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')
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':
72
raise AssertionError("unknown escape character %s" %
79
def fix_person_identifier(text):
80
if "<" in text and ">" in text:
82
return "%s <%s>" % (text, text)
85
def warn_escaped(commit, num_escaped):
86
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
87
"to regenerate the SHA map.", num_escaped, commit)
90
def warn_unusual_mode(commit, path, mode):
91
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
92
"property. ", mode, path, commit)
95
class BzrGitMapping(foreign.VcsMapping):
96
"""Class that maps between Git and Bazaar semantics."""
99
BZR_FILE_IDS_FILE = None
101
BZR_DUMMY_FILE = None
104
super(BzrGitMapping, self).__init__(foreign_vcs_git)
106
def __eq__(self, other):
107
return (type(self) == type(other) and
108
self.revid_prefix == other.revid_prefix)
111
def revision_id_foreign_to_bzr(cls, git_rev_id):
112
"""Convert a git revision id handle to a Bazaar revision id."""
113
from dulwich.protocol import ZERO_SHA
114
if git_rev_id == ZERO_SHA:
116
return "%s:%s" % (cls.revid_prefix, git_rev_id)
119
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
120
"""Convert a Bazaar revision id to a git revision id handle."""
121
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
122
raise errors.InvalidRevisionId(bzr_rev_id, cls)
123
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
125
def generate_file_id(self, path):
126
# Git paths are just bytestrings
127
# We must just hope they are valid UTF-8..
130
if type(path) is unicode:
131
path = path.encode("utf-8")
132
return escape_file_id(path)
134
def is_control_file(self, path):
135
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
137
def parse_file_id(self, file_id):
138
if file_id == ROOT_ID:
140
return unescape_file_id(file_id)
142
def revid_as_refname(self, revid):
144
return "refs/bzr/%s" % urllib.quote(revid)
146
def import_unusual_file_modes(self, rev, unusual_file_modes):
147
if unusual_file_modes:
148
ret = [(path, unusual_file_modes[path])
149
for path in sorted(unusual_file_modes.keys())]
150
rev.properties['file-modes'] = bencode.bencode(ret)
152
def export_unusual_file_modes(self, rev):
154
file_modes = rev.properties['file-modes']
158
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
160
def _generate_git_svn_metadata(self, rev, encoding):
162
git_svn_id = rev.properties["git-svn-id"]
166
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
168
def _generate_hg_message_tail(self, rev):
172
for name in rev.properties:
173
if name == 'hg:extra:branch':
174
branch = rev.properties['hg:extra:branch']
175
elif name.startswith('hg:extra'):
176
extra[name[len('hg:extra:'):]] = base64.b64decode(
177
rev.properties[name])
178
elif name == 'hg:renames':
179
renames = bencode.bdecode(base64.b64decode(
180
rev.properties['hg:renames']))
181
# TODO: Export other properties as 'bzr:' extras?
182
ret = format_hg_metadata(renames, branch, extra)
183
assert isinstance(ret, str)
186
def _extract_git_svn_metadata(self, rev, message):
187
lines = message.split("\n")
188
if not (lines[-1] == "" and len(lines) >= 2 and lines[-2].startswith("git-svn-id:")):
190
git_svn_id = lines[-2].split(": ", 1)[1]
191
rev.properties['git-svn-id'] = git_svn_id
192
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
193
# FIXME: Convert this to converted-from property somehow..
194
ret = "\n".join(lines[:-2])
195
assert isinstance(ret, str)
198
def _extract_hg_metadata(self, rev, message):
199
(message, renames, branch, extra) = extract_hg_metadata(message)
200
if branch is not None:
201
rev.properties['hg:extra:branch'] = branch
202
for name, value in extra.iteritems():
203
rev.properties['hg:extra:' + name] = base64.b64encode(value)
205
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
206
[(new, old) for (old, new) in renames.iteritems()]))
209
def _extract_bzr_metadata(self, rev, message):
210
(message, metadata) = extract_bzr_metadata(message)
211
return message, metadata
213
def _decode_commit_message(self, rev, message, encoding):
214
return message.decode(encoding), CommitSupplement()
216
def _encode_commit_message(self, rev, message, encoding):
217
return message.encode(encoding)
219
def export_fileid_map(self, fileid_map):
220
"""Export a file id map to a fileid map.
