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 warn_escaped(commit, num_escaped):
137
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
138
"to regenerate the SHA map.", num_escaped, commit)
141
def warn_unusual_mode(commit, path, mode):
142
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
143
"property. ", mode, path, commit)
146
class BzrGitMapping(foreign.VcsMapping):
147
"""Class that maps between Git and Bazaar semantics."""
150
BZR_DUMMY_FILE = None
152
def is_special_file(self, filename):
153
return (filename in (self.BZR_DUMMY_FILE, ))
156
super(BzrGitMapping, self).__init__(foreign_vcs_git)
158
def __eq__(self, other):
159
return (type(self) == type(other)
160
and self.revid_prefix == other.revid_prefix)
163
def revision_id_foreign_to_bzr(cls, git_rev_id):
164
"""Convert a git revision id handle to a Bazaar revision id."""
165
from dulwich.protocol import ZERO_SHA
166
if git_rev_id == ZERO_SHA:
168
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
171
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
172
"""Convert a Bazaar revision id to a git revision id handle."""
173
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
174
raise errors.InvalidRevisionId(bzr_rev_id, cls)
175
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
177
def generate_file_id(self, path):
178
# Git paths are just bytestrings
179
# We must just hope they are valid UTF-8..
180
if isinstance(path, text_type):
181
path = path.encode("utf-8")
184
return FILE_ID_PREFIX + escape_file_id(path)
186
def parse_file_id(self, file_id):
187
if file_id == ROOT_ID:
189
if not file_id.startswith(FILE_ID_PREFIX):
191
return unescape_file_id(file_id[len(FILE_ID_PREFIX):]).decode('utf-8')
193
def revid_as_refname(self, revid):
194
if not isinstance(revid, bytes):
195
raise TypeError(revid)
197
revid = revid.decode('utf-8')
198
quoted_revid = urlutils.quote(revid)
199
return b"refs/bzr/" + quoted_revid.encode('utf-8')
201
def import_unusual_file_modes(self, rev, unusual_file_modes):
202
if unusual_file_modes:
203
ret = [(path, unusual_file_modes[path])
204
for path in sorted(unusual_file_modes.keys())]
205
rev.properties[u'file-modes'] = bencode.bencode(ret)
207
def export_unusual_file_modes(self, rev):
209
file_modes = rev.properties[u'file-modes']
213
return dict(bencode.bdecode(file_modes.encode("utf-8")))
215
def _generate_git_svn_metadata(self, rev, encoding):
217
git_svn_id = rev.properties[u"git-svn-id"]
221
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
223
def _generate_hg_message_tail(self, rev):
227
for name in rev.properties:
228
if name == u'hg:extra:branch':
229
branch = rev.properties[u'hg:extra:branch']
230
elif name.startswith(u'hg:extra'):
231
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
232
rev.properties[name])
233
elif name == u'hg:renames':
234
renames = bencode.bdecode(base64.b64decode(
235
rev.properties[u'hg:renames']))
236
# TODO: Export other properties as 'bzr:' extras?
237
ret = format_hg_metadata(renames, branch, extra)
238
if not isinstance(ret, bytes):
242
def _extract_git_svn_metadata(self, rev, message):
243
lines = message.split("\n")
244
if not (lines[-1] == "" and len(lines) >= 2 and
245
lines[-2].startswith("git-svn-id:")):
247
git_svn_id = lines[-2].split(": ", 1)[1]
248
rev.properties[u'git-svn-id'] = git_svn_id
249
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
250
# FIXME: Convert this to converted-from property somehow..
251
return "\n".join(lines[:-2])
253
def _extract_hg_metadata(self, rev, message):
254
(message, renames, branch, extra) = extract_hg_metadata(message)
255
if branch is not None:
256
rev.properties[u'hg:extra:branch'] = branch
257
for name, value in viewitems(extra):
258
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
260
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
261
[(new, old) for (old, new) in viewitems(renames)]))
264
def _extract_bzr_metadata(self, rev, message):
265
(message, metadata) = extract_bzr_metadata(message)
266
return message, metadata
268
def _decode_commit_message(self, rev, message, encoding):
269
return message.decode(encoding), CommitSupplement()
271
def _encode_commit_message(self, rev, message, encoding):
272
return message.encode(encoding)
274
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
276
"""Turn a Bazaar revision in to a Git commit
278
:param tree_sha: Tree sha for the commit
279
:param parent_lookup: Function for looking up the GIT sha equiv of a
281
:param lossy: Whether to store roundtripping information.
