To get this branch, use:
bzr branch
http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar
« back to all changes in this revision
Viewing changes to mapping.py
- browse files at revision 0.200.920
- compare with another revision
- download diff
- download tarball
- view history from revision 0.200.920
- Committer: Jelmer Vernooij
- Date: 2010-05-22 22:55:03 UTC
- mto: (0.312.1 master) (6883.23.1 bundle-git)
- mto: This revision was merged to the branch mainline in revision 6960.
- Revision ID: jelmer@samba.org-20100522225503-dv038fqxo3r7gcjd
Fix some more tests.
- 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
47
45
format_hg_metadata,
48
46
extract_hg_metadata,
49
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
)
50
55
51
56
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
52
57
67
72
elif file_id[i+1] == 's':
68
73
ret.append(" ")
69
74
else:
70
raise AssertionError("unknown escape character %s" % file_id[i+1])
75
raise AssertionError("unknown escape character %s" %
76
file_id[i+1])
71
77
i += 1
72
78
i += 1
73
79
return "".join(ret)
85
91
86
92
87
93
def warn_unusual_mode(commit, path, mode):
88
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision property. ",
89
mode, path, commit)
90
91
92
def squash_revision(target_repo, rev):
93
"""Remove characters that can't be stored from a revision, if necessary.
94
95
:param target_repo: Repository in which the revision will be stored
96
:param rev: Revision object, will be modified in-place
97
"""
98
if not getattr(target_repo._serializer, "squashes_xml_invalid_characters", True):
99
return
100
from bzrlib.xml_serializer import escape_invalid_chars
101
rev.message, num_escaped = escape_invalid_chars(rev.message)
102
if num_escaped:
103
warn_escaped(rev.foreign_revid, num_escaped)
104
if 'author' in rev.properties:
105
rev.properties['author'], num_escaped = escape_invalid_chars(
106
rev.properties['author'])
107
if num_escaped:
108
warn_escaped(rev.foreign_revid, num_escaped)
109
rev.committer, num_escaped = escape_invalid_chars(rev.committer)
110
if num_escaped:
111
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)
112
96
113
97
114
98
class BzrGitMapping(foreign.VcsMapping):
115
99
"""Class that maps between Git and Bazaar semantics."""
116
100
experimental = False
117
101
102
BZR_FILE_IDS_FILE = None
103
104
BZR_DUMMY_FILE = None
105
118
106
def __init__(self):
119
107
super(BzrGitMapping, self).__init__(foreign_git)
120
108
125
113
@classmethod
126
114
def revision_id_foreign_to_bzr(cls, git_rev_id):
127
115
"""Convert a git revision id handle to a Bazaar revision id."""
128
if git_rev_id == "0" * 40:
116
from dulwich.protocol import ZERO_SHA
117
if git_rev_id == ZERO_SHA:
129
118
return NULL_REVISION
130
119
return "%s:%s" % (cls.revid_prefix, git_rev_id)
131
120
143
132
return ROOT_ID
144
133
return escape_file_id(path)
145
134
135
def is_control_file(self, path):
136
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
137
146
138
def parse_file_id(self, file_id):
147
139
if file_id == ROOT_ID:
148
140
return ""
149
141
return unescape_file_id(file_id)
150
142
143
def revid_as_refname(self, revid):
144
import urllib
145
return "refs/bzr/%s" % urllib.quote(revid)
146
151
147
def import_unusual_file_modes(self, rev, unusual_file_modes):
152
148
if unusual_file_modes:
153
ret = [(name, unusual_file_modes[name])
154
for name in sorted(unusual_file_modes.keys())]
149
ret = [(path, unusual_file_modes[path])
150
for path in sorted(unusual_file_modes.keys())]
155
151
rev.properties['file-modes'] = bencode.bencode(ret)
156
152
157
153
def export_unusual_file_modes(self, rev):
158
154
try:
159
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']
160
156
except KeyError:
161
157
return {}
158
else:
159
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
162
160
163
161
def _generate_git_svn_metadata(self, rev, encoding):
164
162
try:
165
return "\ngit-svn-id: %s\n" % rev.properties["git-svn-id"].encode(encoding)
163
git_svn_id = rev.properties["git-svn-id"]
166
164
except KeyError:
167
165
return ""
166
else:
167
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
168
168
169
169
def _generate_hg_message_tail(self, rev):
170
170
extra = {}
174
174
if name == 'hg:extra:branch':
175
175
branch = rev.properties['hg:extra:branch']
176
176
elif name.startswith('hg:extra'):
177
extra[name[len('hg:extra:'):]] = base64.b64decode(rev.properties[name])
177
extra[name[len('hg:extra:'):]] = base64.b64decode(
178
rev.properties[name])
178
179
elif name == 'hg:renames':
179
renames = bencode.bdecode(base64.b64decode(rev.properties['hg:renames']))
180
renames = bencode.bdecode(base64.b64decode(
181
rev.properties['hg:renames']))
180
182
# TODO: Export other properties as 'bzr:' extras?
181
183
ret = format_hg_metadata(renames, branch, extra)
182
184
assert isinstance(ret, str)
201
203
for name, value in extra.iteritems():
202
204
rev.properties['hg:extra:' + name] = base64.b64encode(value)
203
205
if renames:
204
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()]))
205
208
return message
206
209
210
def _extract_bzr_metadata(self, rev, message):
211
(message, metadata) = extract_bzr_metadata(message)
212
return message, metadata
213
207
214
def _decode_commit_message(self, rev, message, encoding):
208
return message.decode(encoding)
215
return message.decode(encoding), BzrGitRevisionMetadata()
209
216
210
217
def _encode_commit_message(self, rev, message, encoding):
211
218
return message.encode(encoding)
212
219
213
def export_commit(self, rev, tree_sha, parent_lookup):
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):
214
232
"""Turn a Bazaar revision in to a Git commit
215
233
216
234
:param tree_sha: Tree sha for the commit
217
: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
218
237
:return dulwich.objects.Commit represent the revision:
219
238
"""
220
239
from dulwich.objects import Commit
221
240
commit = Commit()
222
241
commit.tree = tree_sha
242
if roundtrip:
243
metadata = BzrGitRevisionMetadata()
244
else:
245
metadata = None
246
parents = []
223
247
for p in rev.parent_ids:
224
248
try:
225
249
git_p = parent_lookup(p)
226
250
except KeyError:
227
251
git_p = None
252
if metadata is not None:
253
metadata.explicit_parent_ids = rev.parent_ids
228
254
if git_p is not None:
229
255
assert len(git_p) == 40, "unexpected length for %r" % git_p
230
commit.parents.append(git_p)
256
parents.append(git_p)
257
commit.parents = parents
231
258
try:
232
259
encoding = rev.properties['git-explicit-encoding']
233
260
except KeyError:
242
269
commit.author_time = long(rev.properties['author-timestamp'])
243
270
else:
244
271
commit.author_time = commit.commit_time
272
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
245
273
commit.commit_timezone = rev.timezone
274
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
246
275
if 'author-timezone' in rev.properties:
247
276
commit.author_timezone = int(rev.properties['author-timezone'])
248
277
else:
249
278
commit.author_timezone = commit.commit_timezone
250
279
commit.message = self._encode_commit_message(rev, rev.message,
251
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
252
298
return commit
253
299
254
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):
255
309
"""Convert a git commit to a bzr revision.
256
310
257
:return: a `bzrlib.revision.Revision` object.
311
:return: a `bzrlib.revision.Revision` object and a
312
dictionary of path -> file ids
258
313
"""
259
314
if commit is None:
260
315
raise AssertionError("Commit object can't be None")
261
rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
262
rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
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
263
320
def decode_using_encoding(rev, commit, encoding):
264
321
rev.committer = str(commit.committer).decode(encoding)
265
322
if commit.committer != commit.author:
266
323
rev.properties['author'] = str(commit.author).decode(encoding)
267
rev.message = self._decode_commit_message(rev, commit.message,
268
encoding)
324
rev.message, rev.git_metadata = self._decode_commit_message(
325
rev, commit.message, encoding)
269
326
if commit.encoding is not None:
270
327
rev.properties['git-explicit-encoding'] = commit.encoding
271
328
decode_using_encoding(rev, commit, commit.encoding)
282
339
if commit.commit_time != commit.author_time:
283
340
rev.properties['author-timestamp'] = str(commit.author_time)
284
341
if commit.commit_timezone != commit.author_timezone:
285
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'] = ""
286
347
rev.timestamp = commit.commit_time
287
348
rev.timezone = commit.commit_timezone
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)
288
356
return rev
289
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
290
373
291
374
class BzrGitMappingv1(BzrGitMapping):
292
375
revid_prefix = 'git-v1'
299
382
class BzrGitMappingExperimental(BzrGitMappingv1):
300
383
revid_prefix = 'git-experimental'
301
384
experimental = True
385
roundtripping = True
386
387
BZR_FILE_IDS_FILE = '.bzrfileids'
388
389
BZR_DUMMY_FILE = '.bzrdummy'
302
390
303
391
def _decode_commit_message(self, rev, message, encoding):
304
392
message = self._extract_hg_metadata(rev, message)
305
393
message = self._extract_git_svn_metadata(rev, message)
306
return message.decode(encoding)
394
message, metadata = self._extract_bzr_metadata(rev, message)
395
return message.decode(encoding), metadata
307
396
308
397
def _encode_commit_message(self, rev, message, encoding):
309
398
ret = message.encode(encoding)
311
400
ret += self._generate_git_svn_metadata(rev, encoding)
312
401
return ret
313
402
314
def import_commit(self, commit):
315
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)
316
405
rev.properties['converted_revision'] = "git %s\n" % commit.id
317
return rev
406
return rev, file_ids
318
407
319
408
320
409
class GitMappingRegistry(VcsMappingRegistry):
322
411
323
412
def revision_id_bzr_to_foreign(self, bzr_revid):
324
413
if bzr_revid == NULL_REVISION:
325
return "0" * 20, None
414
from dulwich.protocol import ZERO_SHA
415
return ZERO_SHA, None
326
416
if not bzr_revid.startswith("git-"):
327
417
raise errors.InvalidRevisionId(bzr_revid, None)
328
418
(mapping_version, git_sha) = bzr_revid.split(":", 1)
334
424
335
425
mapping_registry = GitMappingRegistry()
336
426
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
337
"BzrGitMappingv1")
338
mapping_registry.register_lazy('git-experimental', "bzrlib.plugins.git.mapping",
339
"BzrGitMappingExperimental")
427
"BzrGitMappingv1")
428
mapping_registry.register_lazy('git-experimental',
429
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
340
430
mapping_registry.set_default('git-v1')
341
431
342
432
431
521
return object_mode(entry.kind, entry.executable)
432
522
433
523
434
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes):
435
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
436
534
tree = Tree()
437
535
for name, value in entry.children.iteritems():
438
536
ie = entry.children[name]
445
543
tree.add(mode, name.encode("utf-8"), hexsha)
446
544
if entry.parent_id is not None and len(tree) == 0:
447
545
# Only the root can be an empty tree
448
return None
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
449
551
return tree
450
552
451
553
452
554
def extract_unusual_modes(rev):
453
555
try:
454
foreign_revid, mapping = mapping_registry.parse_revision_id(rev.revision_id)
556
foreign_revid, mapping = mapping_registry.parse_revision_id(
557
rev.revision_id)
455
558
except errors.InvalidRevisionId:
456
559
return {}
457
560
else:
458
561
return mapping.export_unusual_file_modes(rev)
459
562
460
563
461
def inventory_to_tree_and_blobs(inventory, texts, mapping, unusual_modes, cur=None):
462
"""Convert a Bazaar tree to a Git tree.
463
464
:return: Yields tuples with object sha1, object and path
465
"""
466
from dulwich.objects import Tree
467
import stat
468
stack = []
469
if cur is None:
470
cur = ""
471
tree = Tree()
472
473
# stack contains the set of trees that we haven't
474
# finished constructing
475
for path, entry in inventory.iter_entries():
476
while stack and not path.startswith(osutils.pathjoin(cur, "")):
477
# We've hit a file that's not a child of the previous path
478
sha = tree.id
479
yield sha, tree, cur.encode("utf-8")
480
mode = unusual_modes.get(cur.encode("utf-8"), stat.S_IFDIR)
481
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
482
cur, tree = stack.pop()
483
tree.add(*t)
484
485
if entry.kind == "directory":
486
stack.append((cur, tree))
487
cur = path
488
tree = Tree()
489
else:
490
if entry.kind == "file":
491
from dulwich.objects import Blob
492
text = texts.get_record_stream([(entry.file_id, entry.revision)], 'unordered', True).next().get_bytes_as('fulltext')
493
blob = Blob()
494
blob.data = text
495
elif entry.kind == "symlink":
496
blob = symlink_to_blob(entry)
497
else:
498
raise AssertionError("Unknown kind %s" % entry.kind)
499
sha = blob.id
500
yield sha, blob, path.encode("utf-8")
501
name = urlutils.basename(path).encode("utf-8")
502
mode = unusual_modes.get(path.encode("utf-8"), entry_mode(entry))
503
tree.add(mode, name, sha)
504
505
while len(stack) > 1:
506
sha = tree.id
507
yield sha, tree, cur.encode("utf-8")
508
mode = unusual_modes.get(cur.encode('utf-8'), stat.S_IFDIR)
509
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
510
cur, tree = stack.pop()
511
tree.add(*t)
512
513
yield tree.id, tree, cur.encode("utf-8")
514
515
516
564
def parse_git_svn_id(text):
517
565
(head, uuid) = text.rsplit(" ", 1)
518
566
(full_url, rev) = head.rsplit("@", 1)
519
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: 2.0.2.dev0