202
191
self.random_revid = False
204
def will_record_deletes(self):
205
"""Tell the commit builder that deletes are being notified.
207
This enables the accumulation of an inventory delta; for the resulting
208
commit to be valid, deletes against the basis MUST be recorded via
209
builder.record_delete().
211
raise NotImplementedError(self.will_record_deletes)
213
def record_iter_changes(self, tree, basis_revision_id, iter_changes):
214
"""Record a new tree via iter_changes.
216
:param tree: The tree to obtain text contents from for changed objects.
217
:param basis_revision_id: The revision id of the tree the iter_changes
218
has been generated against. Currently assumed to be the same
219
as self.parents[0] - if it is not, errors may occur.
220
:param iter_changes: An iter_changes iterator with the changes to apply
221
to basis_revision_id. The iterator must not include any items with
222
a current kind of None - missing items must be either filtered out
223
or errored-on beefore record_iter_changes sees the item.
224
:return: A generator of (file_id, relpath, fs_hash) tuples for use with
227
raise NotImplementedError(self.record_iter_changes)
230
class RepositoryWriteLockResult(LogicalLockResult):
231
"""The result of write locking a repository.
233
:ivar repository_token: The token obtained from the underlying lock, or
235
:ivar unlock: A callable which will unlock the lock.
238
def __init__(self, unlock, repository_token):
239
LogicalLockResult.__init__(self, unlock)
240
self.repository_token = repository_token
243
return "RepositoryWriteLockResult(%s, %s)" % (self.repository_token,
193
def _check_root(self, ie, parent_invs, tree):
194
"""Helper for record_entry_contents.
196
:param ie: An entry being added.
197
:param parent_invs: The inventories of the parent revisions of the
199
:param tree: The tree that is being committed.
201
# In this revision format, root entries have no knit or weave When
202
# serializing out to disk and back in root.revision is always
204
ie.revision = self._new_revision_id
206
def _get_delta(self, ie, basis_inv, path):
207
"""Get a delta against the basis inventory for ie."""
208
if ie.file_id not in basis_inv:
210
return (None, path, ie.file_id, ie)
211
elif ie != basis_inv[ie.file_id]:
213
# TODO: avoid tis id2path call.
214
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
219
def record_entry_contents(self, ie, parent_invs, path, tree,
221
"""Record the content of ie from tree into the commit if needed.
223
Side effect: sets ie.revision when unchanged
225
:param ie: An inventory entry present in the commit.
226
:param parent_invs: The inventories of the parent revisions of the
228
:param path: The path the entry is at in the tree.
229
:param tree: The tree which contains this entry and should be used to
231
:param content_summary: Summary data from the tree about the paths
232
content - stat, length, exec, sha/link target. This is only
233
accessed when the entry has a revision of None - that is when it is
234
a candidate to commit.
235
:return: A tuple (change_delta, version_recorded). change_delta is
236
an inventory_delta change for this entry against the basis tree of
237
the commit, or None if no change occured against the basis tree.
238
version_recorded is True if a new version of the entry has been
239
recorded. For instance, committing a merge where a file was only
240
changed on the other side will return (delta, False).
242
if self.new_inventory.root is None:
243
if ie.parent_id is not None:
244
raise errors.RootMissing()
245
self._check_root(ie, parent_invs, tree)
246
if ie.revision is None:
247
kind = content_summary[0]
249
# ie is carried over from a prior commit
251
# XXX: repository specific check for nested tree support goes here - if
252
# the repo doesn't want nested trees we skip it ?
253
if (kind == 'tree-reference' and
254
not self.repository._format.supports_tree_reference):
255
# mismatch between commit builder logic and repository:
256
# this needs the entry creation pushed down into the builder.
257
raise NotImplementedError('Missing repository subtree support.')
258
self.new_inventory.add(ie)
260
# TODO: slow, take it out of the inner loop.
262
basis_inv = parent_invs[0]
264
basis_inv = Inventory(root_id=None)
266
# ie.revision is always None if the InventoryEntry is considered
267
# for committing. We may record the previous parents revision if the
268
# content is actually unchanged against a sole head.
269
if ie.revision is not None:
270
if not self._versioned_root and path == '':
271
# repositories that do not version the root set the root's
272
# revision to the new commit even when no change occurs, and
273
# this masks when a change may have occurred against the basis,
274
# so calculate if one happened.
275
if ie.file_id in basis_inv:
276
delta = (basis_inv.id2path(ie.file_id), path,
280
delta = (None, path, ie.file_id, ie)
283
# we don't need to commit this, because the caller already
284
# determined that an existing revision of this file is
286
return None, (ie.revision == self._new_revision_id)
287
# XXX: Friction: parent_candidates should return a list not a dict
288
# so that we don't have to walk the inventories again.
289
parent_candiate_entries = ie.parent_candidates(parent_invs)
290
head_set = self._heads(parent_candiate_entries.keys())
292
for inv in parent_invs:
293
if ie.file_id in inv:
294
old_rev = inv[ie.file_id].revision
295
if old_rev in head_set:
296
heads.append(inv[ie.file_id].revision)
297
head_set.remove(inv[ie.file_id].revision)
300
# now we check to see if we need to write a new record to the
302
# We write a new entry unless there is one head to the ancestors, and
303
# the kind-derived content is unchanged.
305
# Cheapest check first: no ancestors, or more the one head in the
306
# ancestors, we write a new node.
310
# There is a single head, look it up for comparison
311
parent_entry = parent_candiate_entries[heads[0]]
312
# if the non-content specific data has changed, we'll be writing a
314
if (parent_entry.parent_id != ie.parent_id or
315
parent_entry.name != ie.name):
317
# now we need to do content specific checks:
319
# if the kind changed the content obviously has
320
if kind != parent_entry.kind:
323
assert content_summary[2] is not None, \
324
"Files must not have executable = None"
326
if (# if the file length changed we have to store:
327
parent_entry.text_size != content_summary[1] or
328
# if the exec bit has changed we have to store:
329
parent_entry.executable != content_summary[2]):
331
elif parent_entry.text_sha1 == content_summary[3]:
332
# all meta and content is unchanged (using a hash cache
333
# hit to check the sha)
334
ie.revision = parent_entry.revision
335
ie.text_size = parent_entry.text_size
336
ie.text_sha1 = parent_entry.text_sha1
337
ie.executable = parent_entry.executable
338
return self._get_delta(ie, basis_inv, path), False
340
# Either there is only a hash change(no hash cache entry,
341
# or same size content change), or there is no change on
343
# Provide the parent's hash to the store layer, so that the
344
# content is unchanged we will not store a new node.
345
nostore_sha = parent_entry.text_sha1
347
# We want to record a new node regardless of the presence or
348
# absence of a content change in the file.
350
ie.executable = content_summary[2]
351
lines = tree.get_file(ie.file_id, path).readlines()
353
ie.text_sha1, ie.text_size = self._add_text_to_weave(
354
ie.file_id, lines, heads, nostore_sha)
355
except errors.ExistingContent:
356
# Turns out that the file content was unchanged, and we were
357
# only going to store a new node if it was changed. Carry over
359
ie.revision = parent_entry.revision
360
ie.text_size = parent_entry.text_size
361
ie.text_sha1 = parent_entry.text_sha1
362
ie.executable = parent_entry.executable
363
return self._get_delta(ie, basis_inv, path), False
364
elif kind == 'directory':
366
# all data is meta here, nothing specific to directory, so
368
ie.revision = parent_entry.revision
369
return self._get_delta(ie, basis_inv, path), False
371
self._add_text_to_weave(ie.file_id, lines, heads, None)
372
elif kind == 'symlink':
373
current_link_target = content_summary[3]
375
# symlink target is not generic metadata, check if it has
377
if current_link_target != parent_entry.symlink_target:
380
# unchanged, carry over.
381
ie.revision = parent_entry.revision
382
ie.symlink_target = parent_entry.symlink_target
383
return self._get_delta(ie, basis_inv, path), False
384
ie.symlink_target = current_link_target
386
self._add_text_to_weave(ie.file_id, lines, heads, None)
387
elif kind == 'tree-reference':
389
if content_summary[3] != parent_entry.reference_revision:
392
# unchanged, carry over.
393
ie.reference_revision = parent_entry.reference_revision
394
ie.revision = parent_entry.revision
395
return self._get_delta(ie, basis_inv, path), False
396
ie.reference_revision = content_summary[3]
398
self._add_text_to_weave(ie.file_id, lines, heads, None)
400
raise NotImplementedError('unknown kind')
401
ie.revision = self._new_revision_id
402
return self._get_delta(ie, basis_inv, path), True
404
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
405
versionedfile = self.repository.weave_store.get_weave_or_empty(
406
file_id, self.repository.get_transaction())
407
# Don't change this to add_lines - add_lines_with_ghosts is cheaper
408
# than add_lines, and allows committing when a parent is ghosted for
410
# Note: as we read the content directly from the tree, we know its not
411
# been turned into unicode or badly split - but a broken tree
412
# implementation could give us bad output from readlines() so this is
413
# not a guarantee of safety. What would be better is always checking
414
# the content during test suite execution. RBC 20070912
416
return versionedfile.add_lines_with_ghosts(
417
self._new_revision_id, parents, new_lines,
418
nostore_sha=nostore_sha, random_id=self.random_revid,
419
check_content=False)[0:2]
421
versionedfile.clear_cache()
424
class RootCommitBuilder(CommitBuilder):
425
"""This commitbuilder actually records the root id"""
427
# the root entry gets versioned properly by this builder.
428
_versioned_root = True
430
def _check_root(self, ie, parent_invs, tree):
431
"""Helper for record_entry_contents.