282
:param verifiers: Verifiers info
283
:return dulwich.objects.Commit represent the revision:
285
from dulwich.objects import Commit, Tag
287
commit.tree = tree_sha
289
metadata = CommitSupplement()
290
metadata.verifiers = verifiers
294
for p in rev.parent_ids:
296
git_p = parent_lookup(p)
299
if metadata is not None:
300
metadata.explicit_parent_ids = rev.parent_ids
301
if git_p is not None:
303
raise AssertionError("unexpected length for %r" % git_p)
304
parents.append(git_p)
305
commit.parents = parents
307
encoding = rev.properties[u'git-explicit-encoding']
309
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
311
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
315
commit.committer = fix_person_identifier(rev.committer.encode(
317
commit.author = fix_person_identifier(
318
rev.get_apparent_authors()[0].encode(encoding))
319
# TODO(jelmer): Don't use this hack.
320
long = getattr(__builtins__, 'long', int)
321
commit.commit_time = long(rev.timestamp)
322
if u'author-timestamp' in rev.properties:
323
commit.author_time = long(rev.properties[u'author-timestamp'])
325
commit.author_time = commit.commit_time
326
commit._commit_timezone_neg_utc = (
327
u"commit-timezone-neg-utc" in rev.properties)
328
commit.commit_timezone = rev.timezone
329
commit._author_timezone_neg_utc = (
330
u"author-timezone-neg-utc" in rev.properties)
331
if u'author-timezone' in rev.properties:
332
commit.author_timezone = int(rev.properties[u'author-timezone'])
334
commit.author_timezone = commit.commit_timezone
335
if u'git-gpg-signature' in rev.properties:
336
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
338
if u'git-gpg-signature-b64' in rev.properties:
339
commit.gpgsig = base64.b64decode(rev.properties[u'git-gpg-signature-b64'])
340
commit.message = self._encode_commit_message(rev, rev.message,
342
if not isinstance(commit.message, bytes):
343
raise TypeError(commit.message)
344
if metadata is not None:
346
mapping_registry.parse_revision_id(rev.revision_id)
347
except errors.InvalidRevisionId:
348
metadata.revision_id = rev.revision_id
349
mapping_properties = set(
350
[u'author', u'author-timezone', u'author-timezone-neg-utc',
351
u'commit-timezone-neg-utc', u'git-implicit-encoding',
352
u'git-gpg-signature', u'git-gpg-signature-b64',
353
u'git-explicit-encoding',
354
u'author-timestamp', u'file-modes'])
355
for k, v in viewitems(rev.properties):
356
if k not in mapping_properties:
357
metadata.properties[k] = v
358
if not lossy and metadata:
359
if self.roundtripping:
360
commit.message = inject_bzr_metadata(commit.message, metadata,
364
None, None, self, revision_id=rev.revision_id)
365
if not isinstance(commit.message, bytes):
366
raise TypeError(commit.message)
368
propname = u'git-mergetag-0'
369
while propname in rev.properties:
370
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
372
propname = u'git-mergetag-%d' % i
373
if u'git-extra' in rev.properties:
376
for l in rev.properties[u'git-extra'].splitlines()])
379
def get_revision_id(self, commit):
381
encoding = commit.encoding.decode('ascii')
385
message, metadata = self._decode_commit_message(
386
None, commit.message, encoding)
387
except UnicodeDecodeError:
390
if metadata.revision_id:
391
return metadata.revision_id
392
return self.revision_id_foreign_to_bzr(commit.id)
394
def import_commit(self, commit, lookup_parent_revid, strict=True):
395
"""Convert a git commit to a bzr revision.
