/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 fetch.py

Add more details for commit, to avoid checksum from changing.

Show diffs side-by-side

added added

removed removed

Lines of Context:
fetch.py
1
 
# Copyright (C) 2008 Canonical Ltd
 
1
# Copyright (C) 2008 Jelmer Vernooij <jelmer@samba.org>
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
16
 
17
 
from bzrlib import osutils, ui, urlutils
18
 
from bzrlib.errors import InvalidRevisionId
 
17
from cStringIO import StringIO
 
18
import dulwich as git
 
19
from dulwich.client import SimpleFetchGraphWalker
 
20
from dulwich.objects import Commit
 
21
 
 
22
from bzrlib import (
 
23
    osutils,
 
24
    ui,
 
25
    urlutils,
 
26
    )
 
27
from bzrlib.errors import (
 
28
    InvalidRevisionId,
 
29
    NoSuchRevision,
 
30
    )
19
31
from bzrlib.inventory import Inventory
20
32
from bzrlib.repository import InterRepository
21
33
from bzrlib.trace import info
22
34
from bzrlib.tsort import topo_sort
23
35
 
24
 
from bzrlib.plugins.git import git
25
36
from bzrlib.plugins.git.repository import (
26
37
        LocalGitRepository, 
27
38
        GitRepository, 
28
39
        GitFormat,
29
40
        )
 
41
from bzrlib.plugins.git.converter import GitObjectConverter
30
42
from bzrlib.plugins.git.remote import RemoteGitRepository
31
43
 
32
 
from dulwich.objects import Commit
33
 
 
34
 
from cStringIO import StringIO
35
44
 
36
45
 
37
46
class BzrFetchGraphWalker(object):
 
47
    """GraphWalker implementation that uses a Bazaar repository."""
38
48
 
39
49
    def __init__(self, repository, mapping):
40
50
        self.repository = repository
43
53
        self.heads = set(repository.all_revision_ids())
44
54
        self.parents = {}
45
55
 
 
56
    def __iter__(self):
 
57
        return iter(self.next, None)
 
58
 
46
59
    def ack(self, sha):
47
60
        revid = self.mapping.revision_id_foreign_to_bzr(sha)
48
61
        self.remove(revid)
49
62
 
50
63
    def remove(self, revid):
51
64
        self.done.add(revid)
52
 
        if ref in self.heads:
 
65
        if revid in self.heads:
53
66
            self.heads.remove(revid)
54
67
        if revid in self.parents:
55
68
            for p in self.parents[revid]:
63
76
            self.heads.update([p for p in ps if not p in self.done])
64
77
            try:
65
78
                self.done.add(ret)
66
 
                return self.mapping.revision_id_bzr_to_foreign(ret)
 
79
                return self.mapping.revision_id_bzr_to_foreign(ret)[0]
67
80
            except InvalidRevisionId:
68
81
                pass
69
82
        return None
70
83
 
71
84
 
72
 
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
 
85
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, gitmap, executable):
73
86
    """Import a git blob object into a bzr repository.
74
87
 
75
88
    :param repo: bzr repository
86
99
    ie.text_size = len(blob.data)
87
100
    ie.text_sha1 = osutils.sha_string(blob.data)
88
101
    ie.executable = executable
89
 
 
90
 
 
91
 
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
 
102
    gitmap._idmap.add_entry(blob.sha().hexdigest(), "blob", (ie.file_id, ie.revision))
 
103
 
 
104
 
 
105
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, 
 
106
                    gitmap, lookup_object):
92
107
    """Import a git tree object into a bzr repository.
93
108
 
94
109
    :param repo: A Bzr repository object
103
118
        [])
104
119
    ie = inv.add_path(path, "directory", file_id)
105
120
    ie.revision = text_revision
 
121
    gitmap._idmap.add_entry(tree.sha().hexdigest(), "tree", (file_id, text_revision))
106
122
    for mode, name, hexsha in tree.entries():
107
123
        entry_kind = (mode & 0700000) / 0100000
108
124
        basename = name.decode("utf-8")
112
128
            child_path = urlutils.join(path, name)
113
129
        if entry_kind == 0:
114
130
            tree = lookup_object(hexsha)
115
 
            import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
 
131
            import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, gitmap, lookup_object)
116
132
        elif entry_kind == 1:
117
133
            blob = lookup_object(hexsha)
118
134
            fs_mode = mode & 0777
119
 
            import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
 
135
            import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, gitmap, bool(fs_mode & 0111))
120
136
        else:
121
137
            raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
122
138
 
123
139
 
124
 
def import_git_objects(repo, mapping, object_iter, pb=None):
 
140
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever, 
 
141
        pb=None):
125
142
    """Import a set of git objects into a bzr repository.
126
143
 
127
144
    :param repo: Bazaar repository
129
146
    :param object_iter: Iterator over Git objects.
130
147
    """
131
148
    # TODO: a more (memory-)efficient implementation of this
132
 
    objects = {}
133
 
    for i, o in enumerate(object_iter):
134
 
        if pb is not None:
135
 
            pb.update("fetching objects", i) 
136
 
        objects[o.id] = o
137
149
    graph = []
138
150
    root_trees = {}
139
151
    revisions = {}
140
152
    # Find and convert commit objects
141
 
    for o in objects.itervalues():
 
153
    for o in object_iter.iterobjects():
142
154
        if isinstance(o, Commit):
143
155
            rev = mapping.import_commit(o)
144
 
            root_trees[rev.revision_id] = objects[o.tree]
 
156
            root_trees[rev.revision_id] = object_iter[o.tree]
145
157
            revisions[rev.revision_id] = rev
146
158
            graph.append((rev.revision_id, rev.parent_ids))
 
159
            target_git_object_retriever._idmap.add_entry(o.sha().hexdigest(), "commit", (rev.revision_id, o._tree))
147
160
    # Order the revisions
148
161
    # Create the inventory objects
149
162
    for i, revid in enumerate(topo_sort(graph)):
157
170
        inv = Inventory()
158
171
        inv.revision_id = rev.revision_id
159
172
        def lookup_object(sha):
160
 
            if sha in objects:
161
 
                return objects[sha]
162
 
            return reconstruct_git_object(repo, mapping, sha)
 
173
            if sha in object_iter:
 
174
                return object_iter[sha]
 
175
            return target_git_object_retriever[sha]
163
176
        parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
164
 
        import_git_tree(repo, mapping, "", root_tree, inv, parent_invs, lookup_object)
 
177
        import_git_tree(repo, mapping, "", root_tree, inv, parent_invs, 
 
178
            target_git_object_retriever, lookup_object)
165
179
        repo.add_revision(rev.revision_id, rev, inv)
166
180
 
167
181
 
168
 
def reconstruct_git_commit(repo, rev):
169
 
    raise NotImplementedError(self.reconstruct_git_commit)
170
 
 
171
 
 
172
 
def reconstruct_git_object(repo, mapping, sha):
173
 
    # Commit
174
 
    revid = mapping.revision_id_foreign_to_bzr(sha)
175
 
    try:
176
 
        rev = repo.get_revision(revid)
177
 
    except NoSuchRevision:
178
 
        pass
179
 
    else:
180
 
        return reconstruct_git_commit(rev)
181
 
 
182
 
    # TODO: Tree
183
 
    # TODO: Blob
184
 
    raise KeyError("No such object %s" % sha)
185
 
 
186
 
 
187
 
class InterGitRepository(InterRepository):
 
182
class InterGitNonGitRepository(InterRepository):
188
183
 
189
184
    _matching_repo_format = GitFormat()
190
185
 
196
191
        """See InterRepository.copy_content."""
197
192
        self.fetch(revision_id, pb, find_ghosts=False)
198
193
 
199
 
    def fetch(self, revision_id=None, pb=None, find_ghosts=False, 
200
 
              mapping=None):
201
 
        if mapping is None:
202
 
            mapping = self.source.get_mapping()
 
194
    def fetch_objects(self, determine_wants, mapping, pb=None):
203
195
        def progress(text):
204
 
            pb.note("git: %s", text)
205
 
        def determine_wants(heads):
206
 
            if revision_id is None:
207
 
                ret = heads.values()
208
 
            else:
209
 
                ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
210
 
            return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
 
196
            pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
211
197
        graph_walker = BzrFetchGraphWalker(self.target, mapping)
212
198
        create_pb = None
213
199
        if pb is None:
214
200
            create_pb = pb = ui.ui_factory.nested_progress_bar()
 
