1
# Copyright (C) 2009-2012 Jelmer Vernooij <jelmer@samba.org>
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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
"""Map from Git sha's to Bazaar objects."""
20
from __future__ import absolute_import
22
from dulwich.objects import (
29
from dulwich.object_store import (
40
from ...lock import LogicalLockResult
41
from ...revision import (
44
from ...testament import(
49
from_repository as cache_from_repository,
51
from .mapping import (
54
extract_unusual_modes,
58
from .unpeel_map import (
66
def get_object_store(repo, mapping=None):
67
git = getattr(repo, "_git", None)
69
git.object_store.unlock = lambda: None
70
git.object_store.lock_read = lambda: LogicalLockResult(lambda: None)
71
git.object_store.lock_write = lambda: LogicalLockResult(lambda: None)
72
return git.object_store
73
return BazaarObjectStore(repo, mapping)
76
MAX_TREE_CACHE_SIZE = 50 * 1024 * 1024
79
class LRUTreeCache(object):
81
def __init__(self, repository):
82
def approx_tree_size(tree):
83
# Very rough estimate, 250 per inventory entry
85
inv = tree.root_inventory
86
except AttributeError:
89
self.repository = repository
90
self._cache = lru_cache.LRUSizeCache(max_size=MAX_TREE_CACHE_SIZE,
91
after_cleanup_size=None, compute_size=approx_tree_size)
93
def revision_tree(self, revid):
95
tree = self._cache[revid]
97
tree = self.repository.revision_tree(revid)
101
def iter_revision_trees(self, revids):
106
tree = self._cache[revid]
110
assert tree.get_revision_id() == revid
112
for tree in self.repository.revision_trees(todo):
113
trees[tree.get_revision_id()] = tree
115
return (trees[r] for r in revids)
117
def revision_trees(self, revids):
118
return list(self.iter_revision_trees(revids))
121
self._cache[tree.get_revision_id()] = tree
124
def _find_missing_bzr_revids(graph, want, have):
125
"""Find the revisions that have to be pushed.
127
:param get_parent_map: Function that returns the parents for a sequence
129
:param want: Revisions the target wants
130
:param have: Revisions the target already has
131
:return: Set of revisions to fetch
136
extra_todo = graph.find_unique_ancestors(rev, handled)
137
todo.update(extra_todo)
138
handled.update(extra_todo)
139
if NULL_REVISION in todo:
140
todo.remove(NULL_REVISION)
144
def _check_expected_sha(expected_sha, object):
145
"""Check whether an object matches an expected SHA.
147
:param expected_sha: None or expected SHA as either binary or as hex digest
148
:param object: Object to verify
150
if expected_sha is None:
152
if len(expected_sha) == 40:
153
if expected_sha != object.sha().hexdigest():
154
raise AssertionError("Invalid sha for %r: %s" % (object,
156
elif len(expected_sha) == 20:
157
if expected_sha != object.sha().digest():
158
raise AssertionError("Invalid sha for %r: %s" % (object,
159
sha_to_hex(expected_sha)))
161
raise AssertionError("Unknown length %d for %r" % (len(expected_sha),
165
def _tree_to_objects(tree, parent_trees, idmap, unusual_modes,
166
dummy_file_name=None):
167
"""Iterate over the objects that were introduced in a revision.
