192
202
self.random_revid = False
194
def _heads(self, file_id, revision_ids):
195
"""Calculate the graph heads for revision_ids in the graph of file_id.
197
This can use either a per-file graph or a global revision graph as we
198
have an identity relationship between the two graphs.
200
return self.__heads(revision_ids)
202
def _check_root(self, ie, parent_invs, tree):
203
"""Helper for record_entry_contents.
205
:param ie: An entry being added.
206
:param parent_invs: The inventories of the parent revisions of the
208
:param tree: The tree that is being committed.
210
# In this revision format, root entries have no knit or weave When
211
# serializing out to disk and back in root.revision is always
213
ie.revision = self._new_revision_id
215
def _get_delta(self, ie, basis_inv, path):
216
"""Get a delta against the basis inventory for ie."""
217
if ie.file_id not in basis_inv:
219
return (None, path, ie.file_id, ie)
220
elif ie != basis_inv[ie.file_id]:
222
# TODO: avoid tis id2path call.
223
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
228
def record_entry_contents(self, ie, parent_invs, path, tree,
230
"""Record the content of ie from tree into the commit if needed.
232
Side effect: sets ie.revision when unchanged
234
:param ie: An inventory entry present in the commit.
235
:param parent_invs: The inventories of the parent revisions of the
237
:param path: The path the entry is at in the tree.
238
:param tree: The tree which contains this entry and should be used to
240
:param content_summary: Summary data from the tree about the paths
241
content - stat, length, exec, sha/link target. This is only
242
accessed when the entry has a revision of None - that is when it is
243
a candidate to commit.
244
:return: A tuple (change_delta, version_recorded). change_delta is
245
an inventory_delta change for this entry against the basis tree of
246
the commit, or None if no change occured against the basis tree.
247
version_recorded is True if a new version of the entry has been
248
recorded. For instance, committing a merge where a file was only
249
changed on the other side will return (delta, False).
251
if self.new_inventory.root is None:
252
if ie.parent_id is not None:
253
raise errors.RootMissing()
254
self._check_root(ie, parent_invs, tree)
255
if ie.revision is None:
256
kind = content_summary[0]
258
# ie is carried over from a prior commit
260
# XXX: repository specific check for nested tree support goes here - if
261
# the repo doesn't want nested trees we skip it ?
262
if (kind == 'tree-reference' and
263
not self.repository._format.supports_tree_reference):
264
# mismatch between commit builder logic and repository:
265
# this needs the entry creation pushed down into the builder.
266
raise NotImplementedError('Missing repository subtree support.')
267
self.new_inventory.add(ie)
269
# TODO: slow, take it out of the inner loop.
271
basis_inv = parent_invs[0]
273
basis_inv = Inventory(root_id=None)
275
# ie.revision is always None if the InventoryEntry is considered
276
# for committing. We may record the previous parents revision if the
277
# content is actually unchanged against a sole head.
278
if ie.revision is not None:
279
if not self._versioned_root and path == '':
280
# repositories that do not version the root set the root's
281
# revision to the new commit even when no change occurs, and
282
# this masks when a change may have occurred against the basis,
283
# so calculate if one happened.
284
if ie.file_id in basis_inv:
285
delta = (basis_inv.id2path(ie.file_id), path,
289
delta = (None, path, ie.file_id, ie)
292
# we don't need to commit this, because the caller already
293
# determined that an existing revision of this file is
294
# appropriate. If its not being considered for committing then
295
# it and all its parents to the root must be unaltered so
296
# no-change against the basis.
297
if ie.revision == self._new_revision_id:
298
raise AssertionError("Impossible situation, a skipped "
299
"inventory entry (%r) claims to be modified in this "
300
"commit (%r).", (ie, self._new_revision_id))
302
# XXX: Friction: parent_candidates should return a list not a dict
303
# so that we don't have to walk the inventories again.
304
parent_candiate_entries = ie.parent_candidates(parent_invs)
305
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
307
for inv in parent_invs:
308
if ie.file_id in inv:
309
old_rev = inv[ie.file_id].revision
310
if old_rev in head_set:
311
heads.append(inv[ie.file_id].revision)
312
head_set.remove(inv[ie.file_id].revision)
315
# now we check to see if we need to write a new record to the
317
# We write a new entry unless there is one head to the ancestors, and
318
# the kind-derived content is unchanged.
320
# Cheapest check first: no ancestors, or more the one head in the
321
# ancestors, we write a new node.
325
# There is a single head, look it up for comparison
326
parent_entry = parent_candiate_entries[heads[0]]
327
# if the non-content specific data has changed, we'll be writing a
329
if (parent_entry.parent_id != ie.parent_id or
330
parent_entry.name != ie.name):
332
# now we need to do content specific checks:
334
# if the kind changed the content obviously has
335
if kind != parent_entry.kind:
338
if content_summary[2] is None:
339
raise ValueError("Files must not have executable = None")
341
if (# if the file length changed we have to store:
342
parent_entry.text_size != content_summary[1] or
343
# if the exec bit has changed we have to store:
344
parent_entry.executable != content_summary[2]):
346
elif parent_entry.text_sha1 == content_summary[3]:
347
# all meta and content is unchanged (using a hash cache
348
# hit to check the sha)
349
ie.revision = parent_entry.revision
350
ie.text_size = parent_entry.text_size
351
ie.text_sha1 = parent_entry.text_sha1
352
ie.executable = parent_entry.executable
353
return self._get_delta(ie, basis_inv, path), False
355
# Either there is only a hash change(no hash cache entry,
356
# or same size content change), or there is no change on
358
# Provide the parent's hash to the store layer, so that the
359
# content is unchanged we will not store a new node.
360
nostore_sha = parent_entry.text_sha1
362
# We want to record a new node regardless of the presence or
363
# absence of a content change in the file.
365
ie.executable = content_summary[2]
366
lines = tree.get_file(ie.file_id, path).readlines()
368
ie.text_sha1, ie.text_size = self._add_text_to_weave(
369
ie.file_id, lines, heads, nostore_sha)
370
except errors.ExistingContent:
371
# Turns out that the file content was unchanged, and we were
372
# only going to store a new node if it was changed. Carry over
374
ie.revision = parent_entry.revision
375
ie.text_size = parent_entry.text_size
376
ie.text_sha1 = parent_entry.text_sha1
377
ie.executable = parent_entry.executable
378
return self._get_delta(ie, basis_inv, path), False
379
elif kind == 'directory':
381
# all data is meta here, nothing specific to directory, so
383
ie.revision = parent_entry.revision
384
return self._get_delta(ie, basis_inv, path), False
386
self._add_text_to_weave(ie.file_id, lines, heads, None)
387
elif kind == 'symlink':
388
current_link_target = content_summary[3]
390
# symlink target is not generic metadata, check if it has
392
if current_link_target != parent_entry.symlink_target:
395
# unchanged, carry over.
396
ie.revision = parent_entry.revision
397
ie.symlink_target = parent_entry.symlink_target
398
return self._get_delta(ie, basis_inv, path), False
399
ie.symlink_target = current_link_target
401
self._add_text_to_weave(ie.file_id, lines, heads, None)
402
elif kind == 'tree-reference':
404
if content_summary[3] != parent_entry.reference_revision:
