1
# Copyright (C) 2009 Jelmer Vernooij <jelmer@samba.org>
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""Map from Git sha's to Bazaar objects."""
19
from dulwich.objects import (
25
from dulwich.object_store import (
36
from bzrlib.revision import (
39
from bzrlib.testament import(
43
from bzrlib.plugins.git.mapping import (
46
extract_unusual_modes,
50
from bzrlib.plugins.git.cache import (
51
from_repository as cache_from_repository,
58
def get_object_store(repo, mapping=None):
59
git = getattr(repo, "_git", None)
61
return git.object_store
62
return BazaarObjectStore(repo, mapping)
65
MAX_TREE_CACHE_SIZE = 50 * 1024 * 1024
68
class LRUTreeCache(object):
70
def __init__(self, repository):
71
def approx_tree_size(tree):
72
# Very rough estimate, 1k per inventory entry
73
return len(tree.inventory) * 1024
74
self.repository = repository
75
self._cache = lru_cache.LRUSizeCache(max_size=MAX_TREE_CACHE_SIZE,
76
after_cleanup_size=None, compute_size=approx_tree_size)
78
def revision_tree(self, revid):
80
tree = self._cache[revid]
82
tree = self.repository.revision_tree(revid)
84
assert tree.get_revision_id() == tree.inventory.revision_id
87
def iter_revision_trees(self, revids):
92
tree = self._cache[revid]
96
assert tree.get_revision_id() == revid
97
assert tree.inventory.revision_id == revid
99
for tree in self.repository.revision_trees(todo):
100
trees[tree.get_revision_id()] = tree
102
return (trees[r] for r in revids)
104
def revision_trees(self, revids):
105
return list(self.iter_revision_trees(revids))
108
self._cache.add(tree.get_revision_id(), tree)
111
def _find_missing_bzr_revids(get_parent_map, want, have):
112
"""Find the revisions that have to be pushed.
114
:param get_parent_map: Function that returns the parents for a sequence
116
:param want: Revisions the target wants
117
:param have: Revisions the target already has
118
:return: Set of revisions to fetch
120
pending = want - have
124
processed.update(pending)
125
next_map = get_parent_map(pending)
127
for item in next_map.iteritems():
131
next_pending.update(p for p in item[1] if p not in processed)
132
pending = next_pending
133
if NULL_REVISION in todo:
134
todo.remove(NULL_REVISION)
138
def _check_expected_sha(expected_sha, object):
139
"""Check whether an object matches an expected SHA.
141
:param expected_sha: None or expected SHA as either binary or as hex digest
142
:param object: Object to verify
144
if expected_sha is None:
146
if len(expected_sha) == 40:
147
if expected_sha != object.sha().hexdigest():
148
raise AssertionError("Invalid sha for %r: %s" % (object,
150
elif len(expected_sha) == 20:
151
if expected_sha != object.sha().digest():
152
raise AssertionError("Invalid sha for %r: %s" % (object,
153
sha_to_hex(expected_sha)))
155
raise AssertionError("Unknown length %d for %r" % (len(expected_sha),
159
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes,
160
dummy_file_name=None):
161
"""Iterate over the objects that were introduced in a revision.
164
:param parent_trees: Parent revision trees
165
:param unusual_modes: Unusual file modes dictionary
166
:param dummy_file_name: File name to use for dummy files
167
in empty directories. None to skip empty directories
168
:return: Yields (path, object, ie) entries
174
base_tree = parent_trees[0]
175
other_parent_trees = parent_trees[1:]
177
base_tree = tree._repository.revision_tree(NULL_REVISION)
178
other_parent_trees = []
179
def find_unchanged_parent_ie(ie, parent_trees):
180
assert ie.kind in ("symlink", "file")
181
for ptree in parent_trees:
183
pie = ptree.inventory[ie.file_id]
184
except errors.NoSuchId:
187
if (pie.text_sha1 == ie.text_sha1 and
188
pie.kind == ie.kind and
189
pie.symlink_target == ie.symlink_target):
193
# Find all the changed blobs
194
for (file_id, path, changed_content, versioned, parent, name, kind,
195
executable) in tree.iter_changes(base_tree):
196
if kind[1] == "file":
197
ie = tree.inventory[file_id]
200
pie = find_unchanged_parent_ie(ie, other_parent_trees)
205
shamap[ie.file_id] = idmap.lookup_blob_id(
206
pie.file_id, pie.revision)
210
blob.data = tree.get_file_text(ie.file_id)
211
shamap[ie.file_id] = blob.id
212
if not file_id in shamap:
213
new_blobs.append((path[1], ie))
214
new_trees[posixpath.dirname(path[1])] = parent[1]
215
elif kind[1] == "symlink":
216
ie = tree.inventory[file_id]
218
blob = symlink_to_blob(ie)
219
shamap[file_id] = blob.id
221
find_unchanged_parent_ie(ie, other_parent_trees)
223
yield path[1], blob, ie
224
new_trees[posixpath.dirname(path[1])] = parent[1]
225
elif kind[1] not in (None, "directory"):
226
raise AssertionError(kind[1])
227
if (path[0] not in (None, "") and
228
parent[0] in tree.inventory and
229
tree.inventory[parent[0]].kind == "directory"):
231
new_trees[posixpath.dirname(path[0])] = parent[0]
233
# Fetch contents of the blobs that were changed
234
for (path, ie), chunks in tree.iter_files_bytes(
235
[(ie.file_id, (path, ie)) for (path, ie) in new_blobs]):
239
shamap[ie.file_id] = obj.id
241
for path in unusual_modes:
242
parent_path = posixpath.dirname(path)
243
new_trees[parent_path] = tree.path2id(parent_path)
247
items = new_trees.items()
249
for path, file_id in items:
250
parent_id = tree.inventory[file_id].parent_id
251
if parent_id is not None:
252
parent_path = urlutils.dirname(path)
253
new_trees[parent_path] = parent_id
254
trees[path] = file_id
256
def ie_to_hexsha(ie):
258
return shamap[ie.file_id]
260
# FIXME: Should be the same as in parent
261
if ie.kind in ("file", "symlink"):
263
return idmap.lookup_blob_id(ie.file_id, ie.revision)
267
blob.data = tree.get_file_text(ie.file_id)
269
elif ie.kind == "directory":
270
# Not all cache backends store the tree information,
271
# calculate again from scratch
272
ret = directory_to_tree(ie, ie_to_hexsha, unusual_modes,
280
for path in sorted(trees.keys(), reverse=True):
281
ie = tree.inventory[trees[path]]
282
assert ie.kind == "directory"
283
obj = directory_to_tree(ie, ie_to_hexsha, unusual_modes,
287
shamap[ie.file_id] = obj.id
290
class BazaarObjectStore(BaseObjectStore):
291
"""A Git-style object store backed onto a Bazaar repository."""
293
def __init__(self, repository, mapping=None):
294
self.repository = repository
296
self.mapping = default_mapping
298
self.mapping = mapping
299
self._cache = cache_from_repository(repository)
300
self._content_cache_types = ("tree")
301
self.start_write_group = self._cache.idmap.start_write_group
302
self.abort_write_group = self._cache.idmap.abort_write_group
303
self.commit_write_group = self._cache.idmap.commit_write_group
304
self.tree_cache = LRUTreeCache(self.repository)
306
def _update_sha_map(self, stop_revision=None):
307
graph = self.repository.get_graph()
308
if stop_revision is None:
309
heads = graph.heads(self.repository.all_revision_ids())
311
heads = set([stop_revision])
312
missing_revids = self._cache.idmap.missing_revisions(heads)
314
parents = graph.get_parent_map(heads)
316
for p in parents.values():
317
todo.update([x for x in p if x not in missing_revids])
318
heads = self._cache.idmap.missing_revisions(todo)
319
missing_revids.update(heads)
320
if NULL_REVISION in missing_revids:
321
missing_revids.remove(NULL_REVISION)
322
missing_revids = self.repository.has_revisions(missing_revids)
323
if not missing_revids:
325
self.start_write_group()
327
pb = ui.ui_factory.nested_progress_bar()
329
for i, revid in enumerate(graph.iter_topo_order(missing_revids)):
330
trace.mutter('processing %r', revid)
331
pb.update("updating git map", i, len(missing_revids))
332
self._update_sha_map_revision(revid)
336
self.abort_write_group()
339
self.commit_write_group()
342
self._update_sha_map()
343
return iter(self._cache.idmap.sha1s())
345
def _reconstruct_commit(self, rev, tree_sha, roundtrip, verifiers):
346
"""Reconstruct a Commit object.
348
:param rev: Revision object
349
:param tree_sha: SHA1 of the root tree object
350
:param roundtrip: Whether or not to roundtrip bzr metadata
351
:param verifiers: Verifiers for the commits
352
:return: Commit object
354
def parent_lookup(revid):
356
return self._lookup_revision_sha1(revid)
357
except errors.NoSuchRevision:
359
return self.mapping.export_commit(rev, tree_sha, parent_lookup,
360
roundtrip, verifiers)
362
def _create_fileid_map_blob(self, inv):
363
# FIXME: This can probably be a lot more efficient,
364
# not all files necessarily have to be processed.
366
for (path, ie) in inv.iter_entries():
367
if self.mapping.generate_file_id(path) != ie.file_id:
368
file_ids[path] = ie.file_id
369
return self.mapping.export_fileid_map(file_ids)
371
def _revision_to_objects(self, rev, tree, roundtrip):
372
"""Convert a revision to a set of git objects.
374
:param rev: Bazaar revision object
375
:param tree: Bazaar revision tree
376
:param roundtrip: Whether to roundtrip all Bazaar revision data
378
unusual_modes = extract_unusual_modes(rev)
379
present_parents = self.repository.has_revisions(rev.parent_ids)
380
parent_trees = self.tree_cache.revision_trees(
381
[p for p in rev.parent_ids if p in present_parents])
383
for path, obj, ie in _tree_to_objects(tree, parent_trees,
384
self._cache.idmap, unusual_modes, self.mapping.BZR_DUMMY_FILE):
388
# Don't yield just yet
391
if root_tree is None:
392
# Pointless commit - get the tree sha elsewhere
393
if not rev.parent_ids:
396
base_sha1 = self._lookup_revision_sha1(rev.parent_ids[0])
397
root_tree = self[self[base_sha1].tree]
398
root_ie = tree.inventory.root
399
if roundtrip and self.mapping.BZR_FILE_IDS_FILE is not None:
400
b = self._create_fileid_map_blob(tree.inventory)
402
root_tree[self.mapping.BZR_FILE_IDS_FILE] = ((stat.S_IFREG | 0644), b.id)
403
yield self.mapping.BZR_FILE_IDS_FILE, b, None
404
yield "", root_tree, root_ie
406
testament3 = StrictTestament3(rev, tree.inventory)
407
verifiers = { "testament3-sha1": testament3.as_sha1() }
410
commit_obj = self._reconstruct_commit(rev, root_tree.id,
411
roundtrip=roundtrip, verifiers=verifiers)
413
foreign_revid, mapping = mapping_registry.parse_revision_id(
415
except errors.InvalidRevisionId:
418
_check_expected_sha(foreign_revid, commit_obj)
419
yield None, commit_obj, None
421
def _get_updater(self, rev):
422
return self._cache.get_updater(rev)
424
def _update_sha_map_revision(self, revid):
425
rev = self.repository.get_revision(revid)
426
tree = self.tree_cache.revision_tree(rev.revision_id)
427
updater = self._get_updater(rev)
428
for path, obj, ie in self._revision_to_objects(rev, tree,
430
if isinstance(obj, Commit):
431
testament3 = StrictTestament3(rev, tree.inventory)
432
ie = { "testament3-sha1": testament3.as_sha1() }
433
updater.add_object(obj, ie, path)
434
commit_obj = updater.finish()
437
def _reconstruct_blobs(self, keys):
438
"""Return a Git Blob object from a fileid and revision stored in bzr.
440
:param fileid: File id of the text
441
:param revision: Revision of the text
443
stream = self.repository.iter_files_bytes(
444
((key[0], key[1], key) for key in keys))
445
for (fileid, revision, expected_sha), chunks in stream:
447
blob.chunked = chunks
448
if blob.id != expected_sha and blob.data == "":
449
# Perhaps it's a symlink ?
450
tree = self.tree_cache.revision_tree(revision)
451
entry = tree.inventory[fileid]
452
if entry.kind == 'symlink':
453
blob = symlink_to_blob(entry)
454
_check_expected_sha(expected_sha, blob)
457
def _reconstruct_tree(self, fileid, revid, inv, unusual_modes,
459
"""Return a Git Tree object from a file id and a revision stored in bzr.
461
:param fileid: fileid in the tree.
462
:param revision: Revision of the tree.
464
def get_ie_sha1(entry):
465
if entry.kind == "directory":
467
return self._cache.idmap.lookup_tree_id(entry.file_id,
469
except (NotImplementedError, KeyError):
470
obj = self._reconstruct_tree(entry.file_id, revid, inv,
476
elif entry.kind in ("file", "symlink"):
478
return self._cache.idmap.lookup_blob_id(entry.file_id,
482
return self._reconstruct_blobs(
483
[(entry.file_id, entry.revision, None)]).next().id
485
raise AssertionError("unknown entry kind '%s'" % entry.kind)
486
tree = directory_to_tree(inv[fileid], get_ie_sha1, unusual_modes,
487
self.mapping.BZR_DUMMY_FILE)
488
if (inv.root.file_id == fileid and
489
self.mapping.BZR_FILE_IDS_FILE is not None):
490
b = self._create_fileid_map_blob(inv)
491
# If this is the root tree, add the file ids
492
tree[self.mapping.BZR_FILE_IDS_FILE] = ((stat.S_IFREG | 0644), b.id)
493
_check_expected_sha(expected_sha, tree)
496
def get_parents(self, sha):
497
"""Retrieve the parents of a Git commit by SHA1.
499
:param sha: SHA1 of the commit
500
:raises: KeyError, NotCommitError
502
return self[sha].parents
504
def _lookup_revision_sha1(self, revid):
505
"""Return the SHA1 matching a Bazaar revision."""
506
from dulwich.protocol import ZERO_SHA
507
if revid == NULL_REVISION:
510
return self._cache.idmap.lookup_commit(revid)
513
return mapping_registry.parse_revision_id(revid)[0]
514
except errors.InvalidRevisionId:
515
self.repository.lock_read()
517
self._update_sha_map(revid)
519
self.repository.unlock()
520
return self._cache.idmap.lookup_commit(revid)
522
def get_raw(self, sha):
523
"""Get the raw representation of a Git object by SHA1.
525
:param sha: SHA1 of the git object
528
return (obj.type, obj.as_raw_string())
530
def __contains__(self, sha):
531
# See if sha is in map
533
(type, type_data) = self.lookup_git_sha(sha)
535
return self.repository.has_revision(type_data[0])
537
return self.repository.texts.has_key(type_data)
539
return self.repository.has_revision(type_data[1])
541
raise AssertionError("Unknown object type '%s'" % type)
545
def lookup_git_shas(self, shas, update_map=True):
546
from dulwich.protocol import ZERO_SHA
550
ret[sha] = ("commit", (NULL_REVISION, None, {}))
553
ret[sha] = self._cache.idmap.lookup_git_sha(sha)
556
# if not, see if there are any unconverted revisions and add
557
# them to the map, search for sha in map again
558
self._update_sha_map()
561
ret[sha] = self._cache.idmap.lookup_git_sha(sha)
566
def lookup_git_sha(self, sha, update_map=True):
567
return self.lookup_git_shas([sha], update_map=update_map)[sha]
569
def __getitem__(self, sha):
570
if self._cache.content_cache is not None:
572
return self._cache.content_cache[sha]
575
(type, type_data) = self.lookup_git_sha(sha)
576
# convert object to git object
578
(revid, tree_sha, verifiers) = type_data
580
rev = self.repository.get_revision(revid)
581
except errors.NoSuchRevision:
582
trace.mutter('entry for %s %s in shamap: %r, but not found in '
583
'repository', type, sha, type_data)
585
commit = self._reconstruct_commit(rev, tree_sha, roundtrip=True,
587
_check_expected_sha(sha, commit)
590
(fileid, revision) = type_data
591
return self._reconstruct_blobs([(fileid, revision, sha)]).next()
593
(fileid, revid) = type_data
595
tree = self.tree_cache.revision_tree(revid)
596
rev = self.repository.get_revision(revid)
597
except errors.NoSuchRevision:
598
trace.mutter('entry for %s %s in shamap: %r, but not found in repository', type, sha, type_data)
600
unusual_modes = extract_unusual_modes(rev)
602
return self._reconstruct_tree(fileid, revid, tree.inventory,
603
unusual_modes, expected_sha=sha)
604
except errors.NoSuchRevision:
607
raise AssertionError("Unknown object type '%s'" % type)
609
def generate_lossy_pack_contents(self, have, want, progress=None,
611
return self.generate_pack_contents(have, want, progress, get_tagged,
614
def generate_pack_contents(self, have, want, progress=None,
615
get_tagged=None, lossy=False):
616
"""Iterate over the contents of a pack file.
618
:param have: List of SHA1s of objects that should not be sent
619
:param want: List of SHA1s of objects that should be sent
622
ret = self.lookup_git_shas(have + want)
623
for commit_sha in have:
625
(type, (revid, tree_sha)) = ret[commit_sha]
629
assert type == "commit"
632
for commit_sha in want:
633
if commit_sha in have:
636
(type, (revid, tree_sha)) = ret[commit_sha]
640
assert type == "commit"
643
todo = _find_missing_bzr_revids(self.repository.get_parent_map,
645
trace.mutter('sending revisions %r', todo)
647
pb = ui.ui_factory.nested_progress_bar()
649
for i, revid in enumerate(todo):
650
pb.update("generating git objects", i, len(todo))
651
rev = self.repository.get_revision(revid)
652
tree = self.tree_cache.revision_tree(revid)
653
for path, obj, ie in self._revision_to_objects(rev, tree,
654
roundtrip=not lossy):
655
ret.append((obj, path))
660
def add_thin_pack(self):
663
fd, path = tempfile.mkstemp(suffix=".pack")
664
f = os.fdopen(fd, 'wb')
666
from dulwich.pack import PackData, Pack
667
from bzrlib.plugins.git.fetch import import_git_objects
670
if os.path.getsize(path) == 0:
673
pd.create_index_v2(path[:-5]+".idx", self.object_store.get_raw)
676
self.repository.lock_write()
678
self.repository.start_write_group()
680
import_git_objects(self.repository, self.mapping,
681
p.iterobjects(get_raw=self.get_raw),
684
self.repository.abort_write_group()
687
self.repository.commit_write_group()
689
self.repository.unlock()
692
# The pack isn't kept around anyway, so no point
693
# in treating full packs different from thin packs
694
add_pack = add_thin_pack
added
removed
object_store.py
