/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

Implement new git server backend API.

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 (
51
 
    from_repository as cache_from_repository,
 
46
from bzrlib.plugins.git.shamap import (
 
47
    from_repository as idmap_from_repository,
52
48
    )
53
49
 
54
 
import posixpath
55
 
import stat
56
 
 
57
50
 
58
51
def get_object_store(repo, mapping=None):
59
52
    git = getattr(repo, "_git", None)
62
55
    return BazaarObjectStore(repo, mapping)
63
56
 
64
57
 
65
 
MAX_TREE_CACHE_SIZE = 50 * 1024 * 1024
66
 
 
67
 
 
68
 
class LRUTreeCache(object):
 
58
MAX_INV_CACHE_SIZE = 50 * 1024 * 1024
 
59
 
 
60
 
 
61
class LRUInventoryCache(object):
69
62
 
70
63
    def __init__(self, repository):
71
 
        def approx_tree_size(tree):
 
64
        def approx_inv_size(inv):
72
65
            # Very rough estimate, 1k per inventory entry
73
 
            return len(tree.inventory) * 1024
 
66
            return len(inv) * 1024
74
67
        self.repository = repository
75
 
        self._cache = lru_cache.LRUSizeCache(max_size=MAX_TREE_CACHE_SIZE,
76
 
            after_cleanup_size=None, compute_size=approx_tree_size)
 
68
        self._cache = lru_cache.LRUSizeCache(max_size=MAX_INV_CACHE_SIZE,
 
69
            after_cleanup_size=None, compute_size=approx_inv_size)
77
70
 
78
 
    def revision_tree(self, revid):
 
71
    def get_inventory(self, revid):            
79
72
        try:
80
 
            tree = self._cache[revid]
 
73
            return self._cache[revid] 
81
74
        except KeyError:
82
 
            tree = self.repository.revision_tree(revid)
83
 
            self.add(tree)
84
 
        assert tree.get_revision_id() == tree.inventory.revision_id
85
 
        return tree
86
 
 
87
 
    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):
100
 
            trees[tree.get_revision_id()] = tree
101
 
            self.add(tree)
102
 
        return (trees[r] for r in revids)
103
 
 
104
 
    def revision_trees(self, revids):
105
 
        return list(self.iter_revision_trees(revids))
106
 
 
107
 
    def add(self, tree):
108
 
        self._cache.add(tree.get_revision_id(), tree)
109
 
 
110
 
 
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
 
75
            inv = self.repository.get_inventory(revid)
 
76
            self._cache.add(revid, inv)
 
77
            return inv
 
78
 
 
79
    def iter_inventories(self, revids):
 
80
        invs = dict([(k, self._cache.get(k)) for k in revids]) 
 
81
        for inv in self.repository.iter_inventories(
 
82
                [r for r, v in invs.iteritems() if v is None]):
 
83
            invs[inv.revision_id] = inv
 
84
            self._cache.add(inv.revision_id, inv)
 
85
        return (invs[r] for r in revids)
 
86
 
 
87
    def get_inventories(self, revids):
 
88
        return list(self.iter_inventories(revids))
 
89
 
 
90
    def add(self, revid, inv):
 
91
        self._cache.add(revid, inv)
136
92
 
137
93
 
138
94
def _check_expected_sha(expected_sha, object):
156
112
            expected_sha))
157
113
 
158
114
 
159
 
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes,
160
 
                     dummy_file_name=None):
 
115
def _inventory_to_objects(inv, parent_invs, parent_invshamaps,
 
116
        unusual_modes, iter_files_bytes, has_ghost_parents):
161
117
    """Iterate over the objects that were introduced in a revision.
162
118
 
163
 
    :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
168
 
    :return: Yields (path, object, ie) entries
 
119
    :param inv: Inventory to process
 
120
    :param parent_invs: parent inventory SHA maps
 
121
    :param parent_invshamaps: parent inventory SHA Map
 
122
    :param unusual_modes: Unusual file modes
 
123
    :param iter_files_bytes: Repository.iter_files_bytes-like callback
 
124
    :return: Yields (path, object) entries
169
125
    """
170
126
    new_trees = {}
171
127
    new_blobs = []
172
128
    shamap = {}
173
 
    try:
174
 
        base_tree = parent_trees[0]
175
 
        other_parent_trees = parent_trees[1:]
176
 
    except IndexError:
177
 
        base_tree = tree._repository.revision_tree(NULL_REVISION)
178
 
        other_parent_trees = []
179
 
    def find_unchanged_parent_ie(ie, parent_trees):
180
 
        assert ie.kind in ("symlink", "file")
181
 
        for ptree in parent_trees:
182
 
            try:
183
 
                pie = ptree.inventory[ie.file_id]
184
 
            except errors.NoSuchId:
185
 
                pass
186
 
            else:
187
 
                if (pie.text_sha1 == ie.text_sha1 and 
188
 
                    pie.kind == ie.kind and
189
 
                    pie.symlink_target == ie.symlink_target):
190
 
                    return pie
191
 
        raise KeyError
192
 
 
193
 
    # Find all the changed blobs
194
 
    for (file_id, path, changed_content, versioned, parent, name, kind,
195
 
         executable) in tree.iter_changes(base_tree):
196
 
        if kind[1] == "file":
197
 
            ie = tree.inventory[file_id]
198
 
            if changed_content:
199
 
                try:
200
 
                    pie = find_unchanged_parent_ie(ie, other_parent_trees)
201
 
                except KeyError:
202
 
                    pass
203
 
                else:
 
129
    for path, ie in inv.entries():
 
130
        if ie.kind == "file":
 
131
            if ie.revision != inv.revision_id:
 
132
                for (pinv, pinvshamap) in zip(parent_invs, parent_invshamaps):
204
133
                    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
212
 
            if not file_id in shamap:
213
 
                new_blobs.append((path[1], ie))
214
 
            new_trees[posixpath.dirname(path[1])] = parent[1]
215
 
        elif kind[1] == "symlink":
216
 
            ie = tree.inventory[file_id]
217
 
            if changed_content:
218
 
                blob = symlink_to_blob(ie)
219
 
                shamap[file_id] = blob.id
220
 
                try:
221
 
                    find_unchanged_parent_ie(ie, other_parent_trees)
222
 
                except KeyError:
223
 
                    yield path[1], blob, ie
224
 
            new_trees[posixpath.dirname(path[1])] = parent[1]
225
 
        elif kind[1] not in (None, "directory"):
226
 
            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
231
 
            new_trees[posixpath.dirname(path[0])] = parent[0]
 
134
                        pie = pinv[ie.file_id]
 
135
                    except errors.NoSuchId:
 
136
                        pass
 
137
                    else:
 
138
                        if (pie.text_sha1 == ie.text_sha1 and 
 
139
                            pie.kind == ie.kind):
 
140
                            shamap[ie.file_id] = pinvshamap.lookup_blob(
 
141
                                pie.file_id, pie.revision)
 
142
                            break
 
143
            if not ie.file_id in shamap:
 
144
                new_blobs.append((path, ie))
 
145
                new_trees[urlutils.dirname(path)] = ie.parent_id
 
146
        elif ie.kind == "symlink":
 
147
            blob = symlink_to_blob(ie)
 
148
            for pinv in parent_invs:
 
149
                try:
 
150
                    pie = pinv[ie.file_id]
 
151
                except errors.NoSuchId:
 
152
                    pass
 
153
                else:
 
154
                    if (ie.kind == pie.kind and
 
155
                        ie.symlink_target == pie.symlink_target):
 
156
                        break
 
157
            else:
 
158
                yield path, blob
 
159
                new_trees[urlutils.dirname(path)] = ie.parent_id
 
160
            shamap[ie.file_id] = blob.id
 
161
        elif ie.kind == "directory":
 
162
            for (pinv, pinvshamap) in zip(parent_invs, parent_invshamaps):
 
163
                try:
 
164
                    pie = pinv[ie.file_id]
 
165
                except errors.NoSuchId:
 
166
                    pass
 
167
                else:
 
168
                    if (pie.kind == ie.kind and 
 
169
                        pie.children.keys() == ie.children.keys()):
 
170
                        try:
 
171
                            shamap[ie.file_id] = pinvshamap.lookup_tree(
 
172
                                ie.file_id)
 
173
                        except (NotImplementedError, KeyError):
 
174
                            pass
 
175
                        else:
 
176
                            break
 
177
            else:
 
178
                new_trees[path] = ie.file_id
 
179
        else:
 
180
            raise AssertionError(ie.kind)
232
181
    
233
 
    # Fetch contents of the blobs that were changed
234
 
    for (path, ie), chunks in tree.iter_files_bytes(
235
 
        [(ie.file_id, (path, ie)) for (path, ie) in new_blobs]):
 
182
    for (path, fid), chunks in iter_files_bytes(
 
183
        [(ie.file_id, ie.revision, (path, ie.file_id))
 
184
            for (path, ie) in new_blobs]):
236
185
        obj = Blob()
237
 
        obj.chunked = chunks
238
 
        yield path, obj, ie
239
 
        shamap[ie.file_id] = obj.id
240
 
 
241
 
    for path in unusual_modes:
242
 
        parent_path = posixpath.dirname(path)
243
 
        new_trees[parent_path] = tree.path2id(parent_path)
244
 
 
 
186
        obj.data = "".join(chunks)
 
187
        yield path, obj
 
188
        shamap[fid] = obj.id
 
189
 
 
190
    for fid in unusual_modes:
 
191
        new_trees[inv.id2path(fid)] = inv[fid].parent_id
 
192
    
245
193
    trees = {}
246
194
    while new_trees:
247
195
        items = new_trees.items()
248
196
        new_trees = {}
249
197
        for path, file_id in items:
250
 
            parent_id = tree.inventory[file_id].parent_id
 
198
            parent_id = inv[file_id].parent_id
251
199
            if parent_id is not None:
252
200
                parent_path = urlutils.dirname(path)
253
201
                new_trees[parent_path] = parent_id
257
205
        try:
258
206
            return shamap[ie.file_id]
259
207
        except KeyError:
260
 
            # FIXME: Should be the same as in parent
261
 
            if ie.kind in ("file", "symlink"):
262
 
                try:
263
 
                    return idmap.lookup_blob_id(ie.file_id, ie.revision)
264
 
                except KeyError:
265
 
                    # no-change merge ?
266
 
                    blob = Blob()
267
 
                    blob.data = tree.get_file_text(ie.file_id)
268
 
                    return blob.id
269
 
            elif ie.kind == "directory":
270
 
                # Not all cache backends store the tree information, 
271
 
                # calculate again from scratch
272
 
                ret = directory_to_tree(ie, ie_to_hexsha, unusual_modes,
273
 
                    dummy_file_name)
274
 
                if ret is None:
275
 
                    return ret
276
 
                return ret.id
277
 
            else:
278
 
                raise AssertionError
 
208
            # Not all cache backends store the tree information, 
 
209
            # calculate again from scratch
 
210
            ret = directory_to_tree(ie, ie_to_hexsha, unusual_modes)
 
211
            if ret is None:
 
212
                return ret
 
213
            return ret.id
279
214
 
280
215
    for path in sorted(trees.keys(), reverse=True):
281
 
        ie = tree.inventory[trees[path]]
 
216
        ie = inv[trees[path]]
282
217
        assert ie.kind == "directory"
283
 
        obj = directory_to_tree(ie, ie_to_hexsha, unusual_modes,
284
 
            dummy_file_name)
 
218
        obj = directory_to_tree(ie, ie_to_hexsha, unusual_modes)
285
219
        if obj is not None:
286
 
            yield path, obj, ie
 
220
            yield path, obj
287
221
            shamap[ie.file_id] = obj.id
288
222
 
289
223
 
296
230
            self.mapping = default_mapping
297
231
        else:
298
232
            self.mapping = mapping
299
 
        self._cache = cache_from_repository(repository)
300
 
        self._content_cache_types = ("tree")
301
 
        self.start_write_group = self._cache.idmap.start_write_group
302
 
        self.abort_write_group = self._cache.idmap.abort_write_group
303
 
        self.commit_write_group = self._cache.idmap.commit_write_group
304
 
        self.tree_cache = LRUTreeCache(self.repository)
 
233
        self._idmap = idmap_from_repository(repository)
 
234
        self.start_write_group = self._idmap.start_write_group
 
235
        self.abort_write_group = self._idmap.abort_write_group
 
236
        self.commit_write_group = self._idmap.commit_write_group
 
237
        self.parent_invs_cache = LRUInventoryCache(self.repository)
305
238
 
306
239
    def _update_sha_map(self, stop_revision=None):
307
240
        graph = self.repository.get_graph()
309
242
            heads = graph.heads(self.repository.all_revision_ids())
310
243
        else:
311
244
            heads = set([stop_revision])
312
 
        missing_revids = self._cache.idmap.missing_revisions(heads)
 
245
        missing_revids = self._idmap.missing_revisions(heads)
313
246
        while heads:
314
247
            parents = graph.get_parent_map(heads)
315
248
            todo = set()
316
249
            for p in parents.values():
317
250
                todo.update([x for x in p if x not in missing_revids])
318
 
            heads = self._cache.idmap.missing_revisions(todo)
 
251
            heads = self._idmap.missing_revisions(todo)
319
252
            missing_revids.update(heads)
320
253
        if NULL_REVISION in missing_revids:
321
254
            missing_revids.remove(NULL_REVISION)
340
273
 
341
274
    def __iter__(self):
342
275
        self._update_sha_map()
343
 
        return iter(self._cache.idmap.sha1s())
344
 
 
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
 
        """
 
276
        return iter(self._idmap.sha1s())
 
277
 
 
278
    def _revision_to_commit(self, rev, tree_sha):
354
279
        def parent_lookup(revid):
355
280
            try:
356
281
                return self._lookup_revision_sha1(revid)
357
282
            except errors.NoSuchRevision:
 
283
                trace.warning("Ignoring ghost parent %s", revid)
358
284
                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
 
        """
 
285
        return self.mapping.export_commit(rev, tree_sha, parent_lookup)
 
286
 
 
287
    def _revision_to_objects(self, rev, inv):
378
288
        unusual_modes = extract_unusual_modes(rev)
379
289
        present_parents = self.repository.has_revisions(rev.parent_ids)
380
 
        parent_trees = self.tree_cache.revision_trees(
 
290
        has_ghost_parents = (len(rev.parent_ids) < len(present_parents))
 
291
        parent_invs = self.parent_invs_cache.get_inventories(
381
292
            [p for p in rev.parent_ids if p in present_parents])
382
 
        root_tree = None
383
 
        for path, obj, ie in _tree_to_objects(tree, parent_trees,
384
 
                self._cache.idmap, unusual_modes, self.mapping.BZR_DUMMY_FILE):
 
293
        parent_invshamaps = [self._idmap.get_inventory_sha_map(r) for r in rev.parent_ids if r in present_parents]
 
294
        tree_sha = None
 
295
        for path, obj in _inventory_to_objects(inv, parent_invs,
 
296
                parent_invshamaps, unusual_modes,
 
297
                self.repository.iter_files_bytes, has_ghost_parents):
 
298
            yield path, obj
385
299
            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:
392
 
            # Pointless commit - get the tree sha elsewhere
 
300
                tree_sha = obj.id
 
301
        if tree_sha is None:
393
302
            if not rev.parent_ids:
394
 
                root_tree = Tree()
 
303
                tree_sha = Tree().id
395
304
            else:
396
 
                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)
 
305
                tree_sha = parent_invshamaps[0][inv.root.file_id]
 
306
        commit_obj = self._revision_to_commit(rev, tree_sha)
412
307
        try:
413
 
            foreign_revid, mapping = mapping_registry.parse_revision_id(
414
 
                rev.revision_id)
 
308
            foreign_revid, mapping = mapping_registry.parse_revision_id(rev.revision_id)
415
309
        except errors.InvalidRevisionId:
416
310
            pass
417
311
        else:
418
312
            _check_expected_sha(foreign_revid, commit_obj)
419
 
        yield None, commit_obj, None
420
 
 
421
 
    def _get_updater(self, rev):
422
 
        return self._cache.get_updater(rev)
 
313
        yield None, commit_obj
423
314
 
424
315
    def _update_sha_map_revision(self, revid):
425
316
        rev = self.repository.get_revision(revid)
426
 
        tree = self.tree_cache.revision_tree(rev.revision_id)
427
 
        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)
434
 
        commit_obj = updater.finish()
 
317
        inv = self.parent_invs_cache.get_inventory(rev.revision_id)
 
318
        commit_obj = None
 
319
        entries = []
 
320
        for path, obj in self._revision_to_objects(rev, inv):
 
321
            if obj.type_name == "commit":
 
322
                commit_obj = obj
 
323
            elif obj.type_name in ("blob", "tree"):
 
324
                file_id = inv.path2id(path)
 
325
                ie = inv[file_id]
 
326
                if obj.type_name == "blob":
 
327
                    revision = ie.revision
 
328
                else:
 
329
                    revision = revid
 
330
                entries.append((file_id, obj.type_name, obj.id, revision))
 
331
            else:
 
332
                raise AssertionError
 
333
        self._idmap.add_entries(revid, rev.parent_ids, commit_obj.id, 
 
334
            commit_obj.tree, entries)
435
335
        return commit_obj.id
436
336
 
437
 
    def _reconstruct_blobs(self, keys):
 
337
    def _get_blob(self, fileid, revision, expected_sha):
438
338
        """Return a Git Blob object from a fileid and revision stored in bzr.
439
339
 
440
340
        :param fileid: File id of the text
441
341
        :param revision: Revision of the text
442
342
        """
443
 
        stream = self.repository.iter_files_bytes(
444
 
            ((key[0], key[1], key) for key in keys))
445
 
        for (fileid, revision, expected_sha), chunks in stream:
446
 
            blob = Blob()
447
 
            blob.chunked = chunks
448
 
            if blob.id != expected_sha and blob.data == "":
449
 
                # Perhaps it's a symlink ?
450
 
                tree = self.tree_cache.revision_tree(revision)
451
 
                entry = tree.inventory[fileid]
452
 
                if entry.kind == 'symlink':
453
 
                    blob = symlink_to_blob(entry)
454
 
            _check_expected_sha(expected_sha, blob)
455
 
            yield blob
 
343
        blob = Blob()
 
344
        chunks = self.repository.iter_files_bytes([(fileid, revision, None)]).next()[1]
 
345
        blob.data = "".join(chunks)
 
346
        if blob.id != expected_sha:
 
347
            # Perhaps it's a symlink ?
 
348
            inv = self.parent_invs_cache.get_inventory(revision)
 
349
            entry = inv[fileid]
 
350
            assert entry.kind == 'symlink'
 
351
            blob = symlink_to_blob(entry)
 
352
        _check_expected_sha(expected_sha, blob)
 
353
        return blob
456
354
 
457
 
    def _reconstruct_tree(self, fileid, revid, inv, unusual_modes,
458
 
        expected_sha=None):
 
355
    def _get_tree(self, fileid, revid, inv, unusual_modes, expected_sha=None):
459
356
        """Return a Git Tree object from a file id and a revision stored in bzr.
460
357
 
461
358
        :param fileid: fileid in the tree.
462
359
        :param revision: Revision of the tree.
463
360
        """
 
361
        invshamap = self._idmap.get_inventory_sha_map(inv.revision_id)
464
362
        def get_ie_sha1(entry):
465
363
            if entry.kind == "directory":
466
364
                try:
467
 
                    return self._cache.idmap.lookup_tree_id(entry.file_id,
468
 
                        revid)
 
365
                    return invshamap.lookup_tree(entry.file_id)
469
366
                except (NotImplementedError, KeyError):
470
 
                    obj = self._reconstruct_tree(entry.file_id, revid, inv,
 
367
                    obj = self._get_tree(entry.file_id, revid, inv,
471
368
                        unusual_modes)
472
369
                    if obj is None:
473
370
                        return None
474
371
                    else:
475
372
                        return obj.id
476
373
            elif entry.kind in ("file", "symlink"):
477
 
                try:
478
 
                    return self._cache.idmap.lookup_blob_id(entry.file_id,
479
 
                        entry.revision)
480
 
                except KeyError:
481
 
                    # no-change merge?
482
 
                    return self._reconstruct_blobs(
483
 
                        [(entry.file_id, entry.revision, None)]).next().id
 
374
                return invshamap.lookup_blob(entry.file_id, entry.revision)
484
375
            else:
485
376
                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)
 
377
        tree = directory_to_tree(inv[fileid], get_ie_sha1, unusual_modes)
493
378
        _check_expected_sha(expected_sha, tree)
494
379
        return tree
495
380
 
503
388
 
504
389
    def _lookup_revision_sha1(self, revid):
505
390
        """Return the SHA1 matching a Bazaar revision."""
506
 
        from dulwich.protocol import ZERO_SHA
507
391
        if revid == NULL_REVISION:
508
 
            return ZERO_SHA
 
392
            return "0" * 40
509
393
        try:
510
 
            return self._cache.idmap.lookup_commit(revid)
 
394
            return self._idmap.lookup_commit(revid)
511
395
        except KeyError:
512
396
            try:
513
397
                return mapping_registry.parse_revision_id(revid)[0]
514
398
            except errors.InvalidRevisionId:
515
 
                self.repository.lock_read()
516
 
                try:
517
 
                    self._update_sha_map(revid)
518
 
                finally:
519
 
                    self.repository.unlock()
520
 
                return self._cache.idmap.lookup_commit(revid)
 
399
                self._update_sha_map(revid)
 
400
                return self._idmap.lookup_commit(revid)
521
401
 
522
402
    def get_raw(self, sha):
523
403
        """Get the raw representation of a Git object by SHA1.
530
410
    def __contains__(self, sha):
531
411
        # See if sha is in map
532
412
        try:
533
 
            (type, type_data) = self.lookup_git_sha(sha)
 
413
            (type, type_data) = self._lookup_git_sha(sha)
534
414
            if type == "commit":
535
415
                return self.repository.has_revision(type_data[0])
536
416
            elif type == "blob":
537
 
                return self.repository.texts.has_key(type_data)
 
417
                return self.repository.texts.has_version(type_data)
538
418
            elif type == "tree":
539
419
                return self.repository.has_revision(type_data[1])
540
420
            else:
542
422
        except KeyError:
543
423
            return False
544
424
 
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]
 
425
    def _lookup_git_sha(self, sha):
 
426
        # See if sha is in map
 
427
        try:
 
428
            return self._idmap.lookup_git_sha(sha)
 
429
        except KeyError:
 
430
            # if not, see if there are any unconverted revisions and add them
 
431
            # to the map, search for sha in map again
 
432
            self._update_sha_map()
 
433
            return self._idmap.lookup_git_sha(sha)
568
434
 
569
435
    def __getitem__(self, sha):
570
 
        if self._cache.content_cache is not None:
571
 
            try:
572
 
                return self._cache.content_cache[sha]
573
 
            except KeyError:
574
 
                pass
575
 
        (type, type_data) = self.lookup_git_sha(sha)
 
436
        (type, type_data) = self._lookup_git_sha(sha)
576
437
        # convert object to git object
577
438
        if type == "commit":
578
 
            (revid, tree_sha, verifiers) = type_data
 
439
            (revid, tree_sha) = type_data
579
440
            try:
580
441
                rev = self.repository.get_revision(revid)
581
442
            except errors.NoSuchRevision:
582
 
                trace.mutter('entry for %s %s in shamap: %r, but not found in '
583
 
                             'repository', type, sha, type_data)
 
443
                trace.mutter('entry for %s %s in shamap: %r, but not found in repository', type, sha, type_data)
584
444
                raise KeyError(sha)
585
 
            commit = self._reconstruct_commit(rev, tree_sha, roundtrip=True,
586
 
                verifiers=verifiers)
 
445
            commit = self._revision_to_commit(rev, tree_sha)
587
446
            _check_expected_sha(sha, commit)
588
447
            return commit
589
448
        elif type == "blob":
590
449
            (fileid, revision) = type_data
591
 
            return self._reconstruct_blobs([(fileid, revision, sha)]).next()
 
450
            return self._get_blob(fileid, revision, expected_sha=sha)
592
451
        elif type == "tree":
593
452
            (fileid, revid) = type_data
594
453
            try:
595
 
                tree = self.tree_cache.revision_tree(revid)
 