407
# unchanged, carry over.
408
ie.reference_revision = parent_entry.reference_revision
409
ie.revision = parent_entry.revision
410
return self._get_delta(ie, basis_inv, path), False
411
ie.reference_revision = content_summary[3]
413
self._add_text_to_weave(ie.file_id, lines, heads, None)
415
raise NotImplementedError('unknown kind')
416
ie.revision = self._new_revision_id
417
return self._get_delta(ie, basis_inv, path), True
419
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
420
# Note: as we read the content directly from the tree, we know its not
421
# been turned into unicode or badly split - but a broken tree
422
# implementation could give us bad output from readlines() so this is
423
# not a guarantee of safety. What would be better is always checking
424
# the content during test suite execution. RBC 20070912
425
parent_keys = tuple((file_id, parent) for parent in parents)
426
return self.repository.texts.add_lines(
427
(file_id, self._new_revision_id), parent_keys, new_lines,
428
nostore_sha=nostore_sha, random_id=self.random_revid,
429
check_content=False)[0:2]
432
class RootCommitBuilder(CommitBuilder):
433
"""This commitbuilder actually records the root id"""
435
# the root entry gets versioned properly by this builder.
436
_versioned_root = True
438
def _check_root(self, ie, parent_invs, tree):
439
"""Helper for record_entry_contents.
441
:param ie: An entry being added.
442
:param parent_invs: The inventories of the parent revisions of the
444
:param tree: The tree that is being committed.
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,
448
247
######################################################################
451
class Repository(object):
251
class Repository(_RelockDebugMixin, controldir.ControlComponent):
452
252
"""Repository holding history for one or more branches.
454
254
The repository holds and retrieves historical information including
455
255
revisions and file history. It's normally accessed only by the Branch,
456
256
which views a particular line of development through that history.
458
The Repository builds on top of some byte storage facilies (the revisions,
459
signatures, inventories and texts attributes) and a Transport, which
460
respectively provide byte storage and a means to access the (possibly
463
The byte storage facilities are addressed via tuples, which we refer to
464
as 'keys' throughout the code base. Revision_keys, inventory_keys and
465
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
466
(file_id, revision_id). We use this interface because it allows low
467
friction with the underlying code that implements disk indices, network
468
encoding and other parts of bzrlib.
470
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
471
the serialised revisions for the repository. This can be used to obtain
472
revision graph information or to access raw serialised revisions.
473
The result of trying to insert data into the repository via this store
474
is undefined: it should be considered read-only except for implementors
476
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
477
the serialised signatures for the repository. This can be used to
478
obtain access to raw serialised signatures. The result of trying to
479
insert data into the repository via this store is undefined: it should
480
be considered read-only except for implementors of repositories.
481
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
482
the serialised inventories for the repository. This can be used to
483
obtain unserialised inventories. The result of trying to insert data
484
into the repository via this store is undefined: it should be
485
considered read-only except for implementors of repositories.
486
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
487
texts of files and directories for the repository. This can be used to
488
obtain file texts or file graphs. Note that Repository.iter_file_bytes
489
is usually a better interface for accessing file texts.
490
The result of trying to insert data into the repository via this store
491
is undefined: it should be considered read-only except for implementors
493
:ivar _transport: Transport for file access to repository, typically
494
pointing to .bzr/repository.
258
See VersionedFileRepository in bzrlib.vf_repository for the
259
base class for most Bazaar repositories.
497
# What class to use for a CommitBuilder. Often its simpler to change this
498
# in a Repository class subclass rather than to override
499
# get_commit_builder.
500
_commit_builder_class = CommitBuilder
501
# The search regex used by xml based repositories to determine what things
502
# where changed in a single commit.
503
_file_ids_altered_regex = lazy_regex.lazy_compile(
504
r'file_id="(?P<file_id>[^"]+)"'
505
r'.* revision="(?P<revision_id>[^"]+)"'
508
def abort_write_group(self):
262
def abort_write_group(self, suppress_errors=False):
509
263
"""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.
511
269
:seealso: start_write_group.
513
271
if self._write_group is not self.get_transaction():
514
272
# has an unlock or relock occured ?
515
raise errors.BzrError('mismatched lock context and write group.')
516
self._abort_write_group()
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)
517
290
self._write_group = None
519
292
def _abort_write_group(self):
520
293
"""Template method for per-repository write group cleanup.
522
This is called during abort before the write group is considered to be
295
This is called during abort before the write group is considered to be
523
296
finished and should cleanup any internal state accrued during the write
524
297
group. There is no requirement that data handed to the repository be
525
298
*not* made available - this is not a rollback - but neither should any
939
622
"""Commit the contents accrued within the current write group.
941
624
:seealso: start_write_group.
626
:return: it may return an opaque hint that can be passed to 'pack'.
943
628
if self._write_group is not self.get_transaction():
944
629
# has an unlock or relock occured ?
945
630
raise errors.BzrError('mismatched lock context %r and '
946
631
'write group %r.' %
947
632
(self.get_transaction(), self._write_group))
948
self._commit_write_group()
633
result = self._commit_write_group()
949
634
self._write_group = None
951
637
def _commit_write_group(self):
952
638
"""Template method for per-repository write group cleanup.
954
This is called before the write group is considered to be
640
This is called before the write group is considered to be
955
641
finished and should ensure that all data handed to the repository
956
for writing during the write group is safely committed (to the
642
for writing during the write group is safely committed (to the
957
643
extent possible considering file system caching etc).
960
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
646
def suspend_write_group(self):
647
"""Suspend a write group.
649
:raise UnsuspendableWriteGroup: If the write group can not be
651
:return: List of tokens
653
raise errors.UnsuspendableWriteGroup(self)
655
def refresh_data(self):
656
"""Re-read any data needed to synchronise with disk.
658
This method is intended to be called after another repository instance
659
(such as one used by a smart server) has inserted data into the
660
repository. On all repositories this will work outside of write groups.
661
Some repository formats (pack and newer for bzrlib native formats)
662
support refresh_data inside write groups. If called inside a write
663
group on a repository that does not support refreshing in a write group
664
IsInWriteGroupError will be raised.
668
def resume_write_group(self, tokens):
669
if not self.is_write_locked():
670
raise errors.NotWriteLocked(self)
671
if self._write_group:
672
raise errors.BzrError('already in a write group')
673
self._resume_write_group(tokens)
674
# so we can detect unlock/relock - the write group is now entered.
675
self._write_group = self.get_transaction()
677
def _resume_write_group(self, tokens):
678
raise errors.UnsuspendableWriteGroup(self)
680
def fetch(self, source, revision_id=None, find_ghosts=False):
961
681
"""Fetch the content required to construct revision_id from source.
963
If revision_id is None all content is copied.
683
If revision_id is None, then all content is copied.
685
fetch() may not be used when the repository is in a write group -
686
either finish the current write group before using fetch, or use
687
fetch before starting the write group.
964
689
:param find_ghosts: Find and copy revisions in the source that are
965
690
ghosts in the target (and not reachable directly by walking out to
966
691
the first-present revision in target from revision_id).
