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
103
def is_special_file(self, filename):
104
return (filename in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE))
107
super(BzrGitMapping, self).__init__(foreign_vcs_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
if type(path) is unicode:
134
path = path.encode("utf-8")
135
return escape_file_id(path)
137
def is_control_file(self, path):
138
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
140
def parse_file_id(self, file_id):
141
if file_id == ROOT_ID:
143
return unescape_file_id(file_id)
145
def revid_as_refname(self, revid):
147
return "refs/bzr/%s" % urllib.quote(revid)
149
def import_unusual_file_modes(self, rev, unusual_file_modes):
150
if unusual_file_modes:
151
ret = [(path, unusual_file_modes[path])
152
for path in sorted(unusual_file_modes.keys())]
153
rev.properties['file-modes'] = bencode.bencode(ret)
155
def export_unusual_file_modes(self, rev):
157
file_modes = rev.properties['file-modes']
161
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
163
def _generate_git_svn_metadata(self, rev, encoding):
165
git_svn_id = rev.properties["git-svn-id"]
169
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
171
def _generate_hg_message_tail(self, rev):
175
for name in rev.properties:
176
if name == 'hg:extra:branch':
177
branch = rev.properties['hg:extra:branch']
178
elif name.startswith('hg:extra'):
179
extra[name[len('hg:extra:'):]] = base64.b64decode(
180
rev.properties[name])
181
elif name == 'hg:renames':
182
renames = bencode.bdecode(base64.b64decode(
183
rev.properties['hg:renames']))
184
# TODO: Export other properties as 'bzr:' extras?
185
ret = format_hg_metadata(renames, branch, extra)
186
assert isinstance(ret, str)
189
def _extract_git_svn_metadata(self, rev, message):
190
lines = message.split("\n")
191
if not (lines[-1] == "" and len(lines) >= 2 and lines[-2].startswith("git-svn-id:")):
193
git_svn_id = lines[-2].split(": ", 1)[1]
194
rev.properties['git-svn-id'] = git_svn_id
195
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
196
# FIXME: Convert this to converted-from property somehow..
197
ret = "\n".join(lines[:-2])
198
assert isinstance(ret, str)
201
def _extract_hg_metadata(self, rev, message):
202
(message, renames, branch, extra) = extract_hg_metadata(message)
203
if branch is not None:
204
rev.properties['hg:extra:branch'] = branch
205
for name, value in extra.iteritems():
206
rev.properties['hg:extra:' + name] = base64.b64encode(value)
208
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
209
[(new, old) for (old, new) in renames.iteritems()]))
212
def _extract_bzr_metadata(self, rev, message):
213
(message, metadata) = extract_bzr_metadata(message)
214
return message, metadata
216
def _decode_commit_message(self, rev, message, encoding):
217
return message.decode(encoding), CommitSupplement()
219
def _encode_commit_message(self, rev, message, encoding):
220
return message.encode(encoding)
222
def export_fileid_map(self, fileid_map):
223
"""Export a file id map to a fileid map.
225
:param fileid_map: File id map, mapping paths to file ids
226
:return: A Git blob object
228
from dulwich.objects import Blob
230
b.set_raw_chunks(serialize_fileid_map(fileid_map))
233
def export_commit(self, rev, tree_sha, parent_lookup, roundtrip,
235
"""Turn a Bazaar revision in to a Git commit
237
:param tree_sha: Tree sha for the commit
238
:param parent_lookup: Function for looking up the GIT sha equiv of a
240
:param roundtrip: Whether to store roundtripping information.
241
:param verifiers: Verifiers info
242
:return dulwich.objects.Commit represent the revision:
244
from dulwich.objects import Commit
246
commit.tree = tree_sha
248
metadata = CommitSupplement()
249
metadata.verifiers = verifiers
253
for p in rev.parent_ids:
255
git_p = parent_lookup(p)
258
if metadata is not None:
259
metadata.explicit_parent_ids = rev.parent_ids
260
if git_p is not None:
261
assert len(git_p) == 40, "unexpected length for %r" % git_p
262
parents.append(git_p)
263
commit.parents = parents
265
encoding = rev.properties['git-explicit-encoding']
267
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
268
commit.encoding = rev.properties.get('git-explicit-encoding')
269
commit.committer = fix_person_identifier(rev.committer.encode(
271
commit.author = fix_person_identifier(
272
rev.get_apparent_authors()[0].encode(encoding))
273
commit.commit_time = long(rev.timestamp)
274
if 'author-timestamp' in rev.properties:
275
commit.author_time = long(rev.properties['author-timestamp'])
277
commit.author_time = commit.commit_time
278
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
279
commit.commit_timezone = rev.timezone
280
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
281
if 'author-timezone' in rev.properties:
282
commit.author_timezone = int(rev.properties['author-timezone'])
284
commit.author_timezone = commit.commit_timezone
285
commit.message = self._encode_commit_message(rev, rev.message,
287
assert type(commit.message) == str
288
if metadata is not None:
290
mapping_registry.parse_revision_id(rev.revision_id)
291
except errors.InvalidRevisionId:
292
metadata.revision_id = rev.revision_id
293
mapping_properties = set(
294
['author', 'author-timezone', 'author-timezone-neg-utc',
295
'commit-timezone-neg-utc', 'git-implicit-encoding',
296
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
297
for k, v in rev.properties.iteritems():
298
if not k in mapping_properties:
299
metadata.properties[k] = v
300
if self.roundtripping:
301
commit.message = inject_bzr_metadata(commit.message, metadata,
303
assert type(commit.message) == str
306
def import_fileid_map(self, blob):
307
"""Convert a git file id map blob.