170
:param parent_trees: Parent revision trees
171
:param unusual_modes: Unusual file modes dictionary
172
:param dummy_file_name: File name to use for dummy files
173
in empty directories. None to skip empty directories
174
:return: Yields (path, object, ie) entries
180
base_tree = parent_trees[0]
181
other_parent_trees = parent_trees[1:]
183
base_tree = tree._repository.revision_tree(NULL_REVISION)
184
other_parent_trees = []
185
def find_unchanged_parent_ie(file_id, kind, other, parent_trees):
186
for ptree in parent_trees:
188
pkind = ptree.kind(file_id)
189
except errors.NoSuchId:
193
if (pkind == "file" and
194
ptree.get_file_sha1(file_id) == other):
195
return (file_id, ptree.get_file_revision(file_id))
196
if kind == "symlink":
197
if (pkind == "symlink" and
198
ptree.get_symlink_target(file_id) == other):
199
return (file_id, ptree.get_file_revision(file_id))
202
# Find all the changed blobs
203
for (file_id, path, changed_content, versioned, parent, name, kind,
204
executable) in tree.iter_changes(base_tree):
205
if kind[1] == "file":
208
(pfile_id, prevision) = find_unchanged_parent_ie(file_id, kind[1], tree.get_file_sha1(file_id), other_parent_trees)
213
shamap[file_id] = idmap.lookup_blob_id(
218
blob.data = tree.get_file_text(file_id)
219
shamap[file_id] = blob.id
220
if not file_id in shamap:
221
new_blobs.append((path[1], file_id))
222
elif kind[1] == "symlink":
224
target = tree.get_symlink_target(file_id)
225
blob = symlink_to_blob(target)
226
shamap[file_id] = blob.id
228
find_unchanged_parent_ie(file_id, kind[1], target, other_parent_trees)
230
yield path[1], blob, (file_id, tree.get_file_revision(file_id, path[1]))
231
elif kind[1] not in (None, "directory"):
232
raise AssertionError(kind[1])
234
if p and tree.has_id(p) and tree.kind(p) == "directory":
237
# Fetch contents of the blobs that were changed
238
for (path, file_id), chunks in tree.iter_files_bytes(
239
[(file_id, (path, file_id)) for (path, file_id) in new_blobs]):
242
yield path, obj, (file_id, tree.get_file_revision(file_id, path))
243
shamap[file_id] = obj.id
245
for path in unusual_modes:
246
parent_path = posixpath.dirname(path)
247
file_id = tree.path2id(parent_path)
248
assert file_id is not None, "Unable to find file id for %r" % parent_path
249
dirty_dirs.add(file_id)
252
inv = tree.root_inventory
253
except AttributeError:
259
for file_id in dirty_dirs:
260
if file_id is None or not inv.has_id(file_id):
262
trees[inv.id2path(file_id)] = file_id
264
if ie.parent_id is not None:
265
new_dirs.add(ie.parent_id)
266
dirty_dirs = new_dirs
268
def ie_to_hexsha(ie):
270
return shamap[ie.file_id]
272
# FIXME: Should be the same as in parent
273
if ie.kind in ("file", "symlink"):
275
return idmap.lookup_blob_id(ie.file_id, ie.revision)
279
blob.data = tree.get_file_text(ie.file_id)
281
elif ie.kind == "directory":
282
# Not all cache backends store the tree information,
283
# calculate again from scratch
284
ret = directory_to_tree(ie.children, ie_to_hexsha,
285
unusual_modes, dummy_file_name, ie.parent_id is None)
292
for path in sorted(trees.keys(), reverse=True):
293
file_id = trees[path]
294
assert tree.kind(file_id) == 'directory'
296
obj = directory_to_tree(ie.children, ie_to_hexsha, unusual_modes,
297
dummy_file_name, path == "")
299
yield path, obj, (file_id, )
300
shamap[file_id] = obj.id
303
class PackTupleIterable(object):
305
def __init__(self, store):
307
self.store.lock_read()
313
def add(self, sha, path):
314
self.objects[sha] = path
317
return len(self.objects)
320
return ((self.store[object_id], path) for (object_id, path) in
321
self.objects.iteritems())
324
class BazaarObjectStore(BaseObjectStore):
325
"""A Git-style object store backed onto a Bazaar repository."""