692
:param revision_id: If specified, all the content needed for this
693
revision ID will be copied to the target. Fetch will determine for
694
itself which content needs to be copied.
696
if self.is_in_write_group():
697
raise errors.InternalBzrError(
698
"May not fetch while in a write group.")
968
699
# fast path same-url fetch operations
969
if self.has_same_location(source):
700
# TODO: lift out to somewhere common with RemoteRepository
701
# <https://bugs.launchpad.net/bzr/+bug/401646>
702
if (self.has_same_location(source)
703
and self._has_same_fallbacks(source)):
970
704
# check that last_revision is in 'from' and then return a
972
706
if (revision_id is not None and
973
707
not _mod_revision.is_null(revision_id)):
974
708
self.get_revision(revision_id)
976
# if there is no specific appropriate InterRepository, this will get
977
# the InterRepository base class, which raises an
978
# IncompatibleRepositories when asked to fetch.
979
710
inter = InterRepository.get(source, self)
980
return inter.fetch(revision_id=revision_id, pb=pb,
981
find_ghosts=find_ghosts)
711
return inter.fetch(revision_id=revision_id, find_ghosts=find_ghosts)
983
713
def create_bundle(self, target, base, fileobj, format=None):
984
714
return serializer.write_bundle(self, target, base, fileobj, format)
986
def get_commit_builder(self, branch, parents, config, timestamp=None,
716
def get_commit_builder(self, branch, parents, config_stack, timestamp=None,
987
717
timezone=None, committer=None, revprops=None,
718
revision_id=None, lossy=False):
989
719
"""Obtain a CommitBuilder for this repository.
991
721
:param branch: Branch to commit to.
992
722
:param parents: Revision ids of the parents of the new revision.
993
:param config: Configuration to use.
723
:param config_stack: Configuration stack to use.
994
724
:param timestamp: Optional timestamp recorded for commit.
995
725
:param timezone: Optional timezone for timestamp.
996
726
:param committer: Optional committer to set for commit.
997
727
:param revprops: Optional dictionary of revision properties.
998
728
:param revision_id: Optional revision id.
729
:param lossy: Whether to discard data that can not be natively
730
represented, when pushing to a foreign VCS
1000
result = self._commit_builder_class(self, parents, config,
1001
timestamp, timezone, committer, revprops, revision_id)
1002
self.start_write_group()
732
raise NotImplementedError(self.get_commit_builder)
734
@only_raises(errors.LockNotHeld, errors.LockBroken)
1005
735
def unlock(self):
1006
736
if (self.control_files._lock_count == 1 and
1007
737
self.control_files._lock_mode == 'w'):
1093
825
:param revision_ids: An iterable of revision_ids.
1094
826
:return: A set of the revision_ids that were present.
1096
parent_map = self.revisions.get_parent_map(
1097
[(rev_id,) for rev_id in revision_ids])
1099
if _mod_revision.NULL_REVISION in revision_ids:
1100
result.add(_mod_revision.NULL_REVISION)
1101
result.update([key[0] for key in parent_map])
828
raise NotImplementedError(self.has_revisions)
1104
830
@needs_read_lock
1105
831
def get_revision(self, revision_id):
1106
832
"""Return the Revision object for a named revision."""
1107
833
return self.get_revisions([revision_id])[0]
1110
835
def get_revision_reconcile(self, revision_id):
1111
836
"""'reconcile' helper routine that allows access to a revision always.
1113
838
This variant of get_revision does not cross check the weave graph
1114
839
against the revision one as get_revision does: but it should only
1115
840
be used by reconcile, or reconcile-alike commands that are correcting
1116
841
or testing the revision graph.
1118
return self._get_revisions([revision_id])[0]
843
raise NotImplementedError(self.get_revision_reconcile)
1121
845
def get_revisions(self, revision_ids):
1122
"""Get many revisions at once."""
1123
return self._get_revisions(revision_ids)
1126
def _get_revisions(self, revision_ids):
1127
"""Core work logic to get many revisions without sanity checks."""
1128
for rev_id in revision_ids:
1129
if not rev_id or not isinstance(rev_id, basestring):
1130
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1131
keys = [(key,) for key in revision_ids]
1132
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1134
for record in stream:
1135
if record.storage_kind == 'absent':
1136
raise errors.NoSuchRevision(self, record.key[0])
1137
text = record.get_bytes_as('fulltext')
1138
rev = self._serializer.read_revision_from_string(text)
1139
revs[record.key[0]] = rev
1140
return [revs[revid] for revid in revision_ids]
1143
def get_revision_xml(self, revision_id):
1144
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1145
# would have already do it.
1146
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1147
rev = self.get_revision(revision_id)
1148
rev_tmp = StringIO()
1149
# the current serializer..
1150
self._serializer.write_revision(rev, rev_tmp)
1152
return rev_tmp.getvalue()
1154
def get_deltas_for_revisions(self, revisions):
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):
1155
854
"""Produce a generator of revision deltas.
1157
856
Note that the input is a sequence of REVISIONS, not revision_ids.
1158
857
Trees will be held in memory until the generator exits.
1159
858
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
1161
865
required_trees = set()
1162
866
for revision in revisions:
1163
867
required_trees.add(revision.revision_id)
1164
868
required_trees.update(revision.parent_ids[:1])
1165
trees = dict((t.get_revision_id(), t) for
1166
t in self.revision_trees(required_trees))
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
1167
885
for revision in revisions:
1168
886
if not revision.parent_ids:
1169
old_tree = self.revision_tree(None)
887
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
1171
889
old_tree = trees[revision.parent_ids[0]]
1172
890
yield trees[revision.revision_id].changes_from(old_tree)
1174
892
@needs_read_lock
1175
def get_revision_delta(self, revision_id):
893
def get_revision_delta(self, revision_id, specific_fileids=None):
1176
894
"""Return the delta for one revision.
1178
896
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.
1181
903
r = self.get_revision(revision_id)
1182
return list(self.get_deltas_for_revisions([r]))[0]
904
return list(self.get_deltas_for_revisions([r],
905
specific_fileids=specific_fileids))[0]
1184
907
@needs_write_lock
1185
908
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1186
909
signature = gpg_strategy.sign(plaintext)
1187
910
self.add_signature_text(revision_id, signature)
1190
912
def add_signature_text(self, revision_id, signature):
1191
self.signatures.add_lines((revision_id,), (),
1192
osutils.split_lines(signature))
1194
def find_text_key_references(self):
1195
"""Find the text key references within the repository.
1197
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1198
revision_ids. Each altered file-ids has the exact revision_ids that
1199
altered it listed explicitly.
1200
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1201
to whether they were referred to by the inventory of the
1202
revision_id that they contain. The inventory texts from all present
1203
revision ids are assessed to generate this report.
1205
revision_keys = self.revisions.keys()
1206
w = self.inventories
1207
pb = ui.ui_factory.nested_progress_bar()
1209
return self._find_text_key_references_from_xml_inventory_lines(
1210
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1214
def _find_text_key_references_from_xml_inventory_lines(self,
1216
"""Core routine for extracting references to texts from inventories.
