1
# Copyright (C) 2009-2018 Jelmer Vernooij <jelmer@jelmer.uk>
2
# Copyright (C) 2012 Canonical Ltd
4
# This program is free software; you can redistribute it and/or modify
5
# it under the terms of the GNU General Public License as published by
6
# the Free Software Foundation; either version 2 of the License, or
7
# (at your option) any later version.
9
# This program is distributed in the hope that it will be useful,
10
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
# GNU General Public License for more details.
14
# You should have received a copy of the GNU General Public License
15
# along with this program; if not, write to the Free Software
16
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18
"""Map from Git sha's to Bazaar objects."""
20
from dulwich.objects import (
27
from dulwich.object_store import (
30
from dulwich.pack import (
43
from ..lock import LogicalLockResult
44
from ..revision import (
47
from ..tree import InterTree
48
from ..bzr.testament import (
53
from_repository as cache_from_repository,
55
from .mapping import (
58
extract_unusual_modes,
62
from .unpeel_map import (
70
BANNED_FILENAMES = ['.git']
73
def get_object_store(repo, mapping=None):
74
git = getattr(repo, "_git", None)
76
git.object_store.unlock = lambda: None
77
git.object_store.lock_read = lambda: LogicalLockResult(lambda: None)
78
git.object_store.lock_write = lambda: LogicalLockResult(lambda: None)
79
return git.object_store
80
return BazaarObjectStore(repo, mapping)
83
MAX_TREE_CACHE_SIZE = 50 * 1024 * 1024
86
class LRUTreeCache(object):
88
def __init__(self, repository):
89
def approx_tree_size(tree):
90
# Very rough estimate, 250 per inventory entry
92
inv = tree.root_inventory
93
except AttributeError:
96
self.repository = repository
97
self._cache = lru_cache.LRUSizeCache(
98
max_size=MAX_TREE_CACHE_SIZE, after_cleanup_size=None,
99
compute_size=approx_tree_size)
101
def revision_tree(self, revid):
103
tree = self._cache[revid]
105
tree = self.repository.revision_tree(revid)
109
def iter_revision_trees(self, revids):
114
tree = self._cache[revid]
118
if tree.get_revision_id() != revid:
119
raise AssertionError(
120
"revision id did not match: %s != %s" % (
121
tree.get_revision_id(), revid))
123
for tree in self.repository.revision_trees(todo):
124
trees[tree.get_revision_id()] = tree
126
return (trees[r] for r in revids)
128
def revision_trees(self, revids):
129
return list(self.iter_revision_trees(revids))
132
self._cache[tree.get_revision_id()] = tree
135
def _find_missing_bzr_revids(graph, want, have):
136
"""Find the revisions that have to be pushed.
138
:param get_parent_map: Function that returns the parents for a sequence
140
:param want: Revisions the target wants
141
:param have: Revisions the target already has
142
:return: Set of revisions to fetch
147
extra_todo = graph.find_unique_ancestors(rev, handled)
148
todo.update(extra_todo)
149
handled.update(extra_todo)
150
if NULL_REVISION in todo:
151
todo.remove(NULL_REVISION)
155
def _check_expected_sha(expected_sha, object):
156
"""Check whether an object matches an expected SHA.
158
:param expected_sha: None or expected SHA as either binary or as hex digest
159
:param object: Object to verify
161
if expected_sha is None:
163
if len(expected_sha) == 40:
164
if expected_sha != object.sha().hexdigest().encode('ascii'):
165
raise AssertionError("Invalid sha for %r: %s" % (object,
167
elif len(expected_sha) == 20:
168
if expected_sha != object.sha().digest():
169
raise AssertionError("Invalid sha for %r: %s" % (
170
object, sha_to_hex(expected_sha)))
172
raise AssertionError("Unknown length %d for %r" % (len(expected_sha),
176
def directory_to_tree(path, children, lookup_ie_sha1, unusual_modes,
177
empty_file_name, allow_empty=False):
178
"""Create a Git Tree object from a Bazaar directory.
