/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

SupportĀ tagĀ refs.

Show diffs side-by-side

added added

removed removed

Lines of Context:
19
19
from dulwich.objects import (
20
20
    Blob,
21
21
    Tree,
 
22
    sha_to_hex,
22
23
    )
23
24
from dulwich.object_store import (
24
25
    BaseObjectStore,
25
 
    ObjectStoreIterator,
26
26
    )
27
 
import stat
28
27
 
29
28
from bzrlib import (
30
 
    debug,
31
29
    errors,
 
30
    lru_cache,
32
31
    trace,
33
32
    ui,
 
33
    urlutils,
34
34
    )
35
35
from bzrlib.revision import (
36
36
    NULL_REVISION,
37
37
    )
38
38
 
39
 
from bzrlib.plugins.git.errors import (
40
 
    GhostRevision,
41
 
    )
42
 
 
43
39
from bzrlib.plugins.git.mapping import (
44
40
    default_mapping,
45
41
    directory_to_tree,
46
42
    extract_unusual_modes,
47
43
    mapping_registry,
48
 
    revision_to_commit,
 
44
    symlink_to_blob,
49
45
    )
50
46
from bzrlib.plugins.git.shamap import (
51
 
    SqliteGitShaMap,
52
 
    TdbGitShaMap,
 
47
    from_repository as cache_from_repository,
53
48
    )
54
49
 
55
50
 
60
55
    return BazaarObjectStore(repo, mapping)
61
56
 
62
57
 
 
58
MAX_TREE_CACHE_SIZE = 50 * 1024 * 1024
 
59
 
 
60
 
 
61
class LRUTreeCache(object):
 
62
 
 
63
    def __init__(self, repository):
 
64
        def approx_tree_size(tree):
 
65
            # Very rough estimate, 1k per inventory entry
 
66
            return len(tree.inventory) * 1024
 
67
        self.repository = repository
 
68
        self._cache = lru_cache.LRUSizeCache(max_size=MAX_TREE_CACHE_SIZE,
 
69
            after_cleanup_size=None, compute_size=approx_tree_size)
 
70
 
 
71
    def revision_tree(self, revid):            
 
72
        try:
 
73
            return self._cache[revid] 
 
74
        except KeyError:
 
75
            tree = self.repository.revision_tree(revid)
 
76
            self.add(tree)
 
77
            return tree
 
78
 
 
79
    def iter_revision_trees(self, revids):
 
80
        trees = dict([(k, self._cache.get(k)) for k in revids]) 
 
81
        for tree in self.repository.revision_trees(
 
82
                [r for r, v in trees.iteritems() if v is None]):
 
83
            trees[tree.get_revision_id()] = tree
 
84
            self.add(tree)
 
85
        return (trees[r] for r in revids)
 
86
 
 
87
    def revision_trees(self, revids):
 
88
        return list(self.iter_revision_trees(revids))
 
89
 
 
90
    def add(self, tree):
 
91
        self._cache.add(tree.get_revision_id(), tree)
 
92
 
 
93
 
 
94
def _check_expected_sha(expected_sha, object):
 
95
    """Check whether an object matches an expected SHA.
 
96
 
 
97
    :param expected_sha: None or expected SHA as either binary or as hex digest
 
98
    :param object: Object to verify
 
99
    """
 
100
    if expected_sha is None:
 
101
        return
 
102
    if len(expected_sha) == 40:
 
103
        if expected_sha != object.sha().hexdigest():
 
104
            raise AssertionError("Invalid sha for %r: %s" % (object,
 
105
                expected_sha))
 
106
    elif len(expected_sha) == 20:
 
107
        if expected_sha != object.sha().digest():
 
108
            raise AssertionError("Invalid sha for %r: %s" % (object,
 
109
                sha_to_hex(expected_sha)))
 
110
    else:
 
111
        raise AssertionError("Unknown length %d for %r" % (len(expected_sha),
 
112
            expected_sha))
 
113
 
 
114
 
 
115
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes):
 
116
    """Iterate over the objects that were introduced in a revision.
 
117
 
 
118
    :param idmap: id map
 
119
    :param unusual_modes: Unusual file modes
 
120
    :return: Yields (path, object, ie) entries
 
121
    """
 
122
    new_trees = {}
 
123
    new_blobs = []
 
124
    shamap = {}
 
125
    try:
 
126
        base_tree = parent_trees[0]
 
127
        other_parent_trees = parent_trees[1:]
 
128
    except IndexError:
 
129
        base_tree = tree._repository.revision_tree(NULL_REVISION)
 
130
        other_parent_trees = []
 
131
    def find_unchanged_parent_ie(ie, parent_trees):
 
132
        assert ie.kind in ("symlink", "file")
 
133
        for ptree in parent_trees:
 
134
            try:
 
135
                pie = ptree.inventory[ie.file_id]
 
136
            except errors.NoSuchId:
 
137
                pass
 
138
            else:
 
139
                if (pie.text_sha1 == ie.text_sha1 and 
 
140
                    pie.kind == ie.kind and
 
141
                    pie.symlink_target == ie.symlink_target):
 
142
                    return pie
 
143
        raise KeyError
 
144
    for (file_id, path, changed_content, versioned, parent, name, kind,
 
145
         executable) in tree.iter_changes(base_tree):
 
146
        if kind[1] == "file":
 
147
            ie = tree.inventory[file_id]
 
148
            if changed_content:
 
149
                
 
150
                try:
 
151
                    pie = find_unchanged_parent_ie(ie, other_parent_trees)
 
152
                except KeyError:
 
153
                    pass
 
154
                else:
 
155
                    shamap[ie.file_id] = idmap.lookup_blob_id(
 
156
                        pie.file_id, pie.revision)
 
157
            if not file_id in shamap:
 
158
                new_blobs.append((path[1], ie))
 
159
            new_trees[urlutils.dirname(path[1])] = parent[1]
 
160
        elif kind[1] == "symlink":
 
161
            ie = tree.inventory[file_id]
 
162
            if changed_content:
 
163
                blob = symlink_to_blob(ie)
 
164
                shamap[file_id] = blob.id
 
165
                try:
 
166
                    find_unchanged_parent_ie(ie, other_parent_trees)
 
167
                except KeyError:
 
168
                    yield path[1], blob, ie
 
169
            new_trees[urlutils.dirname(path[1])] = parent[1]
 
170
        elif kind[1] not in (None, "directory"):
 
171
            raise AssertionError(kind[1])
 
172
        if path[0] is not None:
 
173
            new_trees[urlutils.dirname(path[0])] = parent[0]
 
174
    
 
175
    for (path, ie), chunks in tree.iter_files_bytes(
 
176
        [(ie.file_id, (path, ie)) for (path, ie) in new_blobs]):
 
177
        obj = Blob()
 
178
        obj.chunked = chunks
 
179
        yield path, obj, ie
 
180
        shamap[ie.file_id] = obj.id
 
181
 
 
182
    for fid in unusual_modes:
 
183
        new_trees[tree.id2path(fid)] = tree.inventory[fid].parent_id
 
184
    
 
185
    trees = {}
 
186
    while new_trees:
 
187
        items = new_trees.items()
 
188
        new_trees = {}
 
189
        for path, file_id in items:
 
190
            try:
 
191
                parent_id = tree.inventory[file_id].parent_id
 
192
            except errors.NoSuchId:
 
193
                # Directory was removed recursively perhaps ?
 
194
                continue
 
195
            if parent_id is not None:
 
196
                parent_path = urlutils.dirname(path)
 
197
                new_trees[parent_path] = parent_id
 
198
            trees[path] = file_id
 
199
 
 
200
    def ie_to_hexsha(ie):
 
201
        try:
 
202
            return shamap[ie.file_id]
 
203
        except KeyError:
 
204
            if ie.kind in ("file", "symlink"):
 
205
                try:
 
206
                    return idmap.lookup_blob_id(ie.file_id, ie.revision)
 
207
                except KeyError:
 
208
                    # no-change merge ?
 