327
def __init__(self, repository, mapping=None):
328
self.repository = repository
329
self._map_updated = False
332
self.mapping = default_mapping
334
self.mapping = mapping
335
self._cache = cache_from_repository(repository)
336
self._content_cache_types = ("tree",)
337
self.start_write_group = self._cache.idmap.start_write_group
338
self.abort_write_group = self._cache.idmap.abort_write_group
339
self.commit_write_group = self._cache.idmap.commit_write_group
340
self.tree_cache = LRUTreeCache(self.repository)
341
self.unpeel_map = UnpeelMap.from_repository(self.repository)
343
def _missing_revisions(self, revisions):
344
return self._cache.idmap.missing_revisions(revisions)
346
def _update_sha_map(self, stop_revision=None):
347
if not self.is_locked():
348
raise AssertionError()
349
if self._map_updated:
351
if (stop_revision is not None and
352
not self._missing_revisions([stop_revision])):
354
graph = self.repository.get_graph()
355
if stop_revision is None:
356
all_revids = self.repository.all_revision_ids()
357
missing_revids = self._missing_revisions(all_revids)
359
heads = set([stop_revision])
360
missing_revids = self._missing_revisions(heads)
362
parents = graph.get_parent_map(heads)
364
for p in parents.values():
365
todo.update([x for x in p if x not in missing_revids])
366
heads = self._missing_revisions(todo)
367
missing_revids.update(heads)
368
if NULL_REVISION in missing_revids:
369
missing_revids.remove(NULL_REVISION)
370
missing_revids = self.repository.has_revisions(missing_revids)
371
if not missing_revids:
372
if stop_revision is None:
373
self._map_updated = True
375
self.start_write_group()
377
pb = ui.ui_factory.nested_progress_bar()
379
for i, revid in enumerate(graph.iter_topo_order(missing_revids)):
380
trace.mutter('processing %r', revid)
381
pb.update("updating git map", i, len(missing_revids))
382
self._update_sha_map_revision(revid)
385
if stop_revision is None:
386
self._map_updated = True
388
self.abort_write_group()
391
self.commit_write_group()
394
self._update_sha_map()
395
return iter(self._cache.idmap.sha1s())
397
def _reconstruct_commit(self, rev, tree_sha, lossy, verifiers):
398
"""Reconstruct a Commit object.
400
:param rev: Revision object
401
:param tree_sha: SHA1 of the root tree object
402
:param lossy: Whether or not to roundtrip bzr metadata
403
:param verifiers: Verifiers for the commits
404
:return: Commit object
406
def parent_lookup(revid):
408
return self._lookup_revision_sha1(revid)
409
except errors.NoSuchRevision:
411
return self.mapping.export_commit(rev, tree_sha, parent_lookup,
414
def _create_fileid_map_blob(self, tree):
415
# FIXME: This can probably be a lot more efficient,
416
# not all files necessarily have to be processed.
418
for (path, ie) in tree.inventory.iter_entries():
419
if self.mapping.generate_file_id(path) != ie.file_id:
420
file_ids[path] = ie.file_id
421
return self.mapping.export_fileid_map(file_ids)
423
def _revision_to_objects(self, rev, tree, lossy):
424
"""Convert a revision to a set of git objects.
426
:param rev: Bazaar revision object
427
:param tree: Bazaar revision tree
428
:param lossy: Whether to not roundtrip all Bazaar revision data
430
unusual_modes = extract_unusual_modes(rev)
431
present_parents = self.repository.has_revisions(rev.parent_ids)
432
parent_trees = self.tree_cache.revision_trees(
433
[p for p in rev.parent_ids if p in present_parents])
435
for path, obj, bzr_key_data in _tree_to_objects(tree, parent_trees,
436
self._cache.idmap, unusual_modes, self.mapping.BZR_DUMMY_FILE):
439
root_key_data = bzr_key_data
440
# Don't yield just yet
442
yield path, obj, bzr_key_data
443
if root_tree is None:
444
# Pointless commit - get the tree sha elsewhere
445
if not rev.parent_ids:
448
base_sha1 = self._lookup_revision_sha1(rev.parent_ids[0])
449
root_tree = self[self[base_sha1].tree]
450
root_key_data = (tree.get_root_id(), )
451
if not lossy and self.mapping.BZR_FILE_IDS_FILE is not None:
452
b = self._create_fileid_map_blob(tree)
454
root_tree[self.mapping.BZR_FILE_IDS_FILE] = (
455
(stat.S_IFREG | 0644), b.id)
456
yield self.mapping.BZR_FILE_IDS_FILE, b, None
457
yield "", root_tree, root_key_data
459
testament3 = StrictTestament3(rev, tree)
460
verifiers = { "testament3-sha1": testament3.as_sha1() }
463
commit_obj = self._reconstruct_commit(rev, root_tree.id,
464
lossy=lossy, verifiers=verifiers)
466
foreign_revid, mapping = mapping_registry.parse_revision_id(
468
except errors.InvalidRevisionId:
471
_check_expected_sha(foreign_revid, commit_obj)
472
yield None, commit_obj, None
474
def _get_updater(self, rev):
475
return self._cache.get_updater(rev)
477
def _update_sha_map_revision(self, revid):
478
rev = self.repository.get_revision(revid)
479
tree = self.tree_cache.revision_tree(rev.revision_id)
480
updater = self._get_updater(rev)
481
# FIXME JRV 2011-12-15: Shouldn't we try both values for lossy ?
482
for path, obj, ie in self._revision_to_objects(rev, tree, lossy=(not self.mapping.roundtripping)):
483
if isinstance(obj, Commit):
484
testament3 = StrictTestament3(rev, tree)
485
ie = { "testament3-sha1": testament3.as_sha1() }
486
updater.add_object(obj, ie, path)
487
commit_obj = updater.finish()
490
def _reconstruct_blobs(self, keys):
491
"""Return a Git Blob object from a fileid and revision stored in bzr.