397
:return: a `breezy.revision.Revision` object, foreign revid and a
401
raise AssertionError("Commit object can't be None")
402
rev = ForeignRevision(commit.id, self,
403
self.revision_id_foreign_to_bzr(commit.id))
404
rev.git_metadata = None
406
def decode_using_encoding(rev, commit, encoding):
407
rev.committer = commit.committer.decode(encoding)
408
if commit.committer != commit.author:
409
rev.properties[u'author'] = commit.author.decode(encoding)
410
rev.message, rev.git_metadata = self._decode_commit_message(
411
rev, commit.message, encoding)
412
if commit.encoding is not None:
413
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode(
415
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
417
for encoding in ('utf-8', 'latin1'):
419
decode_using_encoding(rev, commit, encoding)
420
except UnicodeDecodeError:
423
if encoding != 'utf-8':
424
rev.properties[u'git-implicit-encoding'] = encoding
426
if commit.commit_time != commit.author_time:
427
rev.properties[u'author-timestamp'] = str(commit.author_time)
428
if commit.commit_timezone != commit.author_timezone:
429
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
430
if commit._author_timezone_neg_utc:
431
rev.properties[u'author-timezone-neg-utc'] = ""
432
if commit._commit_timezone_neg_utc:
433
rev.properties[u'commit-timezone-neg-utc'] = ""
436
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode(
438
except UnicodeDecodeError:
439
rev.properties[u'git-gpg-signature-b64'] = base64.b64encode(
442
for i, tag in enumerate(commit.mergetag):
443
rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string()
444
rev.timestamp = commit.commit_time
445
rev.timezone = commit.commit_timezone
446
rev.parent_ids = None
447
if rev.git_metadata is not None:
448
md = rev.git_metadata
449
roundtrip_revid = md.revision_id
450
if md.explicit_parent_ids:
451
rev.parent_ids = md.explicit_parent_ids
452
rev.properties.update(md.properties)
453
verifiers = md.verifiers
455
roundtrip_revid = None
457
if rev.parent_ids is None:
459
for p in commit.parents:
461
parents.append(lookup_parent_revid(p))
463
parents.append(self.revision_id_foreign_to_bzr(p))
464
rev.parent_ids = list(parents)
465
unknown_extra_fields = []
467
for k, v in commit.extra:
468
if k == HG_RENAME_SOURCE:
469
extra_lines.append(k + b' ' + v + b'\n')
471
hgk, hgv = v.split(b':', 1)
472
if hgk not in (HG_EXTRA_AMEND_SOURCE, ) and strict:
473
raise UnknownMercurialCommitExtra(commit, [hgk])
474
extra_lines.append(k + b' ' + v + b'\n')
476
unknown_extra_fields.append(k)
477
if unknown_extra_fields and strict:
478
raise UnknownCommitExtra(
480
[f.decode('ascii', 'replace') for f in unknown_extra_fields])
482
rev.properties[u'git-extra'] = b''.join(extra_lines)
483
return rev, roundtrip_revid, verifiers
486
class BzrGitMappingv1(BzrGitMapping):
487
revid_prefix = b'git-v1'
491
return self.revid_prefix
494
class BzrGitMappingExperimental(BzrGitMappingv1):
495
revid_prefix = b'git-experimental'
497
roundtripping = False
499
BZR_DUMMY_FILE = '.bzrdummy'
501
def _decode_commit_message(self, rev, message, encoding):
504
message = self._extract_hg_metadata(rev, message)
505
message = self._extract_git_svn_metadata(rev, message)
506
message, metadata = self._extract_bzr_metadata(rev, message)
507
return message.decode(encoding), metadata
509
def _encode_commit_message(self, rev, message, encoding):
510
ret = message.encode(encoding)
511
ret += self._generate_hg_message_tail(rev)
512
ret += self._generate_git_svn_metadata(rev, encoding)
515
def import_commit(self, commit, lookup_parent_revid, strict=True):
516
rev, roundtrip_revid, verifiers = super(
517
BzrGitMappingExperimental, self).import_commit(
518
commit, lookup_parent_revid, strict)
519
rev.properties[u'converted_revision'] = "git %s\n" % commit.id
520
return rev, roundtrip_revid, verifiers
523
class GitMappingRegistry(VcsMappingRegistry):
524
"""Registry with available git mappings."""