1218
This performs the translation of xml lines to revision ids.
1220
:param line_iterator: An iterator of lines, origin_version_id
1221
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1222
to whether they were referred to by the inventory of the
1223
revision_id that they contain. Note that if that revision_id was
1224
not part of the line_iterator's output then False will be given -
1225
even though it may actually refer to that key.
1227
if not self._serializer.support_altered_by_hack:
1228
raise AssertionError(
1229
"_find_text_key_references_from_xml_inventory_lines only "
1230
"supported for branches which store inventory as unnested xml"
1231
", not on %r" % self)
1234
# this code needs to read every new line in every inventory for the
1235
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1236
# not present in one of those inventories is unnecessary but not
1237
# harmful because we are filtering by the revision id marker in the
1238
# inventory lines : we only select file ids altered in one of those
1239
# revisions. We don't need to see all lines in the inventory because
1240
# only those added in an inventory in rev X can contain a revision=X
1242
unescape_revid_cache = {}
1243
unescape_fileid_cache = {}
1245
# jam 20061218 In a big fetch, this handles hundreds of thousands
1246
# of lines, so it has had a lot of inlining and optimizing done.
1247
# Sorry that it is a little bit messy.
1248
# Move several functions to be local variables, since this is a long
1250
search = self._file_ids_altered_regex.search
1251
unescape = _unescape_xml
1252
setdefault = result.setdefault
1253
for line, line_key in line_iterator:
1254
match = search(line)
1257
# One call to match.group() returning multiple items is quite a
1258
# bit faster than 2 calls to match.group() each returning 1
1259
file_id, revision_id = match.group('file_id', 'revision_id')
1261
# Inlining the cache lookups helps a lot when you make 170,000
1262
# lines and 350k ids, versus 8.4 unique ids.
1263
# Using a cache helps in 2 ways:
1264
# 1) Avoids unnecessary decoding calls
1265
# 2) Re-uses cached strings, which helps in future set and
1267
# (2) is enough that removing encoding entirely along with
1268
# the cache (so we are using plain strings) results in no
1269
# performance improvement.
1271
revision_id = unescape_revid_cache[revision_id]
1273
unescaped = unescape(revision_id)
1274
unescape_revid_cache[revision_id] = unescaped
1275
revision_id = unescaped
1277
# Note that unconditionally unescaping means that we deserialise
1278
# every fileid, which for general 'pull' is not great, but we don't
1279
# really want to have some many fulltexts that this matters anyway.
1282
file_id = unescape_fileid_cache[file_id]
1284
unescaped = unescape(file_id)
1285
unescape_fileid_cache[file_id] = unescaped
1288
key = (file_id, revision_id)
1289
setdefault(key, False)
1290
if revision_id == line_key[-1]:
1294
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1296
"""Helper routine for fileids_altered_by_revision_ids.
1298
This performs the translation of xml lines to revision ids.
1300
:param line_iterator: An iterator of lines, origin_version_id
1301
:param revision_ids: The revision ids to filter for. This should be a
1302
set or other type which supports efficient __contains__ lookups, as
1303
the revision id from each parsed line will be looked up in the
1304
revision_ids filter.
1305
:return: a dictionary mapping altered file-ids to an iterable of
1306
revision_ids. Each altered file-ids has the exact revision_ids that
1307
altered it listed explicitly.
1310
setdefault = result.setdefault
1312
self._find_text_key_references_from_xml_inventory_lines(
1313
line_iterator).iterkeys():
1314
# once data is all ensured-consistent; then this is
1315
# if revision_id == version_id
1316
if key[-1:] in revision_ids:
1317
setdefault(key[0], set()).add(key[-1])
1320
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1321
"""Find the file ids and versions affected by revisions.
1323
:param revisions: an iterable containing revision ids.
1324
:param _inv_weave: The inventory weave from this repository or None.
1325
If None, the inventory weave will be opened automatically.
1326
:return: a dictionary mapping altered file-ids to an iterable of
1327
revision_ids. Each altered file-ids has the exact revision_ids that
1328
altered it listed explicitly.
1330
selected_keys = set((revid,) for revid in revision_ids)
1331
w = _inv_weave or self.inventories
1332
pb = ui.ui_factory.nested_progress_bar()
1334
return self._find_file_ids_from_xml_inventory_lines(
1335
w.iter_lines_added_or_present_in_keys(
1336
selected_keys, pb=pb),
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)
1341
933
def iter_files_bytes(self, desired_files):
1342
934
"""Iterate through file versions.
1349
941
uniquely identify the file version in the caller's context. (Examples:
1350
942
an index number or a TreeTransform trans_id.)
1352
bytes_iterator is an iterable of bytestrings for the file. The
1353
kind of iterable and length of the bytestrings are unspecified, but for
1354
this implementation, it is a list of bytes produced by
1355
VersionedFile.get_record_stream().
1357
944
:param desired_files: a list of (file_id, revision_id, identifier)
1360
transaction = self.get_transaction()
1362
for file_id, revision_id, callable_data in desired_files:
1363
text_keys[(file_id, revision_id)] = callable_data
1364
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1365
if record.storage_kind == 'absent':
1366
raise errors.RevisionNotPresent(record.key, self)
1367
yield text_keys[record.key], record.get_bytes_as('fulltext')
1369
def _generate_text_key_index(self, text_key_references=None,
1371
"""Generate a new text key index for the repository.
1373
This is an expensive function that will take considerable time to run.
1375
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1376
list of parents, also text keys. When a given key has no parents,
1377
the parents list will be [NULL_REVISION].
1379
# All revisions, to find inventory parents.
1380
if ancestors is None:
1381
graph = self.get_graph()
1382
ancestors = graph.get_parent_map(self.all_revision_ids())
1383
if text_key_references is None:
1384
text_key_references = self.find_text_key_references()
1385
pb = ui.ui_factory.nested_progress_bar()
1387
return self._do_generate_text_key_index(ancestors,
1388
text_key_references, pb)
1392
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1393
"""Helper for _generate_text_key_index to avoid deep nesting."""
1394
revision_order = tsort.topo_sort(ancestors)
1395
invalid_keys = set()
1397
for revision_id in revision_order:
1398
revision_keys[revision_id] = set()
1399
text_count = len(text_key_references)
1400
# a cache of the text keys to allow reuse; costs a dict of all the
1401
# keys, but saves a 2-tuple for every child of a given key.
1403
for text_key, valid in text_key_references.iteritems():
1405
invalid_keys.add(text_key)
1407
revision_keys[text_key[1]].add(text_key)
