« back to all changes in this revision
Viewing changes to mapping.py
- browse files at revision 57
- compare with another revision
- download diff
- download tarball
- view history from revision 57
- Committer: Jelmer Vernooij
- Date: 2010-05-22 23:40:04 UTC
- mfrom: (17.25.364 trunk)
- Revision ID: jelmer@samba.org-20100522234004-48r0bfkodp3xspgv
* New upstream release.
* Bump standards version to 3.8.4.
* Bump standards version to 3.8.4.
- files added:
- notes/git-serve.txt
- files modified:
- NEWS
added
removed
mapping.py
1
1
# Copyright (C) 2007 Canonical Ltd
2
# Copyright (C) 2008-2009 Jelmer Vernooij <jelmer@samba.org>
2
# Copyright (C) 2008-2010 Jelmer Vernooij <jelmer@samba.org>
3
3
# Copyright (C) 2008 John Carr
4
4
#
5
5
# This program is free software; you can redistribute it and/or modify
24
24
from bzrlib import (
25
25
errors,
26
26
foreign,
27
osutils,
28
27
trace,
29
urlutils,
30
28
)
31
29
try:
32
30
from bzrlib import bencode
40
38
VcsMappingRegistry,
41
39
ForeignRevision,
42
40
)
41
from bzrlib.revision import (
42
NULL_REVISION,
43
)
43
44
from bzrlib.plugins.git.hg import (
44
45
format_hg_metadata,
45
46
extract_hg_metadata,
46
47
)
48
from bzrlib.plugins.git.roundtrip import (
49
extract_bzr_metadata,
50
inject_bzr_metadata,
51
BzrGitRevisionMetadata,
52
deserialize_fileid_map,
53
serialize_fileid_map,
54
)
47
55
48
56
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
49
57
64
72
elif file_id[i+1] == 's':
65
73
ret.append(" ")
66
74
else:
67
raise AssertionError("unknown escape character %s" % file_id[i+1])
75
raise AssertionError("unknown escape character %s" %
76
file_id[i+1])
68
77
i += 1
69
78
i += 1
70
79
return "".join(ret)
82
91
83
92
84
93
def warn_unusual_mode(commit, path, mode):
85
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision property. ",
86
mode, path, commit)
87
88
89
def squash_revision(target_repo, rev):
90
"""Remove characters that can't be stored from a revision, if necessary.
91
92
:param target_repo: Repository in which the revision will be stored
93
:param rev: Revision object, will be modified in-place
94
"""
95
if not getattr(target_repo._serializer, "squashes_xml_invalid_characters", True):
96
return
97
from bzrlib.xml_serializer import escape_invalid_chars
98
rev.message, num_escaped = escape_invalid_chars(rev.message)
99
if num_escaped:
100
warn_escaped(rev.foreign_revid, num_escaped)
101
if 'author' in rev.properties:
102
rev.properties['author'], num_escaped = escape_invalid_chars(
103
rev.properties['author'])
104
if num_escaped:
105
warn_escaped(rev.foreign_revid, num_escaped)
106
rev.committer, num_escaped = escape_invalid_chars(rev.committer)
107
if num_escaped:
108
warn_escaped(rev.foreign_revid, num_escaped)
94
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
95
"property. ", mode, path, commit)
109
96
110
97
111
98
class BzrGitMapping(foreign.VcsMapping):
112
99
"""Class that maps between Git and Bazaar semantics."""
113
100
experimental = False
114
101
102
BZR_FILE_IDS_FILE = None
103
104
BZR_DUMMY_FILE = None
105
115
106
def __init__(self):
116
107
super(BzrGitMapping, self).__init__(foreign_git)
117
108
118
109
def __eq__(self, other):
119
return type(self) == type(other) and self.revid_prefix == other.revid_prefix
110
return (type(self) == type(other) and
111
self.revid_prefix == other.revid_prefix)
120
112
121
113
@classmethod
122
114
def revision_id_foreign_to_bzr(cls, git_rev_id):
123
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:
118
return NULL_REVISION
124
119
return "%s:%s" % (cls.revid_prefix, git_rev_id)
125
120
126
121
@classmethod
137
132
return ROOT_ID
138
133
return escape_file_id(path)
139
134
135
def is_control_file(self, path):
136
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
137
140
138
def parse_file_id(self, file_id):
141
139
if file_id == ROOT_ID:
142
140
return ""
143
141
return unescape_file_id(file_id)
144
142
143
def revid_as_refname(self, revid):
144
import urllib
145
return "refs/bzr/%s" % urllib.quote(revid)
146
145
147
def import_unusual_file_modes(self, rev, unusual_file_modes):
146
148
if unusual_file_modes:
147
ret = [(name, unusual_file_modes[name]) for name in sorted(unusual_file_modes.keys())]
149
ret = [(path, unusual_file_modes[path])
150
for path in sorted(unusual_file_modes.keys())]
148
151
rev.properties['file-modes'] = bencode.bencode(ret)
149
152
150
153
def export_unusual_file_modes(self, rev):
151
154
try:
152
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(rev.properties['file-modes'].encode("utf-8"))])
155
file_modes = rev.properties['file-modes']
153
156
except KeyError:
154
157
return {}
158
else:
159
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
155
160
156
def _generate_git_svn_metadata(self, rev):
161
def _generate_git_svn_metadata(self, rev, encoding):
157
162
try:
158
return "\ngit-svn-id: %s\n" % rev.properties["git-svn-id"].encode("utf-8")
163
git_svn_id = rev.properties["git-svn-id"]
159
164
except KeyError:
160
165
return ""
166
else:
167
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
161
168
162
169
def _generate_hg_message_tail(self, rev):
163
170
extra = {}
167
174
if name == 'hg:extra:branch':
168
175
branch = rev.properties['hg:extra:branch']
169
176
elif name.startswith('hg:extra'):
170
extra[name[len('hg:extra:'):]] = base64.b64decode(rev.properties[name])
177
extra[name[len('hg:extra:'):]] = base64.b64decode(
178
rev.properties[name])
171
179
elif name == 'hg:renames':
172
renames = bencode.bdecode(base64.b64decode(rev.properties['hg:renames']))
180
renames = bencode.bdecode(base64.b64decode(
181
rev.properties['hg:renames']))
173
182
# TODO: Export other properties as 'bzr:' extras?
174
183
ret = format_hg_metadata(renames, branch, extra)
175
184
assert isinstance(ret, str)
194
203
for name, value in extra.iteritems():
195
204
rev.properties['hg:extra:' + name] = base64.b64encode(value)
196
205
if renames:
197
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode([(new, old) for (old, new) in renames.iteritems()]))
206
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
207
[(new, old) for (old, new) in renames.iteritems()]))
198
208
return message
199
209
200
def _decode_commit_message(self, rev, message):
201
return message.decode("utf-8", "replace")
202
203
def _encode_commit_message(self, rev, message):
204
return message.encode("utf-8")
205
206
def export_commit(self, rev, tree_sha, parent_lookup):
210
def _extract_bzr_metadata(self, rev, message):
211
(message, metadata) = extract_bzr_metadata(message)
212
return message, metadata
213
214
def _decode_commit_message(self, rev, message, encoding):
215
return message.decode(encoding), BzrGitRevisionMetadata()
216
217
def _encode_commit_message(self, rev, message, encoding):
218
return message.encode(encoding)
219
220
def export_fileid_map(self, fileid_map):
221
"""Export a file id map to a fileid map.
