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 (
48
BzrGitRevisionMetadata,
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_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
return escape_file_id(path)
132
def is_control_file(self, path):
133
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
135
def parse_file_id(self, file_id):
136
if file_id == ROOT_ID:
138
return unescape_file_id(file_id)
140
def revid_as_refname(self, revid):
142
return "refs/bzr/%s" % urllib.quote(revid)
144
def import_unusual_file_modes(self, rev, unusual_file_modes):
145
if unusual_file_modes:
146
ret = [(path, unusual_file_modes[path])
147
for path in sorted(unusual_file_modes.keys())]
148
rev.properties['file-modes'] = bencode.bencode(ret)
150
def export_unusual_file_modes(self, rev):
152
file_modes = rev.properties['file-modes']
156
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
158
def _generate_git_svn_metadata(self, rev, encoding):
160
git_svn_id = rev.properties["git-svn-id"]
164
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
166
def _generate_hg_message_tail(self, rev):
170
for name in rev.properties:
171
if name == 'hg:extra:branch':
172
branch = rev.properties['hg:extra:branch']
173
elif name.startswith('hg:extra'):
174
extra[name[len('hg:extra:'):]] = base64.b64decode(
175
rev.properties[name])
176
elif name == 'hg:renames':
177
renames = bencode.bdecode(base64.b64decode(
178
rev.properties['hg:renames']))
179
# TODO: Export other properties as 'bzr:' extras?
180
ret = format_hg_metadata(renames, branch, extra)
181
assert isinstance(ret, str)
184
def _extract_git_svn_metadata(self, rev, message):
185
lines = message.split("\n")
186
if not (lines[-1] == "" and lines[-2].startswith("git-svn-id:")):
188
git_svn_id = lines[-2].split(": ", 1)[1]
189
rev.properties['git-svn-id'] = git_svn_id
190
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
191
# FIXME: Convert this to converted-from property somehow..
192
ret = "\n".join(lines[:-2])
193
assert isinstance(ret, str)
196
def _extract_hg_metadata(self, rev, message):
197
(message, renames, branch, extra) = extract_hg_metadata(message)
198
if branch is not None:
199
rev.properties['hg:extra:branch'] = branch
200
for name, value in extra.iteritems():
201
rev.properties['hg:extra:' + name] = base64.b64encode(value)
203
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
204
[(new, old) for (old, new) in renames.iteritems()]))
207
def _extract_bzr_metadata(self, rev, message):
208
(message, metadata) = extract_bzr_metadata(message)
209
return message, metadata
211
def _decode_commit_message(self, rev, message, encoding):
212
return message.decode(encoding), BzrGitRevisionMetadata()
214
def _encode_commit_message(self, rev, message, encoding):
215
return message.encode(encoding)
217
def export_fileid_map(self, fileid_map):
218
"""Export a file id map to a fileid map.
220
:param fileid_map: File id map, mapping paths to file ids
221
:return: A Git blob object
223
from dulwich.objects import Blob
225
b.set_raw_chunks(serialize_fileid_map(fileid_map))
228
def export_commit(self, rev, tree_sha, parent_lookup, roundtrip):
229
"""Turn a Bazaar revision in to a Git commit
231
:param tree_sha: Tree sha for the commit
232
:param parent_lookup: Function for looking up the GIT sha equiv of a
234
:return dulwich.objects.Commit represent the revision:
236
from dulwich.objects import Commit
238
commit.tree = tree_sha
240
metadata = BzrGitRevisionMetadata()
244
for p in rev.parent_ids:
246
git_p = parent_lookup(p)
249
if metadata is not None:
250
metadata.explicit_parent_ids = rev.parent_ids
251
if git_p is not None:
252
assert len(git_p) == 40, "unexpected length for %r" % git_p
253
parents.append(git_p)
254
commit.parents = parents
256
encoding = rev.properties['git-explicit-encoding']
258
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
259
commit.encoding = rev.properties.get('git-explicit-encoding')
260
commit.committer = fix_person_identifier(rev.committer.encode(
262
commit.author = fix_person_identifier(
263
rev.get_apparent_authors()[0].encode(encoding))
264
commit.commit_time = long(rev.timestamp)
265
if 'author-timestamp' in rev.properties:
266
commit.author_time = long(rev.properties['author-timestamp'])
268
commit.author_time = commit.commit_time
269
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
270
commit.commit_timezone = rev.timezone
271
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
272
if 'author-timezone' in rev.properties:
273
commit.author_timezone = int(rev.properties['author-timezone'])
275
commit.author_timezone = commit.commit_timezone
276
commit.message = self._encode_commit_message(rev, rev.message,
278
assert type(commit.message) == str
279
if metadata is not None:
281
mapping_registry.parse_revision_id(rev.revision_id)
282
except errors.InvalidRevisionId:
283
metadata.revision_id = rev.revision_id
284
mapping_properties = set(
285
['author', 'author-timezone', 'author-timezone-neg-utc',
286
'commit-timezone-neg-utc', 'git-implicit-encoding',
287
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
288
for k, v in rev.properties.iteritems():
289
if not k in mapping_properties:
290
metadata.properties[k] = v
291
if self.roundtripping:
292
commit.message = inject_bzr_metadata(commit.message, metadata,
294
assert type(commit.message) == str
297
def import_fileid_map(self, blob):
298
"""Convert a git file id map blob.