1408
text_key_cache[text_key] = text_key
1409
del text_key_references
1411
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1412
NULL_REVISION = _mod_revision.NULL_REVISION
1413
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1414
# too small for large or very branchy trees. However, for 55K path
1415
# trees, it would be easy to use too much memory trivially. Ideally we
1416
# could gauge this by looking at available real memory etc, but this is
1417
# always a tricky proposition.
1418
inventory_cache = lru_cache.LRUCache(10)
1419
batch_size = 10 # should be ~150MB on a 55K path tree
1420
batch_count = len(revision_order) / batch_size + 1
1422
pb.update("Calculating text parents.", processed_texts, text_count)
1423
for offset in xrange(batch_count):
1424
to_query = revision_order[offset * batch_size:(offset + 1) *
1428
for rev_tree in self.revision_trees(to_query):
1429
revision_id = rev_tree.get_revision_id()
1430
parent_ids = ancestors[revision_id]
1431
for text_key in revision_keys[revision_id]:
1432
pb.update("Calculating text parents.", processed_texts)
1433
processed_texts += 1
1434
candidate_parents = []
1435
for parent_id in parent_ids:
1436
parent_text_key = (text_key[0], parent_id)
1438
check_parent = parent_text_key not in \
1439
revision_keys[parent_id]
1441
# the parent parent_id is a ghost:
1442
check_parent = False
1443
# truncate the derived graph against this ghost.
1444
parent_text_key = None
1446
# look at the parent commit details inventories to
1447
# determine possible candidates in the per file graph.
1450
inv = inventory_cache[parent_id]
1452
inv = self.revision_tree(parent_id).inventory
1453
inventory_cache[parent_id] = inv
1454
parent_entry = inv._byid.get(text_key[0], None)
1455
if parent_entry is not None:
1457
text_key[0], parent_entry.revision)
1459
parent_text_key = None
1460
if parent_text_key is not None:
1461
candidate_parents.append(
1462
text_key_cache[parent_text_key])
1463
parent_heads = text_graph.heads(candidate_parents)
1464
new_parents = list(parent_heads)
1465
new_parents.sort(key=lambda x:candidate_parents.index(x))
1466
if new_parents == []:
1467
new_parents = [NULL_REVISION]
1468
text_index[text_key] = new_parents
1470
for text_key in invalid_keys:
1471
text_index[text_key] = [NULL_REVISION]
1474
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1475
"""Get an iterable listing the keys of all the data introduced by a set
1478
The keys will be ordered so that the corresponding items can be safely
1479
fetched and inserted in that order.
1481
:returns: An iterable producing tuples of (knit-kind, file-id,
1482
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1483
'revisions'. file-id is None unless knit-kind is 'file'.
1485
# XXX: it's a bit weird to control the inventory weave caching in this
1486
# generator. Ideally the caching would be done in fetch.py I think. Or
1487
# maybe this generator should explicitly have the contract that it
1488
# should not be iterated until the previously yielded item has been
1490
inv_w = self.inventories
1492
# file ids that changed
1493
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1495
num_file_ids = len(file_ids)
1496
for file_id, altered_versions in file_ids.iteritems():
1497
if _files_pb is not None:
1498
_files_pb.update("fetch texts", count, num_file_ids)
1500
yield ("file", file_id, altered_versions)
1501
# We're done with the files_pb. Note that it finished by the caller,
1502
# just as it was created by the caller.
1506
yield ("inventory", None, revision_ids)
1509
revisions_with_signatures = set()
1510
for rev_id in revision_ids:
1512
self.get_signature_text(rev_id)
1513
except errors.NoSuchRevision:
1517
revisions_with_signatures.add(rev_id)
1518
yield ("signatures", None, revisions_with_signatures)
1521
yield ("revisions", None, revision_ids)
1524
def get_inventory(self, revision_id):
1525
"""Get Inventory object by revision id."""
1526
return self.iter_inventories([revision_id]).next()
1528
def iter_inventories(self, revision_ids):
1529
"""Get many inventories by revision_ids.
1531
This will buffer some or all of the texts used in constructing the
1532
inventories in memory, but will only parse a single inventory at a
1535
:return: An iterator of inventories.
1537
if ((None in revision_ids)
1538
or (_mod_revision.NULL_REVISION in revision_ids)):
1539
raise ValueError('cannot get null revision inventory')
1540
return self._iter_inventories(revision_ids)
1542
def _iter_inventories(self, revision_ids):
1543
"""single-document based inventory iteration."""
1544
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1545
yield self.deserialise_inventory(revision_id, text)
1547
def _iter_inventory_xmls(self, revision_ids):
1548
keys = [(revision_id,) for revision_id in revision_ids]
1549
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1551
for record in stream:
1552
if record.storage_kind != 'absent':
1553
texts[record.key] = record.get_bytes_as('fulltext')
1555
raise errors.NoSuchRevision(self, record.key)
1557
yield texts[key], key[-1]
1559
def deserialise_inventory(self, revision_id, xml):
1560
"""Transform the xml into an inventory object.
1562
:param revision_id: The expected revision id of the inventory.
1563
:param xml: A serialised inventory.
1565
result = self._serializer.read_inventory_from_string(xml, revision_id)
1566
if result.revision_id != revision_id:
1567
raise AssertionError('revision id mismatch %s != %s' % (
1568
result.revision_id, revision_id))
1571
def serialise_inventory(self, inv):
1572
return self._serializer.write_inventory_to_string(inv)
1574
def _serialise_inventory_to_lines(self, inv):
1575
return self._serializer.write_inventory_to_lines(inv)
1577
def get_serializer_format(self):
1578
return self._serializer.format_num
1581
def get_inventory_xml(self, revision_id):
1582
"""Get inventory XML as a file object."""
1583
texts = self._iter_inventory_xmls([revision_id])
1585
text, revision_id = texts.next()
1586
except StopIteration:
1587
raise errors.HistoryMissing(self, 'inventory', revision_id)
1591
def get_inventory_sha1(self, revision_id):
1592
"""Return the sha1 hash of the inventory entry
1594
return self.get_revision(revision_id).inventory_sha1
1596
def iter_reverse_revision_history(self, revision_id):
1597
"""Iterate backwards through revision ids in the lefthand history
1599
:param revision_id: The revision id to start with. All its lefthand
1600
ancestors will be traversed.
1602
graph = self.get_graph()
1603
next_id = revision_id
1605
if next_id in (None, _mod_revision.NULL_REVISION):
1608
# Note: The following line may raise KeyError in the event of
1609
# truncated history. We decided not to have a try:except:raise
1610
# RevisionNotPresent here until we see a use for it, because of the
1611
# cost in an inner loop that is by its very nature O(history).
1612
# Robert Collins 20080326
1613
parents = graph.get_parent_map([next_id])[next_id]
1614
if len(parents) == 0:
1617
next_id = parents[0]
1620
def get_revision_inventory(self, revision_id):
1621
"""Return inventory of a past revision."""
1622
# TODO: Unify this with get_inventory()
1623
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1624
# must be the same as its revision, so this is trivial.
1625
if revision_id is None:
1626
# This does not make sense: if there is no revision,
1627
# then it is the current tree inventory surely ?!
1628
# and thus get_root_id() is something that looks at the last
1629
# commit on the branch, and the get_root_id is an inventory check.
1630
raise NotImplementedError
1631
# return Inventory(self.get_root_id())
1633
return self.get_inventory(revision_id)
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])
1635
984
def is_shared(self):
1636
985
"""Return True if this repository is flagged as a shared repository."""
1651
1000
for repositories to maintain loaded indices across multiple locks
1652
1001
by checking inside their implementation of this method to see
1653
1002
whether their indices are still valid. This depends of course on
1654
the disk format being validatable in this manner.
