/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

  • Committer: Jelmer Vernooij
  • Date: 2010-05-13 12:34:24 UTC
  • mto: (0.200.912 trunk)
  • mto: This revision was merged to the branch mainline in revision 6960.
  • Revision ID: jelmer@samba.org-20100513123424-c1sk9vcg2ekrcsol
Some refactoring, support proper file ids in revision deltas.

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
 
75
71
        self._cache = lru_cache.LRUSizeCache(max_size=MAX_TREE_CACHE_SIZE,
76
72
            after_cleanup_size=None, compute_size=approx_tree_size)
77
73
 
78
 
    def revision_tree(self, revid):
 
74
    def revision_tree(self, revid):            
79
75
        try:
80
 
            tree = self._cache[revid]
 
76
            return self._cache[revid] 
81
77
        except KeyError:
82
78
            tree = self.repository.revision_tree(revid)
83
79
            self.add(tree)
84
 
        assert tree.get_revision_id() == tree.inventory.revision_id
85
 
        return tree
 
80
            return tree
86
81
 
87
82
    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):
 
83
        trees = dict([(k, self._cache.get(k)) for k in revids]) 
 
84
        for tree in self.repository.revision_trees(
 
85
                [r for r, v in trees.iteritems() if v is None]):
100
86
            trees[tree.get_revision_id()] = tree
101
87
            self.add(tree)
102
88
        return (trees[r] for r in revids)
156
142
            expected_sha))
157
143
 
158
144
 
159
 
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes,
160
 
                     dummy_file_name=None):
 
145
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes, dummy_file_name=None):
161
146
    """Iterate over the objects that were introduced in a revision.
162
147
 
163
148
    :param idmap: id map
164
 
    :param parent_trees: Parent revision trees
165
 
    :param unusual_modes: Unusual file modes dictionary
 
149
    :param unusual_modes: Unusual file modes
166
150
    :param dummy_file_name: File name to use for dummy files
167
151
        in empty directories. None to skip empty directories
168
152
    :return: Yields (path, object, ie) entries
189
173
                    pie.symlink_target == ie.symlink_target):
190
174
                    return pie
191
175
        raise KeyError
192
 
 
193
 
    # Find all the changed blobs
194
176
    for (file_id, path, changed_content, versioned, parent, name, kind,
195
177
         executable) in tree.iter_changes(base_tree):
196
178
        if kind[1] == "file":
224
206
            new_trees[posixpath.dirname(path[1])] = parent[1]
225
207
        elif kind[1] not in (None, "directory"):
226
208
            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
 
209
        if path[0] is not None:
231
210
            new_trees[posixpath.dirname(path[0])] = parent[0]
232
211
    
233
 
    # Fetch contents of the blobs that were changed
234
212
    for (path, ie), chunks in tree.iter_files_bytes(
235
213
        [(ie.file_id, (path, ie)) for (path, ie) in new_blobs]):
236
214
        obj = Blob()
241
219
    for path in unusual_modes:
242
220
        parent_path = posixpath.dirname(path)
243
221
        new_trees[parent_path] = tree.path2id(parent_path)
244
 
 
 
222
    
245
223
    trees = {}
246
224
    while new_trees:
247
225
        items = new_trees.items()
248
226
        new_trees = {}
249
227
        for path, file_id in items:
250
 
            parent_id = tree.inventory[file_id].parent_id
 
228
            try:
 
229
                parent_id = tree.inventory[file_id].parent_id
 
230
            except errors.NoSuchId:
 
231
                # Directory was removed recursively perhaps ?
 
232
                continue
251
233
            if parent_id is not None:
252
234
                parent_path = urlutils.dirname(path)
253
235
                new_trees[parent_path] = parent_id
342
324
        self._update_sha_map()
343
325
        return iter(self._cache.idmap.sha1s())
344
326
 
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
 
        """
 
327
    def _reconstruct_commit(self, rev, tree_sha, roundtrip):
354
328
        def parent_lookup(revid):
355
329
            try:
356
330
                return self._lookup_revision_sha1(revid)
357
331
            except errors.NoSuchRevision:
358
332
                return None
359
333
        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)
 
334
            roundtrip)
370
335
 
371
336
    def _revision_to_objects(self, rev, tree, roundtrip):
372
337
        """Convert a revision to a set of git objects.
396
361
                base_sha1 = self._lookup_revision_sha1(rev.parent_ids[0])
397
362
                root_tree = self[self[base_sha1].tree]
398
363
            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)
 
364
        if roundtrip:
 
365
            # FIXME: This can probably be a lot more efficient, 
 
366
            # not all files necessarily have to be processed.
 
