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."""
30
from bzrlib.inventory import (
33
from bzrlib.foreign import (
38
from bzrlib.revision import (
41
from bzrlib.plugins.git.hg import (
45
from bzrlib.plugins.git.roundtrip import (
49
deserialize_fileid_map,
53
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
56
def escape_file_id(file_id):
57
return file_id.replace('_', '__').replace(' ', '_s').replace('\x0c', '_c')
60
def unescape_file_id(file_id):
63
while i < len(file_id):
65
ret.append(file_id[i])
67
if file_id[i+1] == '_':
69
elif file_id[i+1] == 's':
71
elif file_id[i+1] == 'c':
74
raise AssertionError("unknown escape character %s" %
81
def fix_person_identifier(text):
82
if "<" in text and ">" in text:
84
return "%s <%s>" % (text, text)
87
def warn_escaped(commit, num_escaped):
88
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
89
"to regenerate the SHA map.", num_escaped, commit)
92
def warn_unusual_mode(commit, path, mode):
93
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
94
"property. ", mode, path, commit)
97
class BzrGitMapping(foreign.VcsMapping):
98
"""Class that maps between Git and Bazaar semantics."""
101
BZR_FILE_IDS_FILE = None
103
BZR_DUMMY_FILE = None
105
def is_special_file(self, filename):
106
return (filename in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE))
109
super(BzrGitMapping, self).__init__(foreign_vcs_git)
111
def __eq__(self, other):
112
return (type(self) == type(other) and
113
self.revid_prefix == other.revid_prefix)
116
def revision_id_foreign_to_bzr(cls, git_rev_id):
117
"""Convert a git revision id handle to a Bazaar revision id."""
118
from dulwich.protocol import ZERO_SHA
119
if git_rev_id == ZERO_SHA:
121
return "%s:%s" % (cls.revid_prefix, git_rev_id)
124
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
125
"""Convert a Bazaar revision id to a git revision id handle."""
126
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
127
raise errors.InvalidRevisionId(bzr_rev_id, cls)
128
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
130
def generate_file_id(self, path):
131
# Git paths are just bytestrings
132
# We must just hope they are valid UTF-8..
135
if type(path) is unicode:
136
path = path.encode("utf-8")
137
return escape_file_id(path)
139
def is_control_file(self, path):
140
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
142
def parse_file_id(self, file_id):
143
if file_id == ROOT_ID:
145
return unescape_file_id(file_id)
147
def revid_as_refname(self, revid):
149
return "refs/bzr/%s" % urllib.quote(revid)
151
def import_unusual_file_modes(self, rev, unusual_file_modes):
152
if unusual_file_modes:
153
ret = [(path, unusual_file_modes[path])
154
for path in sorted(unusual_file_modes.keys())]
155
rev.properties['file-modes'] = bencode.bencode(ret)
157
def export_unusual_file_modes(self, rev):
159
file_modes = rev.properties['file-modes']
163
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
165
def _generate_git_svn_metadata(self, rev, encoding):
167
git_svn_id = rev.properties["git-svn-id"]
171
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
173
def _generate_hg_message_tail(self, rev):
177
for name in rev.properties:
178
if name == 'hg:extra:branch':
179
branch = rev.properties['hg:extra:branch']
180
elif name.startswith('hg:extra'):
181
extra[name[len('hg:extra:'):]] = base64.b64decode(
182
rev.properties[name])
183
elif name == 'hg:renames':
184
renames = bencode.bdecode(base64.b64decode(
185
rev.properties['hg:renames']))
186
# TODO: Export other properties as 'bzr:' extras?
187
ret = format_hg_metadata(renames, branch, extra)
188
assert isinstance(ret, str)
191
def _extract_git_svn_metadata(self, rev, message):
192
lines = message.split("\n")
193
if not (lines[-1] == "" and len(lines) >= 2 and lines[-2].startswith("git-svn-id:")):
195
git_svn_id = lines[-2].split(": ", 1)[1]
196
rev.properties['git-svn-id'] = git_svn_id
197
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
198
# FIXME: Convert this to converted-from property somehow..
199
ret = "\n".join(lines[:-2])
200
assert isinstance(ret, str)
203
def _extract_hg_metadata(self, rev, message):
204
(message, renames, branch, extra) = extract_hg_metadata(message)
205
if branch is not None:
206
rev.properties['hg:extra:branch'] = branch
207
for name, value in extra.iteritems():
208
rev.properties['hg:extra:' + name] = base64.b64encode(value)
210
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
211
[(new, old) for (old, new) in renames.iteritems()]))
214
def _extract_bzr_metadata(self, rev, message):
215
(message, metadata) = extract_bzr_metadata(message)
216
return message, metadata
218
def _decode_commit_message(self, rev, message, encoding):
219
return message.decode(encoding), CommitSupplement()
221
def _encode_commit_message(self, rev, message, encoding):
222
return message.encode(encoding)
224
def export_fileid_map(self, fileid_map):
225
"""Export a file id map to a fileid map.
227
:param fileid_map: File id map, mapping paths to file ids
228
:return: A Git blob object
230
from dulwich.objects import Blob
232
b.set_raw_chunks(serialize_fileid_map(fileid_map))
235
def export_commit(self, rev, tree_sha, parent_lookup, roundtrip,
237
"""Turn a Bazaar revision in to a Git commit
239
:param tree_sha: Tree sha for the commit
240
:param parent_lookup: Function for looking up the GIT sha equiv of a
242
:param roundtrip: Whether to store roundtripping information.
243
:param verifiers: Verifiers info
244
:return dulwich.objects.Commit represent the revision:
246
from dulwich.objects import Commit
248
commit.tree = tree_sha
250
metadata = CommitSupplement()
251
metadata.verifiers = verifiers
255
for p in rev.parent_ids:
257
git_p = parent_lookup(p)
260
if metadata is not None:
261
metadata.explicit_parent_ids = rev.parent_ids
262
if git_p is not None:
263
assert len(git_p) == 40, "unexpected length for %r" % git_p
264
parents.append(git_p)
265
commit.parents = parents
267
encoding = rev.properties['git-explicit-encoding']
269
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
270
commit.encoding = rev.properties.get('git-explicit-encoding')
271
commit.committer = fix_person_identifier(rev.committer.encode(
273
commit.author = fix_person_identifier(
274
rev.get_apparent_authors()[0].encode(encoding))
275
commit.commit_time = long(rev.timestamp)
276
if 'author-timestamp' in rev.properties:
277
commit.author_time = long(rev.properties['author-timestamp'])
279
commit.author_time = commit.commit_time
280
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
281
commit.commit_timezone = rev.timezone
282
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
283
if 'author-timezone' in rev.properties:
284
commit.author_timezone = int(rev.properties['author-timezone'])
286
commit.author_timezone = commit.commit_timezone
287
commit.message = self._encode_commit_message(rev, rev.message,
289
assert type(commit.message) == str
290
if metadata is not None:
292
mapping_registry.parse_revision_id(rev.revision_id)
293
except errors.InvalidRevisionId:
294
metadata.revision_id = rev.revision_id
295
mapping_properties = set(
296
['author', 'author-timezone', 'author-timezone-neg-utc',
297
'commit-timezone-neg-utc', 'git-implicit-encoding',
298
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
299
for k, v in rev.properties.iteritems():
300
if not k in mapping_properties:
301
metadata.properties[k] = v
302
if self.roundtripping:
303
commit.message = inject_bzr_metadata(commit.message, metadata,
305
assert type(commit.message) == str
308
def import_fileid_map(self, blob):
309
"""Convert a git file id map blob.
311
:param blob: Git blob object with fileid map
312
:return: Dictionary mapping paths to file ids
314
return deserialize_fileid_map(blob.data)
316
def import_commit(self, commit, lookup_parent_revid):
317
"""Convert a git commit to a bzr revision.