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
1657
1010
@needs_read_lock
1658
1011
def revision_tree(self, revision_id):
1659
1012
"""Return Tree for a revision on this branch.
1661
`revision_id` may be None for the empty tree revision.
1014
`revision_id` may be NULL_REVISION for the empty tree revision.
1663
# TODO: refactor this to use an existing revision object
1664
# so we don't need to read it in twice.
1665
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1666
return RevisionTree(self, Inventory(root_id=None),
1667
_mod_revision.NULL_REVISION)
1669
inv = self.get_revision_inventory(revision_id)
1670
return RevisionTree(self, inv, revision_id)
1016
raise NotImplementedError(self.revision_tree)
1672
1018
def revision_trees(self, revision_ids):
1673
"""Return Tree for a revision on this branch.
1675
`revision_id` may not be None or 'null:'"""
1676
inventories = self.iter_inventories(revision_ids)
1677
for inv in inventories:
1678
yield RevisionTree(self, inv, inv.revision_id)
1681
def get_ancestry(self, revision_id, topo_sorted=True):
1682
"""Return a list of revision-ids integrated by a revision.
1684
The first element of the list is always None, indicating the origin
1685
revision. This might change when we have history horizons, or
1686
perhaps we should have a new API.
1688
This is topologically sorted.
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:'
1690
if _mod_revision.is_null(revision_id):
1692
if not self.has_revision(revision_id):
1693
raise errors.NoSuchRevision(self, revision_id)
1694
graph = self.get_graph()
1696
search = graph._make_breadth_first_searcher([revision_id])
1699
found, ghosts = search.next_with_ghosts()
1700
except StopIteration:
1703
if _mod_revision.NULL_REVISION in keys:
1704
keys.remove(_mod_revision.NULL_REVISION)
1706
parent_map = graph.get_parent_map(keys)
1707
keys = tsort.topo_sort(parent_map)
1708
return [None] + list(keys)
1024
raise NotImplementedError(self.revision_trees)
1026
def pack(self, hint=None, clean_obsolete_packs=False):
1711
1027
"""Compress the data within the repository.
1713
1029
This operation only makes sense for some repository types. For other
1714
1030
types it should be a no-op that just returns.
1716
1032
This stub method does not require a lock, but subclasses should use
1717
@needs_write_lock as this is a long running call its reasonable to
1033
@needs_write_lock as this is a long running call it's reasonable to
1718
1034
implicitly lock for the user.
1722
@deprecated_method(one_six)
1723
def print_file(self, file, revision_id):
1724
"""Print `file` to stdout.
1726
FIXME RBC 20060125 as John Meinel points out this is a bad api
1727
- it writes to stdout, it assumes that that is valid etc. Fix
1728
by creating a new more flexible convenience function.
1730
tree = self.revision_tree(revision_id)
1731
# use inventory as it was in that revision
1732
file_id = tree.inventory.path2id(file)
1734
# TODO: jam 20060427 Write a test for this code path
1735
# it had a bug in it, and was raising the wrong
1737
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1738
tree.print_file(file_id)
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
1740
1047
def get_transaction(self):
1741
1048
return self.control_files.get_transaction()
1743
@deprecated_method(one_one)
1744
def get_parents(self, revision_ids):
1745
"""See StackedParentsProvider.get_parents"""
1746
parent_map = self.get_parent_map(revision_ids)
1747
return [parent_map.get(r, None) for r in revision_ids]
1749
1050
def get_parent_map(self, revision_ids):
1750
"""See graph._StackedParentsProvider.get_parent_map"""
1051
"""See graph.StackedParentsProvider.get_parent_map"""
1052
raise NotImplementedError(self.get_parent_map)
1054
def _get_parent_map_no_fallbacks(self, revision_ids):
1055
"""Same as Repository.get_parent_map except doesn't query fallbacks."""
1751
1056
# revisions index works in keys; this just works in revisions
1752
1057
# therefore wrap and unwrap
1753
1058
query_keys = []
2347
1632
_optimisers = []
2348
1633
"""The available optimised InterRepository types."""
2350
1636
def copy_content(self, revision_id=None):
2351
raise NotImplementedError(self.copy_content)
2353
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
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):
2354
1653
"""Fetch the content required to construct revision_id.
2356
1655
The content is copied from self.source to self.target.
2358
1657
:param revision_id: if None all content is copied, if NULL_REVISION no
2359
1658
content is copied.
2360
:param pb: optional progress bar to use for progress reports. If not
2361
provided a default one will be created.
2363
:returns: (copied_revision_count, failures).
2365
# Normally we should find a specific InterRepository subclass to do
2366
# the fetch; if nothing else then at least InterSameDataRepository.
2367
# If none of them is suitable it looks like fetching is not possible;
2368
# we try to give a good message why. _assert_same_model will probably
2369
# give a helpful message; otherwise a generic one.
2370
self._assert_same_model(self.source, self.target)
2371
raise errors.IncompatibleRepositories(self.source, self.target,
2372
"no suitableInterRepository found")
2374
def _walk_to_common_revisions(self, revision_ids):
2375
"""Walk out from revision_ids in source to revisions target has.
2377
:param revision_ids: The start point for the search.
2378
:return: A set of revision ids.
2380
target_graph = self.target.get_graph()
2381
revision_ids = frozenset(revision_ids)
2382
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2383
return graph.SearchResult(revision_ids, set(), 0, set())
2384
missing_revs = set()
2385
source_graph = self.source.get_graph()
2386
# ensure we don't pay silly lookup costs.
2387
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2388
null_set = frozenset([_mod_revision.NULL_REVISION])
2391
next_revs, ghosts = searcher.next_with_ghosts()
2392
except StopIteration:
2394
if revision_ids.intersection(ghosts):
2395
absent_ids = set(revision_ids.intersection(ghosts))
2396
# If all absent_ids are present in target, no error is needed.
2397
absent_ids.difference_update(
2398
set(target_graph.get_parent_map(absent_ids)))
2400
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2401
# we don't care about other ghosts as we can't fetch them and
2402
# haven't been asked to.
2403
next_revs = set(next_revs)
2404
# we always have NULL_REVISION present.
2405
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2406
missing_revs.update(next_revs - have_revs)
2407
searcher.stop_searching_any(have_revs)
2408
return searcher.get_result()
2410
@deprecated_method(one_two)
2412
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2413
"""Return the revision ids that source has that target does not.
2415
These are returned in topological order.
2417
:param revision_id: only return revision ids included by this
2419
:param find_ghosts: If True find missing revisions in deep history
2420
rather than just finding the surface difference.
2422
return list(self.search_missing_revision_ids(
2423
revision_id, find_ghosts).get_keys())
2426
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2427
"""Return the revision ids that source has that target does not.
2429
:param revision_id: only return revision ids included by this
2431
:param find_ghosts: If True find missing revisions in deep history
2432
rather than just finding the surface difference.
1661
raise NotImplementedError(self.fetch)
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,
1668
"""Return the revision ids that source has that target does not.
1670
: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
2433
1683
:return: A bzrlib.graph.SearchResult.
