/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 repository.py

Add FOSDEM roundtripping notes.

Show diffs side-by-side

added added

removed removed

Lines of Context:
repository.py
16
16
 
17
17
"""An adapter between a Git Repository and a Bazaar Branch"""
18
18
 
19
 
import git
20
19
import os
21
20
import time
22
21
 
23
22
import bzrlib
24
23
from bzrlib import (
25
 
    deprecated_graph,
26
24
    errors,
 
25
    graph,
27
26
    inventory,
28
27
    osutils,
29
28
    repository,
30
29
    revision,
31
30
    revisiontree,
 
31
    ui,
32
32
    urlutils,
33
 
    versionedfile,
34
33
    )
 
34
from bzrlib.foreign import (
 
35
        ForeignRepository,
 
36
        )
 
37
from bzrlib.trace import mutter
35
38
from bzrlib.transport import get_transport
36
39
 
37
40
from bzrlib.plugins.git.foreign import (
38
 
    versionedfiles
 
41
    versionedfiles,
39
42
    )
40
 
from bzrlib.plugins.git.mapping import default_mapping
41
 
 
42
 
 
43
 
class GitRepository(repository.Repository):
 
43
from bzrlib.plugins.git.mapping import default_mapping, mapping_registry, inventory_to_tree_and_blobs, revision_to_commit
 
44
from bzrlib.plugins.git.versionedfiles import GitTexts
 
45
 
 
46
import dulwich as git
 
47
 
 
48
 
 
49
class GitTags(object):
 
50
 
 
51
    def __init__(self, tags):
 
52
        self._tags = tags
 
53
 
 
54
    def __iter__(self):
 
55
        return iter(self._tags)
 
56
 
 
57
 
 
58
class GitRepository(ForeignRepository):
44
59
    """An adapter to git repositories for bzr."""
45
60
 
46
61
    _serializer = None
47
62
 
48
63
    def __init__(self, gitdir, lockfiles):
 
64
        ForeignRepository.__init__(self, GitFormat(), gitdir, lockfiles)
 
65
        from bzrlib.plugins.git import fetch
 
66
        repository.InterRepository.register_optimiser(fetch.InterGitRepository)
 
67
        repository.InterRepository.register_optimiser(fetch.InterGitNonGitRepository)
 
68
 
 
69
    def is_shared(self):
 
70
        return True
 
71
 
 
72
    def supports_rich_root(self):
 
73
        return True
 
74
 
 
75
    def _warn_if_deprecated(self):
 
76
        # This class isn't deprecated
 
77
        pass
 
78
 
 
79
    def get_mapping(self):
 
80
        return default_mapping
 
81
 
 
82
    def make_working_trees(self):
 
83
        return True
 
84
 
 
85
 
 
86
class LocalGitRepository(GitRepository):
 
87
 
 
88
    def __init__(self, gitdir, lockfiles):
 
89
        # FIXME: This also caches negatives. Need to be more careful 
 
90
        # about this once we start writing to git
 
91
        self._parents_provider = graph.CachingParentsProvider(self)
 
92
        GitRepository.__init__(self, gitdir, lockfiles)
49
93
        self.base = gitdir.root_transport.base
50
 
        self.bzrdir = gitdir
51
 
        self.control_files = lockfiles
52
94
        self._git = gitdir._git
53
95
        self.texts = None
54
96
        self.signatures = versionedfiles.VirtualSignatureTexts(self)
55
97
        self.revisions = versionedfiles.VirtualRevisionTexts(self)
56
 
        self._format = GitFormat()
57
 
        self._fallback_repositories = []
 
98
        self.inventories = versionedfiles.VirtualInventoryTexts(self)
 
99
        self.texts = GitTexts(self)
 
100
        self.tags = GitTags(self._git.get_tags())
58
101
 
59
 
    def _all_revision_ids(self):
60
 
        if self._git.heads == []:
61
 
            return set()
62
 
        ret = set()
63
 
        skip = 0
64
 
        max_count = 1000
65
 
        cms = None
66
 
        while cms != []:
67
 
            cms = self._git.commits("--all", max_count=max_count, skip=skip)
68
 
            skip += max_count
69
 
            ret.update([default_mapping.revision_id_foreign_to_bzr(cm.id) for cm in cms])
 
102
    def all_revision_ids(self):
 
103
        ret = set([revision.NULL_REVISION])
 
104
        if self._git.heads() == []:
 