493
:param fileid: File id of the text
494
:param revision: Revision of the text
496
stream = self.repository.iter_files_bytes(
497
((key[0], key[1], key) for key in keys))
498
for (fileid, revision, expected_sha), chunks in stream:
500
blob.chunked = chunks
501
if blob.id != expected_sha and blob.data == "":
502
# Perhaps it's a symlink ?
503
tree = self.tree_cache.revision_tree(revision)
504
if tree.kind(fileid) == 'symlink':
505
blob = symlink_to_blob(tree.get_symlink_target(fileid))
506
_check_expected_sha(expected_sha, blob)
509
def _reconstruct_tree(self, fileid, revid, bzr_tree, unusual_modes,
511
"""Return a Git Tree object from a file id and a revision stored in bzr.
513
:param fileid: fileid in the tree.
514
:param revision: Revision of the tree.
516
def get_ie_sha1(entry):
517
if entry.kind == "directory":
519
return self._cache.idmap.lookup_tree_id(entry.file_id,
521
except (NotImplementedError, KeyError):
522
obj = self._reconstruct_tree(entry.file_id, revid, bzr_tree,
528
elif entry.kind in ("file", "symlink"):
530
return self._cache.idmap.lookup_blob_id(entry.file_id,
534
return self._reconstruct_blobs(
535
[(entry.file_id, entry.revision, None)]).next().id
536
elif entry.kind == 'tree-reference':
537
# FIXME: Make sure the file id is the root id
538
return self._lookup_revision_sha1(entry.reference_revision)
540
raise AssertionError("unknown entry kind '%s'" % entry.kind)
542
inv = bzr_tree.root_inventory
543
except AttributeError:
544
inv = bzr_tree.inventory
545
tree = directory_to_tree(inv[fileid].children,
546
get_ie_sha1, unusual_modes, self.mapping.BZR_DUMMY_FILE,
547
bzr_tree.get_root_id() == fileid)
548
if (bzr_tree.get_root_id() == fileid and
549
self.mapping.BZR_FILE_IDS_FILE is not None):
552
b = self._create_fileid_map_blob(bzr_tree)
553
# If this is the root tree, add the file ids
554
tree[self.mapping.BZR_FILE_IDS_FILE] = (
555
(stat.S_IFREG | 0644), b.id)
557
_check_expected_sha(expected_sha, tree)
560
def get_parents(self, sha):
561
"""Retrieve the parents of a Git commit by SHA1.
563
:param sha: SHA1 of the commit
564
:raises: KeyError, NotCommitError
566
return self[sha].parents
568
def _lookup_revision_sha1(self, revid):
569
"""Return the SHA1 matching a Bazaar revision."""
570
if revid == NULL_REVISION:
573
return self._cache.idmap.lookup_commit(revid)
576
return mapping_registry.parse_revision_id(revid)[0]
577
except errors.InvalidRevisionId:
578
self._update_sha_map(revid)
579
return self._cache.idmap.lookup_commit(revid)
581
def get_raw(self, sha):
582
"""Get the raw representation of a Git object by SHA1.
