17
18
"""An adapter between a Git Repository and a Bazaar Branch"""
23
20
from bzrlib import (
29
from bzrlib.revisiontree import InventoryRevisionTree
30
except ImportError: # bzr < 2.4
31
from bzrlib.revisiontree import RevisionTree as InventoryRevisionTree
35
32
from bzrlib.foreign import (
38
from bzrlib.trace import mutter
39
from bzrlib.transport import get_transport
41
from bzrlib.plugins.git.foreign import (
44
from bzrlib.plugins.git.mapping import default_mapping
46
from bzrlib.plugins.git import git
49
class GitTags(object):
51
def __init__(self, tags):
55
return iter(self._tags)
36
from bzrlib.plugins.git.commit import (
39
from bzrlib.plugins.git.filegraph import (
40
GitFileLastChangeScanner,
41
GitFileParentProvider,
43
from bzrlib.plugins.git.mapping import (
48
from bzrlib.plugins.git.tree import (
53
from dulwich.objects import (
58
from dulwich.object_store import (
63
class RepoReconciler(object):
64
"""Reconciler that reconciles a repository.
68
def __init__(self, repo, other=None, thorough=False):
69
"""Construct a RepoReconciler.
71
:param thorough: perform a thorough check which may take longer but
72
will correct non-data loss issues such as incorrect
78
"""Perform reconciliation.
80
After reconciliation the following attributes document found issues:
81
inconsistent_parents: The number of revisions in the repository whose
82
ancestry was being reported incorrectly.
83
garbage_inventories: The number of inventory objects without revisions
84
that were garbage collected.
88
class GitCheck(check.Check):
90
def __init__(self, repository, check_repo=True):
91
self.repository = repository
92
self.checked_rev_cnt = 0
94
def check(self, callback_refs=None, check_repo=True):
95
if callback_refs is None:
97
self.repository.lock_read()
98
self.repository.unlock()
100
def report_results(self, verbose):
58
104
class GitRepository(ForeignRepository):
59
105
"""An adapter to git repositories for bzr."""
61
107
_serializer = None
108
vcs = foreign_vcs_git
63
111
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)
112
ForeignRepository.__init__(self, GitRepositoryFormat(), gitdir, lockfiles)
113
from bzrlib.plugins.git import fetch, push
114
for optimiser in [fetch.InterRemoteGitNonGitRepository,
115
fetch.InterLocalGitNonGitRepository,
116
fetch.InterGitGitRepository,
117
push.InterToLocalGitRepository,
118
push.InterToRemoteGitRepository]:
119
repository.InterRepository.register_optimiser(optimiser)
121
def add_fallback_repository(self, basis_url):
122
raise errors.UnstackableRepositoryFormat(self._format, self.control_transport.base)
68
124
def is_shared(self):
127
def reconcile(self, other=None, thorough=False):
128
"""Reconcile this repository."""
129
reconciler = RepoReconciler(self, thorough=thorough)
130
reconciler.reconcile()
71
133
def supports_rich_root(self):
74
def _warn_if_deprecated(self):
136
def _warn_if_deprecated(self, branch=None): # for bzr < 2.4
75
137
# This class isn't deprecated
79
141
return default_mapping
81
143
def make_working_trees(self):
144
return not self._git.bare
146
def revision_graph_can_have_wrong_parents(self):
149
def dfetch(self, source, stop_revision):
150
interrepo = repository.InterRepository.get(source, self)
151
return interrepo.dfetch(stop_revision)
153
def add_signature_text(self, revid, signature):
154
raise errors.UnsupportedOperation(self.add_signature_text, self)
85
157
class LocalGitRepository(GitRepository):
158
"""Git repository on the file system."""
87
160
def __init__(self, gitdir, lockfiles):
88
# FIXME: This also caches negatives. Need to be more careful
89
# about this once we start writing to git
90
self._parents_provider = graph.CachingParentsProvider(self)
91
161
GitRepository.__init__(self, gitdir, lockfiles)
92
162
self.base = gitdir.root_transport.base
93
163
self._git = gitdir._git
95
self.signatures = versionedfiles.VirtualSignatureTexts(self)
96
self.revisions = versionedfiles.VirtualRevisionTexts(self)
97
self.tags = GitTags(self._git.get_tags())
164
self._file_change_scanner = GitFileLastChangeScanner(self)
166
def get_commit_builder(self, branch, parents, config, timestamp=None,
167
timezone=None, committer=None, revprops=None,
168
revision_id=None, lossy=False):
169
"""Obtain a CommitBuilder for this repository.
171
:param branch: Branch to commit to.
172
:param parents: Revision ids of the parents of the new revision.
173
:param config: Configuration to use.
174
:param timestamp: Optional timestamp recorded for commit.