433
:param ie: An entry being added.
434
:param parent_invs: The inventories of the parent revisions of the
436
:param tree: The tree that is being committed.
247
440
######################################################################
251
class Repository(_RelockDebugMixin, controldir.ControlComponent):
443
class Repository(object):
252
444
"""Repository holding history for one or more branches.
254
446
The repository holds and retrieves historical information including
255
447
revisions and file history. It's normally accessed only by the Branch,
256
448
which views a particular line of development through that history.
258
See VersionedFileRepository in bzrlib.vf_repository for the
259
base class for most Bazaar repositories.
450
The Repository builds on top of Stores and a Transport, which respectively
451
describe the disk data format and the way of accessing the (possibly
262
def abort_write_group(self, suppress_errors=False):
455
# What class to use for a CommitBuilder. Often its simpler to change this
456
# in a Repository class subclass rather than to override
457
# get_commit_builder.
458
_commit_builder_class = CommitBuilder
459
# The search regex used by xml based repositories to determine what things
460
# where changed in a single commit.
461
_file_ids_altered_regex = lazy_regex.lazy_compile(
462
r'file_id="(?P<file_id>[^"]+)"'
463
r'.* revision="(?P<revision_id>[^"]+)"'
466
def abort_write_group(self):
263
467
"""Commit the contents accrued within the current write group.
265
:param suppress_errors: if true, abort_write_group will catch and log
266
unexpected errors that happen during the abort, rather than
267
allowing them to propagate. Defaults to False.
269
469
:seealso: start_write_group.
271
471
if self._write_group is not self.get_transaction():
272
472
# has an unlock or relock occured ?
275
'(suppressed) mismatched lock context and write group. %r, %r',
276
self._write_group, self.get_transaction())
278
raise errors.BzrError(
279
'mismatched lock context and write group. %r, %r' %
280
(self._write_group, self.get_transaction()))
282
self._abort_write_group()
283
except Exception, exc:
284
self._write_group = None
285
if not suppress_errors:
287
mutter('abort_write_group failed')
288
log_exception_quietly()
289
note(gettext('bzr: ERROR (ignored): %s'), exc)
473
raise errors.BzrError('mismatched lock context and write group.')
474
self._abort_write_group()
290
475
self._write_group = None
292
477
def _abort_write_group(self):
293
478
"""Template method for per-repository write group cleanup.
295
This is called during abort before the write group is considered to be
480
This is called during abort before the write group is considered to be
296
481
finished and should cleanup any internal state accrued during the write
297
482
group. There is no requirement that data handed to the repository be
298
483
*not* made available - this is not a rollback - but neither should any
817
969
def has_revision(self, revision_id):
818
970
"""True if this repository has a copy of the revision."""
819
return revision_id in self.has_revisions((revision_id,))
822
def has_revisions(self, revision_ids):
823
"""Probe to find out the presence of multiple revisions.
825
:param revision_ids: An iterable of revision_ids.
826
:return: A set of the revision_ids that were present.
828
raise NotImplementedError(self.has_revisions)
971
if 'evil' in debug.debug_flags:
972
mutter_callsite(3, "has_revision is a LBYL symptom.")
973
return self._revision_store.has_revision_id(revision_id,
974
self.get_transaction())
831
977
def get_revision(self, revision_id):
832
978
"""Return the Revision object for a named revision."""
833
979
return self.get_revisions([revision_id])[0]
835
982
def get_revision_reconcile(self, revision_id):
836
983
"""'reconcile' helper routine that allows access to a revision always.
838
985
This variant of get_revision does not cross check the weave graph
839
986
against the revision one as get_revision does: but it should only
840
987
be used by reconcile, or reconcile-alike commands that are correcting
841
988
or testing the revision graph.
843
raise NotImplementedError(self.get_revision_reconcile)
990
return self._get_revisions([revision_id])[0]
845
993
def get_revisions(self, revision_ids):
846
"""Get many revisions at once.
848
Repositories that need to check data on every revision read should
849
subclass this method.
851
raise NotImplementedError(self.get_revisions)
853
def get_deltas_for_revisions(self, revisions, specific_fileids=None):
994
"""Get many revisions at once."""
995
return self._get_revisions(revision_ids)
998
def _get_revisions(self, revision_ids):
999
"""Core work logic to get many revisions without sanity checks."""
1000
for rev_id in revision_ids:
1001
if not rev_id or not isinstance(rev_id, basestring):
1002
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1003
revs = self._revision_store.get_revisions(revision_ids,
1004
self.get_transaction())
1006
assert not isinstance(rev.revision_id, unicode)
1007
for parent_id in rev.parent_ids:
1008
assert not isinstance(parent_id, unicode)
1012
def get_revision_xml(self, revision_id):
1013
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1014
# would have already do it.
1015
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1016
rev = self.get_revision(revision_id)
1017
rev_tmp = StringIO()
1018
# the current serializer..
1019
self._revision_store._serializer.write_revision(rev, rev_tmp)
1021
return rev_tmp.getvalue()
1024
def get_deltas_for_revisions(self, revisions):
854
1025
"""Produce a generator of revision deltas.
856
1027
Note that the input is a sequence of REVISIONS, not revision_ids.
857
1028
Trees will be held in memory until the generator exits.
858
1029
Each delta is relative to the revision's lefthand predecessor.
860
:param specific_fileids: if not None, the result is filtered
861
so that only those file-ids, their parents and their
862
children are included.
864
# Get the revision-ids of interest
865
1031
required_trees = set()
866
1032
for revision in revisions:
867
1033
required_trees.add(revision.revision_id)
868
1034
required_trees.update(revision.parent_ids[:1])
870
# Get the matching filtered trees. Note that it's more
871
# efficient to pass filtered trees to changes_from() rather
872
# than doing the filtering afterwards. changes_from() could
873
# arguably do the filtering itself but it's path-based, not
874
# file-id based, so filtering before or afterwards is
876
if specific_fileids is None:
877
trees = dict((t.get_revision_id(), t) for
878
t in self.revision_trees(required_trees))
880
trees = dict((t.get_revision_id(), t) for
881
t in self._filtered_revision_trees(required_trees,
884
# Calculate the deltas
1035
trees = dict((t.get_revision_id(), t) for
1036
t in self.revision_trees(required_trees))
885
1037
for revision in revisions:
886
1038
if not revision.parent_ids:
887
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
1039
old_tree = self.revision_tree(None)
889
1041
old_tree = trees[revision.parent_ids[0]]
890
1042
yield trees[revision.revision_id].changes_from(old_tree)
892
1044
@needs_read_lock
893
def get_revision_delta(self, revision_id, specific_fileids=None):
1045
def get_revision_delta(self, revision_id):
894
1046
"""Return the delta for one revision.
896
1048
The delta is relative to the left-hand predecessor of the
899
:param specific_fileids: if not None, the result is filtered
900
so that only those file-ids, their parents and their
901
children are included.
903
1051
r = self.get_revision(revision_id)
904
return list(self.get_deltas_for_revisions([r],
905
specific_fileids=specific_fileids))[0]
1052
return list(self.get_deltas_for_revisions([r]))[0]
907
1054
@needs_write_lock
908
1055
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
909
1056
signature = gpg_strategy.sign(plaintext)
910
1057
self.add_signature_text(revision_id, signature)
912
1060
def add_signature_text(self, revision_id, signature):
913
"""Store a signature text for a revision.
915
:param revision_id: Revision id of the revision
916
:param signature: Signature text.
918
raise NotImplementedError(self.add_signature_text)
920
def _find_parent_ids_of_revisions(self, revision_ids):
921
"""Find all parent ids that are mentioned in the revision graph.
923
:return: set of revisions that are parents of revision_ids which are
924
not part of revision_ids themselves
926
parent_map = self.get_parent_map(revision_ids)
928
map(parent_ids.update, parent_map.itervalues())
929
parent_ids.difference_update(revision_ids)
930
parent_ids.discard(_mod_revision.NULL_REVISION)
1061
self._revision_store.add_revision_signature_text(revision_id,
1063
self.get_transaction())
1065
def find_text_key_references(self):
1066
"""Find the text key references within the repository.
1068
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1069
revision_ids. Each altered file-ids has the exact revision_ids that
1070
altered it listed explicitly.
1071
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1072
to whether they were referred to by the inventory of the
1073
revision_id that they contain. The inventory texts from all present
1074
revision ids are assessed to generate this report.