300
:param blob: Git blob object with fileid map
301
:return: Dictionary mapping paths to file ids
303
return deserialize_fileid_map(blob.data)
305
def import_commit(self, commit, lookup_parent_revid):
306
"""Convert a git commit to a bzr revision.
308
:return: a `bzrlib.revision.Revision` object and a
309
dictionary of path -> file ids
312
raise AssertionError("Commit object can't be None")
313
rev = ForeignRevision(commit.id, self,
314
self.revision_id_foreign_to_bzr(commit.id))
315
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
316
rev.git_metadata = None
317
def decode_using_encoding(rev, commit, encoding):
318
rev.committer = str(commit.committer).decode(encoding)
319
if commit.committer != commit.author:
320
rev.properties['author'] = str(commit.author).decode(encoding)
321
rev.message, rev.git_metadata = self._decode_commit_message(
322
rev, commit.message, encoding)
323
if commit.encoding is not None:
324
rev.properties['git-explicit-encoding'] = commit.encoding
325
decode_using_encoding(rev, commit, commit.encoding)
327
for encoding in ('utf-8', 'latin1'):
329
decode_using_encoding(rev, commit, encoding)
330
except UnicodeDecodeError:
333
if encoding != 'utf-8':
334
rev.properties['git-implicit-encoding'] = encoding
336
if commit.commit_time != commit.author_time:
337
rev.properties['author-timestamp'] = str(commit.author_time)
338
if commit.commit_timezone != commit.author_timezone:
339
rev.properties['author-timezone'] = "%d" % commit.author_timezone
340
if commit._author_timezone_neg_utc:
341
rev.properties['author-timezone-neg-utc'] = ""
342
if commit._commit_timezone_neg_utc:
343
rev.properties['commit-timezone-neg-utc'] = ""
344
rev.timestamp = commit.commit_time
345
rev.timezone = commit.commit_timezone
346
if rev.git_metadata is not None:
347
md = rev.git_metadata
349
rev.revision_id = md.revision_id
350
if md.explicit_parent_ids:
351
rev.parent_ids = md.explicit_parent_ids
352
rev.properties.update(md.properties)
355
def get_fileid_map(self, lookup_object, tree_sha):
356
"""Obtain a fileid map for a particular tree.
358
:param lookup_object: Function for looking up an object
359
:param tree_sha: SHA of the root tree
360
:return: GitFileIdMap instance
363
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
367
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
368
return GitFileIdMap(file_ids, self)
371
class BzrGitMappingv1(BzrGitMapping):
372
revid_prefix = 'git-v1'
376
return self.revid_prefix
379
class BzrGitMappingExperimental(BzrGitMappingv1):
380
revid_prefix = 'git-experimental'
384
BZR_FILE_IDS_FILE = '.bzrfileids'
386
BZR_DUMMY_FILE = '.bzrdummy'
388
def _decode_commit_message(self, rev, message, encoding):
389
message = self._extract_hg_metadata(rev, message)
390
message = self._extract_git_svn_metadata(rev, message)
391
message, metadata = self._extract_bzr_metadata(rev, message)
392
return message.decode(encoding), metadata
394
def _encode_commit_message(self, rev, message, encoding):
395
ret = message.encode(encoding)
396
ret += self._generate_hg_message_tail(rev)
397
ret += self._generate_git_svn_metadata(rev, encoding)
400
def import_commit(self, commit, lookup_parent_revid):
401
rev, file_ids = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
402
rev.properties['converted_revision'] = "git %s\n" % commit.id
406
class GitMappingRegistry(VcsMappingRegistry):
407
"""Registry with available git mappings."""