319
:return: a `bzrlib.revision.Revision` object, foreign revid and a
323
raise AssertionError("Commit object can't be None")
324
rev = ForeignRevision(commit.id, self,
325
self.revision_id_foreign_to_bzr(commit.id))
326
rev.git_metadata = None
327
def decode_using_encoding(rev, commit, encoding):
328
rev.committer = str(commit.committer).decode(encoding)
329
if commit.committer != commit.author:
330
rev.properties['author'] = str(commit.author).decode(encoding)
331
rev.message, rev.git_metadata = self._decode_commit_message(
332
rev, commit.message, encoding)
333
if commit.encoding is not None:
334
rev.properties['git-explicit-encoding'] = commit.encoding
335
decode_using_encoding(rev, commit, commit.encoding)
337
for encoding in ('utf-8', 'latin1'):
339
decode_using_encoding(rev, commit, encoding)
340
except UnicodeDecodeError:
343
if encoding != 'utf-8':
344
rev.properties['git-implicit-encoding'] = encoding
346
if commit.commit_time != commit.author_time:
347
rev.properties['author-timestamp'] = str(commit.author_time)
348
if commit.commit_timezone != commit.author_timezone:
349
rev.properties['author-timezone'] = "%d" % commit.author_timezone
350
if commit._author_timezone_neg_utc:
351
rev.properties['author-timezone-neg-utc'] = ""
352
if commit._commit_timezone_neg_utc:
353
rev.properties['commit-timezone-neg-utc'] = ""
354
rev.timestamp = commit.commit_time
355
rev.timezone = commit.commit_timezone
356
rev.parent_ids = None
357
if rev.git_metadata is not None:
358
md = rev.git_metadata
359
roundtrip_revid = md.revision_id
360
if md.explicit_parent_ids:
361
rev.parent_ids = md.explicit_parent_ids
362
rev.properties.update(md.properties)
363
verifiers = md.verifiers
365
roundtrip_revid = None
367
if rev.parent_ids is None:
368
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
369
return rev, roundtrip_revid, verifiers
371
def get_fileid_map(self, lookup_object, tree_sha):
372
"""Obtain a fileid map for a particular tree.
374
:param lookup_object: Function for looking up an object
375
:param tree_sha: SHA of the root tree
376
:return: GitFileIdMap instance
379
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
383
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
384
return GitFileIdMap(file_ids, self)
387
class BzrGitMappingv1(BzrGitMapping):
388
revid_prefix = 'git-v1'
392
return self.revid_prefix
395
class BzrGitMappingExperimental(BzrGitMappingv1):
396
revid_prefix = 'git-experimental'
400
BZR_FILE_IDS_FILE = '.bzrfileids'
402
BZR_DUMMY_FILE = '.bzrdummy'
404
def _decode_commit_message(self, rev, message, encoding):
405
message = self._extract_hg_metadata(rev, message)
406
message = self._extract_git_svn_metadata(rev, message)
407
message, metadata = self._extract_bzr_metadata(rev, message)
408
return message.decode(encoding), metadata
410
def _encode_commit_message(self, rev, message, encoding):
411
ret = message.encode(encoding)
412
ret += self._generate_hg_message_tail(rev)
413
ret += self._generate_git_svn_metadata(rev, encoding)
416
def import_commit(self, commit, lookup_parent_revid):
417
rev, roundtrip_revid, verifiers = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
418
rev.properties['converted_revision'] = "git %s\n" % commit.id
419
return rev, roundtrip_revid, verifiers
422
class GitMappingRegistry(VcsMappingRegistry):
423
"""Registry with available git mappings."""