209
                    blob = Blob()
 
210
                    blob.data = tree.get_file_text(ie.file_id)
 
211
                    return blob.id
 
212
            elif ie.kind == "directory":
 
213
                # Not all cache backends store the tree information, 
 
214
                # calculate again from scratch
 
215
                ret = directory_to_tree(ie, ie_to_hexsha, unusual_modes)
 
216
                if ret is None:
 
217
                    return ret
 
218
                return ret.id
 
219
            else:
 
220
                raise AssertionError
 
221
 
 
222
    for path in sorted(trees.keys(), reverse=True):
 
223
        ie = tree.inventory[trees[path]]
 
224
        assert ie.kind == "directory"
 
225
        obj = directory_to_tree(ie, ie_to_hexsha, unusual_modes)
 
226
        if obj is not None:
 
227
            yield path, obj, ie
 
228
            shamap[ie.file_id] = obj.id
 
229
 
 
230
 
63
231
class BazaarObjectStore(BaseObjectStore):
64
232
    """A Git-style object store backed onto a Bazaar repository."""
65
233
 
69
237
            self.mapping = default_mapping
70
238
        else:
71
239
            self.mapping = mapping
72
 
        try:
73
 
            self._idmap = TdbGitShaMap.from_repository(repository)
74
 
        except ImportError:
75
 
            self._idmap = SqliteGitShaMap.from_repository(repository)
 
240
        self._cache = cache_from_repository(repository)
 
241
        self._content_cache_types = ("tree")
 
242
        self.start_write_group = self._cache.idmap.start_write_group
 
243
        self.abort_write_group = self._cache.idmap.abort_write_group
 
244
        self.commit_write_group = self._cache.idmap.commit_write_group
 
245
        self.tree_cache = LRUTreeCache(self.repository)
76
246
 
77
247
    def _update_sha_map(self, stop_revision=None):
 
248
        graph = self.repository.get_graph()
78
249
        if stop_revision is None:
79
 
            all_revids = self.repository.all_revision_ids()
 
250
            heads = graph.heads(self.repository.all_revision_ids())
80
251
        else:
81
 
            all_revids = self.repository.get_ancestry(stop_revision)
82
 
            first = all_revids.pop(0) # Pop leading None
83
 
            assert first is None
84
 
        graph = self.repository.get_graph()
85
 
        present_revids = set(self._idmap.revids())
86
 
        missing_revids = [revid for revid in graph.iter_topo_order(all_revids) if revid not in present_revids]
87
 
        pb = ui.ui_factory.nested_progress_bar()
 
252
            heads = set([stop_revision])
 
253
        missing_revids = self._cache.idmap.missing_revisions(heads)
 
254
        while heads:
 
255
            parents = graph.get_parent_map(heads)
 
256
            todo = set()
 
257
            for p in parents.values():
 
258
                todo.update([x for x in p if x not in missing_revids])
 
259
            heads = self._cache.idmap.missing_revisions(todo)
 
260
            missing_revids.update(heads)
 
261
        if NULL_REVISION in missing_revids:
 
262
            missing_revids.remove(NULL_REVISION)
 
263
        missing_revids = self.repository.has_revisions(missing_revids)
 
264
        if not missing_revids:
 
265
            return
 
266
        self.start_write_group()
88
267
        try:
89
 
            for i, revid in enumerate(missing_revids):
90
 
                pb.update("updating git map", i, len(missing_revids))
91
 
                self._update_sha_map_revision(revid)
92
 
        finally:
93
 
            self._idmap.commit()
94
 
            pb.finished()
 
268
            pb = ui.ui_factory.nested_progress_bar()
 
269
            try:
 
270
                for i, revid in enumerate(graph.iter_topo_order(missing_revids)):
 
271
                    trace.mutter('processing %r', revid)
 
272
                    pb.update("updating git map", i, len(missing_revids))
 
273
                    self._update_sha_map_revision(revid)
 
274
            finally:
 
275
                pb.finished()
 
276
        except:
 
277
            self.abort_write_group()
 
278
            raise
 
279
        else:
 
280
            self.commit_write_group()
95
281
 
96
282
    def __iter__(self):
97
283
        self._update_sha_map()
98
 
        return iter(self._idmap.sha1s())
99
 
 
100
 
    def _update_sha_map_revision(self, revid):
101
 
        inv = self.repository.get_inventory(revid)
102
 
        rev = self.repository.get_revision(revid)
 