175
:param timezone: Optional timezone for timestamp.
176
:param committer: Optional committer to set for commit.
177
:param revprops: Optional dictionary of revision properties.
178
:param revision_id: Optional revision id.
179
:param lossy: Whether to discard data that can not be natively
180
represented, when pushing to a foreign VCS
182
self.start_write_group()
183
return GitCommitBuilder(self, parents, config,
184
timestamp, timezone, committer, revprops, revision_id,
187
def get_file_graph(self):
188
return _mod_graph.Graph(GitFileParentProvider(self._file_change_scanner))
190
def iter_files_bytes(self, desired_files):
191
"""Iterate through file versions.
193
Files will not necessarily be returned in the order they occur in
194
desired_files. No specific order is guaranteed.
196
Yields pairs of identifier, bytes_iterator. identifier is an opaque
197
value supplied by the caller as part of desired_files. It should
198
uniquely identify the file version in the caller's context. (Examples:
199
an index number or a TreeTransform trans_id.)
201
bytes_iterator is an iterable of bytestrings for the file. The
202
kind of iterable and length of the bytestrings are unspecified, but for
203
this implementation, it is a list of bytes produced by
204
VersionedFile.get_record_stream().
206
:param desired_files: a list of (file_id, revision_id, identifier)
210
for (file_id, revision_id, identifier) in desired_files:
211
per_revision.setdefault(revision_id, []).append((file_id, identifier))
212
for revid, files in per_revision.iteritems():
213
(commit_id, mapping) = self.lookup_bzr_revision_id(revid)
215
commit = self._git.object_store[commit_id]
217
raise errors.RevisionNotPresent(revid, self)
218
root_tree = commit.tree
219
for fileid, identifier in files:
220
path = mapping.parse_file_id(fileid)
222
obj = tree_lookup_path(
223
self._git.object_store.__getitem__, root_tree, path)
224
if isinstance(obj, tuple):
225
(mode, item_id) = obj
226
obj = self._git.object_store[item_id]
228
raise errors.RevisionNotPresent((fileid, revid), self)
230
if obj.type_name == "tree":
231
yield (identifier, [])
232
elif obj.type_name == "blob":
233
yield (identifier, obj.chunked)
235
raise AssertionError("file text resolved to %r" % obj)
238
def _iter_revision_ids(self):
239
mapping = self.get_mapping()
240
for sha in self._git.object_store:
241
o = self._git.object_store[sha]
242
if not isinstance(o, Commit):
244
rev, roundtrip_revid, verifiers = mapping.import_commit(o,
245
mapping.revision_id_foreign_to_bzr)
246
yield o.id, rev.revision_id, roundtrip_revid
99
248
def all_revision_ids(self):
100
ret = set([revision.NULL_REVISION])
101
if self._git.heads() == []:
103
bzr_heads = [self.get_mapping().revision_id_foreign_to_bzr(h) for h in self._git.heads()]
105
graph = self.get_graph()
106
for rev, parents in graph.iter_ancestry(bzr_heads):
250
for git_sha, revid, roundtrip_revid in self._iter_revision_ids():
253
ret.add(roundtrip_revid)
110
#def get_revision_delta(self, revision_id):
111
# parent_revid = self.get_revision(revision_id).parent_ids[0]
112
# diff = self._git.diff(ids.convert_revision_id_bzr_to_git(parent_revid),
113
# ids.convert_revision_id_bzr_to_git(revision_id))
115
def _make_parents_provider(self):
116
"""See Repository._make_parents_provider()."""
117
return self._parents_provider
119
256
def get_parent_map(self, revids):
121
mutter("get_parent_map(%r)", revids)
122
258
for revision_id in revids:
123
259
assert isinstance(revision_id, str)
124
260
if revision_id == revision.NULL_REVISION:
125
261
parent_map[revision_id] = ()
127
hexsha = self.lookup_git_revid(revision_id, self.get_mapping())
128
commit = self._git.commit(hexsha)
263
hexsha, mapping = self.lookup_bzr_revision_id(revision_id)
265
commit = self._git[hexsha]
132
parent_map[revision_id] = [self.get_mapping().revision_id_foreign_to_bzr(p) for p in commit.parents]
269
self.lookup_foreign_revision_id(p, mapping)
270
for p in commit.parents]
272
parents = [revision.NULL_REVISION]
273
parent_map[revision_id] = tuple(parents)
133
274
return parent_map
135
def get_ancestry(self, revision_id, topo_sorted=True):
136
"""See Repository.get_ancestry().
276
def get_known_graph_ancestry(self, revision_ids):
277
"""Return the known graph for a set of revision ids and their ancestors.
