/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 docstring.

Show diffs side-by-side

added added

removed removed

Lines of Context:
18
18
 
19
19
from dulwich.objects import (
20
20
    Blob,
21
 
    Tree,
 
21
    sha_to_hex,
22
22
    )
23
23
from dulwich.object_store import (
24
24
    BaseObjectStore,
25
 
    ObjectStoreIterator,
26
25
    )
27
 
import stat
28
26
 
29
27
from bzrlib import (
30
28
    debug,
36
34
    NULL_REVISION,
37
35
    )
38
36
 
39
 
from bzrlib.plugins.git.errors import (
40
 
    GhostRevision,
41
 
    )
42
 
 
43
37
from bzrlib.plugins.git.mapping import (
44
38
    default_mapping,
45
39
    directory_to_tree,
46
40
    extract_unusual_modes,
47
41
    mapping_registry,
48
 
    revision_to_commit,
49
42
    )
50
43
from bzrlib.plugins.git.shamap import (
51
 
    SqliteGitShaMap,
52
 
    TdbGitShaMap,
 
44
    from_repository as idmap_from_repository,
53
45
    )
54
46
 
55
47
 
69
61
            self.mapping = default_mapping
70
62
        else:
71
63
            self.mapping = mapping
72
 
        try:
73
 
            self._idmap = TdbGitShaMap.from_repository(repository)
74
 
        except ImportError:
75
 
            self._idmap = SqliteGitShaMap.from_repository(repository)
 
64
        self._idmap = idmap_from_repository(repository)
 
65
        self.start_write_group = self._idmap.start_write_group
 
66
        self.abort_write_group = self._idmap.abort_write_group
 
67
        self.commit_write_group = self._idmap.commit_write_group
76
68
 
77
69
    def _update_sha_map(self, stop_revision=None):
 
70
        graph = self.repository.get_graph()
78
71
        if stop_revision is None:
79
 
            all_revids = self.repository.all_revision_ids()
 
72
            heads = graph.heads(self.repository.all_revision_ids())
80
73
        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()
 
74
            heads = set([stop_revision])
 
75
        missing_revids = self._idmap.missing_revisions(heads)
 
76
        while heads:
 
77
            parents = graph.get_parent_map(heads)
 
78
            todo = set()
 
79
            for p in parents.values():
 
80
                todo.update([x for x in p if x not in missing_revids])
 
81
            heads = self._idmap.missing_revisions(todo)
 
82
            missing_revids.update(heads)
 
83
        if NULL_REVISION in missing_revids:
 
84
            missing_revids.remove(NULL_REVISION)
 
85
        missing_revids = self.repository.has_revisions(missing_revids)
 
86
        if not missing_revids:
 
87
            return
 
88
        self.start_write_group()
88
89
        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()
 
90
            pb = ui.ui_factory.nested_progress_bar()
 
91
            try:
 
92
                for i, revid in enumerate(graph.iter_topo_order(missing_revids)):
 
93
                    trace.mutter('processing %r', revid)
 
94
                    pb.update("updating git map", i, len(missing_revids))
 
95
                    self._update_sha_map_revision(revid)
 
96
            finally:
 
97
                pb.finished()
 
98
        except:
 
99
            self.abort_write_group()
 
100
            raise
 
101
        else:
 
102
            self.commit_write_group()
95
103
 
96
104
    def __iter__(self):
97
105
        self._update_sha_map()
98
106
        return iter(self._idmap.sha1s())
99
107
 
 
108
    def _revision_to_commit(self, rev, tree_sha):
 
109
        def parent_lookup(revid):
 
110
            try:
 
111
                return self._lookup_revision_sha1(revid)
 
112
            except errors.NoSuchRevision:
 
113
                trace.warning("Ignoring ghost parent %s", revid)
 
114
                return None
 
115
        return self.mapping.export_commit(rev, tree_sha, parent_lookup)
 
116
 
 
117
    def _inventory_to_objects(self, inv, invshamap, unusual_modes):
 
118
        tree_sha = self._get_ie_sha1(inv.root, inv, invshamap, unusual_modes)
 
119
        return tree_sha, []
 
120
 
100
121
    def _update_sha_map_revision(self, revid):
101
122
        inv = self.repository.get_inventory(revid)
102
123
        rev = self.repository.get_revision(revid)
103
124
        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)
 
125
        invshamap = self._idmap.get_inventory_sha_map(revid)
 
126
        tree_sha, entries = self._inventory_to_objects(inv, invshamap,
 
127
            unusual_modes)
 
128
        commit_obj = self._revision_to_commit(rev, tree_sha)
107
129
        try:
108
130
            foreign_revid, mapping = mapping_registry.parse_revision_id(revid)
109
131
        except errors.InvalidRevisionId:
112
134
            if foreign_revid != commit_obj.id:
113
135
                if not "fix-shamap" in debug.debug_flags:
114
136
                    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))
 
137
        self._idmap.add_entries(revid, rev.parent_ids, commit_obj.id, 
 
138
            tree_sha, entries)
116
139
 
117
140
    def _check_expected_sha(self, expected_sha, object):
118
141
        if expected_sha is None:
119
142
            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]
 
143
        if len(expected_sha) == 40:
 
144
            if expected_sha != object.sha().hexdigest():
 
145
                raise AssertionError("Invalid sha for %r: %s" % (object, expected_sha))
 
146
        elif len(expected_sha) == 20:
 
147
            if expected_sha != object.sha().digest():
 
148
                raise AssertionError("Invalid sha for %r: %s" % (object, sha_to_hex(expected_sha)))
 
149
        else:
 
150
            raise AssertionError("Unknown length %d for %r" % (len(expected_sha), expected_sha))
 
151
 
 
152
    def _get_ie_object(self, entry, inv, unusual_modes):
 
153
        if entry.kind == "directory":
 
154
            return self._get_tree(entry.file_id, inv.revision_id, inv,
 
155
                unusual_modes)
 
156
        elif entry.kind == "symlink":
 
157
            return self._get_blob_for_symlink(entry.symlink_target)
 
158
        elif entry.kind == "file":
 
159
            return self._get_blob_for_file(entry.file_id, entry.revision)
 
160
        else:
 
161
            raise AssertionError("unknown entry kind '%s'" % entry.kind)
 
162
 
 
163
    def _get_ie_object_or_sha1(self, entry, inv, invshamap, unusual_modes):
 
164
        if entry.kind == "directory":
 
165
            try:
 
166
                return invshamap.lookup_tree(entry.file_id), None
 
167
            except (KeyError, NotImplementedError):
 
168
                ret = self._get_ie_object(entry, inv, unusual_modes)
 
169
                if ret is None:
 
170
                    # Empty directory
 
171
                    hexsha = None
 
172
                else:
 
173
                    hexsha = ret.id
 
174
                self._idmap._add_entry(hexsha, "tree",
 
175
                    (entry.file_id, inv.revision_id))
 
176
                return hexsha, ret
 
177
        elif entry.kind in ("file", "symlink"):
 
178
            try:
 
179
                return invshamap.lookup_blob(entry.file_id, entry.revision), None
 
180
            except KeyError:
 
181
                ret = self._get_ie_object(entry, inv, unusual_modes)
 
182
                self._idmap._add_entry(ret.id, "blob", (entry.file_id,
 
183
                    entry.revision))
 
184
                return ret.id, ret
 
185
        else:
 
186
            raise AssertionError("unknown entry kind '%s'" % entry.kind)
 
187
 
 
188
    def _get_ie_sha1(self, entry, inv, invshamap, unusual_modes):
 
189
        return self._get_ie_object_or_sha1(entry, inv, invshamap, unusual_modes)[0]
 
190
 
 
191
    def _get_blob_for_symlink(self, symlink_target, expected_sha=None):
 
192
        """Return a Git Blob object for symlink.
 
193
 
 
194
        :param symlink_target: target of symlink.
 
195
        """
 
196
        if type(symlink_target) == unicode:
 
197
            symlink_target = symlink_target.encode('utf-8')
 
198
        blob = Blob()
 
199
        blob._text = symlink_target
 
200
        self._check_expected_sha(expected_sha, blob)
 
201
        return blob
 
202
 
 
203
    def _get_blob_for_file(self, fileid, revision, expected_sha=None):
 
204
        """Return a Git Blob object from a fileid and revision stored in bzr.
 
205
 
 
206
        :param fileid: File id of the text
 
207
        :param revision: Revision of the text
 
208
        """
 
209
        blob = Blob()
 
210
        chunks = self.repository.iter_files_bytes([(fileid, revision, None)]).next()[1]
 
211
        blob._text = "".join(chunks)
 
212
        self._check_expected_sha(expected_sha, blob)
 
213
        return blob
147
214
 
148
215
    def _get_blob(self, fileid, revision, expected_sha=None):
149
216
        """Return a Git Blob object from a fileid and revision stored in bzr.
150
 
        
 
217
 
151
218
        :param fileid: File id of the text
152
219
        :param revision: Revision of the text
153
220
        """
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
 
221
        inv = self.repository.get_inventory(revision)
 