284
        return iter(self._cache.idmap.sha1s())
 
285
 
 
286
    def _reconstruct_commit(self, rev, tree_sha):
 
287
        def parent_lookup(revid):
 
288
            try:
 
289
                return self._lookup_revision_sha1(revid)
 
290
            except errors.NoSuchRevision:
 
291
                trace.warning("Ignoring ghost parent %s", revid)
 
292
                return None
 
293
        return self.mapping.export_commit(rev, tree_sha, parent_lookup)
 
294
 
 
295
    def _revision_to_objects(self, rev, tree):
103
296
        unusual_modes = extract_unusual_modes(rev)
104
 
        tree_sha = self._get_ie_sha1(inv.root, inv, unusual_modes)
105
 
        commit_obj = revision_to_commit(rev, tree_sha,
106
 
                                        self._idmap.lookup_commit)
 
297
        present_parents = self.repository.has_revisions(rev.parent_ids)
 
298
        parent_trees = self.tree_cache.revision_trees(
 
299
            [p for p in rev.parent_ids if p in present_parents])
 
300
        tree_sha = None
 
301
        for path, obj, ie in _tree_to_objects(tree, parent_trees,
 
302
                self._cache.idmap, unusual_modes):
 
303
            yield path, obj, ie
 
304
            if path == "":
 
305
                tree_sha = obj.id
 
306
        if tree_sha is None:
 
307
            # Pointless commit - get the tree sha elsewhere
 
308
            if not rev.parent_ids:
 
309
                tree_sha = Tree().id
 
310
            else:
 
311
                base_sha1 = self._lookup_revision_sha1(rev.parent_ids[0])
 
312
                tree_sha = self[base_sha1].tree
 
313
        commit_obj = self._reconstruct_commit(rev, tree_sha)
107
314
        try:
108
 
            foreign_revid, mapping = mapping_registry.parse_revision_id(revid)
 
315
            foreign_revid, mapping = mapping_registry.parse_revision_id(
 
316
                rev.revision_id)
109
317
        except errors.InvalidRevisionId:
110
318
            pass
111
319
        else:
112
 
            if foreign_revid != commit_obj.id:
113
 
                if not "fix-shamap" in debug.debug_flags:
114
 
                    raise AssertionError("recreated git commit had different sha1: expected %s, got %s" % (foreign_revid, commit_obj.id))
115
 
        self._idmap.add_entry(commit_obj.id, "commit", (revid, tree_sha))
116
 
 
117
 
    def _check_expected_sha(self, expected_sha, object):
118
 
        if expected_sha is None:
119
 
            return
120
 
        if expected_sha != object.id:
121
 
            raise AssertionError("Invalid sha for %r: %s" % (object, expected_sha))
122
 
 
123
 
    def _get_ie_object(self, entry, inv, unusual_modes):  
124
 
        if entry.kind == "directory":
125
 
            return self._get_tree(entry.file_id, inv.revision_id, inv, unusual_modes)
126
 
        else:
127
 
            return self._get_blob(entry.file_id, entry.revision)
128
 
 
129
 
    def _get_ie_object_or_sha1(self, entry, inv, unusual_modes):
130
 
        if entry.kind == "directory":
131
 
            try:
132
 
                return self._idmap.lookup_tree(entry.file_id, inv.revision_id), None
133
 
            except KeyError:
134
 
                ret = self._get_ie_object(entry, inv, unusual_modes)
135
 
                self._idmap.add_entry(ret.id, "tree", (entry.file_id, inv.revision_id))
136
 
                return ret.id, ret
137
 
        else:
138
 
            try:
139
 
                return self._idmap.lookup_blob(entry.file_id, entry.revision), None
140
 
            except KeyError:
141
 
                ret = self._get_ie_object(entry, inv, unusual_modes)
142
 
                self._idmap.add_entry(ret.id, "blob", (entry.file_id, entry.revision))
143
 
                return ret.id, ret
144
 
 
145
 
    def _get_ie_sha1(self, entry, inv, unusual_modes):
146
 
        return self._get_ie_object_or_sha1(entry, inv, unusual_modes)[0]
147
 
 
148
 
    def _get_blob(self, fileid, revision, expected_sha=None):
 