1076
revision_ids = self.all_revision_ids()
1077
w = self.get_inventory_weave()
1078
pb = ui.ui_factory.nested_progress_bar()
1080
return self._find_text_key_references_from_xml_inventory_lines(
1081
w.iter_lines_added_or_present_in_versions(revision_ids, pb=pb))
1085
def _find_text_key_references_from_xml_inventory_lines(self,
1087
"""Core routine for extracting references to texts from inventories.
1089
This performs the translation of xml lines to revision ids.
1091
:param line_iterator: An iterator of lines, origin_version_id
1092
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1093
to whether they were referred to by the inventory of the
1094
revision_id that they contain. Note that if that revision_id was
1095
not part of the line_iterator's output then False will be given -
1096
even though it may actually refer to that key.
1098
if not self._serializer.support_altered_by_hack:
1099
raise AssertionError(
1100
"_find_text_key_references_from_xml_inventory_lines only "
1101
"supported for branches which store inventory as unnested xml"
1102
", not on %r" % self)
1105
# this code needs to read every new line in every inventory for the
1106
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1107
# not present in one of those inventories is unnecessary but not
1108
# harmful because we are filtering by the revision id marker in the
1109
# inventory lines : we only select file ids altered in one of those
1110
# revisions. We don't need to see all lines in the inventory because
1111
# only those added in an inventory in rev X can contain a revision=X
1113
unescape_revid_cache = {}
1114
unescape_fileid_cache = {}
1116
# jam 20061218 In a big fetch, this handles hundreds of thousands
1117
# of lines, so it has had a lot of inlining and optimizing done.
1118
# Sorry that it is a little bit messy.
1119
# Move several functions to be local variables, since this is a long
1121
search = self._file_ids_altered_regex.search
1122
unescape = _unescape_xml
1123
setdefault = result.setdefault
1124
for line, version_id in line_iterator:
1125
match = search(line)
1128
# One call to match.group() returning multiple items is quite a
1129
# bit faster than 2 calls to match.group() each returning 1
1130
file_id, revision_id = match.group('file_id', 'revision_id')
1132
# Inlining the cache lookups helps a lot when you make 170,000
1133
# lines and 350k ids, versus 8.4 unique ids.
1134
# Using a cache helps in 2 ways:
1135
# 1) Avoids unnecessary decoding calls
1136
# 2) Re-uses cached strings, which helps in future set and
1138
# (2) is enough that removing encoding entirely along with
1139
# the cache (so we are using plain strings) results in no
1140
# performance improvement.
1142
revision_id = unescape_revid_cache[revision_id]
1144
unescaped = unescape(revision_id)
1145
unescape_revid_cache[revision_id] = unescaped
1146
revision_id = unescaped
1148
# Note that unconditionally unescaping means that we deserialise
1149
# every fileid, which for general 'pull' is not great, but we don't
1150
# really want to have some many fulltexts that this matters anyway.
1153
file_id = unescape_fileid_cache[file_id]
1155
unescaped = unescape(file_id)
1156
unescape_fileid_cache[file_id] = unescaped
1159
key = (file_id, revision_id)
1160
setdefault(key, False)
1161
if revision_id == version_id:
1165
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1167
"""Helper routine for fileids_altered_by_revision_ids.
1169
This performs the translation of xml lines to revision ids.
1171
:param line_iterator: An iterator of lines, origin_version_id
1172
:param revision_ids: The revision ids to filter for. This should be a
1173
set or other type which supports efficient __contains__ lookups, as
1174
the revision id from each parsed line will be looked up in the
1175
revision_ids filter.
1176
:return: a dictionary mapping altered file-ids to an iterable of
1177
revision_ids. Each altered file-ids has the exact revision_ids that
1178
altered it listed explicitly.
1181
setdefault = result.setdefault
1182
for file_id, revision_id in \
1183
self._find_text_key_references_from_xml_inventory_lines(
1184
line_iterator).iterkeys():
1185
# once data is all ensured-consistent; then this is
1186
# if revision_id == version_id
1187
if revision_id in revision_ids:
1188
setdefault(file_id, set()).add(revision_id)
1191
def fileids_altered_by_revision_ids(self, revision_ids):
1192
"""Find the file ids and versions affected by revisions.
1194
:param revisions: an iterable containing revision ids.
1195
:return: a dictionary mapping altered file-ids to an iterable of
1196
revision_ids. Each altered file-ids has the exact revision_ids that
1197
altered it listed explicitly.
1199
selected_revision_ids = set(revision_ids)
1200
w = self.get_inventory_weave()
1201
pb = ui.ui_factory.nested_progress_bar()
1203
return self._find_file_ids_from_xml_inventory_lines(
1204
w.iter_lines_added_or_present_in_versions(
1205
selected_revision_ids, pb=pb),
1206
selected_revision_ids)
933
1210
def iter_files_bytes(self, desired_files):
934
1211
"""Iterate through file versions.
941
1218
uniquely identify the file version in the caller's context. (Examples:
942
1219
an index number or a TreeTransform trans_id.)
1221
bytes_iterator is an iterable of bytestrings for the file. The
1222
kind of iterable and length of the bytestrings are unspecified, but for
1223
this implementation, it is a list of lines produced by
1224
VersionedFile.get_lines().
944
1226
:param desired_files: a list of (file_id, revision_id, identifier)
947
raise NotImplementedError(self.iter_files_bytes)
949
def get_rev_id_for_revno(self, revno, known_pair):
950
"""Return the revision id of a revno, given a later (revno, revid)
951
pair in the same history.
953
:return: if found (True, revid). If the available history ran out
954
before reaching the revno, then this returns
955
(False, (closest_revno, closest_revid)).
957
known_revno, known_revid = known_pair
958
partial_history = [known_revid]
959
distance_from_known = known_revno - revno
960
if distance_from_known < 0:
962
'requested revno (%d) is later than given known revno (%d)'
963
% (revno, known_revno))
966
self, partial_history, stop_index=distance_from_known)
967
except errors.RevisionNotPresent, err:
968
if err.revision_id == known_revid:
969
# The start revision (known_revid) wasn't found.
971
# This is a stacked repository with no fallbacks, or a there's a
972
# left-hand ghost. Either way, even though the revision named in
973
# the error isn't in this repo, we know it's the next step in this
975
partial_history.append(err.revision_id)
976
if len(partial_history) <= distance_from_known:
977
# Didn't find enough history to get a revid for the revno.
978
earliest_revno = known_revno - len(partial_history) + 1
979
return (False, (earliest_revno, partial_history[-1]))
980
if len(partial_history) - 1 > distance_from_known:
981
raise AssertionError('_iter_for_revno returned too much history')
982
return (True, partial_history[-1])
1229
transaction = self.get_transaction()
1230
for file_id, revision_id, callable_data in desired_files:
1232
weave = self.weave_store.get_weave(file_id, transaction)
1233
except errors.NoSuchFile:
1234
raise errors.NoSuchIdInRepository(self, file_id)
1235
yield callable_data, weave.get_lines(revision_id)
1237
def _generate_text_key_index(self):
1238
"""Generate a new text key index for the repository.
1240
This is an expensive function that will take considerable time to run.
1242
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1243
list of parents, also text keys. When a given key has no parents,
1244
the parents list will be [NULL_REVISION].
1246
# All revisions, to find inventory parents.
1247
revision_graph = self.get_revision_graph_with_ghosts()
1248
ancestors = revision_graph.get_ancestors()
1249
text_key_references = self.find_text_key_references()
1250
pb = ui.ui_factory.nested_progress_bar()
1252
return self._do_generate_text_key_index(ancestors,
1253
text_key_references, pb)
1257
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1258
"""Helper for _generate_text_key_index to avoid deep nesting."""
1259
revision_order = tsort.topo_sort(ancestors)
1260
invalid_keys = set()
1262
for revision_id in revision_order:
1263
revision_keys[revision_id] = set()
1264
text_count = len(text_key_references)
1265
# a cache of the text keys to allow reuse; costs a dict of all the
1266
# keys, but saves a 2-tuple for every child of a given key.
1268
for text_key, valid in text_key_references.iteritems():
1270
invalid_keys.add(text_key)
1272
revision_keys[text_key[1]].add(text_key)
1273
text_key_cache[text_key] = text_key
1274
del text_key_references
1276
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1277
NULL_REVISION = _mod_revision.NULL_REVISION
1278
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1279
# too small for large or very branchy trees. However, for 55K path
1280
# trees, it would be easy to use too much memory trivially. Ideally we
1281
# could gauge this by looking at available real memory etc, but this is
1282
# always a tricky proposition.
1283
inventory_cache = lru_cache.LRUCache(10)
1284
batch_size = 10 # should be ~150MB on a 55K path tree
1285
batch_count = len(revision_order) / batch_size + 1
1287
pb.update("Calculating text parents.", processed_texts, text_count)
1288
for offset in xrange(batch_count):
1289
to_query = revision_order[offset * batch_size:(offset + 1) *
1293
for rev_tree in self.revision_trees(to_query):
1294
revision_id = rev_tree.get_revision_id()
1295
parent_ids = ancestors[revision_id]
1296
for text_key in revision_keys[revision_id]:
1297
pb.update("Calculating text parents.", processed_texts)
1298
processed_texts += 1
1299
candidate_parents = []
1300
for parent_id in parent_ids:
1301
parent_text_key = (text_key[0], parent_id)
1303
check_parent = parent_text_key not in \
1304
revision_keys[parent_id]
1306
# the parent parent_id is a ghost:
1307
check_parent = False
1308
# truncate the derived graph against this ghost.
1309
parent_text_key = None
1311
# look at the parent commit details inventories to
1312
# determine possible candidates in the per file graph.
1315
inv = inventory_cache[parent_id]
1317
inv = self.revision_tree(parent_id).inventory
1318
inventory_cache[parent_id] = inv
1319
parent_entry = inv._byid.get(text_key[0], None)
1320
if parent_entry is not None:
1322
text_key[0], parent_entry.revision)
1324
parent_text_key = None
1325
if parent_text_key is not None:
1326
candidate_parents.append(
1327
text_key_cache[parent_text_key])
1328
parent_heads = text_graph.heads(candidate_parents)
1329
new_parents = list(parent_heads)
1330
new_parents.sort(key=lambda x:candidate_parents.index(x))
1331
if new_parents == []:
1332
new_parents = [NULL_REVISION]
1333
text_index[text_key] = new_parents
1335
for text_key in invalid_keys:
1336
text_index[text_key] = [NULL_REVISION]
1339
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1340
"""Get an iterable listing the keys of all the data introduced by a set
1343
The keys will be ordered so that the corresponding items can be safely
1344
fetched and inserted in that order.
1346
:returns: An iterable producing tuples of (knit-kind, file-id,
1347
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1348
'revisions'. file-id is None unless knit-kind is 'file'.
1350
# XXX: it's a bit weird to control the inventory weave caching in this
1351
# generator. Ideally the caching would be done in fetch.py I think. Or
1352
# maybe this generator should explicitly have the contract that it
1353
# should not be iterated until the previously yielded item has been
1356
inv_w = self.get_inventory_weave()
1357
inv_w.enable_cache()
1359
# file ids that changed
1360
file_ids = self.fileids_altered_by_revision_ids(revision_ids)
1362
num_file_ids = len(file_ids)
1363
for file_id, altered_versions in file_ids.iteritems():
1364
if _files_pb is not None:
1365
_files_pb.update("fetch texts", count, num_file_ids)
1367
yield ("file", file_id, altered_versions)
1368
# We're done with the files_pb. Note that it finished by the caller,
1369
# just as it was created by the caller.
1373
yield ("inventory", None, revision_ids)
1377
revisions_with_signatures = set()
1378
for rev_id in revision_ids:
1380
self.get_signature_text(rev_id)
1381
except errors.NoSuchRevision:
1385
revisions_with_signatures.add(rev_id)
1387
yield ("signatures", None, revisions_with_signatures)
1390
yield ("revisions", None, revision_ids)
1393
def get_inventory_weave(self):
1394
return self.control_weaves.get_weave('inventory',
1395
self.get_transaction())
1398
def get_inventory(self, revision_id):
1399
"""Get Inventory object by hash."""
1400
return self.deserialise_inventory(
1401
revision_id, self.get_inventory_xml(revision_id))
1403
def deserialise_inventory(self, revision_id, xml):
1404
"""Transform the xml into an inventory object.
1406
:param revision_id: The expected revision id of the inventory.
1407
:param xml: A serialised inventory.
1409
return self._serializer.read_inventory_from_string(xml, revision_id)
1411
def serialise_inventory(self, inv):
1412
return self._serializer.write_inventory_to_string(inv)
1414
def _serialise_inventory_to_lines(self, inv):
1415
return self._serializer.write_inventory_to_lines(inv)
1417
def get_serializer_format(self):
1418
return self._serializer.format_num
1421
def get_inventory_xml(self, revision_id):
1422
"""Get inventory XML as a file object."""
1424
assert isinstance(revision_id, str), type(revision_id)
1425
iw = self.get_inventory_weave()
1426
return iw.get_text(revision_id)
1428
raise errors.HistoryMissing(self, 'inventory', revision_id)
1431
def get_inventory_sha1(self, revision_id):
1432
"""Return the sha1 hash of the inventory entry
1434
return self.get_revision(revision_id).inventory_sha1
1437
def get_revision_graph(self, revision_id=None):
1438
"""Return a dictionary containing the revision graph.
1440
NB: This method should not be used as it accesses the entire graph all
1441
at once, which is much more data than most operations should require.
1443
:param revision_id: The revision_id to get a graph from. If None, then
1444
the entire revision graph is returned. This is a deprecated mode of
1445
operation and will be removed in the future.
1446
:return: a dictionary of revision_id->revision_parents_list.
1448
raise NotImplementedError(self.get_revision_graph)
1451
def get_revision_graph_with_ghosts(self, revision_ids=None):
1452
"""Return a graph of the revisions with ghosts marked as applicable.
1454
:param revision_ids: an iterable of revisions to graph or None for all.
1455
:return: a Graph object with the graph reachable from revision_ids.
1457
if 'evil' in debug.debug_flags:
1459
"get_revision_graph_with_ghosts scales with size of history.")
1460
result = deprecated_graph.Graph()
1461
if not revision_ids:
1462
pending = set(self.all_revision_ids())
1465
pending = set(revision_ids)
1466
# special case NULL_REVISION
1467
if _mod_revision.NULL_REVISION in pending:
1468
pending.remove(_mod_revision.NULL_REVISION)
1469
required = set(pending)
1472
revision_id = pending.pop()
1474
rev = self.get_revision(revision_id)
1475
except errors.NoSuchRevision:
1476
if revision_id in required:
1479
result.add_ghost(revision_id)
1481
for parent_id in rev.parent_ids:
1482
# is this queued or done ?
1483
if (parent_id not in pending and
1484
parent_id not in done):
1486
pending.add(parent_id)
1487
result.add_node(revision_id, rev.parent_ids)
1488
done.add(revision_id)
1491
def _get_history_vf(self):
1492
"""Get a versionedfile whose history graph reflects all revisions.
1494
For weave repositories, this is the inventory weave.
1496
return self.get_inventory_weave()
1498
def iter_reverse_revision_history(self, revision_id):
1499
"""Iterate backwards through revision ids in the lefthand history
1501
:param revision_id: The revision id to start with. All its lefthand
1502
ancestors will be traversed.
1504
if revision_id in (None, _mod_revision.NULL_REVISION):
1506
next_id = revision_id
1507
versionedfile = self._get_history_vf()
1510
parents = versionedfile.get_parents(next_id)
1511
if len(parents) == 0:
1514
next_id = parents[0]
1517
def get_revision_inventory(self, revision_id):
1518
"""Return inventory of a past revision."""
1519
# TODO: Unify this with get_inventory()
1520
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1521
# must be the same as its revision, so this is trivial.
1522
if revision_id is None:
1523
# This does not make sense: if there is no revision,
1524
# then it is the current tree inventory surely ?!
1525
# and thus get_root_id() is something that looks at the last
1526
# commit on the branch, and the get_root_id is an inventory check.
1527
raise NotImplementedError
1528
# return Inventory(self.get_root_id())
1530
return self.get_inventory(revision_id)
984
1533
def is_shared(self):
985
1534
"""Return True if this repository is flagged as a shared repository."""
986
1535
raise NotImplementedError(self.is_shared)
1000
1549
for repositories to maintain loaded indices across multiple locks
1001
1550
by checking inside their implementation of this method to see
1002
1551
whether their indices are still valid. This depends of course on
1003
the disk format being validatable in this manner. This method is
1004
also called by the refresh_data() public interface to cause a refresh
1005
to occur while in a write lock so that data inserted by a smart server
1006
push operation is visible on the client's instance of the physical
1552
the disk format being validatable in this manner.
1010
1555
@needs_read_lock
1011
1556
def revision_tree(self, revision_id):
1012
1557
"""Return Tree for a revision on this branch.
1014
`revision_id` may be NULL_REVISION for the empty tree revision.
1559
`revision_id` may be None for the empty tree revision.
1016
raise NotImplementedError(self.revision_tree)
1561
# TODO: refactor this to use an existing revision object
1562
# so we don't need to read it in twice.
1563
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1564
return RevisionTree(self, Inventory(root_id=None),
1565
_mod_revision.NULL_REVISION)
1567
inv = self.get_revision_inventory(revision_id)
1568
return RevisionTree(self, inv, revision_id)
1018
1571
def revision_trees(self, revision_ids):
1019
"""Return Trees for revisions in this repository.
1021
:param revision_ids: a sequence of revision-ids;
1022
a revision-id may not be None or 'null:'
1572
"""Return Tree for a revision on this branch.
1574
`revision_id` may not be None or 'null:'"""
1575
assert None not in revision_ids
1576
assert _mod_revision.NULL_REVISION not in revision_ids
1577
texts = self.get_inventory_weave().get_texts(revision_ids)
1578
for text, revision_id in zip(texts, revision_ids):
1579
inv = self.deserialise_inventory(revision_id, text)
1580
yield RevisionTree(self, inv, revision_id)
1583
def get_ancestry(self, revision_id, topo_sorted=True):
1584
"""Return a list of revision-ids integrated by a revision.