180
:param path: directory path
181
:param children: Children inventory entries
182
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
183
:param unusual_modes: Dictionary with unusual file modes by file ids
184
:param empty_file_name: Name to use for dummy files in empty directories,
185
None to ignore empty directories.
188
for value in children:
189
if value.name in BANNED_FILENAMES:
191
child_path = osutils.pathjoin(path, value.name)
193
mode = unusual_modes[child_path]
195
mode = entry_mode(value)
196
hexsha = lookup_ie_sha1(child_path, value)
197
if hexsha is not None:
198
tree.add(value.name.encode("utf-8"), mode, hexsha)
199
if not allow_empty and len(tree) == 0:
200
# Only the root can be an empty tree
201
if empty_file_name is not None:
202
tree.add(empty_file_name, stat.S_IFREG | 0o644, Blob().id)
208
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes,
209
dummy_file_name=None, add_cache_entry=None):
210
"""Iterate over the objects that were introduced in a revision.
213
:param parent_trees: Parent revision trees
214
:param unusual_modes: Unusual file modes dictionary
215
:param dummy_file_name: File name to use for dummy files
216
in empty directories. None to skip empty directories
217
:return: Yields (path, object, ie) entries
223
base_tree = parent_trees[0]
224
other_parent_trees = parent_trees[1:]
226
base_tree = tree._repository.revision_tree(NULL_REVISION)
227
other_parent_trees = []
229
def find_unchanged_parent_ie(path, kind, other, parent_trees):
230
for ptree in parent_trees:
231
intertree = InterTree.get(ptree, tree)
232
ppath = intertree.find_source_path(path)
233
if ppath is not None:
234
pkind = ptree.kind(ppath)
236
if (pkind == "file" and
237
ptree.get_file_sha1(ppath) == other):
239
ptree.path2id(ppath), ptree.get_file_revision(ppath))
240
if kind == "symlink":
241
if (pkind == "symlink" and
242
ptree.get_symlink_target(ppath) == other):
244
ptree.path2id(ppath), ptree.get_file_revision(ppath))
247
# Find all the changed blobs
248
for change in tree.iter_changes(base_tree):
249
if change.name[1] in BANNED_FILENAMES:
251
if change.kind[1] == "file":
252
sha1 = tree.get_file_sha1(change.path[1])
255
(pfile_id, prevision) = find_unchanged_parent_ie(
256
change.path[1], change.kind[1], sha1, other_parent_trees)
260
# It existed in one of the parents, with the same contents.
261
# So no need to yield any new git objects.
263
blob_id = idmap.lookup_blob_id(
266
if not change.changed_content:
269
blob.data = tree.get_file_text(change.path[1])
272
new_blobs.append((change.path[1], change.file_id))
