/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 object_store.py

Add function for verifying reconstruction of objects still works.

Show diffs side-by-side

added added

removed removed

Lines of Context:
object_store.py
18
18
 
19
19
from dulwich.objects import (
20
20
    Blob,
21
 
    Commit,
22
21
    Tree,
23
22
    sha_to_hex,
24
23
    )
36
35
from bzrlib.revision import (
37
36
    NULL_REVISION,
38
37
    )
39
 
from bzrlib.testament import(
40
 
    StrictTestament3,
41
 
    )
42
38
 
43
39
from bzrlib.plugins.git.mapping import (
44
40
    default_mapping,
47
43
    mapping_registry,
48
44
    symlink_to_blob,
49
45
    )
50
 
from bzrlib.plugins.git.cache import (
 
46
from bzrlib.plugins.git.shamap import (
51
47
    from_repository as cache_from_repository,
52
48
    )
53
49
 
54
50
import posixpath
55
 
import stat
56
51
 
57
52
 
58
53
def get_object_store(repo, mapping=None):
75
70
        self._cache = lru_cache.LRUSizeCache(max_size=MAX_TREE_CACHE_SIZE,
76
71
            after_cleanup_size=None, compute_size=approx_tree_size)
77
72
 
78
 
    def revision_tree(self, revid):
 
73
    def revision_tree(self, revid):            
79
74
        try:
80
 
            tree = self._cache[revid]
 
75
            return self._cache[revid] 
81
76
        except KeyError:
82
77
            tree = self.repository.revision_tree(revid)
83
78
            self.add(tree)
84
 
        assert tree.get_revision_id() == tree.inventory.revision_id
85
 
        return tree
 
79
            return tree
86
80
 
87
81
    def iter_revision_trees(self, revids):
88
 
        trees = {}
89
 
        todo = []
90
 
        for revid in revids:
91
 
            try:
92
 
                tree = self._cache[revid]
93
 
            except KeyError:
94
 
                todo.append(revid)
95
 
            else:
96
 
                assert tree.get_revision_id() == revid
97
 
                assert tree.inventory.revision_id == revid
98
 
                trees[revid] = tree
99
 
        for tree in self.repository.revision_trees(todo):
 
82
        trees = dict([(k, self._cache.get(k)) for k in revids]) 
 
83
        for tree in self.repository.revision_trees(
 
84
                [r for r, v in trees.iteritems() if v is None]):
100
85
            trees[tree.get_revision_id()] = tree
101
86
            self.add(tree)
102
87
        return (trees[r] for r in revids)
108
93
        self._cache.add(tree.get_revision_id(), tree)
109
94
 
110
95
 
111
 
def _find_missing_bzr_revids(get_parent_map, want, have):
112
 
    """Find the revisions that have to be pushed.
113
 
 
114
 
    :param get_parent_map: Function that returns the parents for a sequence
115
 
        of revisions.
116
 
    :param want: Revisions the target wants
117
 
    :param have: Revisions the target already has
118
 
    :return: Set of revisions to fetch
119
 
    """
120
 
    pending = want - have
121
 
    processed = set()
122
 
    todo = set()
123
 
    while pending:
124
 
        processed.update(pending)
125
 
        next_map = get_parent_map(pending)
126
 
        next_pending = set()
127
 
        for item in next_map.iteritems():
128
 
            if item[0] in have:
129
 
                continue
130
 
            todo.add(item[0])
131
 
            next_pending.update(p for p in item[1] if p not in processed)
132
 
        pending = next_pending
133
 
    if NULL_REVISION in todo:
134
 
        todo.remove(NULL_REVISION)
135
 
    return todo
136
 
 
137
 
 
138
96
def _check_expected_sha(expected_sha, object):
139
97
    """Check whether an object matches an expected SHA.
140
98
 
156
114
            expected_sha))
157
115
 
158
116
 
159
 
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes,
160
 
                     dummy_file_name=None):
 
117
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes):
161
118
    """Iterate over the objects that were introduced in a revision.
162
119
 
163
120
    :param idmap: id map
164
 
    :param parent_trees: Parent revision trees
165
 
    :param unusual_modes: Unusual file modes dictionary
166
 
    :param dummy_file_name: File name to use for dummy files
167
 
        in empty directories. None to skip empty directories
 
121
    :param unusual_modes: Unusual file modes
168
122
    :return: Yields (path, object, ie) entries
169
123
    """
170
124
    new_trees = {}
189
143
                    pie.symlink_target == ie.symlink_target):
190
144
                    return pie
191
145
        raise KeyError
192
 
 
193
 
    # Find all the changed blobs
194
146
    for (file_id, path, changed_content, versioned, parent, name, kind,
195
147
         executable) in tree.iter_changes(base_tree):
196
148
        if kind[1] == "file":
197
149
            ie = tree.inventory[file_id]
198
150
            if changed_content:
 
151
                
199
152
                try:
200
153
                    pie = find_unchanged_parent_ie(ie, other_parent_trees)
201
154
                except KeyError:
202
155
                    pass
203
156
                else:
204
 
                    try:
205
 
                        shamap[ie.file_id] = idmap.lookup_blob_id(
206
 
                            pie.file_id, pie.revision)
207
 
                    except KeyError:
208
 
                        # no-change merge ?
209
 
                        blob = Blob()
210
 
                        blob.data = tree.get_file_text(ie.file_id)
211
 
                        shamap[ie.file_id] = blob.id
 
157
                    shamap[ie.file_id] = idmap.lookup_blob_id(
 
158
                        pie.file_id, pie.revision)
212
159
            if not file_id in shamap:
213
160
                new_blobs.append((path[1], ie))
214
161
            new_trees[posixpath.dirname(path[1])] = parent[1]
224
171
            new_trees[posixpath.dirname(path[1])] = parent[1]
225
172
        elif kind[1] not in (None, "directory"):
226
173
            raise AssertionError(kind[1])
227
 
        if (path[0] not in (None, "") and
228
 
            parent[0] in tree.inventory and
229
 
            tree.inventory[parent[0]].kind == "directory"):
230
 
            # Removal
 
174
        if path[0] is not None:
231
175
            new_trees[posixpath.dirname(path[0])] = parent[0]
232
176
    
233
 
    # Fetch contents of the blobs that were changed
234
177
    for (path, ie), chunks in tree.iter_files_bytes(
235
178
        [(ie.file_id, (path, ie)) for (path, ie) in new_blobs]):
236
179
        obj = Blob()
241
184
    for path in unusual_modes:
242
185
        parent_path = posixpath.dirname(path)
243
186
        new_trees[parent_path] = tree.path2id(parent_path)
244
 
 
 
187
    
245
188
    trees = {}
246
189
    while new_trees:
247
190
        items = new_trees.items()
248
191
        new_trees = {}
249
192
        for path, file_id in items:
250
 
            parent_id = tree.inventory[file_id].parent_id
 
193
            try:
 
194
                parent_id = tree.inventory[file_id].parent_id
 
195
            except errors.NoSuchId:
 
196
                # Directory was removed recursively perhaps ?
 
197
                continue
251
198
            if parent_id is not None:
252
199
                parent_path = urlutils.dirname(path)
253
200
                new_trees[parent_path] = parent_id
257
204
        try:
258
205
            return shamap[ie.file_id]
259
206
        except KeyError:
260
 
            # FIXME: Should be the same as in parent
261
207
            if ie.kind in ("file", "symlink"):
262
208
                try:
263
209
                    return idmap.lookup_blob_id(ie.file_id, ie.revision)
269
215
            elif ie.kind == "directory":
270
216
                # Not all cache backends store the tree information, 
271
217
                # calculate again from scratch
272
 
                ret = directory_to_tree(ie, ie_to_hexsha, unusual_modes,
273
 
                    dummy_file_name)
 
218
                ret = directory_to_tree(ie, ie_to_hexsha, unusual_modes)
274
219
                if ret is None:
275
220
                    return ret
276
221
                return ret.id
280
225
    for path in sorted(trees.keys(), reverse=True):
281
226
        ie = tree.inventory[trees[path]]
282
227
        assert ie.kind == "directory"
283
 
        obj = directory_to_tree(ie, ie_to_hexsha, unusual_modes,
284
 
            dummy_file_name)
 
228
        obj = directory_to_tree(ie, ie_to_hexsha, unusual_modes)
285
229
        if obj is not None:
286
230
            yield path, obj, ie
287
231
            shamap[ie.file_id] = obj.id
342
286
        self._update_sha_map()
343
287
        return iter(self._cache.idmap.sha1s())
344
288
 
345
 
    def _reconstruct_commit(self, rev, tree_sha, roundtrip, verifiers):
346
 
        """Reconstruct a Commit object.
347
 
 
348
 
        :param rev: Revision object
349
 
        :param tree_sha: SHA1 of the root tree object
350
 
        :param roundtrip: Whether or not to roundtrip bzr metadata
351
 
        :param verifiers: Verifiers for the commits
352
 
        :return: Commit object
353
 
        """
 
289
    def _reconstruct_commit(self, rev, tree_sha):
354
290
        def parent_lookup(revid):
355
291
            try:
356
292
                return self._lookup_revision_sha1(revid)
357
293
            except errors.NoSuchRevision:
 
294
                trace.warning("Ignoring ghost parent %s", revid)
358
295
                return None
359
 
        return self.mapping.export_commit(rev, tree_sha, parent_lookup,
360
 
            roundtrip, verifiers)
361
 
 
362
 
    def _create_fileid_map_blob(self, inv):
363
 
        # FIXME: This can probably be a lot more efficient, 
364
 
        # not all files necessarily have to be processed.
365
 
        file_ids = {}
366
 
        for (path, ie) in inv.iter_entries():
367
 
            if self.mapping.generate_file_id(path) != ie.file_id:
368
 
                file_ids[path] = ie.file_id
369
 
        return self.mapping.export_fileid_map(file_ids)
370
 
 
371
 
    def _revision_to_objects(self, rev, tree, roundtrip):
372
 
        """Convert a revision to a set of git objects.
373
 
 
374
 
        :param rev: Bazaar revision object
375
 
        :param tree: Bazaar revision tree
376
 
        :param roundtrip: Whether to roundtrip all Bazaar revision data
377
 
        """
 
296
        return self.mapping.export_commit(rev, tree_sha, parent_lookup)
 
297
 
 
298
    def _revision_to_objects(self, rev, tree):
378
299
        unusual_modes = extract_unusual_modes(rev)
379
300
        present_parents = self.repository.has_revisions(rev.parent_ids)
380
301
        parent_trees = self.tree_cache.revision_trees(
381
302
            [p for p in rev.parent_ids if p in present_parents])
382
 
        root_tree = None
 
303
        tree_sha = None
383
304
        for path, obj, ie in _tree_to_objects(tree, parent_trees,
384
 
                self._cache.idmap, unusual_modes, self.mapping.BZR_DUMMY_FILE):
 
305
                self._cache.idmap, unusual_modes):
 
306
            yield path, obj, ie
385
307
            if path == "":
386
 
                root_tree = obj
387
 
                root_ie = ie
388
 
                # Don't yield just yet
389
 
            else:
390
 
                yield path, obj, ie
391
 
        if root_tree is None:
 
308
                tree_sha = obj.id
 
309
        if tree_sha is None:
392
310
            # Pointless commit - get the tree sha elsewhere
393
311
            if not rev.parent_ids:
394
 
                root_tree = Tree()
 
312
                tree_sha = Tree().id
395
313
            else:
396
314
                base_sha1 = self._lookup_revision_sha1(rev.parent_ids[0])
397
 
                root_tree = self[self[base_sha1].tree]
398
 
            root_ie = tree.inventory.root
399
 
        if roundtrip and self.mapping.BZR_FILE_IDS_FILE is not None:
400
 
            b = self._create_fileid_map_blob(tree.inventory)
401
 
            if b is not None:
402
 
                root_tree[self.mapping.BZR_FILE_IDS_FILE] = ((stat.S_IFREG | 0644), b.id)
403
 
                yield self.mapping.BZR_FILE_IDS_FILE, b, None
404
 
        yield "", root_tree, root_ie
405
 
        if roundtrip:
406
 
            testament3 = StrictTestament3(rev, tree.inventory)
407
 
            verifiers = { "testament3-sha1": testament3.as_sha1() }
408
 
        else:
409
 
            verifiers = {}
410
 
        commit_obj = self._reconstruct_commit(rev, root_tree.id,
411
 
            roundtrip=roundtrip, verifiers=verifiers)
 
