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
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 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), BzrGitRevisionMetadata()
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):
231
"""Turn a Bazaar revision in to a Git commit
233
:param tree_sha: Tree sha for the commit
234
:param parent_lookup: Function for looking up the GIT sha equiv of a
236
:return dulwich.objects.Commit represent the revision:
238
from dulwich.objects import Commit
240
commit.tree = tree_sha
242
metadata = BzrGitRevisionMetadata()
246
for p in rev.parent_ids:
248
git_p = parent_lookup(p)
251
if metadata is not None:
252
metadata.explicit_parent_ids = rev.parent_ids
253
if git_p is not None:
254
assert len(git_p) == 40, "unexpected length for %r" % git_p
255
parents.append(git_p)
256
commit.parents = parents
258
encoding = rev.properties['git-explicit-encoding']
260
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
261
commit.encoding = rev.properties.get('git-explicit-encoding')
262
commit.committer = fix_person_identifier(rev.committer.encode(
264
commit.author = fix_person_identifier(
265
rev.get_apparent_authors()[0].encode(encoding))
266
commit.commit_time = long(rev.timestamp)
267
if 'author-timestamp' in rev.properties:
268
commit.author_time = long(rev.properties['author-timestamp'])
270
commit.author_time = commit.commit_time
271
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
272
commit.commit_timezone = rev.timezone
273
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
274
if 'author-timezone' in rev.properties:
275
commit.author_timezone = int(rev.properties['author-timezone'])
277
commit.author_timezone = commit.commit_timezone
278
commit.message = self._encode_commit_message(rev, rev.message,
280
assert type(commit.message) == str
281
if metadata is not None:
283
mapping_registry.parse_revision_id(rev.revision_id)
284
except errors.InvalidRevisionId:
285
metadata.revision_id = rev.revision_id
286
mapping_properties = set(
287
['author', 'author-timezone', 'author-timezone-neg-utc',
288
'commit-timezone-neg-utc', 'git-implicit-encoding',
289
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
290
for k, v in rev.properties.iteritems():
291
if not k in mapping_properties:
292
metadata.properties[k] = v
293
if self.roundtripping:
294
commit.message = inject_bzr_metadata(commit.message, metadata,
296
assert type(commit.message) == str
299
def import_fileid_map(self, blob):
300
"""Convert a git file id map blob.
302
:param blob: Git blob object with fileid map
303
:return: Dictionary mapping paths to file ids
305
return deserialize_fileid_map(blob.data)
307
def import_commit(self, commit, lookup_parent_revid):
308
"""Convert a git commit to a bzr revision.
310
:return: a `bzrlib.revision.Revision` object and a
311
dictionary of path -> file ids
314
raise AssertionError("Commit object can't be None")
315
rev = ForeignRevision(commit.id, self,
316
self.revision_id_foreign_to_bzr(commit.id))
317
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
318
rev.git_metadata = None
319
def decode_using_encoding(rev, commit, encoding):
320
rev.committer = str(commit.committer).decode(encoding)
321
if commit.committer != commit.author:
322
rev.properties['author'] = str(commit.author).decode(encoding)
323
rev.message, rev.git_metadata = self._decode_commit_message(
324
rev, commit.message, encoding)
325
if commit.encoding is not None:
326
rev.properties['git-explicit-encoding'] = commit.encoding
327
decode_using_encoding(rev, commit, commit.encoding)
329
for encoding in ('utf-8', 'latin1'):
331
decode_using_encoding(rev, commit, encoding)
332
except UnicodeDecodeError:
335
if encoding != 'utf-8':
336
rev.properties['git-implicit-encoding'] = encoding
338
if commit.commit_time != commit.author_time:
339
rev.properties['author-timestamp'] = str(commit.author_time)
340
if commit.commit_timezone != commit.author_timezone:
341
rev.properties['author-timezone'] = "%d" % commit.author_timezone
342
if commit._author_timezone_neg_utc:
343
rev.properties['author-timezone-neg-utc'] = ""
344
if commit._commit_timezone_neg_utc:
345
rev.properties['commit-timezone-neg-utc'] = ""
346
rev.timestamp = commit.commit_time
347
rev.timezone = commit.commit_timezone
348
if rev.git_metadata is not None:
349
md = rev.git_metadata
351
rev.revision_id = md.revision_id
352
if md.explicit_parent_ids:
353
rev.parent_ids = md.explicit_parent_ids
354
rev.properties.update(md.properties)
357
def get_fileid_map(self, lookup_object, tree_sha):
358
"""Obtain a fileid map for a particular tree.