309
:param blob: Git blob object with fileid map
310
:return: Dictionary mapping paths to file ids
312
return deserialize_fileid_map(blob.data)
314
def import_commit(self, commit, lookup_parent_revid):
315
"""Convert a git commit to a bzr revision.
317
:return: a `bzrlib.revision.Revision` object, foreign revid and a
321
raise AssertionError("Commit object can't be None")
322
rev = ForeignRevision(commit.id, self,
323
self.revision_id_foreign_to_bzr(commit.id))
324
rev.git_metadata = None
325
def decode_using_encoding(rev, commit, encoding):
326
rev.committer = str(commit.committer).decode(encoding)
327
if commit.committer != commit.author:
328
rev.properties['author'] = str(commit.author).decode(encoding)
329
rev.message, rev.git_metadata = self._decode_commit_message(
330
rev, commit.message, encoding)
331
if commit.encoding is not None:
332
rev.properties['git-explicit-encoding'] = commit.encoding
333
decode_using_encoding(rev, commit, commit.encoding)
335
for encoding in ('utf-8', 'latin1'):
337
decode_using_encoding(rev, commit, encoding)
338
except UnicodeDecodeError:
341
if encoding != 'utf-8':
342
rev.properties['git-implicit-encoding'] = encoding
344
if commit.commit_time != commit.author_time:
345
rev.properties['author-timestamp'] = str(commit.author_time)
346
if commit.commit_timezone != commit.author_timezone:
347
rev.properties['author-timezone'] = "%d" % commit.author_timezone
348
if commit._author_timezone_neg_utc:
349
rev.properties['author-timezone-neg-utc'] = ""
350
if commit._commit_timezone_neg_utc:
351
rev.properties['commit-timezone-neg-utc'] = ""
352
rev.timestamp = commit.commit_time
353
rev.timezone = commit.commit_timezone
354
rev.parent_ids = None
355
if rev.git_metadata is not None:
356
md = rev.git_metadata
357
roundtrip_revid = md.revision_id
358
if md.explicit_parent_ids:
359
rev.parent_ids = md.explicit_parent_ids
360
rev.properties.update(md.properties)
361
verifiers = md.verifiers
363
roundtrip_revid = None
365
if rev.parent_ids is None:
366
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
367
return rev, roundtrip_revid, verifiers
369
def get_fileid_map(self, lookup_object, tree_sha):
370
"""Obtain a fileid map for a particular tree.
372
:param lookup_object: Function for looking up an object
373
:param tree_sha: SHA of the root tree
374
:return: GitFileIdMap instance
377
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
381
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
382
return GitFileIdMap(file_ids, self)
385
class BzrGitMappingv1(BzrGitMapping):
386
revid_prefix = 'git-v1'
390
return self.revid_prefix
393
class BzrGitMappingExperimental(BzrGitMappingv1):
394
revid_prefix = 'git-experimental'
398
BZR_FILE_IDS_FILE = '.bzrfileids'
400
BZR_DUMMY_FILE = '.bzrdummy'
402
def _decode_commit_message(self, rev, message, encoding):
403
message = self._extract_hg_metadata(rev, message)
404
message = self._extract_git_svn_metadata(rev, message)
405
message, metadata = self._extract_bzr_metadata(rev, message)
406
return message.decode(encoding), metadata
408
def _encode_commit_message(self, rev, message, encoding):
409
ret = message.encode(encoding)
410
ret += self._generate_hg_message_tail(rev)
411
ret += self._generate_git_svn_metadata(rev, encoding)
414
def import_commit(self, commit, lookup_parent_revid):
415
rev, roundtrip_revid, verifiers = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
416
rev.properties['converted_revision'] = "git %s\n" % commit.id
417
return rev, roundtrip_revid, verifiers
420
class GitMappingRegistry(VcsMappingRegistry):
421
"""Registry with available git mappings."""