367
            file_ids = {}
 
368
            for (path, ie) in tree.inventory.iter_entries():
 
369
                if self.mapping.generate_file_id(path) != ie.file_id:
 
370
                    file_ids[path] = ie.file_id
 
371
            b = self.mapping.export_fileid_map(file_ids)
401
372
            if b is not None:
402
373
                root_tree[self.mapping.BZR_FILE_IDS_FILE] = ((stat.S_IFREG | 0644), b.id)
403
374
                yield self.mapping.BZR_FILE_IDS_FILE, b, None
404
375
        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
376
        commit_obj = self._reconstruct_commit(rev, root_tree.id,
411
 
            roundtrip=roundtrip, verifiers=verifiers)
 
377
            roundtrip=roundtrip)
412
378
        try:
413
379
            foreign_revid, mapping = mapping_registry.parse_revision_id(
414
380
                rev.revision_id)
427
393
        updater = self._get_updater(rev)
428
394
        for path, obj, ie in self._revision_to_objects(rev, tree,
429
395
            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)
 
396
            updater.add_object(obj, ie)
434
397
        commit_obj = updater.finish()
435
398
        return commit_obj.id
436
399
 
485
448
                raise AssertionError("unknown entry kind '%s'" % entry.kind)
486
449
        tree = directory_to_tree(inv[fileid], get_ie_sha1, unusual_modes,
487
450
            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)
493
451
        _check_expected_sha(expected_sha, tree)
494
452
        return tree
495
453
 
512
470
            try:
513
471
                return mapping_registry.parse_revision_id(revid)[0]
514
472
            except errors.InvalidRevisionId:
515
 
                self.repository.lock_read()
516
 
                try:
517
 
                    self._update_sha_map(revid)
518
 
                finally:
519
 
                    self.repository.unlock()
 
473
                self._update_sha_map(revid)
520
474
                return self._cache.idmap.lookup_commit(revid)
521
475
 
522
476
    def get_raw(self, sha):
534
488
            if type == "commit":
535
489
                return self.repository.has_revision(type_data[0])
536
490
            elif type == "blob":
537
 
                return self.repository.texts.has_key(type_data)
 
491
                return self.repository.texts.has_version(type_data)
538
492
            elif type == "tree":
539
493
                return self.repository.has_revision(type_data[1])
540
494
            else:
543
497
            return False
544
498
 
545
499
    def lookup_git_shas(self, shas, update_map=True):
546
 
        from dulwich.protocol import ZERO_SHA
547
500
        ret = {}
548
501
        for sha in shas:
549
 
            if sha == ZERO_SHA:
550
 
                ret[sha] = ("commit", (NULL_REVISION, None, {}))
551
 
                continue
552
502
            try:
553
503
                ret[sha] = self._cache.idmap.lookup_git_sha(sha)
554
504
            except KeyError:
575
525
        (type, type_data) = self.lookup_git_sha(sha)
576
526
        # convert object to git object
577
527
        if type == "commit":
578
 
            (revid, tree_sha, verifiers) = type_data
 
528
            (revid, tree_sha) = type_data
579
529
            try:
580
530
                rev = self.repository.get_revision(revid)
581
531
            except errors.NoSuchRevision:
582
532
                trace.mutter('entry for %s %s in shamap: %r, but not found in '
583
533
                             'repository', type, sha, type_data)
584
534
                raise KeyError(sha)
585
 
            commit = self._reconstruct_commit(rev, tree_sha, roundtrip=True,
586
 
                verifiers=verifiers)
 
535
            commit = self._reconstruct_commit(rev, tree_sha, roundtrip=True)
587
536
            _check_expected_sha(sha, commit)
588
537
            return commit
589
538
        elif type == "blob":