222
223
:param fileid_map: File id map, mapping paths to file ids
224
:return: A Git blob object
225
"""
226
from dulwich.objects import Blob
227
b = Blob()
228
b.set_raw_chunks(serialize_fileid_map(fileid_map))
229
return b
230
231
def export_commit(self, rev, tree_sha, parent_lookup, roundtrip):
207
232
"""Turn a Bazaar revision in to a Git commit
208
233
209
234
:param tree_sha: Tree sha for the commit
210
:param parent_lookup: Function for looking up the GIT sha equiv of a bzr revision
235
:param parent_lookup: Function for looking up the GIT sha equiv of a
236
bzr revision
211
237
:return dulwich.objects.Commit represent the revision:
212
238
"""
213
239
from dulwich.objects import Commit
214
240
commit = Commit()
215
241
commit.tree = tree_sha
242
if roundtrip:
243
metadata = BzrGitRevisionMetadata()
244
else:
245
metadata = None
246
parents = []
216
247
for p in rev.parent_ids:
217
git_p = parent_lookup(p)
248
try:
249
git_p = parent_lookup(p)
250
except KeyError:
251
git_p = None
252
if metadata is not None:
253
metadata.explicit_parent_ids = rev.parent_ids
218
254
if git_p is not None:
219
255
assert len(git_p) == 40, "unexpected length for %r" % git_p
220
commit.parents.append(git_p)
221
commit.committer = fix_person_identifier(rev.committer.encode("utf-8"))
222
commit.author = fix_person_identifier(rev.get_apparent_authors()[0].encode("utf-8"))
256
parents.append(git_p)
257
commit.parents = parents
258
try:
259
encoding = rev.properties['git-explicit-encoding']
260
except KeyError:
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(
264
encoding))
265
commit.author = fix_person_identifier(
266
rev.get_apparent_authors()[0].encode(encoding))
223
267
commit.commit_time = long(rev.timestamp)
224
268
if 'author-timestamp' in rev.properties:
225
269
commit.author_time = long(rev.properties['author-timestamp'])
226
270
else:
227
271
commit.author_time = commit.commit_time
272
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
228
273
commit.commit_timezone = rev.timezone
274
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
229
275
if 'author-timezone' in rev.properties:
230
276
commit.author_timezone = int(rev.properties['author-timezone'])
231
277
else:
232
278
commit.author_timezone = commit.commit_timezone
233
commit.message = self._encode_commit_message(rev, rev.message)
279
commit.message = self._encode_commit_message(rev, rev.message,
280
encoding)
281
assert type(commit.message) == str
282
if metadata is not None:
283
try:
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,
296
encoding)
297
assert type(commit.message) == str
234
298
return commit
235
299
236
def import_commit(self, commit):
300
def import_fileid_map(self, blob):
301
"""Convert a git file id map blob.
302
303
:param blob: Git blob object with fileid map
304
:return: Dictionary mapping paths to file ids
305
"""
306
return deserialize_fileid_map(blob.data)
307
308
def import_commit(self, commit, lookup_parent_revid):
237
309
"""Convert a git commit to a bzr revision.
238
310
239
:return: a `bzrlib.revision.Revision` object.
311
:return: a `bzrlib.revision.Revision` object and a
312
dictionary of path -> file ids
240
313
"""
241
314
if commit is None:
242
315
raise AssertionError("Commit object can't be None")
243
rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
244
rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
245
rev.committer = str(commit.committer).decode("utf-8", "replace")
246
if commit.committer != commit.author:
247
rev.properties['author'] = str(commit.author).decode("utf-8", "replace")
248
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)
329
else:
330
for encoding in ('utf-8', 'latin1'):
331
try:
332
decode_using_encoding(rev, commit, encoding)
333
except UnicodeDecodeError:
334
pass
335
else:
336
if encoding != 'utf-8':
337
rev.properties['git-implicit-encoding'] = encoding
338
break
249
339
if commit.commit_time != commit.author_time:
250
340
rev.properties['author-timestamp'] = str(commit.author_time)
251
341
if commit.commit_timezone != commit.author_timezone:
252
rev.properties['author-timezone'] = "%d" % (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'] = ""
253
347
rev.timestamp = commit.commit_time
254
348
rev.timezone = commit.commit_timezone
255
rev.message = self._decode_commit_message(rev, commit.message)
349
if rev.git_metadata is not None:
350
md = rev.git_metadata
351
if md.revision_id:
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)
256
356
return rev
257
357
358
def get_fileid_map(self, lookup_object, tree_sha):
359
"""Obtain a fileid map for a particular tree.
