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 import bencode
32
from bzrlib.util import bencode
33
from bzrlib.inventory import (
36
from bzrlib.foreign import (
41
from bzrlib.revision import (
44
from bzrlib.plugins.git.hg import (
48
from bzrlib.plugins.git.roundtrip import (
51
BzrGitRevisionMetadata,
52
deserialize_fileid_map,
56
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
59
def escape_file_id(file_id):
60
return file_id.replace('_', '__').replace(' ', '_s')
63
def unescape_file_id(file_id):
66
while i < len(file_id):
68
ret.append(file_id[i])
70
if file_id[i+1] == '_':
72
elif file_id[i+1] == 's':
75
raise AssertionError("unknown escape character %s" %
82
def fix_person_identifier(text):
83
if "<" in text and ">" in text:
85
return "%s <%s>" % (text, text)
88
def warn_escaped(commit, num_escaped):
89
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
90
"to regenerate the SHA map.", num_escaped, commit)
93
def warn_unusual_mode(commit, path, mode):
94
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
95
"property. ", mode, path, commit)
98
class BzrGitMapping(foreign.VcsMapping):
99
"""Class that maps between Git and Bazaar semantics."""
102
BZR_FILE_IDS_FILE = None
104
BZR_DUMMY_FILE = None
107
super(BzrGitMapping, self).__init__(foreign_git)
109
def __eq__(self, other):
110
return (type(self) == type(other) and
111
self.revid_prefix == other.revid_prefix)
114
def revision_id_foreign_to_bzr(cls, git_rev_id):
115
"""Convert a git revision id handle to a Bazaar revision id."""
116
from dulwich.protocol import ZERO_SHA
117
if git_rev_id == ZERO_SHA:
119
return "%s:%s" % (cls.revid_prefix, git_rev_id)
122
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
123
"""Convert a Bazaar revision id to a git revision id handle."""
124
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
125
raise errors.InvalidRevisionId(bzr_rev_id, cls)
126
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
128
def generate_file_id(self, path):
129
# Git paths are just bytestrings
130
# We must just hope they are valid UTF-8..
133
return escape_file_id(path)
135
def is_control_file(self, path):
136
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
138
def parse_file_id(self, file_id):
139
if file_id == ROOT_ID:
141
return unescape_file_id(file_id)
143
def revid_as_refname(self, revid):
145
return "refs/bzr/%s" % urllib.quote(revid)
147
def import_unusual_file_modes(self, rev, unusual_file_modes):
148
if unusual_file_modes:
149
ret = [(path, unusual_file_modes[path])
150
for path in sorted(unusual_file_modes.keys())]
151
rev.properties['file-modes'] = bencode.bencode(ret)
153
def export_unusual_file_modes(self, rev):
155
file_modes = rev.properties['file-modes']
159
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
161
def _generate_git_svn_metadata(self, rev, encoding):
163
git_svn_id = rev.properties["git-svn-id"]
167
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
169
def _generate_hg_message_tail(self, rev):
173
for name in rev.properties:
174
if name == 'hg:extra:branch':
175
branch = rev.properties['hg:extra:branch']
176
elif name.startswith('hg:extra'):
177
extra[name[len('hg:extra:'):]] = base64.b64decode(
178
rev.properties[name])
179
elif name == 'hg:renames':
180
renames = bencode.bdecode(base64.b64decode(
181
rev.properties['hg:renames']))
182
# TODO: Export other properties as 'bzr:' extras?
183
ret = format_hg_metadata(renames, branch, extra)
184
assert isinstance(ret, str)
187
def _extract_git_svn_metadata(self, rev, message):
188
lines = message.split("\n")
189
if not (lines[-1] == "" and lines[-2].startswith("git-svn-id:")):
191
git_svn_id = lines[-2].split(": ", 1)[1]
192
rev.properties['git-svn-id'] = git_svn_id
193
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
194
# FIXME: Convert this to converted-from property somehow..
195
ret = "\n".join(lines[:-2])
196
assert isinstance(ret, str)
199
def _extract_hg_metadata(self, rev, message):
200
(message, renames, branch, extra) = extract_hg_metadata(message)
201
if branch is not None:
202
rev.properties['hg:extra:branch'] = branch
203
for name, value in extra.iteritems():
204
rev.properties['hg:extra:' + name] = base64.b64encode(value)
206
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
207
[(new, old) for (old, new) in renames.iteritems()]))
210
def _extract_bzr_metadata(self, rev, message):
211
(message, metadata) = extract_bzr_metadata(message)
212
return message, metadata
214
def _decode_commit_message(self, rev, message, encoding):
215
return message.decode(encoding), BzrGitRevisionMetadata()
217
def _encode_commit_message(self, rev, message, encoding):
218
return message.encode(encoding)
220
def export_fileid_map(self, fileid_map):
221
"""Export a file id map to a fileid map.
