/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

Merge http support.

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
44
    SqliteGitShaMap,
69
62
            self.mapping = default_mapping
70
63
        else:
71
64
            self.mapping = mapping
72
 
        try:
73
 
            self._idmap = TdbGitShaMap.from_repository(repository)
74
 
        except ImportError:
75
 
            self._idmap = SqliteGitShaMap.from_repository(repository)
 
65
        self._idmap = SqliteGitShaMap.from_repository(repository)
76
66
 
77
67
    def _update_sha_map(self, stop_revision=None):
 
68
        graph = self.repository.get_graph()
78
69
        if stop_revision is None:
79
 
            all_revids = self.repository.all_revision_ids()
 
70
            heads = graph.heads(self.repository.all_revision_ids())
80
71
        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]
 
72
            heads = set([stop_revision])
 
73
        missing_revids = self._idmap.missing_revisions(heads)
 
74
        while heads:
 
75
            parents = graph.get_parent_map(heads)
 
76
            todo = set()
 
77
            for p in parents.values():
 
78
                todo.update([x for x in p if x not in missing_revids])
 
79
            heads = self._idmap.missing_revisions(todo)
 
80
            missing_revids.update(heads)
 
81
        if NULL_REVISION in missing_revids:
 
82
            missing_revids.remove(NULL_REVISION)
87
83
        pb = ui.ui_factory.nested_progress_bar()
88
84
        try:
89
 
            for i, revid in enumerate(missing_revids):
 
85
            for i, revid in enumerate(graph.iter_topo_order(missing_revids)):
90
86
                pb.update("updating git map", i, len(missing_revids))
91
87
                self._update_sha_map_revision(revid)
92
88
        finally:
93
 
            self._idmap.commit()
94
89
            pb.finished()
 
90
        self._idmap.commit_write_group()
95
91
 
96
92
    def __iter__(self):
97
93
        self._update_sha_map()
98
94
        return iter(self._idmap.sha1s())
99
95
 
 
96
    def _revision_to_commit(self, rev, tree_sha):
 
97
        def parent_lookup(revid):
 
98
            try:
 
99
                return self._lookup_revision_sha1(revid)
 
100
            except errors.NoSuchRevision:
 
101
                trace.warning("Ignoring ghost parent %s", revid)
 
102
                return None
 
103
        return self.mapping.export_commit(rev, tree_sha, parent_lookup)
 
104
 
100
105
    def _update_sha_map_revision(self, revid):
101
106
        inv = self.repository.get_inventory(revid)
102
107
        rev = self.repository.get_revision(revid)
103
108
        unusual_modes = extract_unusual_modes(rev)
104
109
        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)
 
110
        commit_obj = self._revision_to_commit(rev, tree_sha)
107
111
        try:
108
112
            foreign_revid, mapping = mapping_registry.parse_revision_id(revid)
109
113
        except errors.InvalidRevisionId:
117
121
    def _check_expected_sha(self, expected_sha, object):
118
122
        if expected_sha is None:
119
123
            return
120
 
        if expected_sha != object.id:
121
 
            raise AssertionError("Invalid sha for %r: %s" % (object, expected_sha))
 
124
        if len(expected_sha) == 40:
 
125
            if expected_sha != object.sha().hexdigest():
 
126
                raise AssertionError("Invalid sha for %r: %s" % (object, expected_sha))
 
127
        elif len(expected_sha) == 20:
 
128
            if expected_sha != object.sha().digest():
 
129
                raise AssertionError("Invalid sha for %r: %s" % (object, sha_to_hex(expected_sha)))
 
130
        else:
 
131
            raise AssertionError("Unknown length %d for %r" % (len(expected_sha), expected_sha))
122
132
 
123
 
    def _get_ie_object(self, entry, inv, unusual_modes):  
 
133
    def _get_ie_object(self, entry, inv, unusual_modes):
124
134
        if entry.kind == "directory":
125
135
            return self._get_tree(entry.file_id, inv.revision_id, inv, unusual_modes)
 
136
        elif entry.kind == "symlink":
 
137
            return self._get_blob_for_symlink(entry.symlink_target)
 
138
        elif entry.kind == "file":
 
139
            return self._get_blob_for_file(entry.file_id, entry.revision)
126
140
        else:
127
 
            return self._get_blob(entry.file_id, entry.revision)
 
141
            raise AssertionError("unknown entry kind '%s'" % entry.kind)
128
142
 
129
143
    def _get_ie_object_or_sha1(self, entry, inv, unusual_modes):
130
144
        if entry.kind == "directory":
132
146
                return self._idmap.lookup_tree(entry.file_id, inv.revision_id), None
133
147
            except KeyError:
134
148
                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:
 
149
                if ret is None:
 
150
                    hexsha = None
 
151
                else:
 
152
                    hexsha = ret.id
 
153
                self._idmap.add_entry(hexsha, "tree", (entry.file_id, inv.revision_id))
 
154
                return hexsha, ret
 
155
        elif entry.kind in ("file", "symlink"):
138
156
            try:
139
157
                return self._idmap.lookup_blob(entry.file_id, entry.revision), None
140
158
            except KeyError:
141
159
                ret = self._get_ie_object(entry, inv, unusual_modes)
142
160
                self._idmap.add_entry(ret.id, "blob", (entry.file_id, entry.revision))
143
161
                return ret.id, ret
 
162
        else:
 
163
            raise AssertionError("unknown entry kind '%s'" % entry.kind)
144
164
 
145
165
    def _get_ie_sha1(self, entry, inv, unusual_modes):
146
166
        return self._get_ie_object_or_sha1(entry, inv, unusual_modes)[0]
147
167
 
 
168
    def _get_blob_for_symlink(self, symlink_target, expected_sha=None):
 
169
        """Return a Git Blob object for symlink.
 
