(
424
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
426
for encoding in ('utf-8', 'latin1'):
428
decode_using_encoding(rev, commit, encoding)
429
except UnicodeDecodeError:
432
if encoding != 'utf-8':
433
rev.properties[u'git-implicit-encoding'] = encoding
435
if commit.commit_time != commit.author_time:
436
rev.properties[u'author-timestamp'] = str(commit.author_time)
437
if commit.commit_timezone != commit.author_timezone:
438
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
439
if commit._author_timezone_neg_utc:
440
rev.properties[u'author-timezone-neg-utc'] = ""
441
if commit._commit_timezone_neg_utc:
442
rev.properties[u'commit-timezone-neg-utc'] = ""
444
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode(
445
'utf-8', 'surrogateescape')
447
for i, tag in enumerate(commit.mergetag):
448
rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string()
449
rev.timestamp = commit.commit_time
450
rev.timezone = commit.commit_timezone
451
rev.parent_ids = None
452
if rev.git_metadata is not None:
453
md = rev.git_metadata
454
roundtrip_revid = md.revision_id
455
if md.explicit_parent_ids:
456
rev.parent_ids = md.explicit_parent_ids
457
rev.properties.update(md.properties)
458
verifiers = md.verifiers
460
roundtrip_revid = None
462
if rev.parent_ids is None:
464
for p in commit.parents:
466
parents.append(lookup_parent_revid(p))
468
parents.append(self.revision_id_foreign_to_bzr(p))
469
rev.parent_ids = list(parents)
470
unknown_extra_fields = []
472
for k, v in commit.extra:
473
if k == HG_RENAME_SOURCE:
474
extra_lines.append(k + b' ' + v + b'\n')
476
hgk, hgv = v.split(b':', 1)
477
if hgk not in (HG_EXTRA_AMEND_SOURCE, ) and strict:
478
raise UnknownMercurialCommitExtra(commit, [hgk])
479
extra_lines.append(k + b' ' + v + b'\n')
481
unknown_extra_fields.append(k)
482
if unknown_extra_fields and strict:
483
raise UnknownCommitExtra(
485
[f.decode('ascii', 'replace') for f in unknown_extra_fields])
487
rev.properties[u'git-extra'] = b''.join(extra_lines)
488
return rev, roundtrip_revid, verifiers
491
class BzrGitMappingv1(BzrGitMapping):
492
revid_prefix = b'git-v1'
496
return self.revid_prefix
499
class BzrGitMappingExperimental(BzrGitMappingv1):
500
revid_prefix = b'git-experimental'
502
roundtripping = False
504
BZR_DUMMY_FILE = '.bzrdummy'
506
def _decode_commit_message(self, rev, message, encoding):
509
message = self._extract_hg_metadata(rev, message)
510
message = self._extract_git_svn_metadata(rev, message)
511
message, metadata = self._extract_bzr_metadata(rev, message)
512
return message.decode(encoding), metadata
514
def _encode_commit_message(self, rev, message, encoding):
515
ret = message.encode(encoding)
516
ret += self._generate_hg_message_tail(rev)
517
ret += self._generate_git_svn_metadata(rev, encoding)
520
def import_commit(self, commit, lookup_parent_revid, strict=True):
521
rev, roundtrip_revid, verifiers = super(
522
BzrGitMappingExperimental, self).import_commit(
523
commit, lookup_parent_revid, strict)
524
rev.properties[u'converted_revision'] = "git %s\n" % commit.id
525
return rev, roundtrip_revid, verifiers
528
class GitMappingRegistry(VcsMappingRegistry):
529
"""Registry with available git mappings."""
531
def revision_id_bzr_to_foreign(self, bzr_revid):
532
if bzr_revid == NULL_REVISION:
533
from dulwich.protocol import ZERO_SHA
534
return ZERO_SHA, None
535
if not bzr_revid.startswith(b"git-"):
536
raise errors.InvalidRevisionId(bzr_revid, None)
537
(mapping_version, git_sha) = bzr_revid.split(b":", 1)
538
mapping = self.get(mapping_version)
539
return mapping.revision_id_bzr_to_foreign(bzr_revid)
541
parse_revision_id = revision_id_bzr_to_foreign
544
mapping_registry = GitMappingRegistry()
545
mapping_registry.register_lazy(b'git-v1', __name__,
547
mapping_registry.register_lazy(b'git-experimental',
548
__name__, "BzrGitMappingExperimental")
549
# Uncomment the next line to enable the experimental bzr-git mappings.
550
# This will make sure all bzr metadata is pushed into git, allowing for
551
# full roundtripping later.
552
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
553
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
554
# mapping_registry.set_default('git-experimental')
555
mapping_registry.set_default(b'git-v1')
558
class ForeignGit(ForeignVcs):
559
"""The Git Stupid Content Tracker"""
562
def branch_format(self):
563
from .branch import LocalGitBranchFormat
564
return LocalGitBranchFormat()
567
def repository_format(self):
568
from .repository import GitRepositoryFormat
569
return GitRepositoryFormat()
572
super(ForeignGit, self).__init__(mapping_registry)
573
self.abbreviation = "git"
576
def serialize_foreign_revid(self, foreign_revid):
580
def show_foreign_revid(cls, foreign_revid):
581
return {"git commit": foreign_revid.decode('utf-8')}
584
foreign_vcs_git = ForeignGit()
585
default_mapping = mapping_registry.get_default()()
588
def symlink_to_blob(symlink_target):
589
from dulwich.objects import Blob
591
if isinstance(symlink_target, text_type):
592
symlink_target = encode_git_path(symlink_target)
593
blob.data = symlink_target
597
def mode_is_executable(mode):
598
"""Check if mode should be considered executable."""
599
return bool(mode & 0o111)
603
"""Determine the Bazaar inventory kind based on Unix file mode."""
606
entry_kind = (mode & 0o700000) / 0o100000
609
elif entry_kind == 1:
610
file_kind = (mode & 0o70000) / 0o10000
616
return 'tree-reference'
618
raise AssertionError(
619
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
621
raise AssertionError(
622
"Unknown kind, perms=%r." % (mode,))
625
def object_mode(kind, executable):
626
if kind == 'directory':
628
elif kind == 'symlink':
634
mode = stat.S_IFREG | 0o644
638
elif kind == 'tree-reference':
639
from dulwich.objects import S_IFGITLINK
645
def entry_mode(entry):
646
"""Determine the git file mode for an inventory entry."""
647
return object_mode(entry.kind, getattr(entry, 'executable', False))
650
def extract_unusual_modes(rev):
652
foreign_revid, mapping = mapping_registry.parse_revision_id(
654
except errors.InvalidRevisionId:
657
return mapping.export_unusual_file_modes(rev)
660
def parse_git_svn_id(text):
661
(head, uuid) = text.rsplit(" ", 1)
662
(full_url, rev) = head.rsplit("@", 1)
663
return (full_url, int(rev), uuid)
666
def needs_roundtripping(repo, revid):
668
mapping_registry.parse_revision_id(revid)
669
except errors.InvalidRevisionId: