/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 tests for find_missing_bzr_revids.

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()
 
96
def _find_missing_bzr_revids(get_parent_map, pending, processed):
 
97
    pending -= processed
122
98
    todo = set()
123
99
    while pending:
124
100
        processed.update(pending)
125
101
        next_map = get_parent_map(pending)
126
102
        next_pending = set()
127
103
        for item in next_map.iteritems():
128
 
            if item[0] in have:
129
 
                continue
130
104
            todo.add(item[0])
131
105
            next_pending.update(p for p in item[1] if p not in processed)
132
106
        pending = next_pending
156
130
            expected_sha))
157
131
 
158
132
 
159
 
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes,
160
 
                     dummy_file_name=None):
 
133
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes):
161
134
    """Iterate over the objects that were introduced in a revision.
162
135
 
163
136
    :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
 
137
    :param unusual_modes: Unusual file modes
168
138
    :return: Yields (path, object, ie) entries
169
139
    """
170
140
    new_trees = {}
189
159
                    pie.symlink_target == ie.symlink_target):
190
160
                    return pie
191
161
        raise KeyError
192
 
 
193
 
    # Find all the changed blobs
194
162
    for (file_id, path, changed_content, versioned, parent, name, kind,
195
163
         executable) in tree.iter_changes(base_tree):
196
164
        if kind[1] == "file":
197
165
            ie = tree.inventory[file_id]
198
166
            if changed_content:
 
167
                
199
168
                try:
200
169
                    pie = find_unchanged_parent_ie(ie, other_parent_trees)
201
170
                except KeyError:
202
171
                    pass
203
172
                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
 
173
                    shamap[ie.file_id] = idmap.lookup_blob_id(
 
174
                        pie.file_id, pie.revision)
212
175
            if not file_id in shamap:
213
176
                new_blobs.append((path[1], ie))
214
177
            new_trees[posixpath.dirname(path[1])] = parent[1]
224
187
            new_trees[posixpath.dirname(path[1])] = parent[1]
225
188
        elif kind[1] not in (None, "directory"):
226
189
            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
 
190
        if path[0] is not None:
231
191
            new_trees[posixpath.dirname(path[0])] = parent[0]
232
192
    
233
 
    # Fetch contents of the blobs that were changed
234
193
    for (path, ie), chunks in tree.iter_files_bytes(
235
194
        [(ie.file_id, (path, ie)) for (path, ie) in new_blobs]):
236
195
        obj = Blob()
241
200
    for path in unusual_modes:
242
201
        parent_path = posixpath.dirname(path)
243
202
        new_trees[parent_path] = tree.path2id(parent_path)
244
 
 
 
203
    
245
204
    trees = {}
246
205
    while new_trees:
247
206
        items = new_trees.items()
248
207
        new_trees = {}
249
208
        for path, file_id in items:
250
 
            parent_id = tree.inventory[file_id].parent_id
 
209
            try:
 
210
                parent_id = tree.inventory[file_id].parent_id
 
211
            except errors.NoSuchId:
 
212
                # Directory was removed recursively perhaps ?
 
213
                continue
251
214
            if parent_id is not None:
252
215
                parent_path = urlutils.dirname(path)
253
216
                new_trees[parent_path] = parent_id
269
232
            elif ie.kind == "directory":
270
233
                # Not all cache backends store the tree information, 
271
234
                # calculate again from scratch
272
 
                ret = directory_to_tree(ie, ie_to_hexsha, unusual_modes,
273
 
                    dummy_file_name)
 
235
                ret = directory_to_tree(ie, ie_to_hexsha, unusual_modes)
274
236
                if ret is None:
275
237
                    return ret
276
238
                return ret.id
280
242
    for path in sorted(trees.keys(), reverse=True):
281
243
        ie = tree.inventory[trees[path]]
282
244
        assert ie.kind == "directory"
283
 
        obj = directory_to_tree(ie, ie_to_hexsha, unusual_modes,
284
 
            dummy_file_name)
 
245
        obj = directory_to_tree(ie, ie_to_hexsha, unusual_modes)
285
246
        if obj is not None:
286
247
            yield path, obj, ie
287
248
            shamap[ie.file_id] = obj.id
342
303
        self._update_sha_map()
343
304
        return iter(self._cache.idmap.sha1s())
344
305
 
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
 
        """
 
306
    def _reconstruct_commit(self, rev, tree_sha):
354
307
        def parent_lookup(revid):
355
308
            try:
356
309
                return self._lookup_revision_sha1(revid)
357
310
            except errors.NoSuchRevision:
 
311
                trace.warning("Ignoring ghost parent %s", revid)
358
312
                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
 
        """
 
313
        return self.mapping.export_commit(rev, tree_sha, parent_lookup)
 
314
 
 
315
    def _revision_to_objects(self, rev, tree):
378
316
        unusual_modes = extract_unusual_modes(rev)
379
317
        present_parents = self.repository.has_revisions(rev.parent_ids)
380
318
        parent_trees = self.tree_cache.revision_trees(
381
319
            [p for p in rev.parent_ids if p in present_parents])
382
 
        root_tree = None
 
320
        tree_sha = None
383
321
        for path, obj, ie in _tree_to_objects(tree, parent_trees,
384
 
                self._cache.idmap, unusual_modes, self.mapping.BZR_DUMMY_FILE):
 
322
                self._cache.idmap, unusual_modes):
 
323
            yield path, obj, ie
385
324
            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:
 
325
                tree_sha = obj.id
 
326
        if tree_sha is None:
392
327
            # Pointless commit - get the tree sha elsewhere
393
328
            if not rev.parent_ids:
394
 
                root_tree = Tree()
 
329
                tree_sha = Tree().id
395
330
            else:
396
331
                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)
 
332
                tree_sha = self[base_sha1].tree
 
333
        commit_obj = self._reconstruct_commit(rev, tree_sha)
412
334
        try:
413
335
            foreign_revid, mapping = mapping_registry.parse_revision_id(
414
336
                rev.revision_id)
425
347
        rev = self.repository.get_revision(revid)
426
348
        tree = self.tree_cache.revision_tree(rev.revision_id)
427
349
        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)
 
350
        for path, obj, ie in self._revision_to_objects(rev, tree):
 
351
            updater.add_object(obj, ie)
434
352
        commit_obj = updater.finish()
435
353
        return commit_obj.id
436
354
 
483
401
                        [(entry.file_id, entry.revision, None)]).next().id
484
402
            else:
485
403
                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)
 
404
        tree = directory_to_tree(inv[fileid], get_ie_sha1, unusual_modes)
493
405
        _check_expected_sha(expected_sha, tree)
494
406
        return tree
495
407
 
512
424
            try:
513
425
                return mapping_registry.parse_revision_id(revid)[0]
514
426
            except errors.InvalidRevisionId:
515
 
                self.repository.lock_read()
516
 
                try:
517
 
                    self._update_sha_map(revid)
518
 
                finally:
519
 
                    self.repository.unlock()
 
427
                self._update_sha_map(revid)
520
428
                return self._cache.idmap.lookup_commit(revid)
521
429
 
522
430
    def get_raw(self, sha):
534
442
            if type == "commit":
535
443
                return self.repository.has_revision(type_data[0])
536
444
            elif type == "blob":
537
 
                return self.repository.texts.has_key(type_data)
 
445
                return self.repository.texts.has_version(type_data)
538
446
            elif type == "tree":
539
447
                return self.repository.has_revision(type_data[1])
540
448
            else:
543
451
            return False
544
452
 
545
453
    def lookup_git_shas(self, shas, update_map=True):
546
 
        from dulwich.protocol import ZERO_SHA
547
454
        ret = {}
548
455
        for sha in shas:
549
 
            if sha == ZERO_SHA:
550
 
                ret[sha] = ("commit", (NULL_REVISION, None, {}))
551
 
                continue
552
456
            try:
553
457
                ret[sha] = self._cache.idmap.lookup_git_sha(sha)
554
458
            except KeyError:
575
479
        (type, type_data) = self.lookup_git_sha(sha)
576
480
        # convert object to git object
577
481
        if type == "commit":
578
 
            (revid, tree_sha, verifiers) = type_data
 
482
            (revid, tree_sha) = type_data
579
483
            try:
580
484
                rev = self.repository.get_revision(revid)
581
485
            except errors.NoSuchRevision:
582
486
                trace.mutter('entry for %s %s in shamap: %r, but not found in '
583
487
                             'repository', type, sha, type_data)
584
488
                raise KeyError(sha)
585
 
            commit = self._reconstruct_commit(rev, tree_sha, roundtrip=True,
586
 
                verifiers=verifiers)
 
489
            commit = self._reconstruct_commit(rev, tree_sha)
587
490
            _check_expected_sha(sha, commit)
588
491
            return commit
589
492
        elif type == "blob":
606
509
        else:
607
510
            raise AssertionError("Unknown object type '%s'" % type)
608
511
 
609
 
    def generate_lossy_pack_contents(self, have, want, progress=None,
 
512
    def generate_pack_contents(self, have, want, progress=None,
610
513
            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):
616
514
        """Iterate over the contents of a pack file.
617
515
 
618
516
        :param have: List of SHA1s of objects that should not be sent
650
548
                pb.update("generating git objects", i, len(todo))
651
549
                rev = self.repository.get_revision(revid)
652
550
                tree = self.tree_cache.revision_tree(revid)
653
 
                for path, obj, ie in self._revision_to_objects(rev, tree,
654
 
                    roundtrip=not lossy):
 
551
                for path, obj, ie in self._revision_to_objects(rev, tree):
655
552
                    ret.append((obj, path))
656
553
        finally:
657
554
            pb.finished()