222
        entry = inv[fileid]
 
223
 
 
224
        if entry.kind == 'file':
 
225
            return self._get_blob_for_file(entry.file_id, entry.revision,
 
226
                                           expected_sha=expected_sha)
 
227
        elif entry.kind == 'symlink':
 
228
            return self._get_blob_for_symlink(entry.symlink_target,
 
229
                                              expected_sha=expected_sha)
 
230
        else:
 
231
            raise AssertionError
160
232
 
161
233
    def _get_tree(self, fileid, revid, inv, unusual_modes, expected_sha=None):
162
234
        """Return a Git Tree object from a file id and a revision stored in bzr.
164
236
        :param fileid: fileid in the tree.
165
237
        :param revision: Revision of the tree.
166
238
        """
167
 
        tree = directory_to_tree(inv[fileid], 
168
 
            lambda ie: self._get_ie_sha1(ie, inv, unusual_modes),
 
239
        invshamap = self._idmap.get_inventory_sha_map(inv.revision_id)
 
240
        tree = directory_to_tree(inv[fileid],
 
241
            lambda ie: self._get_ie_sha1(ie, inv, invshamap, unusual_modes),
169
242
            unusual_modes)
170
243
        self._check_expected_sha(expected_sha, tree)
171
244
        return tree
172
245
 
173
246
    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)
 
247
        commit = self._revision_to_commit(rev, tree_sha)
178
248
        self._check_expected_sha(expected_sha, commit)
179
249
        return commit
180
250
 
193
263
        try:
194
264
            return self._idmap.lookup_commit(revid)
195
265
        except KeyError:
196
 
            self._update_sha_map(revid)
197
 
            return self._idmap.lookup_commit(revid)
 
266
            try:
 
267
                return mapping_registry.parse_revision_id(revid)[0]
 
268
            except errors.InvalidRevisionId:
 
269
                self._update_sha_map(revid)
 
270
                return self._idmap.lookup_commit(revid)
198
271
 
199
272
    def get_raw(self, sha):
200
273
        """Get the raw representation of a Git object by SHA1.
201
274
 
202
275
        :param sha: SHA1 of the git object
203
276
        """
204
 
        return self[sha].as_raw_string()
 
277
        obj = self[sha]
 
278
        return (obj.type, obj.as_raw_string())
205
279
 
206
280
    def __contains__(self, sha):
207
281
        # See if sha is in map
208
282
        try:
209
 
            self._lookup_git_sha(sha)
 
283
            (type, type_data) = self._lookup_git_sha(sha)
 
284
            if type == "commit":
 
285
                return self.repository.has_revision(type_data[0])
 
286
            elif type == "blob":
 
287
                return self.repository.texts.has_version(type_data)
 
288
            elif type == "tree":
 
289
                return self.repository.has_revision(type_data[1])
 
290
            else:
 
291
                raise AssertionError("Unknown object type '%s'" % type)
210
292
        except KeyError:
211
293
            return False
212
 
        else:
213
 
            return True
214
294
 
215
295
    def _lookup_git_sha(self, sha):
216
296
        # See if sha is in map
217
297
        try:
218
298
            return self._idmap.lookup_git_sha(sha)
219
299
        except KeyError:
220
 
            # if not, see if there are any unconverted revisions and add them 
 
300
            # if not, see if there are any unconverted revisions and add them
221
301
            # to the map, search for sha in map again
222
302
            self._update_sha_map()
223
303
            return self._idmap.lookup_git_sha(sha)
235
315
        elif type == "blob":
236
316
            return self._get_blob(type_data[0], type_data[1], expected_sha=sha)
237
317
        elif type == "tree":
238
 
            inv = self.repository.get_inventory(type_data[1])
239
 
            rev = self.repository.get_revision(type_data[1])
 
318
            try:
 
319
                inv = self.repository.get_inventory(type_data[1])
 
320
                rev = self.repository.get_revision(type_data[1])
 
321
            except errors.NoSuchRevision:
 
322
                trace.mutter('entry for %s %s in shamap: %r, but not found in repository', type, sha, type_data)
 
323
                raise KeyError(sha)
240
324
            unusual_modes = extract_unusual_modes(rev)
241
325
            try:
242
 
                return self._get_tree(type_data[0], type_data[1], inv, unusual_modes,
243
 
                                      expected_sha=sha)
 
326
                return self._get_tree(type_data[0], type_data[1], inv,
 
327
                    unusual_modes, expected_sha=sha)
244
328
            except errors.NoSuchRevision:
245
329
                raise KeyError(sha)
246
330
        else: