1
# Copyright (C) 2007 Canonical Ltd
2
# Copyright (C) 2008-2010 Jelmer Vernooij <jelmer@samba.org>
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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 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"
68
def escape_file_id(file_id):
69
return file_id.replace('_', '__').replace(' ', '_s').replace('\x0c', '_c')
72
def unescape_file_id(file_id):
75
while i < len(file_id):
77
ret.append(file_id[i])
79
if file_id[i+1] == '_':
81
elif file_id[i+1] == 's':
83
elif file_id[i+1] == 'c':
86
raise AssertionError("unknown escape character %s" %
93
def fix_person_identifier(text):
94
if "<" in text and ">" in text:
95
if not " <" in text and text.count("<") == 1:
96
text = text.replace("<", " <")
98
return "%s <%s>" % (text, text)
101
def warn_escaped(commit, num_escaped):
102
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
103
"to regenerate the SHA map.", num_escaped, commit)
106
def warn_unusual_mode(commit, path, mode):
107
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
108
"property. ", mode, path, commit)
111
class BzrGitMapping(foreign.VcsMapping):
112
"""Class that maps between Git and Bazaar semantics."""
115
BZR_FILE_IDS_FILE = None
117
BZR_DUMMY_FILE = None
119
def is_special_file(self, filename):
120
return (filename in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE))
123
super(BzrGitMapping, self).__init__(foreign_vcs_git)
125
def __eq__(self, other):
126
return (type(self) == type(other) and
127
self.revid_prefix == other.revid_prefix)
130
def revision_id_foreign_to_bzr(cls, git_rev_id):
131
"""Convert a git revision id handle to a Bazaar revision id."""
132
from dulwich.protocol import ZERO_SHA
133
if git_rev_id == ZERO_SHA:
135
return "%s:%s" % (cls.revid_prefix, git_rev_id)
138
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
139
"""Convert a Bazaar revision id to a git revision id handle."""
140
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
141
raise errors.InvalidRevisionId(bzr_rev_id, cls)
142
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
144
def generate_file_id(self, path):
145
# Git paths are just bytestrings
146
# We must just hope they are valid UTF-8..
149
if type(path) is unicode:
150
path = path.encode("utf-8")
151
return escape_file_id(path)
153
def is_control_file(self, path):
154
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
156
def parse_file_id(self, file_id):
157
if file_id == ROOT_ID:
159
return unescape_file_id(file_id)
161
def revid_as_refname(self, revid):
163
return "refs/bzr/%s" % urllib.quote(revid)
165
def import_unusual_file_modes(self, rev, unusual_file_modes):
166
if unusual_file_modes:
167
ret = [(path, unusual_file_modes[path])
168
for path in sorted(unusual_file_modes.keys())]
169
rev.properties['file-modes'] = bencode.bencode(ret)
171
def export_unusual_file_modes(self, rev):
173
file_modes = rev.properties['file-modes']
177
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
179
def _generate_git_svn_metadata(self, rev, encoding):
181
git_svn_id = rev.properties["git-svn-id"]
185
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
187
def _generate_hg_message_tail(self, rev):
191
for name in rev.properties:
192
if name == 'hg:extra:branch':
193
branch = rev.properties['hg:extra:branch']
194
elif name.startswith('hg:extra'):
195
extra[name[len('hg:extra:'):]] = base64.b64decode(
196
rev.properties[name])
197
elif name == 'hg:renames':
198
renames = bencode.bdecode(base64.b64decode(
199
rev.properties['hg:renames']))
200
# TODO: Export other properties as 'bzr:' extras?
201
ret = format_hg_metadata(renames, branch, extra)
202
assert isinstance(ret, str)
205
def _extract_git_svn_metadata(self, rev, message):
206
lines = message.split("\n")
207
if not (lines[-1] == "" and len(lines) >= 2 and lines[-2].startswith("git-svn-id:")):
209
git_svn_id = lines[-2].split(": ", 1)[1]
