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
def squash_revision(target_repo, rev):
99
"""Remove characters that can't be stored from a revision, if necessary.
101
:param target_repo: Repository in which the revision will be stored
102
:param rev: Revision object, will be modified in-place
104
if not getattr(target_repo._serializer, "squashes_xml_invalid_characters", True):
106
from bzrlib.xml_serializer import escape_invalid_chars
107
rev.message, num_escaped = escape_invalid_chars(rev.message)
109
warn_escaped(rev.foreign_revid, num_escaped)
110
if 'author' in rev.properties:
111
rev.properties['author'], num_escaped = escape_invalid_chars(
112
rev.properties['author'])
114
warn_escaped(rev.foreign_revid, num_escaped)
115
rev.committer, num_escaped = escape_invalid_chars(rev.committer)
117
warn_escaped(rev.foreign_revid, num_escaped)
120
class BzrGitMapping(foreign.VcsMapping):
121
"""Class that maps between Git and Bazaar semantics."""
124
BZR_FILE_IDS_FILE = '.bzrfileids'
126
BZR_DUMMY_FILE = '.bzrdummy'
129
super(BzrGitMapping, self).__init__(foreign_git)
131
def __eq__(self, other):
132
return (type(self) == type(other) and
133
self.revid_prefix == other.revid_prefix)
136
def revision_id_foreign_to_bzr(cls, git_rev_id):
137
"""Convert a git revision id handle to a Bazaar revision id."""
138
from dulwich.protocol import ZERO_SHA
139
if git_rev_id == ZERO_SHA:
141
return "%s:%s" % (cls.revid_prefix, git_rev_id)
144
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
145
"""Convert a Bazaar revision id to a git revision id handle."""
146
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
147
raise errors.InvalidRevisionId(bzr_rev_id, cls)
148
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
150
def generate_file_id(self, path):
151
# Git paths are just bytestrings
152
# We must just hope they are valid UTF-8..
155
return escape_file_id(path)
157
def is_control_file(self, path):
158
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
160
def parse_file_id(self, file_id):
161
if file_id == ROOT_ID:
163
return unescape_file_id(file_id)
165
def revid_as_refname(self, revid):
167
return "refs/bzr/%s" % urllib.quote(revid)
169
def import_unusual_file_modes(self, rev, unusual_file_modes):
170
if unusual_file_modes:
171
ret = [(path, unusual_file_modes[path])
172
for path in sorted(unusual_file_modes.keys())]
173
rev.properties['file-modes'] = bencode.bencode(ret)
175
def export_unusual_file_modes(self, rev):
177
file_modes = rev.properties['file-modes']
181
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
183
def _generate_git_svn_metadata(self, rev, encoding):
185
git_svn_id = rev.properties["git-svn-id"]
189
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
191
def _generate_hg_message_tail(self, rev):
195
for name in rev.properties:
196
if name == 'hg:extra:branch':
197
branch = rev.properties['hg:extra:branch']
198
elif name.startswith('hg:extra'):
199
extra[name[len('hg:extra:'):]] = base64.b64decode(
200
rev.properties[name])
201
elif name == 'hg:renames':
202
renames = bencode.bdecode(base64.b64decode(
203
rev.properties['hg:renames']))
204
# TODO: Export other properties as 'bzr:' extras?
205
ret = format_hg_metadata(renames, branch, extra)
206
assert isinstance(ret, str)
209
def _extract_git_svn_metadata(self, rev, message):
210
lines = message.split("\n")
211
if not (lines[-1] == "" and lines[-2].startswith("git-svn-id:")):
213
git_svn_id = lines[-2].split(": ", 1)[1]
214
rev.properties['git-svn-id'] = git_svn_id
215
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
216
# FIXME: Convert this to converted-from property somehow..
217
ret = "\n".join(lines[:-2])
218
assert isinstance(ret, str)
221
def _extract_hg_metadata(self, rev, message):
222
(message, renames, branch, extra) = extract_hg_metadata(message)
223
if branch is not None:
224
rev.properties['hg:extra:branch'] = branch
225
for name, value in extra.iteritems():
226
rev.properties['hg:extra:' + name] = base64.b64encode(value)
228
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
229
[(new, old) for (old, new) in renames.iteritems()]))
232
def _extract_bzr_metadata(self, rev, message):
233
(message, metadata) = extract_bzr_metadata(message)
234
return message, metadata
236
def _decode_commit_message(self, rev, message, encoding):
237
message, metadata = self._extract_bzr_metadata(rev, message)
238
return message.decode(encoding), metadata
240
def _encode_commit_message(self, rev, message, encoding):
241
return message.encode(encoding)
243
def export_fileid_map(self, fileid_map):
244
"""Export a file id map to a fileid map.
246
:param fileid_map: File id map, mapping paths to file ids
247
:return: A Git blob object
249
from dulwich.objects import Blob
251
b.set_raw_chunks(serialize_fileid_map(fileid_map))
254
def export_commit(self, rev, tree_sha, parent_lookup, roundtrip):
255
"""Turn a Bazaar revision in to a Git commit
257
:param tree_sha: Tree sha for the commit
258
:param parent_lookup: Function for looking up the GIT sha equiv of a
260
:return dulwich.objects.Commit represent the revision:
262
from dulwich.objects import Commit
264
commit.tree = tree_sha
266
metadata = BzrGitRevisionMetadata()
270
for p in rev.parent_ids:
272
git_p = parent_lookup(p)
275
if metadata is not None:
276
metadata.explicit_parent_ids = rev.parent_ids
277
if git_p is not None:
278
assert len(git_p) == 40, "unexpected length for %r" % git_p
279
parents.append(git_p)
280
commit.parents = parents
282
encoding = rev.properties['git-explicit-encoding']
284
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
285
commit.encoding = rev.properties.get('git-explicit-encoding')
286
commit.committer = fix_person_identifier(rev.committer.encode(
288
commit.author = fix_person_identifier(
289
rev.get_apparent_authors()[0].encode(encoding))
290
commit.commit_time = long(rev.timestamp)
291
if 'author-timestamp' in rev.properties:
292
commit.author_time = long(rev.properties['author-timestamp'])
294
commit.author_time = commit.commit_time
295
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
296
commit.commit_timezone = rev.timezone
297
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
298
if 'author-timezone' in rev.properties:
299
commit.author_timezone = int(rev.properties['author-timezone'])
301
commit.author_timezone = commit.commit_timezone
302
commit.message = self._encode_commit_message(rev, rev.message,
304
assert type(commit.message) == str
305
if metadata is not None:
307
mapping_registry.parse_revision_id(rev.revision_id)
308
except errors.InvalidRevisionId:
309
metadata.revision_id = rev.revision_id
310
mapping_properties = set(
311
['author', 'author-timezone', 'author-timezone-neg-utc',
312
'commit-timezone-neg-utc', 'git-implicit-encoding',
313
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
314
for k, v in rev.properties.iteritems():
315
if not k in mapping_properties:
316
metadata.properties[k] = v
317
commit.message = inject_bzr_metadata(commit.message, metadata,
319
assert type(commit.message) == str
322
def import_fileid_map(self, blob):
323
"""Convert a git file id map blob.
325
:param blob: Git blob object with fileid map
326
:return: Dictionary mapping paths to file ids
328
return deserialize_fileid_map(blob.data)
330
def import_commit(self, commit):
331
"""Convert a git commit to a bzr revision.
333
:return: a `bzrlib.revision.Revision` object and a
334
dictionary of path -> file ids
337
raise AssertionError("Commit object can't be None")
338
rev = ForeignRevision(commit.id, self,
339
self.revision_id_foreign_to_bzr(commit.id))
340
rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
341
rev.git_metadata = None
342
def decode_using_encoding(rev, commit, encoding):
343
rev.committer = str(commit.committer).decode(encoding)
344
if commit.committer != commit.author:
345
rev.properties['author'] = str(commit.author).decode(encoding)
346
rev.message, rev.git_metadata = self._decode_commit_message(
347
rev, commit.message, encoding)
348
if commit.encoding is not None:
349
rev.properties['git-explicit-encoding'] = commit.encoding
350
decode_using_encoding(rev, commit, commit.encoding)
352
for encoding in ('utf-8', 'latin1'):
354
decode_using_encoding(rev, commit, encoding)
355
except UnicodeDecodeError:
358
if encoding != 'utf-8':
359
rev.properties['git-implicit-encoding'] = encoding
361
if commit.commit_time != commit.author_time:
362
rev.properties['author-timestamp'] = str(commit.author_time)
363
if commit.commit_timezone != commit.author_timezone:
364
rev.properties['author-timezone'] = "%d" % commit.author_timezone
365
if commit._author_timezone_neg_utc:
366
rev.properties['author-timezone-neg-utc'] = ""
367
if commit._commit_timezone_neg_utc:
368
rev.properties['commit-timezone-neg-utc'] = ""
369
rev.timestamp = commit.commit_time
370
rev.timezone = commit.commit_timezone
371
if rev.git_metadata is not None:
372
md = rev.git_metadata
374
rev.revision_id = md.revision_id
375
if md.explicit_parent_ids:
376
rev.parent_ids = md.explicit_parent_ids
377
rev.properties.update(md.properties)
380
def get_fileid_map(self, lookup_object, tree_sha):
381
"""Obtain a fileid map for a particular tree.