222
:param fileid_map: File id map, mapping paths to file ids
223
:return: A Git blob object
225
from dulwich.objects import Blob
227
b.set_raw_chunks(serialize_fileid_map(fileid_map))
230
def export_commit(self, rev, tree_sha, parent_lookup, roundtrip,
232
"""Turn a Bazaar revision in to a Git commit
234
:param tree_sha: Tree sha for the commit
235
:param parent_lookup: Function for looking up the GIT sha equiv of a
237
:param roundtrip: Whether to store roundtripping information.
238
:param verifiers: Verifiers info
239
:return dulwich.objects.Commit represent the revision:
241
from dulwich.objects import Commit
243
commit.tree = tree_sha
245
metadata = CommitSupplement()
246
metadata.verifiers = verifiers
250
for p in rev.parent_ids:
252
git_p = parent_lookup(p)
255
if metadata is not None:
256
metadata.explicit_parent_ids = rev.parent_ids
257
if git_p is not None:
258
assert len(git_p) == 40, "unexpected length for %r" % git_p
259
parents.append(git_p)
260
commit.parents = parents
262
encoding = rev.properties['git-explicit-encoding']
264
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
265
commit.encoding = rev.properties.get('git-explicit-encoding')
266
commit.committer = fix_person_identifier(rev.committer.encode(
268
commit.author = fix_person_identifier(
269
rev.get_apparent_authors()[0].encode(encoding))
270
commit.commit_time = long(rev.timestamp)
271
if 'author-timestamp' in rev.properties:
272
commit.author_time = long(rev.properties['author-timestamp'])
274
commit.author_time = commit.commit_time
275
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
276
commit.commit_timezone = rev.timezone
277
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
278
if 'author-timezone' in rev.properties:
279
commit.author_timezone = int(rev.properties['author-timezone'])
281
commit.author_timezone = commit.commit_timezone
282
commit.message = self._encode_commit_message(rev, rev.message,
284
assert type(commit.message) == str
285
if metadata is not None:
287
mapping_registry.parse_revision_id(rev.revision_id)
288
except errors.InvalidRevisionId:
289
metadata.revision_id = rev.revision_id
290
mapping_properties = set(
291
['author', 'author-timezone', 'author-timezone-neg-utc',
292
'commit-timezone-neg-utc', 'git-implicit-encoding',
293
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
294
for k, v in rev.properties.iteritems():
295
if not k in mapping_properties:
296
metadata.properties[k] = v
297
if self.roundtripping:
298
commit.message = inject_bzr_metadata(commit.message, metadata,
300
assert type(commit.message) == str
303
def import_fileid_map(self, blob):
304
"""Convert a git file id map blob.
306
:param blob: Git blob object with fileid map
307
:return: Dictionary mapping paths to file ids
309
return deserialize_fileid_map(blob.data)
311
def import_commit(self, commit, lookup_parent_revid):
312
"""Convert a git commit to a bzr revision.