425
def revision_id_bzr_to_foreign(self, bzr_revid):
426
if bzr_revid == NULL_REVISION:
427
from dulwich.protocol import ZERO_SHA
428
return ZERO_SHA, None
429
if not bzr_revid.startswith("git-"):
430
raise errors.InvalidRevisionId(bzr_revid, None)
431
(mapping_version, git_sha) = bzr_revid.split(":", 1)
432
mapping = self.get(mapping_version)
433
return mapping.revision_id_bzr_to_foreign(bzr_revid)
435
parse_revision_id = revision_id_bzr_to_foreign
438
mapping_registry = GitMappingRegistry()
439
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
441
mapping_registry.register_lazy('git-experimental',
442
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
443
#mapping_registry.set_default('git-experimental')
444
mapping_registry.set_default('git-v1')
447
class ForeignGit(ForeignVcs):
448
"""The Git Stupid Content Tracker"""
451
def branch_format(self):
452
from bzrlib.plugins.git.branch import GitBranchFormat
453
return GitBranchFormat()
456
def repository_format(self):
457
from bzrlib.plugins.git.repository import GitRepositoryFormat
458
return GitRepositoryFormat()
461
super(ForeignGit, self).__init__(mapping_registry)
462
self.abbreviation = "git"
465
def serialize_foreign_revid(self, foreign_revid):
469
def show_foreign_revid(cls, foreign_revid):
470
return { "git commit": foreign_revid }
473
foreign_vcs_git = ForeignGit()
474
default_mapping = mapping_registry.get_default()()
477
def symlink_to_blob(entry):
478
from dulwich.objects import Blob
480
symlink_target = entry.symlink_target
481
if type(symlink_target) == unicode:
482
symlink_target = symlink_target.encode('utf-8')
483
blob.data = symlink_target
487
def mode_is_executable(mode):
488
"""Check if mode should be considered executable."""
489
return bool(mode & 0111)
493
"""Determine the Bazaar inventory kind based on Unix file mode."""
494
entry_kind = (mode & 0700000) / 0100000
497
elif entry_kind == 1:
498
file_kind = (mode & 070000) / 010000
504
return 'tree-reference'
506
raise AssertionError(
507
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
509
raise AssertionError(
510
"Unknown kind, perms=%r." % (mode,))
513
def object_mode(kind, executable):
514
if kind == 'directory':
516
elif kind == 'symlink':
522
mode = stat.S_IFREG | 0644
526
elif kind == 'tree-reference':
527
from dulwich.objects import S_IFGITLINK
533
def entry_mode(entry):
534
"""Determine the git file mode for an inventory entry."""
535
return object_mode(entry.kind, entry.executable)
538
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes, empty_file_name):
539
"""Create a Git Tree object from a Bazaar directory.
541
:param entry: Inventory entry
542
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
543
:param unusual_modes: Dictionary with unusual file modes by file ids
544
:param empty_file_name: Name to use for dummy files in empty directories,
545
None to ignore empty directories.
547
from dulwich.objects import Blob, Tree
549
for name, value in entry.children.iteritems():
550
ie = entry.children[name]
552
mode = unusual_modes[ie.file_id]
554
mode = entry_mode(ie)
555
hexsha = lookup_ie_sha1(ie)
556
if hexsha is not None:
557
tree.add(name.encode("utf-8"), mode, hexsha)
558
if entry.parent_id is not None and len(tree) == 0:
559
# Only the root can be an empty tree
560
if empty_file_name is not None:
561
tree.add(empty_file_name, stat.S_IFREG | 0644, Blob().id)
567
def extract_unusual_modes(rev):
569
foreign_revid, mapping = mapping_registry.parse_revision_id(
571
except errors.InvalidRevisionId:
574
return mapping.export_unusual_file_modes(rev)
577
def parse_git_svn_id(text):
578
(head, uuid) = text.rsplit(" ", 1)
579
(full_url, rev) = head.rsplit("@", 1)
580
return (full_url, int(rev), uuid)
583
class GitFileIdMap(object):
585
def __init__(self, file_ids, mapping):
586
self.file_ids = file_ids
588
self.mapping = mapping
590
def set_file_id(self, path, file_id):
591
assert type(path) is str
592
assert type(file_id) is str
593
self.file_ids[path] = file_id
595
def lookup_file_id(self, path):
596
assert type(path) is str
598
file_id = self.file_ids[path]
600
file_id = self.mapping.generate_file_id(path)
601
assert type(file_id) is str
604
def lookup_path(self, file_id):
605
if self.paths is None:
607
for k, v in self.file_ids.iteritems():
610
path = self.paths[file_id]
612
return self.mapping.parse_file_id(file_id)
614
assert type(path) is str
618
return self.__class__(dict(self.file_ids), self.mapping)
added
removed
mapping.py