320
            _check_expected_sha(foreign_revid, commit_obj)
 
321
        yield None, commit_obj, None
 
322
 
 
323
    def _get_updater(self, rev):
 
324
        return self._cache.get_updater(rev)
 
325
 
 
326
    def _update_sha_map_revision(self, revid):
 
327
        rev = self.repository.get_revision(revid)
 
328
        tree = self.tree_cache.revision_tree(rev.revision_id)
 
329
        updater = self._get_updater(rev)
 
330
        for path, obj, ie in self._revision_to_objects(rev, tree):
 
331
            updater.add_object(obj, ie)
 
332
        commit_obj = updater.finish()
 
333
        return commit_obj.id
 
334
 
 
335
    def _reconstruct_blobs(self, keys):
149
336
        """Return a Git Blob object from a fileid and revision stored in bzr.
150
 
        
 
337
 
151
338
        :param fileid: File id of the text
152
339
        :param revision: Revision of the text
153
340
        """
154
 
        text = self.repository.texts.get_record_stream([(fileid, revision)],
155
 
            "unordered", True).next().get_bytes_as("fulltext")
156
 
        blob = Blob()
157
 
        blob._text = text
158
 
        self._check_expected_sha(expected_sha, blob)
159
 
        return blob
 
341
        stream = self.repository.iter_files_bytes(
 
342
            ((key[0], key[1], key) for key in keys))
 
343
        for (fileid, revision, expected_sha), chunks in stream:
 
344
            blob = Blob()
 
345
            blob.chunked = chunks
 
346
            if blob.id != expected_sha and blob.data == "":
 
347
                # Perhaps it's a symlink ?
 
348
                tree = self.tree_cache.revision_tree(revision)
 
349
                entry = tree.inventory[fileid]
 
350
                if entry.kind == 'symlink':
 
351
                    blob = symlink_to_blob(entry)
 
352
            _check_expected_sha(expected_sha, blob)
 
353
            yield blob
160
354
 
161
 
    def _get_tree(self, fileid, revid, inv, unusual_modes, expected_sha=None):
 
355
    def _reconstruct_tree(self, fileid, revid, inv, unusual_modes,
 
356
        expected_sha=None):
162
357
        """Return a Git Tree object from a file id and a revision stored in bzr.
163
358
 
164
359
        :param fileid: fileid in the tree.
165
360
        :param revision: Revision of the tree.
166
361
        """
167
 
        tree = directory_to_tree(inv[fileid], 
168
 
            lambda ie: self._get_ie_sha1(ie, inv, unusual_modes),
169
 
            unusual_modes)
170
 
        self._check_expected_sha(expected_sha, tree)
 
362
        def get_ie_sha1(entry):
 
363
            if entry.kind == "directory":
 
364
                try:
 
365
                    return self._cache.idmap.lookup_tree_id(entry.file_id,
 
366
                        revid)
 
367
                except (NotImplementedError, KeyError):
 
368
                    obj = self._reconstruct_tree(entry.file_id, revid, inv,
 
369
                        unusual_modes)
 
370
                    if obj is None:
 
371
                        return None
 
372
                    else:
 
373
                        return obj.id
 
374
            elif entry.kind in ("file", "symlink"):
 
375
                try:
 
376
                    return self._cache.idmap.lookup_blob_id(entry.file_id,
 
377
                        entry.revision)
 
378
                except KeyError:
 
379
                    # no-change merge?
 
380
                    return self._reconstruct_blobs(
 
381
                        [(entry.file_id, entry.revision, None)]).next().id
 
382
            else:
 
383
                raise AssertionError("unknown entry kind '%s'" % entry.kind)
 
384
        tree = directory_to_tree(inv[fileid], get_ie_sha1, unusual_modes)
 
385
        _check_expected_sha(expected_sha, tree)
171
386
        return tree
172
387
 
173
 
    def _get_commit(self, rev, tree_sha, expected_sha=None):
174
 
        try:
175
 
            commit = revision_to_commit(rev, tree_sha, self._lookup_revision_sha1)
176
 
        except errors.NoSuchRevision, e:
177
 
            raise GhostRevision(e.branch, e.revision)
178
 
        self._check_expected_sha(expected_sha, commit)
179
 
        return commit
180
 
 
181
388
    def get_parents(self, sha):
182
389
        """Retrieve the parents of a Git commit by SHA1.
183
390
 