105
            return ret
 
106
        bzr_heads = [self.get_mapping().revision_id_foreign_to_bzr(h) for h in self._git.heads()]
 
107
        ret = set(bzr_heads)
 
108
        graph = self.get_graph()
 
109
        for rev, parents in graph.iter_ancestry(bzr_heads):
 
110
            ret.add(rev)
70
111
        return ret
71
112
 
72
 
    def is_shared(self):
73
 
        return True
74
 
 
75
 
    def supports_rich_root(self):
76
 
        return False
77
 
 
78
113
    #def get_revision_delta(self, revision_id):
79
114
    #    parent_revid = self.get_revision(revision_id).parent_ids[0]
80
115
    #    diff = self._git.diff(ids.convert_revision_id_bzr_to_git(parent_revid),
81
116
    #                   ids.convert_revision_id_bzr_to_git(revision_id))
82
117
 
83
 
    def get_ancestry(self, revision_id):
84
 
        revision_id = revision.ensure_null(revision_id)
85
 
        ret = []
86
 
        if revision_id != revision.NULL_REVISION:
87
 
            skip = 0
88
 
            max_count = 1000
89
 
            cms = None
90
 
            while cms != []:
91
 
                cms = self._git.commits(self.lookup_git_revid(revision_id, default_mapping), max_count=max_count, skip=skip)
92
 
                skip += max_count
93
 
                ret += [default_mapping.revision_id_foreign_to_bzr(cm.id) for cm in cms]
94
 
        return [None] + ret
 
118
    def _make_parents_provider(self):
 
119
        """See Repository._make_parents_provider()."""
 
120
        return self._parents_provider
 
121
 
 
122
    def get_parent_map(self, revids):
 
123
        parent_map = {}
 
124
        mutter("get_parent_map(%r)", revids)
 
125
        for revision_id in revids:
 
126
            assert isinstance(revision_id, str)
 
127
            if revision_id == revision.NULL_REVISION:
 
128
                parent_map[revision_id] = ()
 
129
                continue
 
130
            hexsha, mapping = self.lookup_git_revid(revision_id)
 
131
            commit  = self._git.commit(hexsha)
 
132
            if commit is None:
 
133
                continue
 
134
            else:
 
135
                parent_map[revision_id] = [mapping.revision_id_foreign_to_bzr(p) for p in commit.parents]
 
136
        return parent_map
 
137
 
 
138
    def get_ancestry(self, revision_id, topo_sorted=True):
 
139
        """See Repository.get_ancestry().
 
140
        """
 
141
        if revision_id is None:
 
142
            return self._all_revision_ids()
 
143
        assert isinstance(revision_id, str)
 
144
        ancestry = []
 
145
        graph = self.get_graph()
 
146
        for rev, parents in graph.iter_ancestry([revision_id]):
 
147
            if rev == revision.NULL_REVISION:
 
148
                rev = None
 
149
            ancestry.append(rev)
 
150
        ancestry.reverse()
 
151
        return ancestry
 
152
 
 
153
    def import_revision_gist(self, source, revid, parent_lookup):
 
154
        """Impor the gist of another revision into this Git repository.
 
155
 
 
156
        """
 
157
        objects = []
 
158
        rev = source.get_revision(revid)
 
159
        for sha, object, path in inventory_to_tree_and_blobs(source, None, revid):
 
160
            if path == "":
 
161
                tree_sha = sha
 
162
            objects.append((object, path))
 
163
        commit = revision_to_commit(rev, tree_sha, parent_lookup)
 
164
        objects.append((commit, None))
 
165
        self._git.object_store.add_objects(objects)
 
166
 
 
167
    def dfetch(self, source, stop_revision):
 
168
        if stop_revision is None:
 
169
            raise NotImplementedError
 
170
        revidmap = {}
 
171
        todo = []
 
172
        source.lock_read()
 
173
        try:
 
174
            graph = source.get_graph()
 
175
            for revid, parents in graph.iter_ancestry([stop_revision]):
 
176
                if not self.has_revision(revid):
 
177
                    todo.append(revid)
 
178
            pb = ui.ui_factory.nested_progress_bar()
 
179
            try:
 
180
                for i, revid in enumerate(reversed(todo)):
 
181
                    pb.update("pushing revisions", i, len(todo))
 
182
                    revidmap[revid] = self.import_revision_gist(source, revid, revidmap.get)
 
183
            finally:
 
184
                pb.finished()
 
185
        finally:
 
186
            source.unlock()
 
187
        return revidmap
95
188
 
96
189
    def get_signature_text(self, revision_id):
97
190
        raise errors.NoSuchRevision(self, revision_id)
98
191
 
 
192
    def lookup_revision_id(self, revid):
 
193
        """Lookup a revision id.
 