314
:return: a `bzrlib.revision.Revision` object, foreign revid and a
318
raise AssertionError("Commit object can't be None")
319
rev = ForeignRevision(commit.id, self,
320
self.revision_id_foreign_to_bzr(commit.id))
321
rev.git_metadata = None
322
def decode_using_encoding(rev, commit, encoding):
323
rev.committer = str(commit.committer).decode(encoding)
324
if commit.committer != commit.author:
325
rev.properties['author'] = str(commit.author).decode(encoding)
326
rev.message, rev.git_metadata = self._decode_commit_message(
327
rev, commit.message, encoding)
328
if commit.encoding is not None:
329
rev.properties['git-explicit-encoding'] = commit.encoding
330
decode_using_encoding(rev, commit, commit.encoding)
332
for encoding in ('utf-8', 'latin1'):
334
decode_using_encoding(rev, commit, encoding)
335
except UnicodeDecodeError:
338
if encoding != 'utf-8':
339
rev.properties['git-implicit-encoding'] = encoding
341
if commit.commit_time != commit.author_time:
342
rev.properties['author-timestamp'] = str(commit.author_time)
343
if commit.commit_timezone != commit.author_timezone:
344
rev.properties['author-timezone'] = "%d" % commit.author_timezone
345
if commit._author_timezone_neg_utc:
346
rev.properties['author-timezone-neg-utc'] = ""
347
if commit._commit_timezone_neg_utc:
348
rev.properties['commit-timezone-neg-utc'] = ""
349
rev.timestamp = commit.commit_time
350
rev.timezone = commit.commit_timezone
351
rev.parent_ids = None
352
if rev.git_metadata is not None:
353
md = rev.git_metadata
354
roundtrip_revid = md.revision_id
355
if md.explicit_parent_ids:
356
rev.parent_ids = md.explicit_parent_ids
357
rev.properties.update(md.properties)
358
verifiers = md.verifiers
360
roundtrip_revid = None
362
if rev.parent_ids is None:
363
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
364
return rev, roundtrip_revid, verifiers
366
def get_fileid_map(self, lookup_object, tree_sha):
367
"""Obtain a fileid map for a particular tree.
369
:param lookup_object: Function for looking up an object
370
:param tree_sha: SHA of the root tree
371
:return: GitFileIdMap instance
374
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
378
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
379
return GitFileIdMap(file_ids, self)
382
class BzrGitMappingv1(BzrGitMapping):
383
revid_prefix = 'git-v1'
387
return self.revid_prefix
390
class BzrGitMappingExperimental(BzrGitMappingv1):
391
revid_prefix = 'git-experimental'
395
BZR_FILE_IDS_FILE = '.bzrfileids'
397
BZR_DUMMY_FILE = '.bzrdummy'
399
def _decode_commit_message(self, rev, message, encoding):
400
message = self._extract_hg_metadata(rev, message)
401
message = self._extract_git_svn_metadata(rev, message)
402
message, metadata = self._extract_bzr_metadata(rev, message)
403
return message.decode(encoding), metadata
405
def _encode_commit_message(self, rev, message, encoding):
406
ret = message.encode(encoding)
407
ret += self._generate_hg_message_tail(rev)
408
ret += self._generate_git_svn_metadata(rev, encoding)
411
def import_commit(self, commit, lookup_parent_revid):
412
rev, roundtrip_revid, verifiers = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
413
rev.properties['converted_revision'] = "git %s\n" % commit.id
414
return rev, roundtrip_revid, verifiers
417
class GitMappingRegistry(VcsMappingRegistry):
418
"""Registry with available git mappings."""