191
398
        if revid == NULL_REVISION:
192
399
            return "0" * 40
193
400
        try:
194
 
            return self._idmap.lookup_commit(revid)
 
401
            return self._cache.idmap.lookup_commit(revid)
195
402
        except KeyError:
196
 
            self._update_sha_map(revid)
197
 
            return self._idmap.lookup_commit(revid)
 
403
            try:
 
404
                return mapping_registry.parse_revision_id(revid)[0]
 
405
            except errors.InvalidRevisionId:
 
406
                self._update_sha_map(revid)
 
407
                return self._cache.idmap.lookup_commit(revid)
198
408
 
199
409
    def get_raw(self, sha):
200
410
        """Get the raw representation of a Git object by SHA1.
201
411
 
202
412
        :param sha: SHA1 of the git object
203
413
        """
204
 
        return self[sha].as_raw_string()
 
414
        obj = self[sha]
 
415
        return (obj.type, obj.as_raw_string())
205
416
 
206
417
    def __contains__(self, sha):
207
418
        # See if sha is in map
208
419
        try:
209
 
            self._lookup_git_sha(sha)
 
420
            (type, type_data) = self._lookup_git_sha(sha)
 
421
            if type == "commit":
 
422
                return self.repository.has_revision(type_data[0])
 
423
            elif type == "blob":
 
424
                return self.repository.texts.has_version(type_data)
 
425
            elif type == "tree":
 
426
                return self.repository.has_revision(type_data[1])
 
427
            else:
 
428
                raise AssertionError("Unknown object type '%s'" % type)
210
429
        except KeyError:
211
430
            return False
212
 
        else:
213
 
            return True
214
431
 
215
432
    def _lookup_git_sha(self, sha):
216
433
        # See if sha is in map
217
434
        try:
218
 
            return self._idmap.lookup_git_sha(sha)
 
435
            return self._cache.idmap.lookup_git_sha(sha)
219
436
        except KeyError:
220
 
            # if not, see if there are any unconverted revisions and add them 
 
437
            # if not, see if there are any unconverted revisions and add them
221
438
            # to the map, search for sha in map again
222
439
            self._update_sha_map()
223
 
            return self._idmap.lookup_git_sha(sha)
 
440
            return self._cache.idmap.lookup_git_sha(sha)
224
441
 
225
442
    def __getitem__(self, sha):
 
443
        if self._cache.content_cache is not None:
 
444
            try:
 
445
                return self._cache.content_cache[sha]
 
446
            except KeyError:
 
447
                pass
226
448
        (type, type_data) = self._lookup_git_sha(sha)
227
449
        # convert object to git object
228
450
        if type == "commit":
 
451
            (revid, tree_sha) = type_data
229
452
            try:
230
 
                rev = self.repository.get_revision(type_data[0])
 
453
                rev = self.repository.get_revision(revid)
231
454
            except errors.NoSuchRevision:
232
 
                trace.mutter('entry for %s %s in shamap: %r, but not found in repository', type, sha, type_data)
 
455
                trace.mutter('entry for %s %s in shamap: %r, but not found in '
 
456
                             'repository', type, sha, type_data)
233
457
                raise KeyError(sha)
234
 
            return self._get_commit(rev, type_data[1], expected_sha=sha)
 
458
            commit = self._reconstruct_commit(rev, tree_sha)
 
459
            _check_expected_sha(sha, commit)
 
460
            return commit
235
461
        elif type == "blob":
236
 
            return self._get_blob(type_data[0], type_data[1], expected_sha=sha)
 
462
            (fileid, revision) = type_data
 
463
            return self._reconstruct_blobs([(fileid, revision, sha)]).next()
237
464
        elif type == "tree":
238
 
            inv = self.repository.get_inventory(type_data[1])
239
 
            rev = self.repository.get_revision(type_data[1])
 
465
            (fileid, revid) = type_data
 
466
            try:
 
467
                tree = self.tree_cache.revision_tree(revid)
 
468
                rev = self.repository.get_revision(revid)
 