454
                inv = self.parent_invs_cache.get_inventory(revid)
596
455
                rev = self.repository.get_revision(revid)
597
456
            except errors.NoSuchRevision:
598
457
                trace.mutter('entry for %s %s in shamap: %r, but not found in repository', type, sha, type_data)
599
458
                raise KeyError(sha)
600
459
            unusual_modes = extract_unusual_modes(rev)
601
460
            try:
602
 
                return self._reconstruct_tree(fileid, revid, tree.inventory,
 
461
                return self._get_tree(fileid, revid, inv,
603
462
                    unusual_modes, expected_sha=sha)
604
463
            except errors.NoSuchRevision:
605
464
                raise KeyError(sha)
606
465
        else:
607
466
            raise AssertionError("Unknown object type '%s'" % type)
608
467
 
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):
 
468
    def generate_pack_contents(self, have, want):
616
469
        """Iterate over the contents of a pack file.
617
470
 
618
471
        :param have: List of SHA1s of objects that should not be sent
619
472
        :param want: List of SHA1s of objects that should be sent
620
473
        """
621
474
        processed = set()
622
 
        ret = self.lookup_git_shas(have + want)
623
475
        for commit_sha in have:
624
476
            try:
625
 
                (type, (revid, tree_sha)) = ret[commit_sha]
 
477
                (type, (revid, tree_sha)) = self._lookup_git_sha(commit_sha)
626
478
            except KeyError:
627
479
                pass
628
480
            else:
632
484
        for commit_sha in want:
633
485
            if commit_sha in have:
634
486
                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)
 
487
            (type, (revid, tree_sha)) = self._lookup_git_sha(commit_sha)
 
488
            assert type == "commit"
 
489
            pending.add(revid)
 
490
        todo = set()
 
491
        while pending:
 
492
            processed.update(pending)
 
493
            next_map = self.repository.get_parent_map(pending)
 
494
            next_pending = set()
 
495
            for item in next_map.iteritems():
 
496
                todo.add(item[0])
 
497
                next_pending.update(p for p in item[1] if p not in processed)
 
498
            pending = next_pending
 
499
        if NULL_REVISION in todo:
 
500
            todo.remove(NULL_REVISION)
645
501
        trace.mutter('sending revisions %r', todo)
646
502
        ret = []
647
503
        pb = ui.ui_factory.nested_progress_bar()
649
505
            for i, revid in enumerate(todo):
650
506
                pb.update("generating git objects", i, len(todo))
651
507
                rev = self.repository.get_revision(revid)
652
 
                tree = self.tree_cache.revision_tree(revid)
653
 
                for path, obj, ie in self._revision_to_objects(rev, tree,
654
 
                    roundtrip=not lossy):
 
508
                inv = self.parent_invs_cache.get_inventory(revid)
 
509
                for path, obj in self._revision_to_objects(rev, inv):
655
510
                    ret.append((obj, path))
656
511
        finally:
657
512
            pb.finished()
658
513
        return ret
659
 
 
660
 
    def add_thin_pack(self):
661
 
        import tempfile
662
 
        import os
663
 
        fd, path = tempfile.mkstemp(suffix=".pack")
664
 
        f = os.fdopen(fd, 'wb')
665
 
        def commit():
666
 
            from dulwich.pack import PackData, Pack
667
 
            from bzrlib.plugins.git.fetch import import_git_objects
668
 
            os.fsync(fd)
669
 
            f.close()
670
 
            if os.path.getsize(path) == 0:
671
 
                return
672
 
            pd = PackData(path)
673
 
            pd.create_index_v2(path[:-5]+".idx", self.object_store.get_raw)
674
 
 
675
 
            p = Pack(path[:-5])
676
 
            self.repository.lock_write()
677
 
            try:
678
 
                self.repository.start_write_group()
679
 
                try:
680
 
                    import_git_objects(self.repository, self.mapping, 
681
 
                        p.iterobjects(get_raw=self.get_raw),
682
 
                        self.object_store)
683
 
                except:
684
 
                    self.repository.abort_write_group()
685
 
                    raise
686
 
                else:
687
 
                    self.repository.commit_write_group()
688
 
            finally:
689
 
                self.repository.unlock()
690
 
        return f, commit
691
 
 
692
 
    # The pack isn't kept around anyway, so no point 
693
 
    # in treating full packs different from thin packs
694
 
    add_pack = add_thin_pack