526
def revision_id_bzr_to_foreign(self, bzr_revid):
527
if bzr_revid == NULL_REVISION:
528
from dulwich.protocol import ZERO_SHA
529
return ZERO_SHA, None
530
if not bzr_revid.startswith(b"git-"):
531
raise errors.InvalidRevisionId(bzr_revid, None)
532
(mapping_version, git_sha) = bzr_revid.split(b":", 1)
533
mapping = self.get(mapping_version)
534
return mapping.revision_id_bzr_to_foreign(bzr_revid)
536
parse_revision_id = revision_id_bzr_to_foreign
539
mapping_registry = GitMappingRegistry()
540
mapping_registry.register_lazy(b'git-v1', __name__,
542
mapping_registry.register_lazy(b'git-experimental',
543
__name__, "BzrGitMappingExperimental")
544
# Uncomment the next line to enable the experimental bzr-git mappings.
545
# This will make sure all bzr metadata is pushed into git, allowing for
546
# full roundtripping later.
547
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
548
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
549
# mapping_registry.set_default('git-experimental')
550
mapping_registry.set_default(b'git-v1')
553
class ForeignGit(ForeignVcs):
554
"""The Git Stupid Content Tracker"""
557
def branch_format(self):
558
from .branch import LocalGitBranchFormat
559
return LocalGitBranchFormat()
562
def repository_format(self):
563
from .repository import GitRepositoryFormat
564
return GitRepositoryFormat()
567
super(ForeignGit, self).__init__(mapping_registry)
568
self.abbreviation = "git"
571
def serialize_foreign_revid(self, foreign_revid):
575
def show_foreign_revid(cls, foreign_revid):
576
return {"git commit": foreign_revid.decode('utf-8')}
579
foreign_vcs_git = ForeignGit()
580
default_mapping = mapping_registry.get_default()()
583
def symlink_to_blob(symlink_target):
584
from dulwich.objects import Blob
586
if isinstance(symlink_target, text_type):
587
symlink_target = symlink_target.encode('utf-8')
588
blob.data = symlink_target
592
def mode_is_executable(mode):
593
"""Check if mode should be considered executable."""
594
return bool(mode & 0o111)
598
"""Determine the Bazaar inventory kind based on Unix file mode."""
601
entry_kind = (mode & 0o700000) / 0o100000
604
elif entry_kind == 1:
605
file_kind = (mode & 0o70000) / 0o10000
611
return 'tree-reference'
613
raise AssertionError(
614
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
616
raise AssertionError(
617
"Unknown kind, perms=%r." % (mode,))
620
def object_mode(kind, executable):
621
if kind == 'directory':
623
elif kind == 'symlink':
629
mode = stat.S_IFREG | 0o644
633
elif kind == 'tree-reference':
634
from dulwich.objects import S_IFGITLINK
640
def entry_mode(entry):
641
"""Determine the git file mode for an inventory entry."""
642
return object_mode(entry.kind, getattr(entry, 'executable', False))
645
def extract_unusual_modes(rev):
647
foreign_revid, mapping = mapping_registry.parse_revision_id(
649
except errors.InvalidRevisionId:
652
return mapping.export_unusual_file_modes(rev)
655
def parse_git_svn_id(text):
656
(head, uuid) = text.rsplit(" ", 1)
657
(full_url, rev) = head.rsplit("@", 1)
658
return (full_url, int(rev), uuid)
661
def needs_roundtripping(repo, revid):
663
mapping_registry.parse_revision_id(revid)
664
except errors.InvalidRevisionId: