/brz/remove-bazaar

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

Clean up trailing whitespace.

Show diffs side-by-side

added added

removed removed

Lines of Context:
40
40
    VcsMappingRegistry,
41
41
    ForeignRevision,
42
42
    )
43
 
from bzrlib.revision import (
44
 
    NULL_REVISION,
45
 
    )
46
43
from bzrlib.plugins.git.hg import (
47
44
    format_hg_metadata,
48
45
    extract_hg_metadata,
67
64
            elif file_id[i+1] == 's':
68
65
                ret.append(" ")
69
66
            else:
70
 
                raise AssertionError("unknown escape character %s" %
71
 
                    file_id[i+1])
 
67
                raise AssertionError("unknown escape character %s" % file_id[i+1])
72
68
            i += 1
73
69
        i += 1
74
70
    return "".join(ret)
86
82
 
87
83
 
88
84
def warn_unusual_mode(commit, path, mode):
89
 
    trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
90
 
                 "property. ", mode, path, commit)
 
85
    trace.mutter("Unusual file mode %o for %s in %s. Storing as revision property. ",
 
86
                 mode, path, commit)
91
87
 
92
88
 
93
89
def squash_revision(target_repo, rev):
120
116
        super(BzrGitMapping, self).__init__(foreign_git)
121
117
 
122
118
    def __eq__(self, other):
123
 
        return (type(self) == type(other) and 
124
 
                self.revid_prefix == other.revid_prefix)
 
119
        return type(self) == type(other) and self.revid_prefix == other.revid_prefix
125
120
 
126
121
    @classmethod
127
122
    def revision_id_foreign_to_bzr(cls, git_rev_id):
128
123
        """Convert a git revision id handle to a Bazaar revision id."""
129
 
        if git_rev_id == "0" * 40:
130
 
            return NULL_REVISION
131
124
        return "%s:%s" % (cls.revid_prefix, git_rev_id)
132
125
 
133
126
    @classmethod
151
144
 
152
145
    def import_unusual_file_modes(self, rev, unusual_file_modes):
153
146
        if unusual_file_modes:
154
 
            ret = [(name, unusual_file_modes[name])
155
 
                   for name in sorted(unusual_file_modes.keys())]
 
147
            ret = [(name, unusual_file_modes[name]) for name in sorted(unusual_file_modes.keys())]
156
148
            rev.properties['file-modes'] = bencode.bencode(ret)
157
149
 
158
150
    def export_unusual_file_modes(self, rev):
161
153
        except KeyError:
162
154
            return {}
163
155
 
164
 
    def _generate_git_svn_metadata(self, rev, encoding):
 
156
    def _generate_git_svn_metadata(self, rev):
165
157
        try:
166
 
            return "\ngit-svn-id: %s\n" % rev.properties["git-svn-id"].encode(
167
 
                encoding)
 
158
            return "\ngit-svn-id: %s\n" % rev.properties["git-svn-id"].encode("utf-8")
168
159
        except KeyError:
169
160
            return ""
170
161
 
176
167
            if name == 'hg:extra:branch':
177
168
                branch = rev.properties['hg:extra:branch']
178
169
            elif name.startswith('hg:extra'):
179
 
                extra[name[len('hg:extra:'):]] = base64.b64decode(
180
 
                    rev.properties[name])
 
170
                extra[name[len('hg:extra:'):]] = base64.b64decode(rev.properties[name])
181
171
            elif name == 'hg:renames':
182
 
                renames = bencode.bdecode(base64.b64decode(
183
 
                    rev.properties['hg:renames']))
 
172
                renames = bencode.bdecode(base64.b64decode(rev.properties['hg:renames']))
184
173
            # TODO: Export other properties as 'bzr:' extras?
185
174
        ret = format_hg_metadata(renames, branch, extra)
186
175
        assert isinstance(ret, str)
205
194
        for name, value in extra.iteritems():
206
195
            rev.properties['hg:extra:' + name] = base64.b64encode(value)
207
196
        if renames:
208
 
            rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
209
 
                [(new, old) for (old, new) in renames.iteritems()]))
 
197
            rev.properties['hg:renames'] = base64.b64encode(bencode.bencode([(new, old) for (old, new) in renames.iteritems()]))
210
198
        return message
211
199
 
212
 
    def _decode_commit_message(self, rev, message, encoding):
213
 
        return message.decode(encoding)
 
200
    def _decode_commit_message(self, rev, message):
 
201
        return message.decode("utf-8", "replace")
214
202
 
215
 
    def _encode_commit_message(self, rev, message, encoding):
216
 
        return message.encode(encoding)
 
203
    def _encode_commit_message(self, rev, message):
 
204
        return message.encode("utf-8")
217
205
 
218
206
    def export_commit(self, rev, tree_sha, parent_lookup):
219
207
        """Turn a Bazaar revision in to a Git commit
220
208
 
221
209
        :param tree_sha: Tree sha for the commit
222
 
        :param parent_lookup: Function for looking up the GIT sha equiv of a
223
 
            bzr revision
 
210
        :param parent_lookup: Function for looking up the GIT sha equiv of a bzr revision
224
211
        :return dulwich.objects.Commit represent the revision:
225
212
        """
226
213
        from dulwich.objects import Commit
227
214
        commit = Commit()
228
215
        commit.tree = tree_sha
229
216
        for p in rev.parent_ids:
230
 
            try:
231
 
                git_p = parent_lookup(p)
232
 
            except KeyError:
233
 
                git_p = None
 
217
            git_p = parent_lookup(p)
234
218
            if git_p is not None:
235
219
                assert len(git_p) == 40, "unexpected length for %r" % git_p
236
220
                commit.parents.append(git_p)
237
 
        try:
238
 
            encoding = rev.properties['git-explicit-encoding']
239
 
        except KeyError:
240
 
            encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
241
 
        commit.encoding = rev.properties.get('git-explicit-encoding')
242
 
        commit.committer = fix_person_identifier(rev.committer.encode(
243
 
            encoding))
244
 
        commit.author = fix_person_identifier(
245
 
            rev.get_apparent_authors()[0].encode(encoding))
 
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"))
246
223
        commit.commit_time = long(rev.timestamp)
247
224
        if 'author-timestamp' in rev.properties:
248
225
            commit.author_time = long(rev.properties['author-timestamp'])
253
230
            commit.author_timezone = int(rev.properties['author-timezone'])
254
231
        else:
255
232
            commit.author_timezone = commit.commit_timezone
256
 
        commit.message = self._encode_commit_message(rev, rev.message, 
257
 
            encoding)
 
233
        commit.message = self._encode_commit_message(rev, rev.message)
258
234
        return commit
259
235
 
260
236
    def import_commit(self, commit):
264
240
        """
265
241
        if commit is None:
266
242
            raise AssertionError("Commit object can't be None")
267
 
        rev = ForeignRevision(commit.id, self,
268
 
                self.revision_id_foreign_to_bzr(commit.id))
 
243
        rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
269
244
        rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
270
 
        def decode_using_encoding(rev, commit, encoding):
271
 
            rev.committer = str(commit.committer).decode(encoding)
272
 
            if commit.committer != commit.author:
273
 
                rev.properties['author'] = str(commit.author).decode(encoding)
274
 
            rev.message = self._decode_commit_message(rev, commit.message, 
275
 
                encoding)
276
 
        if commit.encoding is not None:
277
 
            rev.properties['git-explicit-encoding'] = commit.encoding
278
 
            decode_using_encoding(rev, commit, commit.encoding)
279
 
        else:
280
 
            for encoding in ('utf-8', 'latin1'):
281
 
                try:
282
 
                    decode_using_encoding(rev, commit, encoding)
283
 
                except UnicodeDecodeError:
284
 
                    pass
285
 
                else:
286
 
                    if encoding != 'utf-8':
287
 
                        rev.properties['git-implicit-encoding'] = encoding
288
 
                    break
 
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
 
289
249
        if commit.commit_time != commit.author_time:
290
250
            rev.properties['author-timestamp'] = str(commit.author_time)
291
251
        if commit.commit_timezone != commit.author_timezone:
292
 
            rev.properties['author-timezone'] = "%d" % commit.author_timezone
 
252
            rev.properties['author-timezone'] = "%d" % (commit.author_timezone, )
293
253
        rev.timestamp = commit.commit_time
294
254
        rev.timezone = commit.commit_timezone
 
255
        rev.message = self._decode_commit_message(rev, commit.message)
295
256
        return rev
296
257
 
297
258
 
307
268
    revid_prefix = 'git-experimental'
308
269
    experimental = True
309
270
 
310
 
    def _decode_commit_message(self, rev, message, encoding):
 
271
    def _decode_commit_message(self, rev, message):
311
272
        message = self._extract_hg_metadata(rev, message)
312
273
        message = self._extract_git_svn_metadata(rev, message)
313
 
        return message.decode(encoding)
 
274
        return message.decode("utf-8", "replace")
314
275
 
315
 
    def _encode_commit_message(self, rev, message, encoding):
316
 
        ret = message.encode(encoding)
 
276
    def _encode_commit_message(self, rev, message):
 
277
        ret = message.encode("utf-8")
317
278
        ret += self._generate_hg_message_tail(rev)
318
 
        ret += self._generate_git_svn_metadata(rev, encoding)
 
279
        ret += self._generate_git_svn_metadata(rev)
319
280
        return ret
320
281
 
321
282
    def import_commit(self, commit):
328
289
    """Registry with available git mappings."""
329
290
 
330
291
    def revision_id_bzr_to_foreign(self, bzr_revid):
331
 
        if bzr_revid == NULL_REVISION:
332
 
            return "0" * 20, None
333
292
        if not bzr_revid.startswith("git-"):
334
293
            raise errors.InvalidRevisionId(bzr_revid, None)
335
294
        (mapping_version, git_sha) = bzr_revid.split(":", 1)
341
300
 
342
301
mapping_registry = GitMappingRegistry()
343
302
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
344
 
    "BzrGitMappingv1")
345
 
mapping_registry.register_lazy('git-experimental',
346
 
    "bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
 
303
                                   "BzrGitMappingv1")
 
304
mapping_registry.register_lazy('git-experimental', "bzrlib.plugins.git.mapping",
 
305
                                   "BzrGitMappingExperimental")
347
306
mapping_registry.set_default('git-v1')
348
307
 
349
308
 
377
336
default_mapping = mapping_registry.get_default()()
378
337
 
379
338
 
 
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
 
380
347
def symlink_to_blob(entry):
381
348
    from dulwich.objects import Blob
382
349
    blob = Blob()
383
 
    symlink_target = entry.symlink_target
384
 
    if type(symlink_target) == unicode:
385
 
        symlink_target = symlink_target.encode('utf-8')
386
 
    blob.data = symlink_target
 
350
    blob._text = entry.symlink_target
387
351
    return blob
388
352
 
389
353
 
417
381
    if kind == 'directory':
418
382
        return stat.S_IFDIR
419
383
    elif kind == 'symlink':
420
 
        mode = stat.S_IFLNK
421
 
        if executable:
422
 
            mode |= 0111
423
 
        return mode
 
384
        return stat.S_IFLNK
424
385
    elif kind == 'file':
425
386
        mode = stat.S_IFREG | 0644
426
387
        if executable:
441
402
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes):
442
403
    from dulwich.objects import Tree
443
404
    tree = Tree()
444
 
    for name, value in entry.children.iteritems():
 
405
    for name in sorted(entry.children.keys()):
445
406
        ie = entry.children[name]
446
407
        try:
447
408
            mode = unusual_modes[ie.file_id]
453
414
    if entry.parent_id is not None and len(tree) == 0:
454
415
        # Only the root can be an empty tree
455
416
        return None
 
417
    tree.serialize()
456
418
    return tree
457
419
 
458
420
 
459
421
def extract_unusual_modes(rev):
460
422
    try:
461
 
        foreign_revid, mapping = mapping_registry.parse_revision_id(
462
 
            rev.revision_id)
 
423
        foreign_revid, mapping = mapping_registry.parse_revision_id(rev.revision_id)
463
424
    except errors.InvalidRevisionId:
464
425
        return {}
465
426
    else:
466
427
        return mapping.export_unusual_file_modes(rev)
467
428
 
468
429
 
469
 
def inventory_to_tree_and_blobs(inventory, texts, mapping, unusual_modes,
470
 
                                cur=None):
 
430
def inventory_to_tree_and_blobs(inventory, texts, mapping, unusual_modes, cur=None):
471
431
    """Convert a Bazaar tree to a Git tree.
472
432
 
473
433
    :return: Yields tuples with object sha1, object and path
484
444
    for path, entry in inventory.iter_entries():
485
445
        while stack and not path.startswith(osutils.pathjoin(cur, "")):
486
446
            # We've hit a file that's not a child of the previous path
 
447
            tree.serialize()
487
448
            sha = tree.id
488
449
            yield sha, tree, cur.encode("utf-8")
489
450
            mode = unusual_modes.get(cur.encode("utf-8"), stat.S_IFDIR)
497
458
            tree = Tree()
498
459
        else:
499
460
            if entry.kind == "file":
500
 
                from dulwich.objects import Blob
501
 
                stream = texts.get_record_stream(
502
 
                    [(entry.file_id, entry.revision)], 'unordered', True)
503
 
                blob = Blob()
504
 
                blob.chunked = stream.next().get_bytes_as('chunks')
 
461
                blob = text_to_blob(texts, entry)
505
462
            elif entry.kind == "symlink":
506
463
                blob = symlink_to_blob(entry)
507
464
            else:
513
470
            tree.add(mode, name, sha)
514
471
 
515
472
    while len(stack) > 1:
 
473
        tree.serialize()
516
474
        sha = tree.id
517
475
        yield sha, tree, cur.encode("utf-8")
518
476
        mode = unusual_modes.get(cur.encode('utf-8'), stat.S_IFDIR)
520
478
        cur, tree = stack.pop()
521
479
        tree.add(*t)
522
480
 
 
481
    tree.serialize()
523
482
    yield tree.id, tree, cur.encode("utf-8")
524
483
 
525
484