223
:param fileid_map: File id map, mapping paths to file ids
224
:return: A Git blob object
226
from dulwich.objects import Blob
228
b.set_raw_chunks(serialize_fileid_map(fileid_map))
231
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
:return dulwich.objects.Commit represent the revision:
239
from dulwich.objects import Commit
241
commit.tree = tree_sha
243
metadata = BzrGitRevisionMetadata()
247
for p in rev.parent_ids:
249
git_p = parent_lookup(p)
252
if metadata is not None:
253
metadata.explicit_parent_ids = rev.parent_ids
254
if git_p is not None:
255
assert len(git_p) == 40, "unexpected length for %r" % git_p
256
parents.append(git_p)
257
commit.parents = parents
259
encoding = rev.properties['git-explicit-encoding']
261
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
262
commit.encoding = rev.properties.get('git-explicit-encoding')
263
commit.committer = fix_person_identifier(rev.committer.encode(
265
commit.author = fix_person_identifier(
266
rev.get_apparent_authors()[0].encode(encoding))
267
commit.commit_time = long(rev.timestamp)
268
if 'author-timestamp' in rev.properties:
269
commit.author_time = long(rev.properties['author-timestamp'])
271
commit.author_time = commit.commit_time
272
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
273
commit.commit_timezone = rev.timezone
274
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
275
if 'author-timezone' in rev.properties:
276
commit.author_timezone = int(rev.properties['author-timezone'])
278
commit.author_timezone = commit.commit_timezone
279
commit.message = self._encode_commit_message(rev, rev.message,
281
assert type(commit.message) == str
282
if metadata is not None:
284
mapping_registry.parse_revision_id(rev.revision_id)
285
except errors.InvalidRevisionId:
286
metadata.revision_id = rev.revision_id
287
mapping_properties = set(
288
['author', 'author-timezone', 'author-timezone-neg-utc',
289
'commit-timezone-neg-utc', 'git-implicit-encoding',
290
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
291
for k, v in rev.properties.iteritems():
292
if not k in mapping_properties:
293
metadata.properties[k] = v
294
if self.roundtripping:
295
commit.message = inject_bzr_metadata(commit.message, metadata,
297
assert type(commit.message) == str
300
def import_fileid_map(self, blob):
301
"""Convert a git file id map blob.
303
:param blob: Git blob object with fileid map
304
:return: Dictionary mapping paths to file ids
306
return deserialize_fileid_map(blob.data)
308
def import_commit(self, commit, lookup_parent_revid):
309
"""Convert a git commit to a bzr revision.
311
:return: a `bzrlib.revision.Revision` object and a
312
dictionary of path -> file ids
315
raise AssertionError("Commit object can't be None")
316
rev = ForeignRevision(commit.id, self,
317
self.revision_id_foreign_to_bzr(commit.id))
318
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
319
rev.git_metadata = None
320
def decode_using_encoding(rev, commit, encoding):
321
rev.committer = str(commit.committer).decode(encoding)
322
if commit.committer != commit.author:
323
rev.properties['author'] = str(commit.author).decode(encoding)
324
rev.message, rev.git_metadata = self._decode_commit_message(
325
rev, commit.message, encoding)
326
if commit.encoding is not None:
327
rev.properties['git-explicit-encoding'] = commit.encoding
328
decode_using_encoding(rev, commit, commit.encoding)
330
for encoding in ('utf-8', 'latin1'):
332
decode_using_encoding(rev, commit, encoding)
333
except UnicodeDecodeError:
336
if encoding != 'utf-8':
337
rev.properties['git-implicit-encoding'] = encoding
339
if commit.commit_time != commit.author_time:
340
rev.properties['author-timestamp'] = str(commit.author_time)
341
if commit.commit_timezone != commit.author_timezone:
342
rev.properties['author-timezone'] = "%d" % commit.author_timezone
343
if commit._author_timezone_neg_utc:
344
rev.properties['author-timezone-neg-utc'] = ""
345
if commit._commit_timezone_neg_utc:
346
rev.properties['commit-timezone-neg-utc'] = ""
347
rev.timestamp = commit.commit_time
348
rev.timezone = commit.commit_timezone
349
if rev.git_metadata is not None:
350
md = rev.git_metadata
352
rev.revision_id = md.revision_id
353
if md.explicit_parent_ids:
354
rev.parent_ids = md.explicit_parent_ids
355
rev.properties.update(md.properties)
358
def get_fileid_map(self, lookup_object, tree_sha):
359
"""Obtain a fileid map for a particular tree.
361
:param lookup_object: Function for looking up an object
362
:param tree_sha: SHA of the root tree
363
:return: GitFileIdMap instance
366
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
370
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
371
return GitFileIdMap(file_ids, self)
374
class BzrGitMappingv1(BzrGitMapping):
375
revid_prefix = 'git-v1'
379
return self.revid_prefix
382
class BzrGitMappingExperimental(BzrGitMappingv1):
383
revid_prefix = 'git-experimental'
387
BZR_FILE_IDS_FILE = '.bzrfileids'
389
BZR_DUMMY_FILE = '.bzrdummy'
391
def _decode_commit_message(self, rev, message, encoding):
392
message = self._extract_hg_metadata(rev, message)
393
message = self._extract_git_svn_metadata(rev, message)
394
message, metadata = self._extract_bzr_metadata(rev, message)
395
return message.decode(encoding), metadata
397
def _encode_commit_message(self, rev, message, encoding):
398
ret = message.encode(encoding)
399
ret += self._generate_hg_message_tail(rev)
400
ret += self._generate_git_svn_metadata(rev, encoding)
403
def import_commit(self, commit, lookup_parent_revid):
404
rev, file_ids = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
405
rev.properties['converted_revision'] = "git %s\n" % commit.id
409
class GitMappingRegistry(VcsMappingRegistry):
410
"""Registry with available git mappings."""