469
            except errors.NoSuchRevision:
 
470
                trace.mutter('entry for %s %s in shamap: %r, but not found in repository', type, sha, type_data)
 
471
                raise KeyError(sha)
240
472
            unusual_modes = extract_unusual_modes(rev)
241
473
            try:
242
 
                return self._get_tree(type_data[0], type_data[1], inv, unusual_modes,
243
 
                                      expected_sha=sha)
 
474
                return self._reconstruct_tree(fileid, revid, tree.inventory,
 
475
                    unusual_modes, expected_sha=sha)
244
476
            except errors.NoSuchRevision:
245
477
                raise KeyError(sha)
246
478
        else:
247
479
            raise AssertionError("Unknown object type '%s'" % type)
 
480
 
 
481
    def generate_pack_contents(self, have, want, progress=None, get_tagged=None):
 
482
        """Iterate over the contents of a pack file.
 
483
 
 
484
        :param have: List of SHA1s of objects that should not be sent
 
485
        :param want: List of SHA1s of objects that should be sent
 
486
        """
 
487
        processed = set()
 
488
        for commit_sha in have:
 
489
            try:
 
490
                (type, (revid, tree_sha)) = self._lookup_git_sha(commit_sha)
 
491
            except KeyError:
 
492
                pass
 
493
            else:
 
494
                assert type == "commit"
 
495
                processed.add(revid)
 
496
        pending = set()
 
497
        for commit_sha in want:
 
498
            if commit_sha in have:
 
499
                continue
 
500
            (type, (revid, tree_sha)) = self._lookup_git_sha(commit_sha)
 
501
            assert type == "commit"
 
502
            pending.add(revid)
 
503
        todo = set()
 
504
        while pending:
 
505
            processed.update(pending)
 
506
            next_map = self.repository.get_parent_map(pending)
 
507
            next_pending = set()
 
508
            for item in next_map.iteritems():
 
509
                todo.add(item[0])
 
510
                next_pending.update(p for p in item[1] if p not in processed)
 
511
            pending = next_pending
 
512
        if NULL_REVISION in todo:
 
513
            todo.remove(NULL_REVISION)
 
514
        trace.mutter('sending revisions %r', todo)
 
515
        ret = []
 
516
        pb = ui.ui_factory.nested_progress_bar()
 
517
        try:
 
518
            for i, revid in enumerate(todo):
 
519
                pb.update("generating git objects", i, len(todo))
 
520
                rev = self.repository.get_revision(revid)
 
521
                tree = self.tree_cache.revision_tree(revid)
 
522
                for path, obj, ie in self._revision_to_objects(rev, tree):
 
523
                    ret.append((obj, path))
 
524
        finally:
 
525
            pb.finished()
 
526
        return ret
 
527
 
 
528
    def add_thin_pack(self):
 
529
        import tempfile
 
530
        import os
 
531
        fd, path = tempfile.mkstemp(suffix=".pack")
 
532
        f = os.fdopen(fd, 'wb')
 
533
        def commit():
 
534
            from dulwich.pack import PackData, Pack
 
535
            from bzrlib.plugins.git.fetch import import_git_objects
 
536
            os.fsync(fd)
 
537
            f.close()
 
538
            if os.path.getsize(path) == 0:
 
539
                return
 
540
            pd = PackData(path)
 
541
            pd.create_index_v2(path[:-5]+".idx", self.object_store.get_raw)
 
542
 
 
543
            p = Pack(path[:-5])
 
544
            self.repository.lock_write()
 
545
            try:
 
546
                self.repository.start_write_group()
 
547
                try:
 
548
                    import_git_objects(self.repository, self.mapping, 
 
549
                        p.iterobjects(get_raw=self.get_raw),
 
550
                        self.object_store)
 
551
                except:
 
552
                    self.repository.abort_write_group()
 
553
                    raise
 
554
                else:
 
555
                    self.repository.commit_write_group()
 
556
            finally:
 
557
                self.repository.unlock()
 
558
        return f, commit
 
559
 
 
560
    # The pack isn't kept around anyway, so no point 
 
561
    # in treating full packs different from thin packs
 
562
    add_pack = add_thin_pack