274
# TODO(jelmer): This code path does not have any test coverage.
275
shamap[change.path[1]] = blob_id
276
if add_cache_entry is not None:
279
(change.file_id, tree.get_file_revision(change.path[1])), change.path[1])
280
elif change.kind[1] == "symlink":
281
target = tree.get_symlink_target(change.path[1])
282
blob = symlink_to_blob(target)
283
shamap[change.path[1]] = blob.id
284
if add_cache_entry is not None:
286
blob, (change.file_id, tree.get_file_revision(change.path[1])), change.path[1])
288
find_unchanged_parent_ie(
289
change.path[1], change.kind[1], target, other_parent_trees)
291
if change.changed_content:
292
yield (change.path[1], blob,
293
(change.file_id, tree.get_file_revision(change.path[1])))
294
elif change.kind[1] is None:
295
shamap[change.path[1]] = None
296
elif change.kind[1] != 'directory':
297
raise AssertionError(change.kind[1])
298
for p in change.path:
301
dirty_dirs.add(osutils.dirname(p))
303
# Fetch contents of the blobs that were changed
304
for (path, file_id), chunks in tree.iter_files_bytes(
305
[(path, (path, file_id)) for (path, file_id) in new_blobs]):
308
if add_cache_entry is not None:
309
add_cache_entry(obj, (file_id, tree.get_file_revision(path)), path)
310
yield path, obj, (file_id, tree.get_file_revision(path))
311
shamap[path] = obj.id
313
for path in unusual_modes:
314
dirty_dirs.add(posixpath.dirname(path))
316
for dir in list(dirty_dirs):
317
for parent in osutils.parent_directories(dir):
318
if parent in dirty_dirs:
320
dirty_dirs.add(parent)
325
def ie_to_hexsha(path, ie):
330
# FIXME: Should be the same as in parent
331
if ie.kind == "file":
333
return idmap.lookup_blob_id(ie.file_id, ie.revision)
337
blob.data = tree.get_file_text(path)
338
if add_cache_entry is not None:
339
add_cache_entry(blob, (ie.file_id, ie.revision), path)
341
elif ie.kind == "symlink":
343
return idmap.lookup_blob_id(ie.file_id, ie.revision)
346
target = tree.get_symlink_target(path)
347
blob = symlink_to_blob(target)
348
if add_cache_entry is not None:
349
add_cache_entry(blob, (ie.file_id, ie.revision), path)
351
elif ie.kind == "directory":
352
# Not all cache backends store the tree information,
353
# calculate again from scratch
354
ret = directory_to_tree(
355
path, ie.children.values(), ie_to_hexsha, unusual_modes,
356
dummy_file_name, ie.parent_id is None)
363
for path in sorted(dirty_dirs, reverse=True):
364
if not tree.has_filename(path):
367
if tree.kind(path) != 'directory':
370
obj = directory_to_tree(
371
path, tree.iter_child_entries(path), ie_to_hexsha, unusual_modes,
372
dummy_file_name, path == '')
375
file_id = tree.path2id(path)
376
if add_cache_entry is not None:
377
add_cache_entry(obj, (file_id, tree.get_revision_id()), path)
378
yield path, obj, (file_id, tree.get_revision_id())
379
shamap[path] = obj.id
382
class PackTupleIterable(object):
384
def __init__(self, store):
386
self.store.lock_read()
392
def add(self, sha, path):
393
self.objects[sha] = path
396
return len(self.objects)
399
return ((self.store[object_id], path) for (object_id, path) in
400
self.objects.items())
403
class BazaarObjectStore(BaseObjectStore):
404
"""A Git-style object store backed onto a Bazaar repository."""
406
def __init__(self, repository, mapping=None):
407
self.repository = repository
408
self._map_updated = False
411
self.mapping = default_mapping
413
self.mapping = mapping
414
self._cache = cache_from_repository(repository)
415
self._content_cache_types = ("tree",)
416
self.start_write_group = self._cache.idmap.start_write_group
417
self.abort_write_group = self._cache.idmap.abort_write_group
418
self.commit_write_group = self._cache.idmap.commit_write_group
419
self.tree_cache = LRUTreeCache(self.repository)
420
self.unpeel_map = UnpeelMap.from_repository(self.repository)
422
def _missing_revisions(self, revisions):
423
return self._cache.idmap.missing_revisions(revisions)
425
def _update_sha_map(self, stop_revision=None):
426
if not self.is_locked():
427
raise errors.LockNotHeld(self)
428
if self._map_updated:
430
if (stop_revision is not None and
431
not self._missing_revisions([stop_revision])):
433
graph = self.repository.get_graph()
434
if stop_revision is None:
435
all_revids = self.repository.all_revision_ids()
436
missing_revids = self._missing_revisions(all_revids)
438
heads = set([stop_revision])
439
missing_revids = self._missing_revisions(heads)
441
parents = graph.get_parent_map(heads)
443
for p in parents.values():
444
todo.update([x for x in p if x not in missing_revids])
445
heads = self._missing_revisions(todo)
446
missing_revids.update(heads)
447
if NULL_REVISION in missing_revids:
448
missing_revids.remove(NULL_REVISION)
449
missing_revids = self.repository.has_revisions(missing_revids)
450
if not missing_revids:
451
if stop_revision is None:
452
self._map_updated = True
454
self.start_write_group()
456
with ui.ui_factory.nested_progress_bar() as pb:
457
for i, revid in enumerate(graph.iter_topo_order(
459
trace.mutter('processing %r', revid)
460
pb.update("updating git map", i, len(missing_revids))
461
self._update_sha_map_revision(revid)
462
if stop_revision is None:
463
self._map_updated = True
464
except BaseException:
465
self.abort_write_group()
468
self.commit_write_group()
471
self._update_sha_map()
472
return iter(self._cache.idmap.sha1s())
474
def _reconstruct_commit(self, rev, tree_sha, lossy, verifiers):
475
"""Reconstruct a Commit object.