412
def revision_id_bzr_to_foreign(self, bzr_revid):
413
if bzr_revid == NULL_REVISION:
414
from dulwich.protocol import ZERO_SHA
415
return ZERO_SHA, None
416
if not bzr_revid.startswith("git-"):
417
raise errors.InvalidRevisionId(bzr_revid, None)
418
(mapping_version, git_sha) = bzr_revid.split(":", 1)
419
mapping = self.get(mapping_version)
420
return mapping.revision_id_bzr_to_foreign(bzr_revid)
422
parse_revision_id = revision_id_bzr_to_foreign
425
mapping_registry = GitMappingRegistry()
426
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
428
mapping_registry.register_lazy('git-experimental',
429
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
430
mapping_registry.set_default('git-v1')
433
class ForeignGit(ForeignVcs):
434
"""The Git Stupid Content Tracker"""
437
def branch_format(self):
438
from bzrlib.plugins.git.branch import GitBranchFormat
439
return GitBranchFormat()
442
def repository_format(self):
443
from bzrlib.plugins.git.repository import GitRepositoryFormat
444
return GitRepositoryFormat()
447
super(ForeignGit, self).__init__(mapping_registry)
448
self.abbreviation = "git"
451
def serialize_foreign_revid(self, foreign_revid):
455
def show_foreign_revid(cls, foreign_revid):
456
return { "git commit": foreign_revid }
459
foreign_git = ForeignGit()
460
default_mapping = mapping_registry.get_default()()
463
def symlink_to_blob(entry):
464
from dulwich.objects import Blob
466
symlink_target = entry.symlink_target
467
if type(symlink_target) == unicode:
468
symlink_target = symlink_target.encode('utf-8')
469
blob.data = symlink_target
473
def mode_is_executable(mode):
474
"""Check if mode should be considered executable."""
475
return bool(mode & 0111)
479
"""Determine the Bazaar inventory kind based on Unix file mode."""
480
entry_kind = (mode & 0700000) / 0100000
483
elif entry_kind == 1:
484
file_kind = (mode & 070000) / 010000
490
return 'tree-reference'
492
raise AssertionError(
493
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
495
raise AssertionError(
496
"Unknown kind, perms=%r." % (mode,))
499
def object_mode(kind, executable):
500
if kind == 'directory':
502
elif kind == 'symlink':
508
mode = stat.S_IFREG | 0644
512
elif kind == 'tree-reference':
513
from dulwich.objects import S_IFGITLINK
519
def entry_mode(entry):
520
"""Determine the git file mode for an inventory entry."""
521
return object_mode(entry.kind, entry.executable)
524
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes, empty_file_name):
525
"""Create a Git Tree object from a Bazaar directory.
527
:param entry: Inventory entry
528
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
529
:param unusual_modes: Dictionary with unusual file modes by file ids
530
:param empty_file_name: Name to use for dummy files in empty directories,
531
None to ignore empty directories.
533
from dulwich.objects import Blob, Tree
535
for name, value in entry.children.iteritems():
536
ie = entry.children[name]
538
mode = unusual_modes[ie.file_id]
540
mode = entry_mode(ie)
541
hexsha = lookup_ie_sha1(ie)
542
if hexsha is not None:
543
tree.add(mode, name.encode("utf-8"), hexsha)
544
if entry.parent_id is not None and len(tree) == 0:
545
# Only the root can be an empty tree
546
if empty_file_name is not None:
547
tree.add(stat.S_IFREG | 0644, empty_file_name,
554
def extract_unusual_modes(rev):
556
foreign_revid, mapping = mapping_registry.parse_revision_id(
558
except errors.InvalidRevisionId:
561
return mapping.export_unusual_file_modes(rev)
564
def parse_git_svn_id(text):
565
(head, uuid) = text.rsplit(" ", 1)
566
(full_url, rev) = head.rsplit("@", 1)
567
return (full_url, int(rev), uuid)
570
class GitFileIdMap(object):
572
def __init__(self, file_ids, mapping):
573
self.file_ids = file_ids
575
self.mapping = mapping
577
def lookup_file_id(self, path):
579
return self.file_ids[path]
581
return self.mapping.generate_file_id(path)
583
def lookup_path(self, file_id):
584
if self.paths is None:
586
for k, v in self.file_ids.iteritems():
589
return self.paths[file_id]
591
return self.mapping.parse_file_id(file_id)