360
361
:param lookup_object: Function for looking up an object
362
:param tree_sha: SHA of the root tree
363
:return: GitFileIdMap instance
364
"""
365
try:
366
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
367
except KeyError:
368
file_ids = {}
369
else:
370
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
371
return GitFileIdMap(file_ids, self)
372
258
373
259
374
class BzrGitMappingv1(BzrGitMapping):
260
375
revid_prefix = 'git-v1'
267
382
class BzrGitMappingExperimental(BzrGitMappingv1):
268
383
revid_prefix = 'git-experimental'
269
384
experimental = True
270
271
def _decode_commit_message(self, rev, message):
385
roundtripping = True
386
387
BZR_FILE_IDS_FILE = '.bzrfileids'
388
389
BZR_DUMMY_FILE = '.bzrdummy'
390
391
def _decode_commit_message(self, rev, message, encoding):
272
392
message = self._extract_hg_metadata(rev, message)
273
393
message = self._extract_git_svn_metadata(rev, message)
274
return message.decode("utf-8", "replace")
394
message, metadata = self._extract_bzr_metadata(rev, message)
395
return message.decode(encoding), metadata
275
396
276
def _encode_commit_message(self, rev, message):
277
ret = message.encode("utf-8")
397
def _encode_commit_message(self, rev, message, encoding):
398
ret = message.encode(encoding)
278
399
ret += self._generate_hg_message_tail(rev)
279
ret += self._generate_git_svn_metadata(rev)
400
ret += self._generate_git_svn_metadata(rev, encoding)
280
401
return ret
281
402
282
def import_commit(self, commit):
283
rev = super(BzrGitMappingExperimental, self).import_commit(commit)
403
def import_commit(self, commit, lookup_parent_revid):
404
rev, file_ids = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
284
405
rev.properties['converted_revision'] = "git %s\n" % commit.id
285
return rev
406
return rev, file_ids
286
407
287
408
288
409
class GitMappingRegistry(VcsMappingRegistry):
289
410
"""Registry with available git mappings."""
290
411
291
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
292
416
if not bzr_revid.startswith("git-"):
293
417
raise errors.InvalidRevisionId(bzr_revid, None)
294
418
(mapping_version, git_sha) = bzr_revid.split(":", 1)
300
424
301
425
mapping_registry = GitMappingRegistry()
302
426
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
303
"BzrGitMappingv1")
304
mapping_registry.register_lazy('git-experimental', "bzrlib.plugins.git.mapping",
305
"BzrGitMappingExperimental")
427
"BzrGitMappingv1")
428
mapping_registry.register_lazy('git-experimental',
429
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
306
430
mapping_registry.set_default('git-v1')
307
431
308
432
336
460
default_mapping = mapping_registry.get_default()()
337
461
338
462
339
def text_to_blob(texts, entry):
340
from dulwich.objects import Blob
341
text = texts.get_record_stream([(entry.file_id, entry.revision)], 'unordered', True).next().get_bytes_as('fulltext')
342
blob = Blob()
343
blob._text = text
344
return blob
345
346
347
463
def symlink_to_blob(entry):
348
464
from dulwich.objects import Blob
349
465
blob = Blob()
350
blob._text = entry.symlink_target
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
351
470
return blob
352
471
353
472
381
500
if kind == 'directory':
382
501
return stat.S_IFDIR
383
502
elif kind == 'symlink':
384
return stat.S_IFLNK
503
mode = stat.S_IFLNK
504
if executable:
505
mode |= 0111
506
return mode
385
507
elif kind == 'file':
386
508
mode = stat.S_IFREG | 0644
387
509
if executable:
399
521
return object_mode(entry.kind, entry.executable)
400
522
401
523
402
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes):
403
from dulwich.objects import Tree
524
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes, empty_file_name):
525
"""Create a Git Tree object from a Bazaar directory.
526
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.
532
"""
533
from dulwich.objects import Blob, Tree
404
534
tree = Tree()
405
for name in sorted(entry.children.keys()):
535
for name, value in entry.children.iteritems():
406
536
ie = entry.children[name]
407
537
try:
408
538
mode = unusual_modes[ie.file_id]
413
543
tree.add(mode, name.encode("utf-8"), hexsha)
414
544
if entry.parent_id is not None and len(tree) == 0:
415
545
# Only the root can be an empty tree
416
return None
417
tree.serialize()
546
if empty_file_name is not None:
547
tree.add(stat.S_IFREG | 0644, empty_file_name,
548
Blob().id)
549
else:
550
return None
418
551
return tree
419
552
420
553
421
554
def extract_unusual_modes(rev):
422
555
try:
423
foreign_revid, mapping = mapping_registry.parse_revision_id(rev.revision_id)
556
foreign_revid, mapping = mapping_registry.parse_revision_id(
557
rev.revision_id)
424
558
except errors.InvalidRevisionId:
425
559
return {}
426
560
else:
427
561
return mapping.export_unusual_file_modes(rev)
428
562
429
563
430
def inventory_to_tree_and_blobs(inventory, texts, mapping, unusual_modes, cur=None):
431
"""Convert a Bazaar tree to a Git tree.
432
433
:return: Yields tuples with object sha1, object and path
434
"""
435
from dulwich.objects import Tree
436
import stat
437
stack = []
438
if cur is None:
439
cur = ""
440
tree = Tree()
441
442
# stack contains the set of trees that we haven't
443
# finished constructing
444
for path, entry in inventory.iter_entries():
445
while stack and not path.startswith(osutils.pathjoin(cur, "")):
446
# We've hit a file that's not a child of the previous path
447
tree.serialize()
448
sha = tree.id
449
yield sha, tree, cur.encode("utf-8")
450
mode = unusual_modes.get(cur.encode("utf-8"), stat.S_IFDIR)
451
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
452
cur, tree = stack.pop()
453
tree.add(*t)
454
455
if entry.kind == "directory":
456
stack.append((cur, tree))
457
cur = path
458
tree = Tree()
459
else:
460
if entry.kind == "file":
461
blob = text_to_blob(texts, entry)
462
elif entry.kind == "symlink":
463
blob = symlink_to_blob(entry)
464
else:
465
raise AssertionError("Unknown kind %s" % entry.kind)
466
sha = blob.id
467
yield sha, blob, path.encode("utf-8")
468
name = urlutils.basename(path).encode("utf-8")
469
mode = unusual_modes.get(path.encode("utf-8"), entry_mode(entry))
470
tree.add(mode, name, sha)
471
472
while len(stack) > 1:
473
tree.serialize()
474
sha = tree.id
475
yield sha, tree, cur.encode("utf-8")
476
mode = unusual_modes.get(cur.encode('utf-8'), stat.S_IFDIR)
477
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
478
cur, tree = stack.pop()
479
tree.add(*t)
480
481
tree.serialize()
482
yield tree.id, tree, cur.encode("utf-8")
483
484
485
564
def parse_git_svn_id(text):
486
565
(head, uuid) = text.rsplit(" ", 1)
487
566
(full_url, rev) = head.rsplit("@", 1)
488
567
return (full_url, int(rev), uuid)
568
569
570
class GitFileIdMap(object):
571
572
def __init__(self, file_ids, mapping):
573
self.file_ids = file_ids
574
self.paths = None
575
self.mapping = mapping
576
577
def lookup_file_id(self, path):
578
try:
579
return self.file_ids[path]
580
except KeyError:
581
return self.mapping.generate_file_id(path)
582
583
def lookup_path(self, file_id):
584
if self.paths is None:
585
self.paths = {}
586
for k, v in self.file_ids.iteritems():
587
self.paths[v] = k
588
try:
589
return self.paths[file_id]
590
except KeyError:
591
return self.mapping.parse_file_id(file_id)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1