477
:param rev: Revision object
478
:param tree_sha: SHA1 of the root tree object
479
:param lossy: Whether or not to roundtrip bzr metadata
480
:param verifiers: Verifiers for the commits
481
:return: Commit object
483
def parent_lookup(revid):
485
return self._lookup_revision_sha1(revid)
486
except errors.NoSuchRevision:
488
return self.mapping.export_commit(rev, tree_sha, parent_lookup,
491
def _revision_to_objects(self, rev, tree, lossy, add_cache_entry=None):
492
"""Convert a revision to a set of git objects.
494
:param rev: Bazaar revision object
495
:param tree: Bazaar revision tree
496
:param lossy: Whether to not roundtrip all Bazaar revision data
498
unusual_modes = extract_unusual_modes(rev)
499
present_parents = self.repository.has_revisions(rev.parent_ids)
500
parent_trees = self.tree_cache.revision_trees(
501
[p for p in rev.parent_ids if p in present_parents])
503
for path, obj, bzr_key_data in _tree_to_objects(
504
tree, parent_trees, self._cache.idmap, unusual_modes,
505
self.mapping.BZR_DUMMY_FILE, add_cache_entry):
508
root_key_data = bzr_key_data
509
# Don't yield just yet
512
if root_tree is None:
513
# Pointless commit - get the tree sha elsewhere
514
if not rev.parent_ids:
517
base_sha1 = self._lookup_revision_sha1(rev.parent_ids[0])
518
root_tree = self[self[base_sha1].tree]
519
root_key_data = (tree.path2id(''), tree.get_revision_id())
520
if add_cache_entry is not None:
521
add_cache_entry(root_tree, root_key_data, "")
524
testament3 = StrictTestament3(rev, tree)
525
verifiers = {"testament3-sha1": testament3.as_sha1()}
528
commit_obj = self._reconstruct_commit(rev, root_tree.id,
529
lossy=lossy, verifiers=verifiers)
531
foreign_revid, mapping = mapping_registry.parse_revision_id(
533
except errors.InvalidRevisionId:
536
_check_expected_sha(foreign_revid, commit_obj)
537
if add_cache_entry is not None:
538
add_cache_entry(commit_obj, verifiers, None)
540
yield None, commit_obj
542
def _get_updater(self, rev):
543
return self._cache.get_updater(rev)
545
def _update_sha_map_revision(self, revid):
546
rev = self.repository.get_revision(revid)
547
tree = self.tree_cache.revision_tree(rev.revision_id)
548
updater = self._get_updater(rev)
549
# FIXME JRV 2011-12-15: Shouldn't we try both values for lossy ?
550
for path, obj in self._revision_to_objects(
551
rev, tree, lossy=(not self.mapping.roundtripping),
552
add_cache_entry=updater.add_object):
553
if isinstance(obj, Commit):
555
commit_obj = updater.finish()
558
def _reconstruct_blobs(self, keys):
559
"""Return a Git Blob object from a fileid and revision stored in bzr.
561
:param fileid: File id of the text
562
:param revision: Revision of the text
564
stream = self.repository.iter_files_bytes(
565
((key[0], key[1], key) for key in keys))
566
for (file_id, revision, expected_sha), chunks in stream:
568
blob.chunked = chunks
569
if blob.id != expected_sha and blob.data == b"":
570
# Perhaps it's a symlink ?
571
tree = self.tree_cache.revision_tree(revision)
572
path = tree.id2path(file_id)
573
if tree.kind(path) == 'symlink':
574
blob = symlink_to_blob(tree.get_symlink_target(path))
575
_check_expected_sha(expected_sha, blob)
578
def _reconstruct_tree(self, fileid, revid, bzr_tree, unusual_modes,
580
"""Return a Git Tree object from a file id and a revision stored in bzr.
582
:param fileid: fileid in the tree.
583
:param revision: Revision of the tree.
585
def get_ie_sha1(path, entry):
586
if entry.kind == "directory":
588
return self._cache.idmap.lookup_tree_id(entry.file_id,
590
except (NotImplementedError, KeyError):
591
obj = self._reconstruct_tree(
592
entry.file_id, revid, bzr_tree, unusual_modes)
597
elif entry.kind in ("file", "symlink"):
599
return self._cache.idmap.lookup_blob_id(entry.file_id,
603
return next(self._reconstruct_blobs(
604
[(entry.file_id, entry.revision, None)])).id
605
elif entry.kind == 'tree-reference':
606
# FIXME: Make sure the file id is the root id
607
return self._lookup_revision_sha1(entry.reference_revision)
609
raise AssertionError("unknown entry kind '%s'" % entry.kind)
610
path = bzr_tree.id2path(fileid)
611
tree = directory_to_tree(
613
bzr_tree.iter_child_entries(path),
614
get_ie_sha1, unusual_modes, self.mapping.BZR_DUMMY_FILE,
615
bzr_tree.path2id('') == fileid)
617
_check_expected_sha(expected_sha, tree)
620
def get_parents(self, sha):
621
"""Retrieve the parents of a Git commit by SHA1.
623
:param sha: SHA1 of the commit
624
:raises: KeyError, NotCommitError
626
return self[sha].parents
628
def _lookup_revision_sha1(self, revid):
629
"""Return the SHA1 matching a Bazaar revision."""
630
if revid == NULL_REVISION:
633
return self._cache.idmap.lookup_commit(revid)
636
return mapping_registry.parse_revision_id(revid)[0]
637
except errors.InvalidRevisionId:
638
self._update_sha_map(revid)
639
return self._cache.idmap.lookup_commit(revid)
641
def get_raw(self, sha):
642
"""Get the raw representation of a Git object by SHA1.
644
:param sha: SHA1 of the git object
647
sha = sha_to_hex(sha)
649
return (obj.type, obj.as_raw_string())
651
def __contains__(self, sha):
652
# See if sha is in map
654
for (type, type_data) in self.lookup_git_sha(sha):
656
if self.repository.has_revision(type_data[0]):
659
if type_data in self.repository.texts:
662
if self.repository.has_revision(type_data[1]):
665
raise AssertionError("Unknown object type '%s'" % type)
673
self._map_updated = False
674
self.repository.lock_read()
675
return LogicalLockResult(self.unlock)
677
def lock_write(self):
679
self._map_updated = False
680
self.repository.lock_write()
681
return LogicalLockResult(self.unlock)
684
return (self._locked is not None)
688
self._map_updated = False
689
self.repository.unlock()
691
def lookup_git_shas(self, shas):
695
ret[sha] = [("commit", (NULL_REVISION, None, {}))]
698
ret[sha] = list(self._cache.idmap.lookup_git_sha(sha))
700
# if not, see if there are any unconverted revisions and
701
# add them to the map, search for sha in map again
702
self._update_sha_map()
704
ret[sha] = list(self._cache.idmap.lookup_git_sha(sha))
709
def lookup_git_sha(self, sha):
710
return self.lookup_git_shas([sha])[sha]
712
def __getitem__(self, sha):
713
for (kind, type_data) in self.lookup_git_sha(sha):
714
# convert object to git object
716
(revid, tree_sha, verifiers) = type_data
718
rev = self.repository.get_revision(revid)
719
except errors.NoSuchRevision:
720
if revid == NULL_REVISION:
721
raise AssertionError(
722
"should not try to look up NULL_REVISION")
723
trace.mutter('entry for %s %s in shamap: %r, but not '
724
'found in repository', kind, sha, type_data)
726
# FIXME: the type data should say whether conversion was
728
commit = self._reconstruct_commit(
729
rev, tree_sha, lossy=(not self.mapping.roundtripping),
731
_check_expected_sha(sha, commit)
734
(fileid, revision) = type_data
735
blobs = self._reconstruct_blobs([(fileid, revision, sha)])
738
(fileid, revid) = type_data
740
tree = self.tree_cache.revision_tree(revid)
741
rev = self.repository.get_revision(revid)
742
except errors.NoSuchRevision:
744
'entry for %s %s in shamap: %r, but not found in '
745
'repository', kind, sha, type_data)
747
unusual_modes = extract_unusual_modes(rev)
749
return self._reconstruct_tree(
750
fileid, revid, tree, unusual_modes, expected_sha=sha)
751
except errors.NoSuchRevision:
754
raise AssertionError("Unknown object type '%s'" % kind)
758
def generate_lossy_pack_data(self, have, want, progress=None,
759
get_tagged=None, ofs_delta=False):
760
return pack_objects_to_data(
761
self.generate_pack_contents(have, want, progress, get_tagged,
764
def generate_pack_contents(self, have, want, progress=None,
765
ofs_delta=False, get_tagged=None, lossy=False):
766
"""Iterate over the contents of a pack file.
768
:param have: List of SHA1s of objects that should not be sent
769
:param want: List of SHA1s of objects that should be sent
772
ret = self.lookup_git_shas(have + want)
773
for commit_sha in have:
774
commit_sha = self.unpeel_map.peel_tag(commit_sha, commit_sha)
776
for (type, type_data) in ret[commit_sha]:
778
raise AssertionError("Type was %s, not commit" % type)
779
processed.add(type_data[0])
781
trace.mutter("unable to find remote ref %s", commit_sha)
783
for commit_sha in want:
784
if commit_sha in have:
787
for (type, type_data) in ret[commit_sha]:
789
raise AssertionError("Type was %s, not commit" % type)
790
pending.add(type_data[0])
794
graph = self.repository.get_graph()
795
todo = _find_missing_bzr_revids(graph, pending, processed)
796
ret = PackTupleIterable(self)
797
with ui.ui_factory.nested_progress_bar() as pb:
798
for i, revid in enumerate(graph.iter_topo_order(todo)):
799
pb.update("generating git objects", i, len(todo))
801
rev = self.repository.get_revision(revid)
802
except errors.NoSuchRevision:
804
tree = self.tree_cache.revision_tree(revid)
805
for path, obj in self._revision_to_objects(
806
rev, tree, lossy=lossy):
807
ret.add(obj.id, path)
810
def add_thin_pack(self):
813
fd, path = tempfile.mkstemp(suffix=".pack")
814
f = os.fdopen(fd, 'wb')
817
from .fetch import import_git_objects
820
if os.path.getsize(path) == 0:
823
pd.create_index_v2(path[:-5] + ".idx", self.object_store.get_raw)
826
with self.repository.lock_write():
827
self.repository.start_write_group()
829
import_git_objects(self.repository, self.mapping,
830
p.iterobjects(get_raw=self.get_raw),
832
except BaseException:
833
self.repository.abort_write_group()
836
self.repository.commit_write_group()
839
# The pack isn't kept around anyway, so no point
840
# in treating full packs different from thin packs
841
add_pack = add_thin_pack
added
removed
breezy/git/object_store.py