1586
The first element of the list is always None, indicating the origin
1587
revision. This might change when we have history horizons, or
1588
perhaps we should have a new API.
1590
This is topologically sorted.
1024
raise NotImplementedError(self.revision_trees)
1592
if _mod_revision.is_null(revision_id):
1594
if not self.has_revision(revision_id):
1595
raise errors.NoSuchRevision(self, revision_id)
1596
w = self.get_inventory_weave()
1597
candidates = w.get_ancestry(revision_id, topo_sorted)
1598
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1026
def pack(self, hint=None, clean_obsolete_packs=False):
1027
1601
"""Compress the data within the repository.
1029
1603
This operation only makes sense for some repository types. For other
1030
1604
types it should be a no-op that just returns.
1032
1606
This stub method does not require a lock, but subclasses should use
1033
@needs_write_lock as this is a long running call it's reasonable to
1607
@needs_write_lock as this is a long running call its reasonable to
1034
1608
implicitly lock for the user.
1036
:param hint: If not supplied, the whole repository is packed.
1037
If supplied, the repository may use the hint parameter as a
1038
hint for the parts of the repository to pack. A hint can be
1039
obtained from the result of commit_write_group(). Out of
1040
date hints are simply ignored, because concurrent operations
1041
can obsolete them rapidly.
1043
:param clean_obsolete_packs: Clean obsolete packs immediately after
1612
def print_file(self, file, revision_id):
1613
"""Print `file` to stdout.
1615
FIXME RBC 20060125 as John Meinel points out this is a bad api
1616
- it writes to stdout, it assumes that that is valid etc. Fix
1617
by creating a new more flexible convenience function.
1619
tree = self.revision_tree(revision_id)
1620
# use inventory as it was in that revision
1621
file_id = tree.inventory.path2id(file)
1623
# TODO: jam 20060427 Write a test for this code path
1624
# it had a bug in it, and was raising the wrong
1626
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1627
tree.print_file(file_id)
1047
1629
def get_transaction(self):
1048
1630
return self.control_files.get_transaction()
1050
def get_parent_map(self, revision_ids):
1051
"""See graph.StackedParentsProvider.get_parent_map"""
1052
raise NotImplementedError(self.get_parent_map)
1632
def revision_parents(self, revision_id):
1633
return self.get_inventory_weave().parent_names(revision_id)
1054
def _get_parent_map_no_fallbacks(self, revision_ids):
1055
"""Same as Repository.get_parent_map except doesn't query fallbacks."""
1056
# revisions index works in keys; this just works in revisions
1057
# therefore wrap and unwrap
1635
def get_parents(self, revision_ids):
1636
"""See StackedParentsProvider.get_parents"""
1060
1638
for revision_id in revision_ids:
1061
1639
if revision_id == _mod_revision.NULL_REVISION:
1062
result[revision_id] = ()
1063
elif revision_id is None:
1064
raise ValueError('get_parent_map(None) is not valid')
1066
query_keys.append((revision_id ,))
1067
vf = self.revisions.without_fallbacks()
1068
for ((revision_id,), parent_keys) in \
1069
vf.get_parent_map(query_keys).iteritems():
1071
result[revision_id] = tuple([parent_revid
1072
for (parent_revid,) in parent_keys])
1074
result[revision_id] = (_mod_revision.NULL_REVISION,)
1643
parents = self.get_revision(revision_id).parent_ids
1644
except errors.NoSuchRevision:
1647
if len(parents) == 0:
1648
parents = [_mod_revision.NULL_REVISION]
1649
parents_list.append(parents)
1077
1652
def _make_parents_provider(self):
1078
if not self._format.supports_external_lookups:
1080
return graph.StackedParentsProvider(_LazyListJoin(
1081
[self._make_parents_provider_unstacked()],
1082
self._fallback_repositories))
1084
def _make_parents_provider_unstacked(self):
1085
return graph.CallableToParentsProviderAdapter(
1086
self._get_parent_map_no_fallbacks)
1089
def get_known_graph_ancestry(self, revision_ids):
1090
"""Return the known graph for a set of revision ids and their ancestors.
1092
raise NotImplementedError(self.get_known_graph_ancestry)
1094
def get_file_graph(self):
1095
"""Return the graph walker for files."""
1096
raise NotImplementedError(self.get_file_graph)
1098
1655
def get_graph(self, other_repository=None):
1099
1656
"""Return the graph walker for this repository format"""
1100
1657
parents_provider = self._make_parents_provider()
1101
1658
if (other_repository is not None and
1102
not self.has_same_location(other_repository)):
1103
parents_provider = graph.StackedParentsProvider(
1659
other_repository.bzrdir.transport.base !=
1660
self.bzrdir.transport.base):
1661
parents_provider = graph._StackedParentsProvider(
1104
1662
[parents_provider, other_repository._make_parents_provider()])
1105
1663
return graph.Graph(parents_provider)
1665
def get_versioned_file_checker(self):
1666
"""Return an object suitable for checking versioned files."""
1667
return VersionedFileChecker(self)
1107
1669
@needs_write_lock
1108
1670
def set_make_working_trees(self, new_value):
1109
1671
"""Set the policy flag for making working trees when creating branches.
1220
1741
revision_id.decode('ascii')
1221
1742
except UnicodeDecodeError:
1222
1743
raise errors.NonAsciiRevisionId(method, self)
1745
def revision_graph_can_have_wrong_parents(self):
1746
"""Is it possible for this repository to have a revision graph with
1749
If True, then this repository must also implement
1750
_find_inconsistent_revision_parents so that check and reconcile can
1751
check for inconsistencies before proceeding with other checks that may
1752
depend on the revision index being consistent.
1754
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1756
# remove these delegates a while after bzr 0.15
1757
def __make_delegated(name, from_module):
1758
def _deprecated_repository_forwarder():
1759
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1760
% (name, from_module),
1763
m = __import__(from_module, globals(), locals(), [name])
1765
return getattr(m, name)
1766
except AttributeError:
1767
raise AttributeError('module %s has no name %s'
1769
globals()[name] = _deprecated_repository_forwarder
1772
'AllInOneRepository',
1773
'WeaveMetaDirRepository',
1774
'PreSplitOutRepositoryFormat',
1775
'RepositoryFormat4',
1776
'RepositoryFormat5',
1777
'RepositoryFormat6',
1778
'RepositoryFormat7',
1780
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1784
'RepositoryFormatKnit',
1785
'RepositoryFormatKnit1',
1787
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1790
def install_revision(repository, rev, revision_tree):
1791
"""Install all revision data into a repository."""
1792
install_revisions(repository, [(rev, revision_tree, None)])
1795
def install_revisions(repository, iterable):
1796
"""Install all revision data into a repository.
1798
Accepts an iterable of revision, tree, signature tuples. The signature
1801
repository.start_write_group()
1803
for revision, revision_tree, signature in iterable:
1804
_install_revision(repository, revision, revision_tree, signature)
1806
repository.abort_write_group()
1809
repository.commit_write_group()
1812
def _install_revision(repository, rev, revision_tree, signature):
1813
"""Install all revision data into a repository."""
1814
present_parents = []
1816
for p_id in rev.parent_ids:
1817
if repository.has_revision(p_id):
1818
present_parents.append(p_id)
1819
parent_trees[p_id] = repository.revision_tree(p_id)
1821
parent_trees[p_id] = repository.revision_tree(None)
1823
inv = revision_tree.inventory
1824
entries = inv.iter_entries()
1825
# backwards compatibility hack: skip the root id.
1826
if not repository.supports_rich_root():
1827
path, root = entries.next()
1828
if root.revision != rev.revision_id:
1829
raise errors.IncompatibleRevision(repr(repository))
1830
# Add the texts that are not already present
1831
for path, ie in entries:
1832
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1833
repository.get_transaction())
1834
if ie.revision not in w:
1836
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1837
# with InventoryEntry.find_previous_heads(). if it is, then there
1838
# is a latent bug here where the parents may have ancestors of each
1840
for revision, tree in parent_trees.iteritems():
1841
if ie.file_id not in tree:
1843
parent_id = tree.inventory[ie.file_id].revision
1844
if parent_id in text_parents:
1846
text_parents.append(parent_id)
1848
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1849
repository.get_transaction())
1850
lines = revision_tree.get_file(ie.file_id).readlines()
1851
vfile.add_lines(rev.revision_id, text_parents, lines)
1853
# install the inventory
1854
repository.add_inventory(rev.revision_id, inv, present_parents)
1855
except errors.RevisionAlreadyPresent:
1857
if signature is not None:
1858
repository.add_signature_text(rev.revision_id, signature)
1859
repository.add_revision(rev.revision_id, rev, inv)
1225
1862
class MetaDirRepository(Repository):
1226
"""Repositories in the new meta-dir layout.
1228
:ivar _transport: Transport for access to repository control files,
1229
typically pointing to .bzr/repository.
1232
def __init__(self, _format, a_bzrdir, control_files):
1233
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
1234
self._transport = control_files._transport
1863
"""Repositories in the new meta-dir layout."""
1865
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1866
super(MetaDirRepository, self).__init__(_format,
1872
dir_mode = self.control_files._dir_mode
1873
file_mode = self.control_files._file_mode
1236
1876
def is_shared(self):
1237
1877
"""Return True if this repository is flagged as a shared repository."""
1238
return self._transport.has('shared-storage')
1878
return self.control_files._transport.has('shared-storage')
1240
1880
@needs_write_lock
1241
1881
def set_make_working_trees(self, new_value):
1632
2197
_optimisers = []
1633
2198
"""The available optimised InterRepository types."""
1636
2200
def copy_content(self, revision_id=None):
1637
"""Make a complete copy of the content in self into destination.
1639
This is a destructive operation! Do not use it on existing
1642
:param revision_id: Only copy the content needed to construct
1643
revision_id and its parents.
1646
self.target.set_make_working_trees(self.source.make_working_trees())
1647
except NotImplementedError:
1649
self.target.fetch(self.source, revision_id=revision_id)
1652
def fetch(self, revision_id=None, find_ghosts=False):
2201
raise NotImplementedError(self.copy_content)
2203
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1653
2204
"""Fetch the content required to construct revision_id.
1655
2206
The content is copied from self.source to self.target.
1657
2208
:param revision_id: if None all content is copied, if NULL_REVISION no
1658
2209
content is copied.
2210
:param pb: optional progress bar to use for progress reports. If not
2211
provided a default one will be created.
2213
Returns the copied revision count and the failed revisions in a tuple:
1661
2216
raise NotImplementedError(self.fetch)
1663
2218
@needs_read_lock
1664
def search_missing_revision_ids(self,
1665
revision_id=symbol_versioning.DEPRECATED_PARAMETER,
1666
find_ghosts=True, revision_ids=None, if_present_ids=None,
2219
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
1668
2220
"""Return the revision ids that source has that target does not.
2222
These are returned in topological order.
1670
2224
:param revision_id: only return revision ids included by this
1672
:param revision_ids: return revision ids included by these
1673
revision_ids. NoSuchRevision will be raised if any of these
1674
revisions are not present.
1675
:param if_present_ids: like revision_ids, but will not cause
1676
NoSuchRevision if any of these are absent, instead they will simply
1677
not be in the result. This is useful for e.g. finding revisions
1678
to fetch for tags, which may reference absent revisions.
1679
:param find_ghosts: If True find missing revisions in deep history
1680
rather than just finding the surface difference.
1681
:param limit: Maximum number of revisions to return, topologically
1683
:return: A bzrlib.graph.SearchResult.
1685
raise NotImplementedError(self.search_missing_revision_ids)
2227
# generic, possibly worst case, slow code path.
2228
target_ids = set(self.target.all_revision_ids())
2229
if revision_id is not None:
2230
source_ids = self.source.get_ancestry(revision_id)
2231
assert source_ids[0] is None
2234
source_ids = self.source.all_revision_ids()
2235
result_set = set(source_ids).difference(target_ids)
2236
# this may look like a no-op: its not. It preserves the ordering
2237
# other_ids had while only returning the members from other_ids
2238
# that we've decided we need.
2239
return [rev_id for rev_id in source_ids if rev_id in result_set]
1688
2242
def _same_model(source, target):
1689
"""True if source and target have the same data representation.
1691
Note: this is always called on the base class; overriding it in a
1692
subclass will have no effect.
1695
InterRepository._assert_same_model(source, target)
1697
except errors.IncompatibleRepositories, e:
2243
"""True if source and target have the same data representation."""
2244
if source.supports_rich_root() != target.supports_rich_root():
1701
def _assert_same_model(source, target):
1702
"""Raise an exception if two repositories do not use the same model.
1704
if source.supports_rich_root() != target.supports_rich_root():
1705
raise errors.IncompatibleRepositories(source, target,
1706
"different rich-root support")
1707
2246
if source._serializer != target._serializer:
1708
raise errors.IncompatibleRepositories(source, target,
1709
"different serializers")
2251
class InterSameDataRepository(InterRepository):
2252
"""Code for converting between repositories that represent the same data.
2254
Data format and model must match for this to work.
2258
def _get_repo_format_to_test(self):
2259
"""Repository format for testing with.
2261
InterSameData can pull from subtree to subtree and from non-subtree to
2262
non-subtree, so we test this with the richest repository format.
2264
from bzrlib.repofmt import knitrepo
2265
return knitrepo.RepositoryFormatKnit3()
2268
def is_compatible(source, target):
2269
return InterRepository._same_model(source, target)
2272
def copy_content(self, revision_id=None):
2273
"""Make a complete copy of the content in self into destination.
2275
This copies both the repository's revision data, and configuration information
2276
such as the make_working_trees setting.
2278
This is a destructive operation! Do not use it on existing
2281
:param revision_id: Only copy the content needed to construct
2282
revision_id and its parents.
2285
self.target.set_make_working_trees(self.source.make_working_trees())
2286
except NotImplementedError:
2288
# but don't bother fetching if we have the needed data now.
2289
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2290
self.target.has_revision(revision_id)):
2292
self.target.fetch(self.source, revision_id=revision_id)
2295
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2296
"""See InterRepository.fetch()."""
2297
from bzrlib.fetch import GenericRepoFetcher
2298
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2299
self.source, self.source._format, self.target,
2300
self.target._format)
2301
f = GenericRepoFetcher(to_repository=self.target,
2302
from_repository=self.source,
2303
last_revision=revision_id,
2305
return f.count_copied, f.failed_revisions
2308
class InterWeaveRepo(InterSameDataRepository):
2309
"""Optimised code paths between Weave based repositories.
2311
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2312
implemented lazy inter-object optimisation.
2316
def _get_repo_format_to_test(self):
2317
from bzrlib.repofmt import weaverepo
2318
return weaverepo.RepositoryFormat7()
2321
def is_compatible(source, target):
2322
"""Be compatible with known Weave formats.
2324
We don't test for the stores being of specific types because that
2325
could lead to confusing results, and there is no need to be
2328
from bzrlib.repofmt.weaverepo import (
2334
return (isinstance(source._format, (RepositoryFormat5,
2336
RepositoryFormat7)) and
2337
isinstance(target._format, (RepositoryFormat5,
2339
RepositoryFormat7)))
2340
except AttributeError:
2344
def copy_content(self, revision_id=None):
2345
"""See InterRepository.copy_content()."""
2346
# weave specific optimised path:
2348
self.target.set_make_working_trees(self.source.make_working_trees())
2349
except NotImplementedError:
2351
# FIXME do not peek!
2352
if self.source.control_files._transport.listable():
2353
pb = ui.ui_factory.nested_progress_bar()
2355
self.target.weave_store.copy_all_ids(
2356
self.source.weave_store,
2358
from_transaction=self.source.get_transaction(),
2359
to_transaction=self.target.get_transaction())
2360
pb.update('copying inventory', 0, 1)
2361
self.target.control_weaves.copy_multi(
2362
self.source.control_weaves, ['inventory'],
2363
from_transaction=self.source.get_transaction(),
2364
to_transaction=self.target.get_transaction())
2365
self.target._revision_store.text_store.copy_all_ids(
2366
self.source._revision_store.text_store,
2371
self.target.fetch(self.source, revision_id=revision_id)
2374
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2375
"""See InterRepository.fetch()."""
2376
from bzrlib.fetch import GenericRepoFetcher
2377
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2378
self.source, self.source._format, self.target, self.target._format)
2379
f = GenericRepoFetcher(to_repository=self.target,
2380
from_repository=self.source,
2381
last_revision=revision_id,
2383
return f.count_copied, f.failed_revisions
2386
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2387
"""See InterRepository.missing_revision_ids()."""
2388
# we want all revisions to satisfy revision_id in source.
2389
# but we don't want to stat every file here and there.
2390
# we want then, all revisions other needs to satisfy revision_id
2391
# checked, but not those that we have locally.
2392
# so the first thing is to get a subset of the revisions to
2393
# satisfy revision_id in source, and then eliminate those that
2394
# we do already have.
2395
# this is slow on high latency connection to self, but as as this
2396
# disk format scales terribly for push anyway due to rewriting
2397
# inventory.weave, this is considered acceptable.
2399
if revision_id is not None:
2400
source_ids = self.source.get_ancestry(revision_id)
2401
assert source_ids[0] is None
2404
source_ids = self.source._all_possible_ids()
2405
source_ids_set = set(source_ids)
2406
# source_ids is the worst possible case we may need to pull.
2407
# now we want to filter source_ids against what we actually
2408
# have in target, but don't try to check for existence where we know
2409
# we do not have a revision as that would be pointless.
2410
target_ids = set(self.target._all_possible_ids())
2411
possibly_present_revisions = target_ids.intersection(source_ids_set)
2412
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2413
required_revisions = source_ids_set.difference(actually_present_revisions)
2414
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2415
if revision_id is not None:
2416
# we used get_ancestry to determine source_ids then we are assured all
2417
# revisions referenced are present as they are installed in topological order.
2418
# and the tip revision was validated by get_ancestry.
2419
return required_topo_revisions
2421
# if we just grabbed the possibly available ids, then
2422
# we only have an estimate of whats available and need to validate
2423
# that against the revision records.
2424
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2427
class InterKnitRepo(InterSameDataRepository):
2428
"""Optimised code paths between Knit based repositories."""
2431
def _get_repo_format_to_test(self):
2432
from bzrlib.repofmt import knitrepo
2433
return knitrepo.RepositoryFormatKnit1()
2436
def is_compatible(source, target):
2437
"""Be compatible with known Knit formats.
2439
We don't test for the stores being of specific types because that
2440
could lead to confusing results, and there is no need to be
2443
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2445
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2446
isinstance(target._format, RepositoryFormatKnit))
2447
except AttributeError:
2449
return are_knits and InterRepository._same_model(source, target)
2452
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2453
"""See InterRepository.fetch()."""
2454
from bzrlib.fetch import KnitRepoFetcher
2455
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2456
self.source, self.source._format, self.target, self.target._format)
2457
f = KnitRepoFetcher(to_repository=self.target,
2458
from_repository=self.source,
2459
last_revision=revision_id,
2461
return f.count_copied, f.failed_revisions
2464
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2465
"""See InterRepository.missing_revision_ids()."""
2466
if revision_id is not None:
2467
source_ids = self.source.get_ancestry(revision_id)
2468
assert source_ids[0] is None
2471
source_ids = self.source.all_revision_ids()
2472
source_ids_set = set(source_ids)
2473
# source_ids is the worst possible case we may need to pull.
2474
# now we want to filter source_ids against what we actually
2475
# have in target, but don't try to check for existence where we know
2476
# we do not have a revision as that would be pointless.
2477
target_ids = set(self.target.all_revision_ids())
2478
possibly_present_revisions = target_ids.intersection(source_ids_set)
2479
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2480
required_revisions = source_ids_set.difference(actually_present_revisions)
2481
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2482
if revision_id is not None:
2483
# we used get_ancestry to determine source_ids then we are assured all
2484
# revisions referenced are present as they are installed in topological order.
2485
# and the tip revision was validated by get_ancestry.
2486
return required_topo_revisions
2488
# if we just grabbed the possibly available ids, then
2489
# we only have an estimate of whats available and need to validate
2490
# that against the revision records.
2491
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2494
class InterPackRepo(InterSameDataRepository):
2495
"""Optimised code paths between Pack based repositories."""
2498
def _get_repo_format_to_test(self):
2499
from bzrlib.repofmt import pack_repo
2500
return pack_repo.RepositoryFormatKnitPack1()
2503
def is_compatible(source, target):
2504
"""Be compatible with known Pack formats.
2506
We don't test for the stores being of specific types because that
2507
could lead to confusing results, and there is no need to be
2510
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2512
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2513
isinstance(target._format, RepositoryFormatPack))
2514
except AttributeError:
2516
return are_packs and InterRepository._same_model(source, target)
2519
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2520
"""See InterRepository.fetch()."""
2521
from bzrlib.repofmt.pack_repo import Packer
2522
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2523
self.source, self.source._format, self.target, self.target._format)
2524
self.count_copied = 0
2525
if revision_id is None:
2527
# everything to do - use pack logic
2528
# to fetch from all packs to one without
2529
# inventory parsing etc, IFF nothing to be copied is in the target.
2531
revision_ids = self.source.all_revision_ids()
2532
# implementing the TODO will involve:
2533
# - detecting when all of a pack is selected
2534
# - avoiding as much as possible pre-selection, so the
2535
# more-core routines such as create_pack_from_packs can filter in
2536
# a just-in-time fashion. (though having a HEADS list on a
2537
# repository might make this a lot easier, because we could
2538
# sensibly detect 'new revisions' without doing a full index scan.
2539
elif _mod_revision.is_null(revision_id):
2544
revision_ids = self.missing_revision_ids(revision_id,
2545
find_ghosts=find_ghosts)
2546
except errors.NoSuchRevision:
2547
raise errors.InstallFailed([revision_id])
2548
packs = self.source._pack_collection.all_packs()
2549
pack = Packer(self.target._pack_collection, packs, '.fetch',
2550
revision_ids).pack()
2551
if pack is not None:
2552
self.target._pack_collection._save_pack_names()
2553
# Trigger an autopack. This may duplicate effort as we've just done
2554
# a pack creation, but for now it is simpler to think about as
2555
# 'upload data, then repack if needed'.
2556
self.target._pack_collection.autopack()
2557
return (pack.get_revision_count(), [])
2562
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2563
"""See InterRepository.missing_revision_ids().
2565
:param find_ghosts: Find ghosts throughough the ancestry of
2568
if not find_ghosts and revision_id is not None:
2569
graph = self.source.get_graph()
2570
missing_revs = set()
2571
searcher = graph._make_breadth_first_searcher([revision_id])
2573
self.target._pack_collection.revision_index.combined_index
2574
null_set = frozenset([_mod_revision.NULL_REVISION])
2577
next_revs = set(searcher.next())
2578
except StopIteration:
2580
next_revs.difference_update(null_set)
2581
target_keys = [(key,) for key in next_revs]
2582
have_revs = frozenset(node[1][0] for node in
2583
target_index.iter_entries(target_keys))
2584
missing_revs.update(next_revs - have_revs)
2585
searcher.stop_searching_any(have_revs)
2586
if next_revs - have_revs == set([revision_id]):
2587
# we saw the start rev itself, but no parents from it (or
2588
# next_revs would have been updated to e.g. set(). We remove
2589
# have_revs because if we found revision_id locally we
2590
# stop_searching at the first time around.
2591
raise errors.NoSuchRevision(self.source, revision_id)
2593
elif revision_id is not None:
2594
source_ids = self.source.get_ancestry(revision_id)
2595
assert source_ids[0] is None
2598
source_ids = self.source.all_revision_ids()
2599
# source_ids is the worst possible case we may need to pull.
2600
# now we want to filter source_ids against what we actually
2601
# have in target, but don't try to check for existence where we know
2602
# we do not have a revision as that would be pointless.
2603
target_ids = set(self.target.all_revision_ids())
2604
return [r for r in source_ids if (r not in target_ids)]
2607
class InterModel1and2(InterRepository):
2610
def _get_repo_format_to_test(self):
2614
def is_compatible(source, target):
2615
if not source.supports_rich_root() and target.supports_rich_root():
2621
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2622
"""See InterRepository.fetch()."""
2623
from bzrlib.fetch import Model1toKnit2Fetcher
2624
f = Model1toKnit2Fetcher(to_repository=self.target,
2625
from_repository=self.source,
2626
last_revision=revision_id,
2628
return f.count_copied, f.failed_revisions
2631
def copy_content(self, revision_id=None):
2632
"""Make a complete copy of the content in self into destination.
2634
This is a destructive operation! Do not use it on existing
2637
:param revision_id: Only copy the content needed to construct
2638
revision_id and its parents.
2641
self.target.set_make_working_trees(self.source.make_working_trees())
2642
except NotImplementedError:
2644
# but don't bother fetching if we have the needed data now.
2645
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2646
self.target.has_revision(revision_id)):
2648
self.target.fetch(self.source, revision_id=revision_id)
2651
class InterKnit1and2(InterKnitRepo):
2654
def _get_repo_format_to_test(self):
2658
def is_compatible(source, target):
2659
"""Be compatible with Knit1 source and Knit3 target"""
2660
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
2662
from bzrlib.repofmt.knitrepo import (RepositoryFormatKnit1,
2663
RepositoryFormatKnit3)
2664
from bzrlib.repofmt.pack_repo import (RepositoryFormatKnitPack1,
2665
RepositoryFormatKnitPack3)
2666
return (isinstance(source._format,
2667
(RepositoryFormatKnit1, RepositoryFormatKnitPack1)) and
2668
isinstance(target._format,
2669
(RepositoryFormatKnit3, RepositoryFormatKnitPack3))
2671
except AttributeError:
2675
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2676
"""See InterRepository.fetch()."""
2677
from bzrlib.fetch import Knit1to2Fetcher
2678
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2679
self.source, self.source._format, self.target,
2680
self.target._format)
2681
f = Knit1to2Fetcher(to_repository=self.target,
2682
from_repository=self.source,
2683
last_revision=revision_id,
2685
return f.count_copied, f.failed_revisions
2688
class InterDifferingSerializer(InterKnitRepo):
2691
def _get_repo_format_to_test(self):
2695
def is_compatible(source, target):
2696
"""Be compatible with Knit2 source and Knit3 target"""
2697
if source.supports_rich_root() != target.supports_rich_root():
2699
# Ideally, we'd support fetching if the source had no tree references
2700
# even if it supported them...
2701
if (getattr(source, '_format.supports_tree_reference', False) and
2702
not getattr(target, '_format.supports_tree_reference', False)):
2707
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2708
"""See InterRepository.fetch()."""