2435
# stop searching at found target revisions.
2436
if not find_ghosts and revision_id is not None:
2437
return self._walk_to_common_revisions([revision_id])
2438
# generic, possibly worst case, slow code path.
2439
target_ids = set(self.target.all_revision_ids())
2440
if revision_id is not None:
2441
source_ids = self.source.get_ancestry(revision_id)
2442
if source_ids[0] is not None:
2443
raise AssertionError()
2446
source_ids = self.source.all_revision_ids()
2447
result_set = set(source_ids).difference(target_ids)
2448
return self.source.revision_ids_to_search_result(result_set)
1685
raise NotImplementedError(self.search_missing_revision_ids)
2451
1688
def _same_model(source, target):
2452
1689
"""True if source and target have the same data representation.
2454
1691
Note: this is always called on the base class; overriding it in a
2455
1692
subclass will have no effect.
2472
1709
"different serializers")
2475
class InterSameDataRepository(InterRepository):
2476
"""Code for converting between repositories that represent the same data.
2478
Data format and model must match for this to work.
2482
def _get_repo_format_to_test(self):
2483
"""Repository format for testing with.
2485
InterSameData can pull from subtree to subtree and from non-subtree to
2486
non-subtree, so we test this with the richest repository format.
2488
from bzrlib.repofmt import knitrepo
2489
return knitrepo.RepositoryFormatKnit3()
2492
def is_compatible(source, target):
2493
return InterRepository._same_model(source, target)
2496
def copy_content(self, revision_id=None):
2497
"""Make a complete copy of the content in self into destination.
2499
This copies both the repository's revision data, and configuration information
2500
such as the make_working_trees setting.
2502
This is a destructive operation! Do not use it on existing
2505
:param revision_id: Only copy the content needed to construct
2506
revision_id and its parents.
2509
self.target.set_make_working_trees(self.source.make_working_trees())
2510
except NotImplementedError:
2512
# but don't bother fetching if we have the needed data now.
2513
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2514
self.target.has_revision(revision_id)):
2516
self.target.fetch(self.source, revision_id=revision_id)
2519
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2520
"""See InterRepository.fetch()."""
2521
from bzrlib.fetch import RepoFetcher
2522
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2523
self.source, self.source._format, self.target,
2524
self.target._format)
2525
f = RepoFetcher(to_repository=self.target,
2526
from_repository=self.source,
2527
last_revision=revision_id,
2528
pb=pb, find_ghosts=find_ghosts)
2529
return f.count_copied, f.failed_revisions
2532
class InterWeaveRepo(InterSameDataRepository):
2533
"""Optimised code paths between Weave based repositories.
2535
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2536
implemented lazy inter-object optimisation.
2540
def _get_repo_format_to_test(self):
2541
from bzrlib.repofmt import weaverepo
2542
return weaverepo.RepositoryFormat7()
2545
def is_compatible(source, target):
2546
"""Be compatible with known Weave formats.
2548
We don't test for the stores being of specific types because that
2549
could lead to confusing results, and there is no need to be
2552
from bzrlib.repofmt.weaverepo import (
2558
return (isinstance(source._format, (RepositoryFormat5,
2560
RepositoryFormat7)) and
2561
isinstance(target._format, (RepositoryFormat5,
2563
RepositoryFormat7)))
2564
except AttributeError:
2568
def copy_content(self, revision_id=None):
2569
"""See InterRepository.copy_content()."""
2570
# weave specific optimised path:
2572
self.target.set_make_working_trees(self.source.make_working_trees())
2573
except (errors.RepositoryUpgradeRequired, NotImplemented):
2575
# FIXME do not peek!
2576
if self.source._transport.listable():
2577
pb = ui.ui_factory.nested_progress_bar()
2579
self.target.texts.insert_record_stream(
2580
self.source.texts.get_record_stream(
2581
self.source.texts.keys(), 'topological', False))
2582
pb.update('copying inventory', 0, 1)
2583
self.target.inventories.insert_record_stream(
2584
self.source.inventories.get_record_stream(
2585
self.source.inventories.keys(), 'topological', False))
2586
self.target.signatures.insert_record_stream(
2587
self.source.signatures.get_record_stream(
2588
self.source.signatures.keys(),
2590
self.target.revisions.insert_record_stream(
2591
self.source.revisions.get_record_stream(
2592
self.source.revisions.keys(),
2593
'topological', True))
2597
self.target.fetch(self.source, revision_id=revision_id)
2600
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2601
"""See InterRepository.fetch()."""
2602
from bzrlib.fetch import RepoFetcher
2603
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2604
self.source, self.source._format, self.target, self.target._format)
2605
f = RepoFetcher(to_repository=self.target,
2606
from_repository=self.source,
2607
last_revision=revision_id,
2608
pb=pb, find_ghosts=find_ghosts)
2609
return f.count_copied, f.failed_revisions
2612
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2613
"""See InterRepository.missing_revision_ids()."""
2614
# we want all revisions to satisfy revision_id in source.
2615
# but we don't want to stat every file here and there.
2616
# we want then, all revisions other needs to satisfy revision_id
2617
# checked, but not those that we have locally.
2618
# so the first thing is to get a subset of the revisions to
2619
# satisfy revision_id in source, and then eliminate those that
2620
# we do already have.
2621
# this is slow on high latency connection to self, but as as this
2622
# disk format scales terribly for push anyway due to rewriting
2623
# inventory.weave, this is considered acceptable.
2625
if revision_id is not None:
2626
source_ids = self.source.get_ancestry(revision_id)
2627
if source_ids[0] is not None:
2628
raise AssertionError()
2631
source_ids = self.source._all_possible_ids()
2632
source_ids_set = set(source_ids)
2633
# source_ids is the worst possible case we may need to pull.
2634
# now we want to filter source_ids against what we actually
2635
# have in target, but don't try to check for existence where we know
2636
# we do not have a revision as that would be pointless.
2637
target_ids = set(self.target._all_possible_ids())
2638
possibly_present_revisions = target_ids.intersection(source_ids_set)
2639
actually_present_revisions = set(
2640
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2641
required_revisions = source_ids_set.difference(actually_present_revisions)
2642
if revision_id is not None:
2643
# we used get_ancestry to determine source_ids then we are assured all
2644
# revisions referenced are present as they are installed in topological order.
2645
# and the tip revision was validated by get_ancestry.
2646
result_set = required_revisions
2648
# if we just grabbed the possibly available ids, then
2649
# we only have an estimate of whats available and need to validate
2650
# that against the revision records.
2652
self.source._eliminate_revisions_not_present(required_revisions))
2653
return self.source.revision_ids_to_search_result(result_set)
2656
class InterKnitRepo(InterSameDataRepository):
2657
"""Optimised code paths between Knit based repositories."""
2660
def _get_repo_format_to_test(self):
2661
from bzrlib.repofmt import knitrepo
2662
return knitrepo.RepositoryFormatKnit1()
2665
def is_compatible(source, target):
2666
"""Be compatible with known Knit formats.