210
rev.properties['git-svn-id'] = git_svn_id
211
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
212
# FIXME: Convert this to converted-from property somehow..
213
ret = "\n".join(lines[:-2])
214
assert isinstance(ret, str)
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
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:
277
assert len(git_p) == 40, "unexpected length for %r" % git_p
278
parents.append(git_p)
279
commit.parents = parents
281
encoding = rev.properties['git-explicit-encoding']
283
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
284
commit.encoding = rev.properties.get('git-explicit-encoding')
285
commit.committer = fix_person_identifier(rev.committer.encode(
287
commit.author = fix_person_identifier(
288
rev.get_apparent_authors()[0].encode(encoding))
289
commit.commit_time = long(rev.timestamp)
290
if 'author-timestamp' in rev.properties:
291
commit.author_time = long(rev.properties['author-timestamp'])
293
commit.author_time = commit.commit_time
294
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
295
commit.commit_timezone = rev.timezone
296
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
297
if 'author-timezone' in rev.properties:
298
commit.author_timezone = int(rev.properties['author-timezone'])
300
commit.author_timezone = commit.commit_timezone
301
commit.message = self._encode_commit_message(rev, rev.message,
303
assert type(commit.message) == str
304
if metadata is not None:
306
mapping_registry.parse_revision_id(rev.revision_id)
307
except errors.InvalidRevisionId:
308
metadata.revision_id = rev.revision_id
309
mapping_properties = set(
310
['author', 'author-timezone', 'author-timezone-neg-utc',
311
'commit-timezone-neg-utc', 'git-implicit-encoding',
312
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
313
for k, v in rev.properties.iteritems():
314
if not k in mapping_properties:
315
metadata.properties[k] = v
317
if self.roundtripping:
318
commit.message = inject_bzr_metadata(commit.message, metadata,
321
raise NoPushSupport()
322
assert type(commit.message) == str
323
if 'git-extra' in rev.properties:
324
commit.extra.extend([l.split(' ', 1) for l in rev.properties['git-extra'].splitlines()])
327
def import_fileid_map(self, blob):
328
"""Convert a git file id map blob.
330
:param blob: Git blob object with fileid map
331
:return: Dictionary mapping paths to file ids
333
return deserialize_fileid_map(blob.data)
335
def import_commit(self, commit, lookup_parent_revid):
336
"""Convert a git commit to a bzr revision.
338
:return: a `breezy.revision.Revision` object, foreign revid and a
342
raise AssertionError("Commit object can't be None")
343
rev = ForeignRevision(commit.id, self,
344
self.revision_id_foreign_to_bzr(commit.id))
345
rev.git_metadata = None
346
def decode_using_encoding(rev, commit, encoding):
347
rev.committer = str(commit.committer).decode(encoding)
348
if commit.committer != commit.author:
349
rev.properties['author'] = str(commit.author).decode(encoding)
350
rev.message, rev.git_metadata = self._decode_commit_message(
351
rev, commit.message, encoding)
352
if commit.encoding is not None:
353
rev.properties['git-explicit-encoding'] = commit.encoding
354
decode_using_encoding(rev, commit, commit.encoding)
356
for encoding in ('utf-8', 'latin1'):
358
decode_using_encoding(rev, commit, encoding)
359
except UnicodeDecodeError:
362
if encoding != 'utf-8':
363
rev.properties['git-implicit-encoding'] = encoding
365
if commit.commit_time != commit.author_time:
366