194
        
 
195
        :param revid: Bazaar revision id.
 
196
        :return: Tuple with git revisionid and mapping.
 
197
        """
 
198
        # Yes, this doesn't really work, but good enough as a stub
 
199
        return osutils.sha(rev_id).hexdigest(), self.get_mapping()
 
200
 
99
201
    def has_signature_for_revision_id(self, revision_id):
100
202
        return False
101
203
 
102
 
    def get_parent_map(self, revision_ids):
103
 
        ret = {}
104
 
        for revid in revision_ids:
105
 
            if revid == revision.NULL_REVISION:
106
 
                ret[revid] = ()
107
 
            else:
108
 
                commit = self._git.commit(self.lookup_git_revid(revid, default_mapping))
109
 
                ret[revid] = tuple([default_mapping.revision_id_foreign_to_bzr(p.id) for p in commit.parents])
110
 
        return ret
111
 
 
112
 
    def lookup_git_revid(self, bzr_revid, mapping):
 
204
    def lookup_git_revid(self, bzr_revid):
113
205
        try:
114
 
            return mapping.revision_id_bzr_to_foreign(bzr_revid)
 
206
            return mapping_registry.revision_id_bzr_to_foreign(bzr_revid)
115
207
        except errors.InvalidRevisionId:
116
 
            raise errors.NoSuchRevision(bzr_revid, self)
 
208
            raise errors.NoSuchRevision(self, bzr_revid)
117
209
 
118
210
    def get_revision(self, revision_id):
119
 
        git_commit_id = self.lookup_git_revid(revision_id, default_mapping)
120
 
        commit = self._git.commit(git_commit_id)
 
211
        git_commit_id, mapping = self.lookup_git_revid(revision_id)
 
212
        try:
 
213
            commit = self._git.commit(git_commit_id)
 
214
        except KeyError:
 
215
            raise errors.NoSuchRevision(self, revision_id)
121
216
        # print "fetched revision:", git_commit_id
122
 
        revision = self._parse_rev(commit)
 
217
        revision = mapping.import_commit(commit)
 
218
        assert revision is not None
123
219
        return revision
124
220
 
125
221
    def has_revision(self, revision_id):
126
222
        try:
127
223
            self.get_revision(revision_id)
128
 
        except NoSuchRevision:
 
224
        except errors.NoSuchRevision:
129
225
            return False
130
226
        else:
131
227
            return True
132
228
 
133
 
    def get_revisions(self, revisions):
134
 
        return [self.get_revision(r) for r in revisions]
135
 
 
136
 
    @classmethod
137
 
    def _parse_rev(klass, commit):
138
 
        """Convert a git commit to a bzr revision.
139
 
 
140
 
        :return: a `bzrlib.revision.Revision` object.
141
 
        """
142
 
        rev = revision.Revision(default_mapping.revision_id_foreign_to_bzr(commit.id))
143
 
        rev.parent_ids = tuple([default_mapping.revision_id_foreign_to_bzr(p.id) for p in commit.parents])
144
 
        rev.inventory_sha1 = ""
145
 
        rev.message = commit.message.decode("utf-8", "replace")
146
 
        rev.committer = str(commit.committer)
147
 
        rev.properties['author'] = str(commit.author)
148
 
        rev.timestamp = time.mktime(commit.committed_date)
149
 
        rev.timezone = 0
150
 
        return rev
 
229
    def get_revisions(self, revids):
 
230
        return [self.get_revision(r) for r in revids]
151
231
 
152
232
    def revision_trees(self, revids):
153
233
        for revid in revids:
170
250
    def set_make_working_trees(self, trees):
171
251
        pass
172
252
 
173
 
 
174
 
def escape_file_id(file_id):
175
 
    return file_id.replace('_', '__').replace(' ', '_s')
176
 
 
177
 
 
178
 
def unescape_file_id(file_id):
179
 
    return file_id.replace("_s", " ").replace("__", "_")
 
253
    def fetch_objects(self, determine_wants, graph_walker, progress=None):
 
254
        return self._git.fetch_objects(determine_wants, graph_walker, progress)
180
255
 
181
256
 
182
257
class GitRevisionTree(revisiontree.RevisionTree):
184
259
    def __init__(self, repository, revision_id):
185
260
        self._repository = repository
186
261
        self.revision_id = revision_id
187
 
        git_id = repository.lookup_git_revid(revision_id, default_mapping)
188
 
        self.tree = repository._git.commit(git_id).tree
 
262
        assert isinstance(revision_id, str)
 
263
        git_id, self.mapping = repository.lookup_git_revid(revision_id)
 
264
        try:
 
265
            commit = repository._git.commit(git_id)
 
266
        except KeyError, r:
 
267
            raise errors.NoSuchRevision(repository, revision_id)
 
268
        self.tree = commit.tree
189
269
        self._inventory = inventory.Inventory(revision_id=revision_id)
190
270
        self._inventory.root.revision = revision_id
191
271
        self._build_inventory(self.tree, self._inventory.root, "")
196
276
    def get_file_text(self, file_id):
197
277
        entry = self._inventory[file_id]
198
278
        if entry.kind == 'directory': return ""
199
 
        return self._repository._git.blob(entry.text_id).data
 
279
        return self._repository._git.get_blob(entry.text_id).data
200
280
 
201
 
    def _build_inventory(self, tree, ie, path):
 
281
    def _build_inventory(self, tree_id, ie, path):
202
282
        assert isinstance(path, str)
203
 
        for b in tree.contents:
204
 
            basename = b.name.decode("utf-8")
 
283
        tree = self._repository._git.tree(tree_id)
 
284
        for mode, name, hexsha in tree.entries():
 
285
            basename = name.decode("utf-8")
205
286
            if path == "":
206
 
                child_path = b.name
 
287
                child_path = name
207
288
            else:
208
 
                child_path = urlutils.join(path, b.name)
209
 
            file_id = escape_file_id(child_path.encode('utf-8'))
210
 
            if b.mode[0] == '0':
 
289
                child_path = urlutils.join(path, name)
 
290
            file_id = self.mapping.generate_file_id(child_path)
 
291
            entry_kind = (mode & 0700000) / 0100000
 
292
            if entry_kind == 0:
211
293
                child_ie = inventory.InventoryDirectory(file_id, basename, ie.file_id)
212
 
            elif b.mode[0] == '1':
213
 
                if b.mode[1] == '0':
 
294
            elif entry_kind == 1:
 
295
                file_kind = (mode & 070000) / 010000
 
296
                b = self._repository._git.get_blob(hexsha)
 
297
                if file_kind == 0:
214
298
                    child_ie = inventory.InventoryFile(file_id, basename, ie.file_id)
215
299
                    child_ie.text_sha1 = osutils.sha_string(b.data)
216
 
                elif b.mode[1] == '2':
 
300
                elif file_kind == 2:
217
301
                    child_ie = inventory.InventoryLink(file_id, basename, ie.file_id)
218
302
                    child_ie.text_sha1 = osutils.sha_string("")
219
303
                else:
220
304
                    raise AssertionError(
221
 
                        "Unknown file kind, perms=%r." % (b.mode,))
 
305
                        "Unknown file kind, perms=%o." % (mode,))
222
306
                child_ie.text_id = b.id
223
 
                child_ie.text_size = b.size
 
307
                child_ie.text_size = len(b.data)
224
308
            else:
225
309
                raise AssertionError(
226
 
                    "Unknown blob kind, perms=%r." % (b.mode,))
227
 
            child_ie.executable = bool(int(b.mode[3:], 8) & 0111)
 
310
                    "Unknown blob kind, perms=%r." % (mode,))
 
311
            fs_mode = mode & 0777
 
312
            child_ie.executable = bool(fs_mode & 0111)
228
313
            child_ie.revision = self.revision_id
229
314
            self._inventory.add(child_ie)
230
 
            if b.mode[0] == '0':
231
 
                self._build_inventory(b, child_ie, child_path)
 
315
            if entry_kind == 0:
 
316
                self._build_inventory(hexsha, child_ie, child_path)
232
317
 
233
318
 
234
319
class GitFormat(object):
235
320
 
236
321
    supports_tree_reference = False
 
322
    rich_root_data = True
237
323
 
238
324
    def get_format_description(self):
239
325
        return "Git Repository"
 
326
 
 
327
    def initialize(self, url, shared=False, _internal=False):
 
328
        raise bzr_errors.UninitializableFormat(self)
 
329
 
 
330
    def check_conversion_target(self, target_repo_format):
 
331
        return target_repo_format.rich_root_data