423
def revision_id_bzr_to_foreign(self, bzr_revid):
424
if bzr_revid == NULL_REVISION:
425
from dulwich.protocol import ZERO_SHA
426
return ZERO_SHA, None
427
if not bzr_revid.startswith("git-"):
428
raise errors.InvalidRevisionId(bzr_revid, None)
429
(mapping_version, git_sha) = bzr_revid.split(":", 1)
430
mapping = self.get(mapping_version)
431
return mapping.revision_id_bzr_to_foreign(bzr_revid)
433
parse_revision_id = revision_id_bzr_to_foreign
436
mapping_registry = GitMappingRegistry()
437
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
439
mapping_registry.register_lazy('git-experimental',
440
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
441
# mapping_registry.set_default('git-experimental')
442
mapping_registry.set_default('git-v1')
445
class ForeignGit(ForeignVcs):
446
"""The Git Stupid Content Tracker"""
449
def branch_format(self):
450
from bzrlib.plugins.git.branch import GitBranchFormat
451
return GitBranchFormat()
454
def repository_format(self):
455
from bzrlib.plugins.git.repository import GitRepositoryFormat
456
return GitRepositoryFormat()
459
super(ForeignGit, self).__init__(mapping_registry)
460
self.abbreviation = "git"
463
def serialize_foreign_revid(self, foreign_revid):
467
def show_foreign_revid(cls, foreign_revid):
468
return { "git commit": foreign_revid }
471
foreign_vcs_git = ForeignGit()
472
default_mapping = mapping_registry.get_default()()
475
def symlink_to_blob(entry):
476
from dulwich.objects import Blob
478
symlink_target = entry.symlink_target
479
if type(symlink_target) == unicode:
480
symlink_target = symlink_target.encode('utf-8')
481
blob.data = symlink_target
485
def mode_is_executable(mode):
486
"""Check if mode should be considered executable."""
487
return bool(mode & 0111)
491
"""Determine the Bazaar inventory kind based on Unix file mode."""
492
entry_kind = (mode & 0700000) / 0100000
495
elif entry_kind == 1:
496
file_kind = (mode & 070000) / 010000
502
return 'tree-reference'
504
raise AssertionError(
505
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
507
raise AssertionError(
508
"Unknown kind, perms=%r." % (mode,))
511
def object_mode(kind, executable):
512
if kind == 'directory':
514
elif kind == 'symlink':
520
mode = stat.S_IFREG | 0644
524
elif kind == 'tree-reference':
525
from dulwich.objects import S_IFGITLINK
531
def entry_mode(entry):
532
"""Determine the git file mode for an inventory entry."""
533
return object_mode(entry.kind, entry.executable)
536
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes, empty_file_name):
537
"""Create a Git Tree object from a Bazaar directory.
539
:param entry: Inventory entry
540
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
541
:param unusual_modes: Dictionary with unusual file modes by file ids
542
:param empty_file_name: Name to use for dummy files in empty directories,
543
None to ignore empty directories.
545
from dulwich.objects import Blob, Tree
547
for name, value in entry.children.iteritems():
548
ie = entry.children[name]
550
mode = unusual_modes[ie.file_id]
552
mode = entry_mode(ie)
553
hexsha = lookup_ie_sha1(ie)
554
if hexsha is not None:
555
tree.add(name.encode("utf-8"), mode, hexsha)
556
if entry.parent_id is not None and len(tree) == 0:
557
# Only the root can be an empty tree
558
if empty_file_name is not None:
559
tree.add(empty_file_name, stat.S_IFREG | 0644, Blob().id)
565
def extract_unusual_modes(rev):
567
foreign_revid, mapping = mapping_registry.parse_revision_id(
569
except errors.InvalidRevisionId:
572
return mapping.export_unusual_file_modes(rev)
575
def parse_git_svn_id(text):
576
(head, uuid) = text.rsplit(" ", 1)
577
(full_url, rev) = head.rsplit("@", 1)
578
return (full_url, int(rev), uuid)
581
class GitFileIdMap(object):
583
def __init__(self, file_ids, mapping):
584
self.file_ids = file_ids
586
self.mapping = mapping
588
def set_file_id(self, path, file_id):
589
assert type(path) is str
590
assert type(file_id) is str
591
self.file_ids[path] = file_id
593
def lookup_file_id(self, path):
594
assert type(path) is str
596
file_id = self.file_ids[path]
598
file_id = self.mapping.generate_file_id(path)
599
assert type(file_id) is str
602
def lookup_path(self, file_id):
603
if self.paths is None:
605
for k, v in self.file_ids.iteritems():
608
path = self.paths[file_id]
610
return self.mapping.parse_file_id(file_id)
612
assert type(path) is str
616
return self.__class__(dict(self.file_ids), self.mapping)