383
:param lookup_object: Function for looking up an object
384
:param tree_sha: SHA of the root tree
385
:return: GitFileIdMap instance
388
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
392
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
393
return GitFileIdMap(file_ids, self)
396
class BzrGitMappingv1(BzrGitMapping):
397
revid_prefix = 'git-v1'
401
return self.revid_prefix
404
class BzrGitMappingExperimental(BzrGitMappingv1):
405
revid_prefix = 'git-experimental'
408
def _decode_commit_message(self, rev, message, encoding):
409
message = self._extract_hg_metadata(rev, message)
410
message = self._extract_git_svn_metadata(rev, message)
411
message, metadata = self._extract_bzr_metadata(rev, message)
412
return message.decode(encoding), metadata
414
def _encode_commit_message(self, rev, message, encoding):
415
ret = message.encode(encoding)
416
ret += self._generate_hg_message_tail(rev)
417
ret += self._generate_git_svn_metadata(rev, encoding)
420
def import_commit(self, commit):
421
rev, file_ids = super(BzrGitMappingExperimental, self).import_commit(commit)
422
rev.properties['converted_revision'] = "git %s\n" % commit.id
426
class GitMappingRegistry(VcsMappingRegistry):
427
"""Registry with available git mappings."""
429
def revision_id_bzr_to_foreign(self, bzr_revid):
430
if bzr_revid == NULL_REVISION:
431
from dulwich.protocol import ZERO_SHA
432
return ZERO_SHA, None
433
if not bzr_revid.startswith("git-"):
434
raise errors.InvalidRevisionId(bzr_revid, None)
435
(mapping_version, git_sha) = bzr_revid.split(":", 1)
436
mapping = self.get(mapping_version)
437
return mapping.revision_id_bzr_to_foreign(bzr_revid)
439
parse_revision_id = revision_id_bzr_to_foreign
442
mapping_registry = GitMappingRegistry()
443
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
445
mapping_registry.register_lazy('git-experimental',
446
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
447
mapping_registry.set_default('git-v1')
450
class ForeignGit(ForeignVcs):
451
"""The Git Stupid Content Tracker"""
454
def branch_format(self):
455
from bzrlib.plugins.git.branch import GitBranchFormat
456
return GitBranchFormat()
459
def repository_format(self):
460
from bzrlib.plugins.git.repository import GitRepositoryFormat
461
return GitRepositoryFormat()
464
super(ForeignGit, self).__init__(mapping_registry)
465
self.abbreviation = "git"
468
def serialize_foreign_revid(self, foreign_revid):
472
def show_foreign_revid(cls, foreign_revid):
473
return { "git commit": foreign_revid }
476
foreign_git = ForeignGit()
477
default_mapping = mapping_registry.get_default()()
480
def symlink_to_blob(entry):
481
from dulwich.objects import Blob
483
symlink_target = entry.symlink_target
484
if type(symlink_target) == unicode:
485
symlink_target = symlink_target.encode('utf-8')
486
blob.data = symlink_target
490
def mode_is_executable(mode):
491
"""Check if mode should be considered executable."""
492
return bool(mode & 0111)
496
"""Determine the Bazaar inventory kind based on Unix file mode."""
497
entry_kind = (mode & 0700000) / 0100000
500
elif entry_kind == 1:
501
file_kind = (mode & 070000) / 010000
507
return 'tree-reference'
509
raise AssertionError(
510
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
512
raise AssertionError(
513
"Unknown kind, perms=%r." % (mode,))
516
def object_mode(kind, executable):
517
if kind == 'directory':
519
elif kind == 'symlink':
525
mode = stat.S_IFREG | 0644
529
elif kind == 'tree-reference':
530
from dulwich.objects import S_IFGITLINK
536
def entry_mode(entry):
537
"""Determine the git file mode for an inventory entry."""
538
return object_mode(entry.kind, entry.executable)
541
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes, empty_file_name):
542
"""Create a Git Tree object from a Bazaar directory.
544
:param entry: Inventory entry
545
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
546
:param unusual_modes: Dictionary with unusual file modes by file ids
547
:param empty_file_name: Name to use for dummy files in empty directories,
548
None to ignore empty directories.
550
from dulwich.objects import Blob, Tree
552
for name, value in entry.children.iteritems():
553
ie = entry.children[name]
555
mode = unusual_modes[ie.file_id]
557
mode = entry_mode(ie)
558
hexsha = lookup_ie_sha1(ie)
559
if hexsha is not None:
560
tree.add(mode, name.encode("utf-8"), hexsha)
561
if entry.parent_id is not None and len(tree) == 0:
562
# Only the root can be an empty tree
563
if empty_file_name is not None:
564
tree.add(stat.S_IFREG | 0644, empty_file_name,
571
def extract_unusual_modes(rev):
573
foreign_revid, mapping = mapping_registry.parse_revision_id(
575
except errors.InvalidRevisionId:
578
return mapping.export_unusual_file_modes(rev)
581
def parse_git_svn_id(text):
582
(head, uuid) = text.rsplit(" ", 1)
583
(full_url, rev) = head.rsplit("@", 1)
584
return (full_url, int(rev), uuid)
587
class GitFileIdMap(object):
589
def __init__(self, file_ids, mapping):
590
self.file_ids = file_ids
592
self.mapping = mapping
594
def lookup_file_id(self, path):
596
return self.file_ids[path]
598
return self.mapping.generate_file_id(path)
600
def lookup_path(self, file_id):
601
if self.paths is None:
603
for k, v in self.file_ids.iteritems():
606
return self.paths[file_id]
608
return self.mapping.parse_file_id(file_id)
added
removed
mapping.py