409
def revision_id_bzr_to_foreign(self, bzr_revid):
410
if bzr_revid == NULL_REVISION:
411
from dulwich.protocol import ZERO_SHA
412
return ZERO_SHA, None
413
if not bzr_revid.startswith("git-"):
414
raise errors.InvalidRevisionId(bzr_revid, None)
415
(mapping_version, git_sha) = bzr_revid.split(":", 1)
416
mapping = self.get(mapping_version)
417
return mapping.revision_id_bzr_to_foreign(bzr_revid)
419
parse_revision_id = revision_id_bzr_to_foreign
422
mapping_registry = GitMappingRegistry()
423
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
425
mapping_registry.register_lazy('git-experimental',
426
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
427
mapping_registry.set_default('git-v1')
430
class ForeignGit(ForeignVcs):
431
"""The Git Stupid Content Tracker"""
434
def branch_format(self):
435
from bzrlib.plugins.git.branch import GitBranchFormat
436
return GitBranchFormat()
439
def repository_format(self):
440
from bzrlib.plugins.git.repository import GitRepositoryFormat
441
return GitRepositoryFormat()
444
super(ForeignGit, self).__init__(mapping_registry)
445
self.abbreviation = "git"
448
def serialize_foreign_revid(self, foreign_revid):
452
def show_foreign_revid(cls, foreign_revid):
453
return { "git commit": foreign_revid }
456
foreign_git = ForeignGit()
457
default_mapping = mapping_registry.get_default()()
460
def symlink_to_blob(entry):
461
from dulwich.objects import Blob
463
symlink_target = entry.symlink_target
464
if type(symlink_target) == unicode:
465
symlink_target = symlink_target.encode('utf-8')
466
blob.data = symlink_target
470
def mode_is_executable(mode):
471
"""Check if mode should be considered executable."""
472
return bool(mode & 0111)
476
"""Determine the Bazaar inventory kind based on Unix file mode."""
477
entry_kind = (mode & 0700000) / 0100000
480
elif entry_kind == 1:
481
file_kind = (mode & 070000) / 010000
487
return 'tree-reference'
489
raise AssertionError(
490
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
492
raise AssertionError(
493
"Unknown kind, perms=%r." % (mode,))
496
def object_mode(kind, executable):
497
if kind == 'directory':
499
elif kind == 'symlink':
505
mode = stat.S_IFREG | 0644
509
elif kind == 'tree-reference':
510
from dulwich.objects import S_IFGITLINK
516
def entry_mode(entry):
517
"""Determine the git file mode for an inventory entry."""
518
return object_mode(entry.kind, entry.executable)
521
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes, empty_file_name):
522
"""Create a Git Tree object from a Bazaar directory.
524
:param entry: Inventory entry
525
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
526
:param unusual_modes: Dictionary with unusual file modes by file ids
527
:param empty_file_name: Name to use for dummy files in empty directories,
528
None to ignore empty directories.
530
from dulwich.objects import Blob, Tree
532
for name, value in entry.children.iteritems():
533
ie = entry.children[name]
535
mode = unusual_modes[ie.file_id]
537
mode = entry_mode(ie)
538
hexsha = lookup_ie_sha1(ie)
539
if hexsha is not None:
540
tree.add(mode, name.encode("utf-8"), hexsha)
541
if entry.parent_id is not None and len(tree) == 0:
542
# Only the root can be an empty tree
543
if empty_file_name is not None:
544
tree.add(stat.S_IFREG | 0644, empty_file_name,
551
def extract_unusual_modes(rev):
553
foreign_revid, mapping = mapping_registry.parse_revision_id(
555
except errors.InvalidRevisionId:
558
return mapping.export_unusual_file_modes(rev)
561
def parse_git_svn_id(text):
562
(head, uuid) = text.rsplit(" ", 1)
563
(full_url, rev) = head.rsplit("@", 1)
564
return (full_url, int(rev), uuid)
567
class GitFileIdMap(object):
569
def __init__(self, file_ids, mapping):
570
self.file_ids = file_ids
572
self.mapping = mapping
574
def lookup_file_id(self, path):
576
return self.file_ids[path]
578
return self.mapping.generate_file_id(path)
580
def lookup_path(self, file_id):
581
if self.paths is None:
583
for k, v in self.file_ids.iteritems():
586
return self.paths[file_id]
588
return self.mapping.parse_file_id(file_id)
added
removed
mapping.py