315
                tree_sha = self[base_sha1].tree
 
316
        commit_obj = self._reconstruct_commit(rev, tree_sha)
412
317
        try:
413
318
            foreign_revid, mapping = mapping_registry.parse_revision_id(
414
319
                rev.revision_id)
425
330
        rev = self.repository.get_revision(revid)
426
331
        tree = self.tree_cache.revision_tree(rev.revision_id)
427
332
        updater = self._get_updater(rev)
428
 
        for path, obj, ie in self._revision_to_objects(rev, tree,
429
 
            roundtrip=True):
430
 
            if isinstance(obj, Commit):
431
 
                testament3 = StrictTestament3(rev, tree.inventory)
432
 
                ie = { "testament3-sha1": testament3.as_sha1() }
433
 
            updater.add_object(obj, ie, path)
 
333
        for path, obj, ie in self._revision_to_objects(rev, tree):
 
334
            updater.add_object(obj, ie)
434
335
        commit_obj = updater.finish()
435
336
        return commit_obj.id
436
337
 
483
384
                        [(entry.file_id, entry.revision, None)]).next().id
484
385
            else:
485
386
                raise AssertionError("unknown entry kind '%s'" % entry.kind)
486
 
        tree = directory_to_tree(inv[fileid], get_ie_sha1, unusual_modes,
487
 
            self.mapping.BZR_DUMMY_FILE)
488
 
        if (inv.root.file_id == fileid and
489
 
            self.mapping.BZR_FILE_IDS_FILE is not None):
490
 
            b = self._create_fileid_map_blob(inv)
491
 
            # If this is the root tree, add the file ids
492
 
            tree[self.mapping.BZR_FILE_IDS_FILE] = ((stat.S_IFREG | 0644), b.id)
 
387
        tree = directory_to_tree(inv[fileid], get_ie_sha1, unusual_modes)
493
388
        _check_expected_sha(expected_sha, tree)
494
389
        return tree
495
390
 
503
398
 
504
399
    def _lookup_revision_sha1(self, revid):
505
400
        """Return the SHA1 matching a Bazaar revision."""
506
 
        from dulwich.protocol import ZERO_SHA
507
401
        if revid == NULL_REVISION:
508
 
            return ZERO_SHA
 
402
            return "0" * 40
509
403
        try:
510
404
            return self._cache.idmap.lookup_commit(revid)
511
405
        except KeyError:
512
406
            try:
513
407
                return mapping_registry.parse_revision_id(revid)[0]
514
408
            except errors.InvalidRevisionId:
515
 
                self.repository.lock_read()
516
 
                try:
517
 
                    self._update_sha_map(revid)
518
 
                finally:
519
 
                    self.repository.unlock()
 
409
                self._update_sha_map(revid)
520
410
                return self._cache.idmap.lookup_commit(revid)
521
411
 
522
412
    def get_raw(self, sha):
530
420
    def __contains__(self, sha):
531
421
        # See if sha is in map
532
422
        try:
533
 
            (type, type_data) = self.lookup_git_sha(sha)
 
423
            (type, type_data) = self._lookup_git_sha(sha)
534
424
            if type == "commit":
535
425
                return self.repository.has_revision(type_data[0])
536
426
            elif type == "blob":
537
 
                return self.repository.texts.has_key(type_data)
 
427
                return self.repository.texts.has_version(type_data)
538
428
            elif type == "tree":
539
429
                return self.repository.has_revision(type_data[1])
540
430
            else:
542
432
        except KeyError:
543
433
            return False
544
434
 
545
 
    def lookup_git_shas(self, shas, update_map=True):
546
 
        from dulwich.protocol import ZERO_SHA
547
 
        ret = {}
548
 
        for sha in shas:
549
 
            if sha == ZERO_SHA:
550
 
                ret[sha] = ("commit", (NULL_REVISION, None, {}))
551
 
                continue
552
 
            try:
553
 
                ret[sha] = self._cache.idmap.lookup_git_sha(sha)
554
 
            except KeyError:
555
 
                if update_map:
556
 
                    # if not, see if there are any unconverted revisions and add
557
 
                    # them to the map, search for sha in map again
558
 
                    self._update_sha_map()
559
 
                    update_map = False
560
 
                    try:
561
 
                        ret[sha] = self._cache.idmap.lookup_git_sha(sha)
562
 
                    except KeyError:
563
 
                        pass
564
 
        return ret
565
 
 
566
 
    def lookup_git_sha(self, sha, update_map=True):
567
 
        return self.lookup_git_shas([sha], update_map=update_map)[sha]
 
435
    def _lookup_git_sha(self, sha):
 
436
        # See if sha is in map
 
437
        try:
 
438
            return self._cache.idmap.lookup_git_sha(sha)
 
439
        except KeyError:
 
440
            # if not, see if there are any unconverted revisions and add them
 
441
            # to the map, search for sha in map again
 
442
            self._update_sha_map()
 
443
            return self._cache.idmap.lookup_git_sha(sha)
568
444
 
569
445
    def __getitem__(self, sha):
570
446
        if self._cache.content_cache is not None:
572
448
                return self._cache.content_cache[sha]
573
449
            except KeyError:
574
450
                pass
575
 
        (type, type_data) = self.lookup_git_sha(sha)
 
451
        (type, type_data) = self._lookup_git_sha(sha)
576
452
        # convert object to git object
577
453
        if type == "commit":
578
 
            (revid, tree_sha, verifiers) = type_data
 
454
            (revid, tree_sha) = type_data
579
455
            try:
580
456
                rev = self.repository.get_revision(revid)
581
457
            except errors.NoSuchRevision:
582
458
                trace.mutter('entry for %s %s in shamap: %r, but not found in '
583
459
                             'repository', type, sha, type_data)
584
460
                raise KeyError(sha)
585
 
            commit = self._reconstruct_commit(rev, tree_sha, roundtrip=True,
586
 
                verifiers=verifiers)
 
461
            commit = self._reconstruct_commit(rev, tree_sha)
587
462
            _check_expected_sha(sha, commit)
588
463
            return commit
589
464
        elif type == "blob":
606
481
        else:
607
482
            raise AssertionError("Unknown object type '%s'" % type)
608
483
 
609
 
    def generate_lossy_pack_contents(self, have, want, progress=None,
610
 
            get_tagged=None):
611
 
        return self.generate_pack_contents(have, want, progress, get_tagged,
612
 
            lossy=True)
613
 
 
614
 
    def generate_pack_contents(self, have, want, progress=None,
615
 
            get_tagged=None, lossy=False):
 
484
    def generate_pack_contents(self, have, want, progress=None, get_tagged=None):
616
485
        """Iterate over the contents of a pack file.
617
486
 
618
487
        :param have: List of SHA1s of objects that should not be sent
619
488
        :param want: List of SHA1s of objects that should be sent
620
489
        """
621
490
        processed = set()
622
 
        ret = self.lookup_git_shas(have + want)
623
491
        for commit_sha in have:
624
492
            try:
625
 
                (type, (revid, tree_sha)) = ret[commit_sha]
 
493
                (type, (revid, tree_sha)) = self._lookup_git_sha(commit_sha)
626
494
            except KeyError:
627
495
                pass
628
496
            else:
632
500
        for commit_sha in want:
633
501
            if commit_sha in have:
634
502
                continue
635
 
            try:
636
 
                (type, (revid, tree_sha)) = ret[commit_sha]
637
 
            except KeyError:
638
 
                pass
639
 
            else:
640
 
                assert type == "commit"
641
 
                pending.add(revid)
642
 
 
643
 
        todo = _find_missing_bzr_revids(self.repository.get_parent_map, 
644
 
                                        pending, processed)
 
503
            (type, (revid, tree_sha)) = self._lookup_git_sha(commit_sha)
 
504
            assert type == "commit"
 
505
            pending.add(revid)
 
506
        todo = set()
 
507
        while pending:
 
508
            processed.update(pending)
 
509
            next_map = self.repository.get_parent_map(pending)
 
510
            next_pending = set()
 
511
            for item in next_map.iteritems():
 
512
                todo.add(item[0])
 
513
                next_pending.update(p for p in item[1] if p not in processed)
 
514
            pending = next_pending
 
515
        if NULL_REVISION in todo:
 
516
            todo.remove(NULL_REVISION)
645
517
        trace.mutter('sending revisions %r', todo)
646
518
        ret = []
647
519
        pb = ui.ui_factory.nested_progress_bar()
650
522
                pb.update("generating git objects", i, len(todo))
651
523
                rev = self.repository.get_revision(revid)
652
524
                tree = self.tree_cache.revision_tree(revid)
653
 
                for path, obj, ie in self._revision_to_objects(rev, tree,
654
 
                    roundtrip=not lossy):
 
525
                for path, obj, ie in self._revision_to_objects(rev, tree):
655
526
                    ret.append((obj, path))
656
527
        finally:
657
528
            pb.finished()