584
:param sha: SHA1 of the git object
587
sha = sha_to_hex(sha)
589
return (obj.type, obj.as_raw_string())
591
def __contains__(self, sha):
592
# See if sha is in map
594
for (type, type_data) in self.lookup_git_sha(sha):
596
if self.repository.has_revision(type_data[0]):
599
if type_data in self.repository.texts:
602
if self.repository.has_revision(type_data[1]):
605
raise AssertionError("Unknown object type '%s'" % type)
613
self._map_updated = False
614
self.repository.lock_read()
615
return LogicalLockResult(self.unlock)
617
def lock_write(self):
619
self._map_updated = False
620
self.repository.lock_write()
621
return LogicalLockResult(self.unlock)
624
return (self._locked is not None)
628
self._map_updated = False
629
self.repository.unlock()
631
def lookup_git_shas(self, shas):
635
ret[sha] = [("commit", (NULL_REVISION, None, {}))]
638
ret[sha] = list(self._cache.idmap.lookup_git_sha(sha))
640
# if not, see if there are any unconverted revisions and
641
# add them to the map, search for sha in map again
642
self._update_sha_map()
644
ret[sha] = list(self._cache.idmap.lookup_git_sha(sha))
649
def lookup_git_sha(self, sha):
650
return self.lookup_git_shas([sha])[sha]
652
def __getitem__(self, sha):
653
if self._cache.content_cache is not None:
655
return self._cache.content_cache[sha]
658
for (kind, type_data) in self.lookup_git_sha(sha):
659
# convert object to git object
661
(revid, tree_sha, verifiers) = type_data
663
rev = self.repository.get_revision(revid)
664
except errors.NoSuchRevision:
665
if revid == NULL_REVISION:
666
raise AssertionError(
667
"should not try to look up NULL_REVISION")
668
trace.mutter('entry for %s %s in shamap: %r, but not '
669
'found in repository', kind, sha, type_data)
671
# FIXME: the type data should say whether conversion was lossless
672
commit = self._reconstruct_commit(rev, tree_sha,
673
lossy=(not self.mapping.roundtripping), verifiers=verifiers)
674
_check_expected_sha(sha, commit)
677
(fileid, revision) = type_data
678
blobs = self._reconstruct_blobs([(fileid, revision, sha)])
681
(fileid, revid) = type_data
683
tree = self.tree_cache.revision_tree(revid)
684
rev = self.repository.get_revision(revid)
685
except errors.NoSuchRevision:
686
trace.mutter('entry for %s %s in shamap: %r, but not found in '
687
'repository', kind, sha, type_data)
689
unusual_modes = extract_unusual_modes(rev)
691
return self._reconstruct_tree(fileid, revid,
692
tree, unusual_modes, expected_sha=sha)
693
except errors.NoSuchRevision:
696
raise AssertionError("Unknown object type '%s'" % kind)
700
def generate_lossy_pack_contents(self, have, want, progress=None,
702
return self.generate_pack_contents(have, want, progress, get_tagged,
705
def generate_pack_contents(self, have, want, progress=None,
706
get_tagged=None, lossy=False):
707
"""Iterate over the contents of a pack file.
709
:param have: List of SHA1s of objects that should not be sent
710
:param want: List of SHA1s of objects that should be sent
713
ret = self.lookup_git_shas(have + want)
714
for commit_sha in have:
715
commit_sha = self.unpeel_map.peel_tag(commit_sha, commit_sha)
717
for (type, type_data) in ret[commit_sha]:
718
assert type == "commit"
719
processed.add(type_data[0])
721
trace.mutter("unable to find remote ref %s", commit_sha)
723
for commit_sha in want:
724
if commit_sha in have:
727
for (type, type_data) in ret[commit_sha]:
728
assert type == "commit"
729
pending.add(type_data[0])
733
graph = self.repository.get_graph()
734
todo = _find_missing_bzr_revids(graph, pending, processed)
735
ret = PackTupleIterable(self)
736
pb = ui.ui_factory.nested_progress_bar()
738
for i, revid in enumerate(todo):
739
pb.update("generating git objects", i, len(todo))
741
rev = self.repository.get_revision(revid)
742
except errors.NoSuchRevision:
744
tree = self.tree_cache.revision_tree(revid)
745
for path, obj, ie in self._revision_to_objects(rev, tree, lossy=lossy):
746
ret.add(obj.id, path)
751
def add_thin_pack(self):
754
fd, path = tempfile.mkstemp(suffix=".pack")
755
f = os.fdopen(fd, 'wb')
757
from dulwich.pack import PackData, Pack
758
from .fetch import import_git_objects
761
if os.path.getsize(path) == 0:
764
pd.create_index_v2(path[:-5]+".idx", self.object_store.get_raw)
767
self.repository.lock_write()
769
self.repository.start_write_group()
771
import_git_objects(self.repository, self.mapping,
772
p.iterobjects(get_raw=self.get_raw),
775
self.repository.abort_write_group()
778
self.repository.commit_write_group()
780
self.repository.unlock()
783
# The pack isn't kept around anyway, so no point
784
# in treating full packs different from thin packs
785
add_pack = add_thin_pack
added
removed
object_store.py
