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
32
from ...bzr.inventory import (
35
from ...foreign import (
40
from ...revision import (
46
UnknownMercurialCommitExtra,
52
from .roundtrip import (
56
deserialize_fileid_map,
60
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
61
HG_RENAME_SOURCE = "HG:rename-source"
64
# This HG extra is used to indicate the commit that this commit was based on.
65
HG_EXTRA_AMEND_SOURCE = "amend_source"
67
FILE_ID_PREFIX = b'git:'
70
def escape_file_id(file_id):
71
return file_id.replace('_', '__').replace(' ', '_s').replace('\x0c', '_c')
74
def unescape_file_id(file_id):
77
while i < len(file_id):
79
ret.append(file_id[i])
81
if file_id[i+1] == '_':
83
elif file_id[i+1] == 's':
85
elif file_id[i+1] == 'c':
88
raise ValueError("unknown escape character %s" %
95
def fix_person_identifier(text):
96
if "<" in text and ">" in text:
97
if not " <" in text and text.count("<") == 1:
98
text = text.replace("<", " <")
100
return "%s <%s>" % (text, text)
103
def warn_escaped(commit, num_escaped):
104
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
105
"to regenerate the SHA map.", num_escaped, commit)
108
def warn_unusual_mode(commit, path, mode):
109
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
110
"property. ", mode, path, commit)
113
class BzrGitMapping(foreign.VcsMapping):
114
"""Class that maps between Git and Bazaar semantics."""
117
BZR_FILE_IDS_FILE = None
119
BZR_DUMMY_FILE = None
121
def is_special_file(self, filename):
122
return (filename in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE))
125
super(BzrGitMapping, self).__init__(foreign_vcs_git)
127
def __eq__(self, other):
128
return (type(self) == type(other) and
129
self.revid_prefix == other.revid_prefix)
132
def revision_id_foreign_to_bzr(cls, git_rev_id):
133
"""Convert a git revision id handle to a Bazaar revision id."""
134
from dulwich.protocol import ZERO_SHA
135
if git_rev_id == ZERO_SHA:
137
return "%s:%s" % (cls.revid_prefix, git_rev_id)
140
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
141
"""Convert a Bazaar revision id to a git revision id handle."""
142
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
143
raise errors.InvalidRevisionId(bzr_rev_id, cls)
144
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
146
def generate_file_id(self, path):
147
# Git paths are just bytestrings
148
# We must just hope they are valid UTF-8..
151
if type(path) is unicode:
152
path = path.encode("utf-8")
153
return FILE_ID_PREFIX + escape_file_id(path)
155
def parse_file_id(self, file_id):
156
if file_id == ROOT_ID:
158
if not file_id.startswith(FILE_ID_PREFIX):
160
return unescape_file_id(file_id[len(FILE_ID_PREFIX):])
162
def revid_as_refname(self, revid):
164
return "refs/bzr/%s" % urllib.quote(revid)
166
def import_unusual_file_modes(self, rev, unusual_file_modes):
167
if unusual_file_modes:
168
ret = [(path, unusual_file_modes[path])
169
for path in sorted(unusual_file_modes.keys())]
170
rev.properties['file-modes'] = bencode.bencode(ret)
172
def export_unusual_file_modes(self, rev):
174
file_modes = rev.properties['file-modes']
178
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
180
def _generate_git_svn_metadata(self, rev, encoding):
182
git_svn_id = rev.properties["git-svn-id"]
186
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
188
def _generate_hg_message_tail(self, rev):
192
for name in rev.properties:
193
if name == 'hg:extra:branch':
194
branch = rev.properties['hg:extra:branch']
195
elif name.startswith('hg:extra'):
196
extra[name[len('hg:extra:'):]] = base64.b64decode(
197
rev.properties[name])
198
elif name == 'hg:renames':
199
renames = bencode.bdecode(base64.b64decode(
200
rev.properties['hg:renames']))
201
# TODO: Export other properties as 'bzr:' extras?
202
ret = format_hg_metadata(renames, branch, extra)
203
if type(ret) is not str:
207
def _extract_git_svn_metadata(self, rev, message):
208
lines = message.split("\n")
209
if not (lines[-1] == "" and len(lines) >= 2 and lines[-2].startswith("git-svn-id:")):
211
git_svn_id = lines[-2].split(": ", 1)[1]
212
rev.properties['git-svn-id'] = git_svn_id
213
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
214
# FIXME: Convert this to converted-from property somehow..
215
return "\n".join(lines[:-2])
217
def _extract_hg_metadata(self, rev, message):
218
(message, renames, branch, extra) = extract_hg_metadata(message)
219
if branch is not None:
220
rev.properties['hg:extra:branch'] = branch
221
for name, value in extra.iteritems():
222
rev.properties['hg:extra:' + name] = base64.b64encode(value)
224
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
225
[(new, old) for (old, new) in renames.iteritems()]))
228
def _extract_bzr_metadata(self, rev, message):
229
(message, metadata) = extract_bzr_metadata(message)
230
return message, metadata
232
def _decode_commit_message(self, rev, message, encoding):
233
return message.decode(encoding), CommitSupplement()
235
def _encode_commit_message(self, rev, message, encoding):
236
return message.encode(encoding)
238
def export_fileid_map(self, fileid_map):
239
"""Export a file id map to a fileid map.
241
:param fileid_map: File id map, mapping paths to file ids
242
:return: A Git blob object (or None if there are no entries)
244
from dulwich.objects import Blob
246
b.set_raw_chunks(serialize_fileid_map(fileid_map))
249
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
251
"""Turn a Bazaar revision in to a Git commit
253
:param tree_sha: Tree sha for the commit
254
:param parent_lookup: Function for looking up the GIT sha equiv of a
256
:param lossy: Whether to store roundtripping information.
257
:param verifiers: Verifiers info
258
:return dulwich.objects.Commit represent the revision:
260
from dulwich.objects import Commit
262
commit.tree = tree_sha
264
metadata = CommitSupplement()
265
metadata.verifiers = verifiers
269
for p in rev.parent_ids:
271
git_p = parent_lookup(p)
274
if metadata is not None:
275
metadata.explicit_parent_ids = rev.parent_ids
276
if git_p is not None:
278
raise AssertionError("unexpected length for %r" % git_p)
279
parents.append(git_p)
280
commit.parents = parents
282
encoding = rev.properties['git-explicit-encoding']
284
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
286
commit.encoding = rev.properties['git-explicit-encoding'].encode('ascii')
289
commit.committer = fix_person_identifier(rev.committer.encode(
291
commit.author = fix_person_identifier(
292
rev.get_apparent_authors()[0].encode(encoding))
293
commit.commit_time = long(rev.timestamp)
294
if 'author-timestamp' in rev.properties:
295
commit.author_time = long(rev.properties['author-timestamp'])
297
commit.author_time = commit.commit_time
298
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
299
commit.commit_timezone = rev.timezone
300
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
301
if 'author-timezone' in rev.properties:
302
commit.author_timezone = int(rev.properties['author-timezone'])
304
commit.author_timezone = commit.commit_timezone
305
if 'git-gpg-signature' in rev.properties:
306
commit.gpgsig = rev.properties['git-gpg-signature'].encode('ascii')
307
commit.message = self._encode_commit_message(rev, rev.message,
309
if type(commit.message) is not str:
310
raise TypeError(commit.message)
311
if metadata is not None:
313
mapping_registry.parse_revision_id(rev.revision_id)
314
except errors.InvalidRevisionId:
315
metadata.revision_id = rev.revision_id
316
mapping_properties = set(
317
['author', 'author-timezone', 'author-timezone-neg-utc',
318
'commit-timezone-neg-utc', 'git-implicit-encoding',
319
'git-gpg-signature', 'git-explicit-encoding',
320
'author-timestamp', 'file-modes'])
321
for k, v in rev.properties.iteritems():
322
if not k in mapping_properties:
323
metadata.properties[k] = v
324
if not lossy and metadata:
325
if self.roundtripping:
326
commit.message = inject_bzr_metadata(commit.message, metadata,
329
raise NoPushSupport()
330
if type(commit.message) is not str:
331
raise TypeError(commit.message)
332
if 'git-extra' in rev.properties:
333
commit.extra.extend([l.split(' ', 1) for l in rev.properties['git-extra'].splitlines()])
336
def import_fileid_map(self, blob):
337
"""Convert a git file id map blob.
339
:param blob: Git blob object with fileid map
340
:return: Dictionary mapping paths to file ids
342
return deserialize_fileid_map(blob.data)
344
def import_commit(self, commit, lookup_parent_revid):
345
"""Convert a git commit to a bzr revision.
347
:return: a `breezy.revision.Revision` object, foreign revid and a
351
raise AssertionError("Commit object can't be None")
352
rev = ForeignRevision(commit.id, self,
353
self.revision_id_foreign_to_bzr(commit.id))
354
rev.git_metadata = None
355
def decode_using_encoding(rev, commit, encoding):
356
rev.committer = str(commit.committer).decode(encoding)
357
if commit.committer != commit.author:
358
rev.properties['author'] = str(commit.author).decode(encoding)
359
rev.message, rev.git_metadata = self._decode_commit_message(
360
rev, commit.message, encoding)
361
if commit.encoding is not None:
362
rev.properties['git-explicit-encoding'] = commit.encoding
363
decode_using_encoding(rev, commit, commit.encoding)
365
for encoding in ('utf-8', 'latin1'):
367
decode_using_encoding(rev, commit, encoding)
368
except UnicodeDecodeError:
371
if encoding != 'utf-8':
372
rev.properties['git-implicit-encoding'] = encoding
374
if commit.commit_time != commit.author_time:
375
rev.properties['author-timestamp'] = str(commit.author_time)
376
if commit.commit_timezone != commit.author_timezone:
377
rev.properties['author-timezone'] = "%d" % commit.author_timezone
378
if commit._author_timezone_neg_utc:
379
rev.properties['author-timezone-neg-utc'] = ""
380
if commit._commit_timezone_neg_utc:
381
rev.properties['commit-timezone-neg-utc'] = ""
383
rev.properties['git-gpg-signature'] = commit.gpgsig.decode('ascii')
384
rev.timestamp = commit.commit_time
385
rev.timezone = commit.commit_timezone
386
rev.parent_ids = None
387
if rev.git_metadata is not None:
388
md = rev.git_metadata
389
roundtrip_revid = md.revision_id
390
if md.explicit_parent_ids:
391
rev.parent_ids = md.explicit_parent_ids
392
rev.properties.update(md.properties)
393
verifiers = md.verifiers
395
roundtrip_revid = None
397
if rev.parent_ids is None:
398
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
399
unknown_extra_fields = []
401
for k, v in commit.extra:
402
if k == HG_RENAME_SOURCE:
403
extra_lines.append(k + ' ' + v + '\n')
405
hgk, hgv = v.split(':', 1)
406
if hgk not in (HG_EXTRA_AMEND_SOURCE, ):
407
raise UnknownMercurialCommitExtra(commit, hgk)
408
extra_lines.append(k + ' ' + v + '\n')
410
unknown_extra_fields.append(k)
411
if unknown_extra_fields:
412
raise UnknownCommitExtra(commit, unknown_extra_fields)
414
rev.properties['git-extra'] = ''.join(extra_lines)
415
return rev, roundtrip_revid, verifiers
417
def get_fileid_map(self, lookup_object, tree_sha):
418
"""Obtain a fileid map for a particular tree.
420
:param lookup_object: Function for looking up an object
421
:param tree_sha: SHA of the root tree
422
:return: GitFileIdMap instance
425
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
429
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
430
return GitFileIdMap(file_ids, self)
433
class BzrGitMappingv1(BzrGitMapping):
434
revid_prefix = 'git-v1'
438
return self.revid_prefix
441
class BzrGitMappingExperimental(BzrGitMappingv1):
442
revid_prefix = 'git-experimental'
446
BZR_FILE_IDS_FILE = '.bzrfileids'
448
BZR_DUMMY_FILE = '.bzrdummy'
450
def _decode_commit_message(self, rev, message, encoding):
451
message = self._extract_hg_metadata(rev, message)
452
message = self._extract_git_svn_metadata(rev, message)
453
message, metadata = self._extract_bzr_metadata(rev, message)
454
return message.decode(encoding), metadata
456
def _encode_commit_message(self, rev, message, encoding):
457
ret = message.encode(encoding)
458
ret += self._generate_hg_message_tail(rev)
459
ret += self._generate_git_svn_metadata(rev, encoding)
462
def import_commit(self, commit, lookup_parent_revid):
463
rev, roundtrip_revid, verifiers = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
464
rev.properties['converted_revision'] = "git %s\n" % commit.id
465
return rev, roundtrip_revid, verifiers
468
class GitMappingRegistry(VcsMappingRegistry):
469
"""Registry with available git mappings."""
471
def revision_id_bzr_to_foreign(self, bzr_revid):
472
if bzr_revid == NULL_REVISION:
473
from dulwich.protocol import ZERO_SHA
474
return ZERO_SHA, None
475
if not bzr_revid.startswith("git-"):
476
raise errors.InvalidRevisionId(bzr_revid, None)
477
(mapping_version, git_sha) = bzr_revid.split(":", 1)
478
mapping = self.get(mapping_version)
479
return mapping.revision_id_bzr_to_foreign(bzr_revid)
481
parse_revision_id = revision_id_bzr_to_foreign
484
mapping_registry = GitMappingRegistry()
485
mapping_registry.register_lazy('git-v1', "breezy.plugins.git.mapping",
487
mapping_registry.register_lazy('git-experimental',
488
"breezy.plugins.git.mapping", "BzrGitMappingExperimental")