420
def revision_id_bzr_to_foreign(self, bzr_revid):
421
if bzr_revid == NULL_REVISION:
422
from dulwich.protocol import ZERO_SHA
423
return ZERO_SHA, None
424
if not bzr_revid.startswith("git-"):
425
raise errors.InvalidRevisionId(bzr_revid, None)
426
(mapping_version, git_sha) = bzr_revid.split(":", 1)
427
mapping = self.get(mapping_version)
428
return mapping.revision_id_bzr_to_foreign(bzr_revid)
430
parse_revision_id = revision_id_bzr_to_foreign
433
mapping_registry = GitMappingRegistry()
434
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
436
mapping_registry.register_lazy('git-experimental',
437
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
438
# mapping_registry.set_default('git-experimental')
439
mapping_registry.set_default('git-v1')
442
class ForeignGit(ForeignVcs):
443
"""The Git Stupid Content Tracker"""
446
def branch_format(self):
447
from bzrlib.plugins.git.branch import GitBranchFormat
448
return GitBranchFormat()
451
def repository_format(self):
452
from bzrlib.plugins.git.repository import GitRepositoryFormat
453
return GitRepositoryFormat()
456
super(ForeignGit, self).__init__(mapping_registry)
457
self.abbreviation = "git"
460
def serialize_foreign_revid(self, foreign_revid):
464
def show_foreign_revid(cls, foreign_revid):
465
return { "git commit": foreign_revid }
468
foreign_vcs_git = ForeignGit()
469
default_mapping = mapping_registry.get_default()()
472
def symlink_to_blob(entry):
473
from dulwich.objects import Blob
475
symlink_target = entry.symlink_target
476
if type(symlink_target) == unicode:
477
symlink_target = symlink_target.encode('utf-8')
478
blob.data = symlink_target
482
def mode_is_executable(mode):
483
"""Check if mode should be considered executable."""
484
return bool(mode & 0111)
488
"""Determine the Bazaar inventory kind based on Unix file mode."""
489
entry_kind = (mode & 0700000) / 0100000
492
elif entry_kind == 1:
493
file_kind = (mode & 070000) / 010000
499
return 'tree-reference'
501
raise AssertionError(
502
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
504
raise AssertionError(
505
"Unknown kind, perms=%r." % (mode,))
508
def object_mode(kind, executable):
509
if kind == 'directory':
511
elif kind == 'symlink':
517
mode = stat.S_IFREG | 0644
521
elif kind == 'tree-reference':
522
from dulwich.objects import S_IFGITLINK
528
def entry_mode(entry):
529
"""Determine the git file mode for an inventory entry."""
530
return object_mode(entry.kind, entry.executable)
533
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes, empty_file_name):
534
"""Create a Git Tree object from a Bazaar directory.
536
:param entry: Inventory entry
537
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
538
:param unusual_modes: Dictionary with unusual file modes by file ids
539
:param empty_file_name: Name to use for dummy files in empty directories,
540
None to ignore empty directories.
542
from dulwich.objects import Blob, Tree
544
for name, value in entry.children.iteritems():
545
ie = entry.children[name]
547
mode = unusual_modes[ie.file_id]
549
mode = entry_mode(ie)
550
hexsha = lookup_ie_sha1(ie)
551
if hexsha is not None:
552
tree.add(name.encode("utf-8"), mode, hexsha)
553
if entry.parent_id is not None and len(tree) == 0:
554
# Only the root can be an empty tree
555
if empty_file_name is not None:
556
tree.add(empty_file_name, stat.S_IFREG | 0644, Blob().id)
562
def extract_unusual_modes(rev):
564
foreign_revid, mapping = mapping_registry.parse_revision_id(
566
except errors.InvalidRevisionId:
569
return mapping.export_unusual_file_modes(rev)
572
def parse_git_svn_id(text):
573
(head, uuid) = text.rsplit(" ", 1)
574
(full_url, rev) = head.rsplit("@", 1)
575
return (full_url, int(rev), uuid)
578
class GitFileIdMap(object):
580
def __init__(self, file_ids, mapping):
581
self.file_ids = file_ids
583
self.mapping = mapping
585
def set_file_id(self, path, file_id):
586
assert type(path) is str
587
assert type(file_id) is str
588
self.file_ids[path] = file_id
590
def lookup_file_id(self, path):
591
assert type(path) is str
593
file_id = self.file_ids[path]
595
file_id = self.mapping.generate_file_id(path)
596
assert type(file_id) is str
599
def lookup_path(self, file_id):
600
if self.paths is None:
602
for k, v in self.file_ids.iteritems():
605
path = self.paths[file_id]
607
return self.mapping.parse_file_id(file_id)
609
assert type(path) is str
613
return self.__class__(dict(self.file_ids), self.mapping)
added
removed
mapping.py