138
if revision_id is None:
139
return self._all_revision_ids()
140
assert isinstance(revision_id, str)
142
graph = self.get_graph()
143
for rev, parents in graph.iter_ancestry([revision_id]):
144
if rev == revision.NULL_REVISION:
279
pending = set(revision_ids)
282
this_parent_map = self.get_parent_map(pending)
283
parent_map.update(this_parent_map)
285
map(pending.update, this_parent_map.itervalues())
286
pending = pending.difference(parent_map)
287
return _mod_graph.KnownGraph(parent_map)
150
289
def get_signature_text(self, revision_id):
151
290
raise errors.NoSuchRevision(self, revision_id)
153
def lookup_revision_id(self, revid):
292
def check(self, revision_ids=None, callback_refs=None, check_repo=True):
293
result = GitCheck(self, check_repo=check_repo)
294
result.check(callback_refs)
297
def pack(self, hint=None, clean_obsolete_packs=False):
298
self._git.object_store.pack_loose_objects()
300
def lookup_foreign_revision_id(self, foreign_revid, mapping=None):
154
301
"""Lookup a revision id.
156
:param revid: Bazaar revision id.
157
:return: Tuple with git revisionid and mapping.
159
# Yes, this doesn't really work, but good enough as a stub
160
return osutils.sha(rev_id).hexdigest(), self.get_mapping()
304
assert type(foreign_revid) is str
306
mapping = self.get_mapping()
307
if foreign_revid == ZERO_SHA:
308
return revision.NULL_REVISION
309
commit = self._git.object_store[foreign_revid]
310
while isinstance(commit, Tag):
311
commit = self._git[commit.object[1]]
312
rev, roundtrip_revid, verifiers = mapping.import_commit(commit,
313
mapping.revision_id_foreign_to_bzr)
314
# FIXME: check testament before doing this?
316
return roundtrip_revid
318
return rev.revision_id
162
320
def has_signature_for_revision_id(self, revision_id):
165
def lookup_git_revid(self, bzr_revid, mapping):
323
def lookup_bzr_revision_id(self, bzr_revid, mapping=None):
167
return mapping.revision_id_bzr_to_foreign(bzr_revid)
325
return mapping_registry.revision_id_bzr_to_foreign(bzr_revid)
168
326
except errors.InvalidRevisionId:
169
raise errors.NoSuchRevision(self, bzr_revid)
328
mapping = self.get_mapping()
330
return (self._git.refs[mapping.revid_as_refname(bzr_revid)], mapping)
332
# Update refs from Git commit objects
333
# FIXME: Hitting this a lot will be very inefficient...
334
for git_sha, revid, roundtrip_revid in self._iter_revision_ids():
335
if not roundtrip_revid:
337
refname = mapping.revid_as_refname(roundtrip_revid)
338
self._git.refs[refname] = git_sha
339
if roundtrip_revid == bzr_revid:
340
return git_sha, mapping
341
raise errors.NoSuchRevision(self, bzr_revid)
171
343
def get_revision(self, revision_id):
172
git_commit_id = self.lookup_git_revid(revision_id, self.get_mapping())
344
if not isinstance(revision_id, str):
345
raise errors.InvalidRevisionId(revision_id, self)
346
git_commit_id, mapping = self.lookup_bzr_revision_id(revision_id)
174
commit = self._git.commit(git_commit_id)
348
commit = self._git[git_commit_id]
176
350
raise errors.NoSuchRevision(self, revision_id)
177
# print "fetched revision:", git_commit_id
178
revision = self.get_mapping().import_commit(commit)
351
revision, roundtrip_revid, verifiers = mapping.import_commit(
352
commit, self.lookup_foreign_revision_id)
179
353
assert revision is not None
354
# FIXME: check verifiers ?
356
revision.revision_id = roundtrip_revid
182
359
def has_revision(self, revision_id):
360
"""See Repository.has_revision."""
361
if revision_id == revision.NULL_REVISION:
184
self.get_revision(revision_id)
364
git_commit_id, mapping = self.lookup_bzr_revision_id(revision_id)
185
365
except errors.NoSuchRevision:
367
return (git_commit_id in self._git)
369
def has_revisions(self, revision_ids):
370
"""See Repository.has_revisions."""
371
return set(filter(self.has_revision, revision_ids))
190
373
def get_revisions(self, revids):
374
"""See Repository.get_revisions."""
191
375
return [self.get_revision(r) for r in revids]
193
377
def revision_trees(self, revids):
378
"""See Repository.revision_trees."""
194
379
for revid in revids:
195
380
yield self.revision_tree(revid)
197
382
def revision_tree(self, revision_id):
383
"""See Repository.revision_tree."""
198
384
revision_id = revision.ensure_null(revision_id)
200
385
if revision_id == revision.NULL_REVISION:
201
386
inv = inventory.Inventory(root_id=None)
202
387
inv.revision_id = revision_id
203
return revisiontree.RevisionTree(self, inv, revision_id)
205
return GitRevisionTree(self, self.get_mapping(), revision_id)
388
return InventoryRevisionTree(self, inv, revision_id)
389
return GitRevisionTree(self, revision_id)
207
391
def get_inventory(self, revision_id):
208
assert revision_id != None
209
return self.revision_tree(revision_id).inventory
392
raise NotImplementedError(self.get_inventory)
211
394
def set_make_working_trees(self, trees):
395
# TODO: Set bare= in the configuration bug=777065
396
raise NotImplementedError(self.set_make_working_trees)
214
def fetch_objects(self, determine_wants, graph_walker, progress=None):
398
def fetch_objects(self, determine_wants, graph_walker, resolve_ext_ref,
215
400
return self._git.fetch_objects(determine_wants, graph_walker, progress)
218
class GitRevisionTree(revisiontree.RevisionTree):
220
def __init__(self, repository, mapping, revision_id):
221
self._repository = repository
222
self.revision_id = revision_id
223
assert isinstance(revision_id, str)
224
self.mapping = mapping
225
git_id = repository.lookup_git_revid(revision_id, self.mapping)
227
commit = repository._git.commit(git_id)
229
raise errors.NoSuchRevision(repository, revision_id)
230
self.tree = commit.tree
231
self._inventory = inventory.Inventory(revision_id=revision_id)
232
self._inventory.root.revision = revision_id
233
self._build_inventory(self.tree, self._inventory.root, "")
235
def get_revision_id(self):
236
return self.revision_id
238
def get_file_text(self, file_id):
239
entry = self._inventory[file_id]
240
if entry.kind == 'directory': return ""
241
return self._repository._git.get_blob(entry.text_id).data
243
def _build_inventory(self, tree_id, ie, path):
244
assert isinstance(path, str)
245
tree = self._repository._git.tree(tree_id)
246
for mode, name, hexsha in tree.entries():
247
basename = name.decode("utf-8")
251
child_path = urlutils.join(path, name)
252
file_id = self.mapping.generate_file_id(child_path)
253
entry_kind = (mode & 0700000) / 0100000
255
child_ie = inventory.InventoryDirectory(file_id, basename, ie.file_id)
256
elif entry_kind == 1:
257
file_kind = (mode & 070000) / 010000
258
b = self._repository._git.get_blob(hexsha)
260
child_ie = inventory.InventoryFile(file_id, basename, ie.file_id)
261
child_ie.text_sha1 = osutils.sha_string(b.data)
263
child_ie = inventory.InventoryLink(file_id, basename, ie.file_id)
264
child_ie.text_sha1 = osutils.sha_string("")
266
raise AssertionError(
267
"Unknown file kind, perms=%o." % (mode,))
268
child_ie.text_id = b.id
269
child_ie.text_size = len(b.data)
271
raise AssertionError(
272
"Unknown blob kind, perms=%r." % (mode,))
273
fs_mode = mode & 0777
274
child_ie.executable = bool(fs_mode & 0111)
275
child_ie.revision = self.revision_id
276
self._inventory.add(child_ie)
278
self._build_inventory(hexsha, child_ie, child_path)
281
class GitFormat(object):
403
class GitRepositoryFormat(repository.RepositoryFormat):
404
"""Git repository format."""
406
supports_versioned_directories = False
283
407
supports_tree_reference = False
284
408
rich_root_data = True
409
supports_leaving_lock = False
411
supports_funky_characters = True
412
supports_external_lookups = False
413
supports_full_versioned_files = False
414
supports_revision_signatures = False
415
revision_graph_can_have_wrong_parents = False
418
def _matchingbzrdir(self):
419
from bzrlib.plugins.git.dir import LocalGitControlDirFormat
420
return LocalGitControlDirFormat()
286
422
def get_format_description(self):
287
423
return "Git Repository"
289
def initialize(self, url, shared=False, _internal=False):
290
raise bzr_errors.UninitializableFormat(self)
425
def initialize(self, controldir, shared=False, _internal=False):
426
from bzrlib.plugins.git.dir import GitDir
427
if not isinstance(controldir, GitDir):
428
raise errors.UninitializableFormat(self)
429
return controldir.open_repository()
292
431
def check_conversion_target(self, target_repo_format):
293
432
return target_repo_format.rich_root_data
434
def get_foreign_tests_repository_factory(self):
435
from bzrlib.plugins.git.tests.test_repository import (
436
ForeignTestsRepositoryFactory,
438
return ForeignTestsRepositoryFactory()
440
def network_name(self):
added
removed
repository.py