rev.properties['author-timestamp'] = str(commit.author_time)
367
if commit.commit_timezone != commit.author_timezone:
368
rev.properties['author-timezone'] = "%d" % commit.author_timezone
369
if commit._author_timezone_neg_utc:
370
rev.properties['author-timezone-neg-utc'] = ""
371
if commit._commit_timezone_neg_utc:
372
rev.properties['commit-timezone-neg-utc'] = ""
373
rev.timestamp = commit.commit_time
374
rev.timezone = commit.commit_timezone
375
rev.parent_ids = None
376
if rev.git_metadata is not None:
377
md = rev.git_metadata
378
roundtrip_revid = md.revision_id
379
if md.explicit_parent_ids:
380
rev.parent_ids = md.explicit_parent_ids
381
rev.properties.update(md.properties)
382
verifiers = md.verifiers
384
roundtrip_revid = None
386
if rev.parent_ids is None:
387
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
388
unknown_extra_fields = []
390
for k, v in commit.extra:
391
if k == HG_RENAME_SOURCE:
392
extra_lines.append(k + ' ' + v + '\n')
394
hgk, hgv = v.split(':', 1)
395
if hgk not in (HG_EXTRA_AMEND_SOURCE, ):
396
raise UnknownMercurialCommitExtra(commit, hgk)
397
extra_lines.append(k + ' ' + v + '\n')
399
unknown_extra_fields.append(k)
400
if unknown_extra_fields:
401
raise UnknownCommitExtra(commit, unknown_extra_fields)
403
rev.properties['git-extra'] = ''.join(extra_lines)
404
return rev, roundtrip_revid, verifiers
406
def get_fileid_map(self, lookup_object, tree_sha):
407
"""Obtain a fileid map for a particular tree.
409
:param lookup_object: Function for looking up an object
410
:param tree_sha: SHA of the root tree
411
:return: GitFileIdMap instance
414
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
418
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
419
return GitFileIdMap(file_ids, self)
422
class BzrGitMappingv1(BzrGitMapping):
423
revid_prefix = 'git-v1'
427
return self.revid_prefix
430
class BzrGitMappingExperimental(BzrGitMappingv1):
431
revid_prefix = 'git-experimental'
435
BZR_FILE_IDS_FILE = '.bzrfileids'
437
BZR_DUMMY_FILE = '.bzrdummy'
439
def _decode_commit_message(self, rev, message, encoding):
440
message = self._extract_hg_metadata(rev, message)
441
message = self._extract_git_svn_metadata(rev, message)
442
message, metadata = self._extract_bzr_metadata(rev, message)
443
return message.decode(encoding), metadata
445
def _encode_commit_message(self, rev, message, encoding):
446
ret = message.encode(encoding)
447
ret += self._generate_hg_message_tail(rev)
448
ret += self._generate_git_svn_metadata(rev, encoding)
451
def import_commit(self, commit, lookup_parent_revid):
452
rev, roundtrip_revid, verifiers = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
453
rev.properties['converted_revision'] = "git %s\n" % commit.id
454
return rev, roundtrip_revid, verifiers
457
class GitMappingRegistry(VcsMappingRegistry):
458
"""Registry with available git mappings."""
460
def revision_id_bzr_to_foreign(self, bzr_revid):
461
if bzr_revid == NULL_REVISION:
462
from dulwich.protocol import ZERO_SHA
463
return ZERO_SHA, None
464
if not bzr_revid.startswith("git-"):
465
raise errors.InvalidRevisionId(bzr_revid, None)
466
(mapping_version, git_sha) = bzr_revid.split(":", 1)
467
mapping = self.get(mapping_version)
468
return mapping.revision_id_bzr_to_foreign(bzr_revid)
470
parse_revision_id = revision_id_bzr_to_foreign
473
mapping_registry = GitMappingRegistry()
474
mapping_registry.register_lazy('git-v1', "breezy.plugins.git.mapping",
476
mapping_registry.register_lazy('git-experimental',
477
"breezy.plugins.git.mapping", "BzrGitMappingExperimental")