360
:param lookup_object: Function for looking up an object
361
:param tree_sha: SHA of the root tree
362
:return: GitFileIdMap instance
365
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
369
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
370
return GitFileIdMap(file_ids, self)
373
class BzrGitMappingv1(BzrGitMapping):
374
revid_prefix = 'git-v1'
378
return self.revid_prefix
381
class BzrGitMappingExperimental(BzrGitMappingv1):
382
revid_prefix = 'git-experimental'
386
BZR_FILE_IDS_FILE = '.bzrfileids'
388
BZR_DUMMY_FILE = '.bzrdummy'
390
def _decode_commit_message(self, rev, message, encoding):
391
message = self._extract_hg_metadata(rev, message)
392
message = self._extract_git_svn_metadata(rev, message)
393
message, metadata = self._extract_bzr_metadata(rev, message)
394
return message.decode(encoding), metadata
396
def _encode_commit_message(self, rev, message, encoding):
397
ret = message.encode(encoding)
398
ret += self._generate_hg_message_tail(rev)
399
ret += self._generate_git_svn_metadata(rev, encoding)
402
def import_commit(self, commit, lookup_parent_revid):
403
rev, file_ids = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
404
rev.properties['converted_revision'] = "git %s\n" % commit.id
408
class GitMappingRegistry(VcsMappingRegistry):
409
"""Registry with available git mappings."""
411
def revision_id_bzr_to_foreign(self, bzr_revid):
412
if bzr_revid == NULL_REVISION:
413
from dulwich.protocol import ZERO_SHA
414
return ZERO_SHA, None
415
if not bzr_revid.startswith("git-"):
416
raise errors.InvalidRevisionId(bzr_revid, None)
417
(mapping_version, git_sha) = bzr_revid.split(":", 1)
418
mapping = self.get(mapping_version)
419
return mapping.revision_id_bzr_to_foreign(bzr_revid)
421
parse_revision_id = revision_id_bzr_to_foreign
424
mapping_registry = GitMappingRegistry()
425
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
427
mapping_registry.register_lazy('git-experimental',
428
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
429
mapping_registry.set_default('git-v1')
432
class ForeignGit(ForeignVcs):
433
"""The Git Stupid Content Tracker"""
436
def branch_format(self):
437
from bzrlib.plugins.git.branch import GitBranchFormat
438
return GitBranchFormat()
441
def repository_format(self):
442
from bzrlib.plugins.git.repository import GitRepositoryFormat
443
return GitRepositoryFormat()
446
super(ForeignGit, self).__init__(mapping_registry)
447
self.abbreviation = "git"
450
def serialize_foreign_revid(self, foreign_revid):
454
def show_foreign_revid(cls, foreign_revid):
455
return { "git commit": foreign_revid }
458
foreign_git = ForeignGit()
459
default_mapping = mapping_registry.get_default()()
462
def symlink_to_blob(entry):
463
from dulwich.objects import Blob
465
symlink_target = entry.symlink_target
466
if type(symlink_target) == unicode:
467
symlink_target = symlink_target.encode('utf-8')
468
blob.data = symlink_target
472
def mode_is_executable(mode):
473
"""Check if mode should be considered executable."""
474
return bool(mode & 0111)
478
"""Determine the Bazaar inventory kind based on Unix file mode."""
479
entry_kind = (mode & 0700000) / 0100000
482
elif entry_kind == 1:
483
file_kind = (mode & 070000) / 010000
489
return 'tree-reference'
491
raise AssertionError(
492
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
494
raise AssertionError(
495
"Unknown kind, perms=%r." % (mode,))
498
def object_mode(kind, executable):
499
if kind == 'directory':
501
elif kind == 'symlink':
507
mode = stat.S_IFREG | 0644
511
elif kind == 'tree-reference':
512
from dulwich.objects import S_IFGITLINK
518
def entry_mode(entry):
519
"""Determine the git file mode for an inventory entry."""
520
return object_mode(entry.kind, entry.executable)
523
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes, empty_file_name):
524
"""Create a Git Tree object from a Bazaar directory.
526
:param entry: Inventory entry
527
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
528
:param unusual_modes: Dictionary with unusual file modes by file ids
529
:param empty_file_name: Name to use for dummy files in empty directories,
530
None to ignore empty directories.
532
from dulwich.objects import Blob, Tree
534
for name, value in entry.children.iteritems():
535
ie = entry.children[name]
537
mode = unusual_modes[ie.file_id]
539
mode = entry_mode(ie)
540
hexsha = lookup_ie_sha1(ie)
541
if hexsha is not None:
542
tree.add(mode, name.encode("utf-8"), hexsha)
543
if entry.parent_id is not None and len(tree) == 0:
544
# Only the root can be an empty tree
545
if empty_file_name is not None:
546
tree.add(stat.S_IFREG | 0644, empty_file_name,
553
def extract_unusual_modes(rev):
555
foreign_revid, mapping = mapping_registry.parse_revision_id(
557
except errors.InvalidRevisionId:
560
return mapping.export_unusual_file_modes(rev)
563
def parse_git_svn_id(text):
564
(head, uuid) = text.rsplit(" ", 1)
565
(full_url, rev) = head.rsplit("@", 1)
566
return (full_url, int(rev), uuid)
569
class GitFileIdMap(object):
571
def __init__(self, file_ids, mapping):
572
self.file_ids = file_ids
574
self.mapping = mapping
576
def lookup_file_id(self, path):
577
assert type(path) is str
579
file_id = self.file_ids[path]
581
file_id = self.mapping.generate_file_id(path)
582
assert type(file_id) is str
585
def lookup_path(self, file_id):
586
if self.paths is None:
588
for k, v in self.file_ids.iteritems():
591
path = self.paths[file_id]
593
return self.mapping.parse_file_id(file_id)
595
assert type(path) is str
added
removed
mapping.py