489
# Uncomment the next line to enable the experimental bzr-git mappings.
490
# This will make sure all bzr metadata is pushed into git, allowing for
491
# full roundtripping later.
492
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
493
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
494
#mapping_registry.set_default('git-experimental')
495
mapping_registry.set_default('git-v1')
498
class ForeignGit(ForeignVcs):
499
"""The Git Stupid Content Tracker"""
502
def branch_format(self):
503
from .branch import LocalGitBranchFormat
504
return LocalGitBranchFormat()
507
def repository_format(self):
508
from .repository import GitRepositoryFormat
509
return GitRepositoryFormat()
512
super(ForeignGit, self).__init__(mapping_registry)
513
self.abbreviation = "git"
516
def serialize_foreign_revid(self, foreign_revid):
520
def show_foreign_revid(cls, foreign_revid):
521
return { "git commit": foreign_revid }
524
foreign_vcs_git = ForeignGit()
525
default_mapping = mapping_registry.get_default()()
528
def symlink_to_blob(symlink_target):
529
from dulwich.objects import Blob
531
if type(symlink_target) == unicode:
532
symlink_target = symlink_target.encode('utf-8')
533
blob.data = symlink_target
537
def mode_is_executable(mode):
538
"""Check if mode should be considered executable."""
539
return bool(mode & 0111)
543
"""Determine the Bazaar inventory kind based on Unix file mode."""
546
entry_kind = (mode & 0700000) / 0100000
549
elif entry_kind == 1:
550
file_kind = (mode & 070000) / 010000
556
return 'tree-reference'
558
raise AssertionError(
559
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
561
raise AssertionError(
562
"Unknown kind, perms=%r." % (mode,))
565
def object_mode(kind, executable):
566
if kind == 'directory':
568
elif kind == 'symlink':
574
mode = stat.S_IFREG | 0644
578
elif kind == 'tree-reference':
579
from dulwich.objects import S_IFGITLINK
585
def entry_mode(entry):
586
"""Determine the git file mode for an inventory entry."""
587
return object_mode(entry.kind, entry.executable)
590
def directory_to_tree(children, lookup_ie_sha1, unusual_modes, empty_file_name,
592
"""Create a Git Tree object from a Bazaar directory.
594
:param children: Children inventory entries
595
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
596
:param unusual_modes: Dictionary with unusual file modes by file ids
597
:param empty_file_name: Name to use for dummy files in empty directories,
598
None to ignore empty directories.
600
from dulwich.objects import Blob, Tree
602
for name, value in children.iteritems():
605
mode = unusual_modes[ie.file_id]
607
mode = entry_mode(ie)
608
hexsha = lookup_ie_sha1(ie)
609
if hexsha is not None:
610
tree.add(name.encode("utf-8"), mode, hexsha)
611
if not allow_empty and len(tree) == 0:
612
# Only the root can be an empty tree
613
if empty_file_name is not None:
614
tree.add(empty_file_name, stat.S_IFREG | 0644, Blob().id)
620
def extract_unusual_modes(rev):
622
foreign_revid, mapping = mapping_registry.parse_revision_id(
624
except errors.InvalidRevisionId:
627
return mapping.export_unusual_file_modes(rev)
630
def parse_git_svn_id(text):
631
(head, uuid) = text.rsplit(" ", 1)
632
(full_url, rev) = head.rsplit("@", 1)
633
return (full_url, int(rev), uuid)
636
class GitFileIdMap(object):
638
def __init__(self, file_ids, mapping):
639
self.file_ids = file_ids
641
self.mapping = mapping
643
def all_file_ids(self):
644
return self.file_ids.values()
646
def set_file_id(self, path, file_id):
647
if type(path) is not str:
648
raise TypeError(path)
649
if type(file_id) is not str:
650
raise TypeError(file_id)
651
self.file_ids[path] = file_id
653
def lookup_file_id(self, path):
654
if type(path) is not str:
655
raise TypeError(path)
657
file_id = self.file_ids[path]
659
file_id = self.mapping.generate_file_id(path)
660
if type(file_id) is not str:
661
raise TypeError(file_id)
664
def lookup_path(self, file_id):
665
if self.paths is None:
667
for k, v in self.file_ids.iteritems():
670
path = self.paths[file_id]
672
return self.mapping.parse_file_id(file_id)
674
if type(path) is not str:
675
raise TypeError(path)
679
return self.__class__(dict(self.file_ids), self.mapping)
682
def needs_roundtripping(repo, revid):
684
mapping_registry.parse_revision_id(revid)
685
except errors.InvalidRevisionId:
added
removed
mapping.py