478
# Uncomment the next line to enable the experimental bzr-git mappings.
479
# This will make sure all bzr metadata is pushed into git, allowing for
480
# full roundtripping later.
481
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
482
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
483
#mapping_registry.set_default('git-experimental')
484
mapping_registry.set_default('git-v1')
487
class ForeignGit(ForeignVcs):
488
"""The Git Stupid Content Tracker"""
491
def branch_format(self):
492
from .branch import GitBranchFormat
493
return GitBranchFormat()
496
def repository_format(self):
497
from .repository import GitRepositoryFormat
498
return GitRepositoryFormat()
501
super(ForeignGit, self).__init__(mapping_registry)
502
self.abbreviation = "git"
505
def serialize_foreign_revid(self, foreign_revid):
509
def show_foreign_revid(cls, foreign_revid):
510
return { "git commit": foreign_revid }
513
foreign_vcs_git = ForeignGit()
514
default_mapping = mapping_registry.get_default()()
517
def symlink_to_blob(symlink_target):
518
from dulwich.objects import Blob
520
if type(symlink_target) == unicode:
521
symlink_target = symlink_target.encode('utf-8')
522
blob.data = symlink_target
526
def mode_is_executable(mode):
527
"""Check if mode should be considered executable."""
528
return bool(mode & 0111)
532
"""Determine the Bazaar inventory kind based on Unix file mode."""
535
entry_kind = (mode & 0700000) / 0100000
538
elif entry_kind == 1:
539
file_kind = (mode & 070000) / 010000
545
return 'tree-reference'
547
raise AssertionError(
548
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
550
raise AssertionError(
551
"Unknown kind, perms=%r." % (mode,))
554
def object_mode(kind, executable):
555
if kind == 'directory':
557
elif kind == 'symlink':
563
mode = stat.S_IFREG | 0644
567
elif kind == 'tree-reference':
568
from dulwich.objects import S_IFGITLINK
574
def entry_mode(entry):
575
"""Determine the git file mode for an inventory entry."""
576
return object_mode(entry.kind, entry.executable)
579
def directory_to_tree(children, lookup_ie_sha1, unusual_modes, empty_file_name,
581
"""Create a Git Tree object from a Bazaar directory.
583
:param children: Children inventory entries
584
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
585
:param unusual_modes: Dictionary with unusual file modes by file ids
586
:param empty_file_name: Name to use for dummy files in empty directories,
587
None to ignore empty directories.
589
from dulwich.objects import Blob, Tree
591
for name, value in children.iteritems():
594
mode = unusual_modes[ie.file_id]
596
mode = entry_mode(ie)
597
hexsha = lookup_ie_sha1(ie)
598
if hexsha is not None:
599
tree.add(name.encode("utf-8"), mode, hexsha)
600
if not allow_empty and len(tree) == 0:
601
# Only the root can be an empty tree
602
if empty_file_name is not None:
603
tree.add(empty_file_name, stat.S_IFREG | 0644, Blob().id)
609
def extract_unusual_modes(rev):
611
foreign_revid, mapping = mapping_registry.parse_revision_id(
613
except errors.InvalidRevisionId:
616
return mapping.export_unusual_file_modes(rev)
619
def parse_git_svn_id(text):
620
(head, uuid) = text.rsplit(" ", 1)
621
(full_url, rev) = head.rsplit("@", 1)
622
return (full_url, int(rev), uuid)
625
class GitFileIdMap(object):
627
def __init__(self, file_ids, mapping):
628
self.file_ids = file_ids
630
self.mapping = mapping
632
def all_file_ids(self):
633
return self.file_ids.values()
635
def set_file_id(self, path, file_id):
636
assert type(path) is str
637
assert type(file_id) is str
638
self.file_ids[path] = file_id
640
def lookup_file_id(self, path):
641
assert type(path) is str
643
file_id = self.file_ids[path]
645
file_id = self.mapping.generate_file_id(path)
646
assert type(file_id) is str
649
def lookup_path(self, file_id):
650
if self.paths is None:
652
for k, v in self.file_ids.iteritems():
655
path = self.paths[file_id]
657
return self.mapping.parse_file_id(file_id)
659
assert type(path) is str
663
return self.__class__(dict(self.file_ids), self.mapping)
added
removed
mapping.py