170
 
 
171
        :param symlink_target: target of symlink.
 
172
        """
 
173
        if type(symlink_target) == unicode:
 
174
            symlink_target = symlink_target.encode('utf-8')
 
175
        blob = Blob()
 
176
        blob._text = symlink_target
 
177
        self._check_expected_sha(expected_sha, blob)
 
178
        return blob
 
179
 
 
180
    def _get_blob_for_file(self, fileid, revision, expected_sha=None):
 
181
        """Return a Git Blob object from a fileid and revision stored in bzr.
 
182
 
 
183
        :param fileid: File id of the text
 
184
        :param revision: Revision of the text
 
185
        """
 
186
        blob = Blob()
 
187
        chunks = self.repository.iter_files_bytes([(fileid, revision, None)]).next()[1]
 
188
        blob._text = "".join(chunks)
 
189
        self._check_expected_sha(expected_sha, blob)
 
190
        return blob
 
191
 
148
192
    def _get_blob(self, fileid, revision, expected_sha=None):
149
193
        """Return a Git Blob object from a fileid and revision stored in bzr.
150
 
        
 
194
 
151
195
        :param fileid: File id of the text
152
196
        :param revision: Revision of the text
153
197
        """
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
 
198
        inv = self.repository.get_inventory(revision)
 
199
        entry = inv[fileid]
 
200
 
 
201
        if entry.kind == 'file':
 
202
            return self._get_blob_for_file(entry.file_id, entry.revision,
 
203
                                           expected_sha=expected_sha)
 
204
        elif entry.kind == 'symlink':
 
205
            return self._get_blob_for_symlink(entry.symlink_target,
 
206
                                              expected_sha=expected_sha)
 
207
        else:
 
208
            raise AssertionError
160
209
 
161
210
    def _get_tree(self, fileid, revid, inv, unusual_modes, expected_sha=None):
162
211
        """Return a Git Tree object from a file id and a revision stored in bzr.
164
213
        :param fileid: fileid in the tree.
165
214
        :param revision: Revision of the tree.
166
215
        """
167
 
        tree = directory_to_tree(inv[fileid], 
 
216
        tree = directory_to_tree(inv[fileid],
168
217
            lambda ie: self._get_ie_sha1(ie, inv, unusual_modes),
169
218
            unusual_modes)
170
219
        self._check_expected_sha(expected_sha, tree)
171
220
        return tree
172
221
 
173
222
    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)
 
223
        commit = self._revision_to_commit(rev, tree_sha)
178
224
        self._check_expected_sha(expected_sha, commit)
179
225
        return commit
180
226
 
193
239
        try:
194
240
            return self._idmap.lookup_commit(revid)
195
241
        except KeyError:
196
 
            self._update_sha_map(revid)
197
 
            return self._idmap.lookup_commit(revid)
 
242
            try:
 
243
                return mapping_registry.parse_revision_id(revid)[0]
 
244
            except errors.InvalidRevisionId:
 
245
                self._update_sha_map(revid)
 
246
                return self._idmap.lookup_commit(revid)
198
247
 
199
248
    def get_raw(self, sha):
200
249
        """Get the raw representation of a Git object by SHA1.
201
250
 
202
251
        :param sha: SHA1 of the git object
203
252
        """
204
 
        return self[sha].as_raw_string()
 
253
        obj = self[sha]
 
254
        return (obj.type, obj.as_raw_string())
205
255
 
206
256
    def __contains__(self, sha):
207
257
        # See if sha is in map
208
258
        try:
209
 
            self._lookup_git_sha(sha)
 
259
            (type, type_data) = self._lookup_git_sha(sha)
 
260
            if type == "commit":
 
261
                return self.repository.has_revision(type_data[0])
 
262
            elif type == "blob":
 
263
                return self.repository.texts.has_version(type_data)
 
264
            elif type == "tree":
 
265
                return self.repository.has_revision(type_data[1])
 
266
            else:
 
267
                raise AssertionError("Unknown object type '%s'" % type)
210
268
        except KeyError:
211
269
            return False
212
270
        else:
217
275
        try:
218
276
            return self._idmap.lookup_git_sha(sha)
219
277
        except KeyError:
220
 
            # if not, see if there are any unconverted revisions and add them 
 
278
            # if not, see if there are any unconverted revisions and add them
221
279
            # to the map, search for sha in map again
222
280
            self._update_sha_map()
223
281
            return self._idmap.lookup_git_sha(sha)
235
293
        elif type == "blob":
236
294
            return self._get_blob(type_data[0], type_data[1], expected_sha=sha)
237
295
        elif type == "tree":
238
 
            inv = self.repository.get_inventory(type_data[1])
239
 
            rev = self.repository.get_revision(type_data[1])
 
296
            try:
 
297
                inv = self.repository.get_inventory(type_data[1])
 
298
                rev = self.repository.get_revision(type_data[1])
 
299
            except errors.NoSuchRevision:
 
300
                trace.mutter('entry for %s %s in shamap: %r, but not found in repository', type, sha, type_data)
 
301
                raise KeyError(sha)
240
302
            unusual_modes = extract_unusual_modes(rev)
241
303
            try:
242
304
                return self._get_tree(type_data[0], type_data[1], inv, unusual_modes,