201
        target_git_object_retriever = GitObjectConverter(self.target, mapping)
 
202
        
215
203
        try:
216
204
            self.target.lock_write()
217
205
            try:
218
206
                self.target.start_write_group()
219
207
                try:
220
 
                    import_git_objects(self.target, mapping,
221
 
                        iter(self.source.fetch_objects(determine_wants, graph_walker, 
222
 
                            progress)), pb)
 
208
                    objects_iter = self.source.fetch_objects(determine_wants, 
 
209
                                graph_walker, 
 
210
                                target_git_object_retriever.__getitem__, 
 
211
                                progress)
 
212
                    import_git_objects(self.target, mapping, objects_iter, 
 
213
                            target_git_object_retriever, pb)
223
214
                finally:
224
215
                    self.target.commit_write_group()
225
216
            finally:
228
219
            if create_pb:
229
220
                create_pb.finished()
230
221
 
 
222
    def fetch(self, revision_id=None, pb=None, find_ghosts=False, 
 
223
              mapping=None, fetch_spec=None):
 
224
        self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
 
225
                mapping=mapping, fetch_spec=fetch_spec)
 
226
 
 
227
    def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False, 
 
228
              mapping=None, fetch_spec=None):
 
229
        if mapping is None:
 
230
            mapping = self.source.get_mapping()
 
231
        if revision_id is not None:
 
232
            interesting_heads = [revision_id]
 
233
        elif fetch_spec is not None:
 
234
            interesting_heads = fetch_spec.heads
 
235
        else:
 
236
            interesting_heads = None
 
237
        self._refs = {}
 
238
        def determine_wants(refs):
 
239
            self._refs = refs
 
240
            if interesting_heads is None:
 
241
                ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
 
242
            else:
 
243
                ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads]
 
244
            return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
 
245
        self.fetch_objects(determine_wants, mapping, pb)
 
246
        return self._refs
 
247
 
231
248
    @staticmethod
232
249
    def is_compatible(source, target):
233
250
        """Be compatible with GitRepository."""
234
251
        # FIXME: Also check target uses VersionedFile
235
252
        return (isinstance(source, GitRepository) and 
236
 
                target.supports_rich_root())
 
253
                target.supports_rich_root() and
 
254
                not isinstance(target, GitRepository))
 
255
 
 
256
 
 
257
class InterGitRepository(InterRepository):
 
258
 
 
259
    _matching_repo_format = GitFormat()
 
260
 
 
261
    @staticmethod
 
262
    def _get_repo_format_to_test():
 
263
        return None
 
264
 
 
265
    def copy_content(self, revision_id=None, pb=None):
 
266
        """See InterRepository.copy_content."""
 
267
        self.fetch(revision_id, pb, find_ghosts=False)
 
268
 
 
269
    def fetch(self, revision_id=None, pb=None, find_ghosts=False, 
 
270
              mapping=None, fetch_spec=None):
 
271
        if mapping is None:
 
272
            mapping = self.source.get_mapping()
 
273
        def progress(text):
 
274
            info("git: %s", text)
 
275
        r = self.target._git
 
276
        if revision_id is not None:
 
277
            args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
 
278
        elif fetch_spec is not None:
 
279
            args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
 
280
        if fetch_spec is None and revision_id is None:
 
281
            determine_wants = r.object_store.determine_wants_all
 
282
        else:
 
283
            determine_wants = lambda x: [y for y in args if not y in r.object_store]
 
284
 
 
285
        graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
 
286
        f, commit = r.object_store.add_pack()
 
287
        try:
 
288
            self.source._git.fetch_pack(path, determine_wants, graphwalker, f.write, progress)
 
289
            f.close()
 
290
            commit()
 
291
        except:
 
292
            f.close()
 
293
            raise
 
294
 
 
295
    @staticmethod
 
296
    def is_compatible(source, target):
 
297
        """Be compatible with GitRepository."""
 
298
        return (isinstance(source, GitRepository) and 
 
299
                isinstance(target, GitRepository))