2709
revision_ids = self.target.missing_revision_ids(self.source,
2711
def revisions_iterator():
2712
for current_revision_id in revision_ids:
2713
revision = self.source.get_revision(current_revision_id)
2714
tree = self.source.revision_tree(current_revision_id)
2716
signature = self.source.get_signature_text(
2717
current_revision_id)
2718
except errors.NoSuchRevision:
2720
yield revision, tree, signature
2721
install_revisions(self.target, revisions_iterator())
2722
return len(revision_ids), 0
2725
class InterRemoteToOther(InterRepository):
2727
def __init__(self, source, target):
2728
InterRepository.__init__(self, source, target)
2729
self._real_inter = None
2732
def is_compatible(source, target):
2733
if not isinstance(source, remote.RemoteRepository):
2735
source._ensure_real()
2736
real_source = source._real_repository
2737
# Is source's model compatible with target's model, and are they the
2738
# same format? Currently we can only optimise fetching from an
2739
# identical model & format repo.
2740
assert not isinstance(real_source, remote.RemoteRepository), (
2741
"We don't support remote repos backed by remote repos yet.")
2742
return real_source._format == target._format
2745
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2746
"""See InterRepository.fetch()."""
2747
from bzrlib.fetch import RemoteToOtherFetcher
2748
mutter("Using fetch logic to copy between %s(remote) and %s(%s)",
2749
self.source, self.target, self.target._format)
2750
# TODO: jam 20070210 This should be an assert, not a translate
2751
revision_id = osutils.safe_revision_id(revision_id)
2752
f = RemoteToOtherFetcher(to_repository=self.target,
2753
from_repository=self.source,
2754
last_revision=revision_id,
2756
return f.count_copied, f.failed_revisions
2759
def _get_repo_format_to_test(self):
2763
class InterOtherToRemote(InterRepository):
2765
def __init__(self, source, target):
2766
InterRepository.__init__(self, source, target)
2767
self._real_inter = None
2770
def is_compatible(source, target):
2771
if isinstance(target, remote.RemoteRepository):
2775
def _ensure_real_inter(self):
2776
if self._real_inter is None:
2777
self.target._ensure_real()
2778
real_target = self.target._real_repository
2779
self._real_inter = InterRepository.get(self.source, real_target)
2781
def copy_content(self, revision_id=None):
2782
self._ensure_real_inter()
2783
self._real_inter.copy_content(revision_id=revision_id)
2785
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2786
self._ensure_real_inter()
2787
self._real_inter.fetch(revision_id=revision_id, pb=pb)
2790
def _get_repo_format_to_test(self):
2794
InterRepository.register_optimiser(InterDifferingSerializer)
2795
InterRepository.register_optimiser(InterSameDataRepository)
2796
InterRepository.register_optimiser(InterWeaveRepo)
2797
InterRepository.register_optimiser(InterKnitRepo)
2798
InterRepository.register_optimiser(InterModel1and2)
2799
InterRepository.register_optimiser(InterKnit1and2)
2800
InterRepository.register_optimiser(InterPackRepo)
2801
InterRepository.register_optimiser(InterRemoteToOther)
2802
InterRepository.register_optimiser(InterOtherToRemote)
1712
2805
class CopyConverter(object):
1713
2806
"""A repository conversion tool which just performs a copy of the content.
1715
2808
This is slow but quite reliable.
1721
2814
:param target_format: The format the resulting repository should be.
1723
2816
self.target_format = target_format
1725
2818
def convert(self, repo, pb):
1726
2819
"""Perform the conversion of to_convert, giving feedback via pb.
1728
2821
:param to_convert: The disk object to convert.
1729
2822
:param pb: a progress bar to use for progress information.
1731
pb = ui.ui_factory.nested_progress_bar()
1734
2827
# this is only useful with metadir layouts - separated repo content.
1735
2828
# trigger an assertion if not such
1736
2829
repo._format.get_format_string()
1737
2830
self.repo_dir = repo.bzrdir
1738
pb.update(gettext('Moving repository to repository.backup'))
2831
self.step('Moving repository to repository.backup')
1739
2832
self.repo_dir.transport.move('repository', 'repository.backup')
1740
2833
backup_transport = self.repo_dir.transport.clone('repository.backup')
1741
2834
repo._format.check_conversion_target(self.target_format)
1742
2835
self.source_repo = repo._format.open(self.repo_dir,
1744
2837
_override_transport=backup_transport)
1745
pb.update(gettext('Creating new repository'))
2838
self.step('Creating new repository')
1746
2839
converted = self.target_format.initialize(self.repo_dir,
1747
2840
self.source_repo.is_shared())
1748
2841
converted.lock_write()
1750
pb.update(gettext('Copying content'))
2843
self.step('Copying content into repository.')
1751
2844
self.source_repo.copy_content_into(converted)
1753
2846
converted.unlock()
1754
pb.update(gettext('Deleting old repository content'))
2847
self.step('Deleting old repository content.')
1755
2848
self.repo_dir.transport.delete_tree('repository.backup')
1756
ui.ui_factory.note(gettext('repository converted'))
1760
def _strip_NULL_ghosts(revision_graph):
1761
"""Also don't use this. more compatibility code for unmigrated clients."""
1762
# Filter ghosts, and null:
1763
if _mod_revision.NULL_REVISION in revision_graph:
1764
del revision_graph[_mod_revision.NULL_REVISION]
1765
for key, parents in revision_graph.items():
1766
revision_graph[key] = tuple(parent for parent in parents if parent
1768
return revision_graph
1771
def _iter_for_revno(repo, partial_history_cache, stop_index=None,
1772
stop_revision=None):
1773
"""Extend the partial history to include a given index
1775
If a stop_index is supplied, stop when that index has been reached.
1776
If a stop_revision is supplied, stop when that revision is
1777
encountered. Otherwise, stop when the beginning of history is
1780
:param stop_index: The index which should be present. When it is
1781
present, history extension will stop.
1782
:param stop_revision: The revision id which should be present. When
1783
it is encountered, history extension will stop.
1785
start_revision = partial_history_cache[-1]
1786
graph = repo.get_graph()
1787
iterator = graph.iter_lefthand_ancestry(start_revision,
1788
(_mod_revision.NULL_REVISION,))
2849
self.pb.note('repository converted')
2851
def step(self, message):
2852
"""Update the pb by a step."""
2854
self.pb.update(message, self.count, self.total)
2866
def _unescaper(match, _map=_unescape_map):
2867
code = match.group(1)
1790
# skip the last revision in the list
1793
if (stop_index is not None and
1794
len(partial_history_cache) > stop_index):
1796
if partial_history_cache[-1] == stop_revision:
1798
revision_id = iterator.next()
1799
partial_history_cache.append(revision_id)
1800
except StopIteration:
1805
class _LazyListJoin(object):
1806
"""An iterable yielding the contents of many lists as one list.
1808
Each iterator made from this will reflect the current contents of the lists
1809
at the time the iterator is made.
2871
if not code.startswith('#'):
2873
return unichr(int(code[1:])).encode('utf8')
2879
def _unescape_xml(data):
2880
"""Unescape predefined XML entities in a string of data."""
2882
if _unescape_re is None:
2883
_unescape_re = re.compile('\&([^;]*);')
2884
return _unescape_re.sub(_unescaper, data)
2887
class VersionedFileChecker(object):
2889
def __init__(self, repository):
2890
self.repository = repository
2891
self.text_index = self.repository._generate_text_key_index()
1811
This is used by Repository's _make_parents_provider implementation so that
1814
pp = repo._make_parents_provider() # uses a list of fallback repos
1815
pp.add_fallback_repository(other_repo) # appends to that list
1816
result = pp.get_parent_map(...)
1817
# The result will include revs from other_repo
1820
def __init__(self, *list_parts):
1821
self.list_parts = list_parts
1825
for list_part in self.list_parts:
1826
full_list.extend(list_part)
1827
return iter(full_list)
1830
return "%s.%s(%s)" % (self.__module__, self.__class__.__name__,
2893
def calculate_file_version_parents(self, revision_id, file_id):
2894
"""Calculate the correct parents for a file version according to
2897
parent_keys = self.text_index[(file_id, revision_id)]
2898
if parent_keys == [_mod_revision.NULL_REVISION]:
2900
# strip the file_id, for the weave api
2901
return tuple([revision_id for file_id, revision_id in parent_keys])
2903
def check_file_version_parents(self, weave, file_id, planned_revisions):
2904
"""Check the parents stored in a versioned file are correct.
2906
It also detects file versions that are not referenced by their
2907
corresponding revision's inventory.
2909
:returns: A tuple of (wrong_parents, dangling_file_versions).
2910
wrong_parents is a dict mapping {revision_id: (stored_parents,
2911
correct_parents)} for each revision_id where the stored parents
2912
are not correct. dangling_file_versions is a set of (file_id,
2913
revision_id) tuples for versions that are present in this versioned
2914
file, but not used by the corresponding inventory.
2917
unused_versions = set()
2918
for num, revision_id in enumerate(planned_revisions):
2920
correct_parents = self.calculate_file_version_parents(
2921
revision_id, file_id)
2923
# we were asked to investigate a non-existant version.
2924
unused_versions.add(revision_id)
2927
knit_parents = tuple(weave.get_parents(revision_id))
2928
except errors.RevisionNotPresent:
2930
if correct_parents != knit_parents:
2931
wrong_parents[revision_id] = (knit_parents, correct_parents)
2932
return wrong_parents, unused_versions
added
removed
bzrlib/repository.py