2668
We don't test for the stores being of specific types because that
2669
could lead to confusing results, and there is no need to be
2672
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2674
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2675
isinstance(target._format, RepositoryFormatKnit))
2676
except AttributeError:
2678
return are_knits and InterRepository._same_model(source, target)
2681
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2682
"""See InterRepository.fetch()."""
2683
from bzrlib.fetch import RepoFetcher
2684
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2685
self.source, self.source._format, self.target, self.target._format)
2686
f = RepoFetcher(to_repository=self.target,
2687
from_repository=self.source,
2688
last_revision=revision_id,
2689
pb=pb, find_ghosts=find_ghosts)
2690
return f.count_copied, f.failed_revisions
2693
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2694
"""See InterRepository.missing_revision_ids()."""
2695
if revision_id is not None:
2696
source_ids = self.source.get_ancestry(revision_id)
2697
if source_ids[0] is not None:
2698
raise AssertionError()
2701
source_ids = self.source.all_revision_ids()
2702
source_ids_set = set(source_ids)
2703
# source_ids is the worst possible case we may need to pull.
2704
# now we want to filter source_ids against what we actually
2705
# have in target, but don't try to check for existence where we know
2706
# we do not have a revision as that would be pointless.
2707
target_ids = set(self.target.all_revision_ids())
2708
possibly_present_revisions = target_ids.intersection(source_ids_set)
2709
actually_present_revisions = set(
2710
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2711
required_revisions = source_ids_set.difference(actually_present_revisions)
2712
if revision_id is not None:
2713
# we used get_ancestry to determine source_ids then we are assured all
2714
# revisions referenced are present as they are installed in topological order.
2715
# and the tip revision was validated by get_ancestry.
2716
result_set = required_revisions
2718
# if we just grabbed the possibly available ids, then
2719
# we only have an estimate of whats available and need to validate
2720
# that against the revision records.
2722
self.source._eliminate_revisions_not_present(required_revisions))
2723
return self.source.revision_ids_to_search_result(result_set)
2726
class InterPackRepo(InterSameDataRepository):
2727
"""Optimised code paths between Pack based repositories."""
2730
def _get_repo_format_to_test(self):
2731
from bzrlib.repofmt import pack_repo
2732
return pack_repo.RepositoryFormatKnitPack1()
2735
def is_compatible(source, target):
2736
"""Be compatible with known Pack formats.
2738
We don't test for the stores being of specific types because that
2739
could lead to confusing results, and there is no need to be
2742
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2744
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2745
isinstance(target._format, RepositoryFormatPack))
2746
except AttributeError:
2748
return are_packs and InterRepository._same_model(source, target)
2751
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2752
"""See InterRepository.fetch()."""
2753
if (len(self.source._fallback_repositories) > 0 or
2754
len(self.target._fallback_repositories) > 0):
2755
# The pack layer is not aware of fallback repositories, so when
2756
# fetching from a stacked repository or into a stacked repository
2757
# we use the generic fetch logic which uses the VersionedFiles
2758
# attributes on repository.
2759
from bzrlib.fetch import RepoFetcher
2760
fetcher = RepoFetcher(self.target, self.source, revision_id,
2762
return fetcher.count_copied, fetcher.failed_revisions
2763
from bzrlib.repofmt.pack_repo import Packer
2764
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2765
self.source, self.source._format, self.target, self.target._format)
2766
self.count_copied = 0
2767
if revision_id is None:
2769
# everything to do - use pack logic
2770
# to fetch from all packs to one without
2771
# inventory parsing etc, IFF nothing to be copied is in the target.
2773
source_revision_ids = frozenset(self.source.all_revision_ids())
2774
revision_ids = source_revision_ids - \
2775
frozenset(self.target.get_parent_map(source_revision_ids))
2776
revision_keys = [(revid,) for revid in revision_ids]
2777
index = self.target._pack_collection.revision_index.combined_index
2778
present_revision_ids = set(item[1][0] for item in
2779
index.iter_entries(revision_keys))
2780
revision_ids = set(revision_ids) - present_revision_ids
2781
# implementing the TODO will involve:
2782
# - detecting when all of a pack is selected
2783
# - avoiding as much as possible pre-selection, so the
2784
# more-core routines such as create_pack_from_packs can filter in
2785
# a just-in-time fashion. (though having a HEADS list on a
2786
# repository might make this a lot easier, because we could
2787
# sensibly detect 'new revisions' without doing a full index scan.
2788
elif _mod_revision.is_null(revision_id):
2793
revision_ids = self.search_missing_revision_ids(revision_id,
2794
find_ghosts=find_ghosts).get_keys()
2795
except errors.NoSuchRevision:
2796
raise errors.InstallFailed([revision_id])
2797
if len(revision_ids) == 0:
2799
packs = self.source._pack_collection.all_packs()
2800
pack = Packer(self.target._pack_collection, packs, '.fetch',
2801
revision_ids).pack()
2802
if pack is not None:
2803
self.target._pack_collection._save_pack_names()
2804
# Trigger an autopack. This may duplicate effort as we've just done
2805
# a pack creation, but for now it is simpler to think about as
2806
# 'upload data, then repack if needed'.
2807
self.target._pack_collection.autopack()
2808
return (pack.get_revision_count(), [])
2813
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2814
"""See InterRepository.missing_revision_ids().
2816
:param find_ghosts: Find ghosts throughout the ancestry of
2819
if not find_ghosts and revision_id is not None:
2820
return self._walk_to_common_revisions([revision_id])
2821
elif revision_id is not None:
2822
# Find ghosts: search for revisions pointing from one repository to
2823
# the other, and vice versa, anywhere in the history of revision_id.
2824
graph = self.target.get_graph(other_repository=self.source)
2825
searcher = graph._make_breadth_first_searcher([revision_id])
2829
next_revs, ghosts = searcher.next_with_ghosts()
2830
except StopIteration:
2832
if revision_id in ghosts:
2833
raise errors.NoSuchRevision(self.source, revision_id)
2834
found_ids.update(next_revs)
2835
found_ids.update(ghosts)
2836
found_ids = frozenset(found_ids)
2837
# Double query here: should be able to avoid this by changing the
2838
# graph api further.
2839
result_set = found_ids - frozenset(
2840
self.target.get_parent_map(found_ids))
2842
source_ids = self.source.all_revision_ids()
2843
# source_ids is the worst possible case we may need to pull.
2844
# now we want to filter source_ids against what we actually
2845
# have in target, but don't try to check for existence where we know
2846
# we do not have a revision as that would be pointless.
2847
target_ids = set(self.target.all_revision_ids())
2848
result_set = set(source_ids).difference(target_ids)
2849
return self.source.revision_ids_to_search_result(result_set)
2852
class InterModel1and2(InterRepository):
2855
def _get_repo_format_to_test(self):
2859
def is_compatible(source, target):
2860
if not source.supports_rich_root() and target.supports_rich_root():
2866
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2867
"""See InterRepository.fetch()."""
2868
from bzrlib.fetch import Model1toKnit2Fetcher
2869
f = Model1toKnit2Fetcher(to_repository=self.target,
2870
from_repository=self.source,
2871
last_revision=revision_id,
2872
pb=pb, find_ghosts=find_ghosts)
2873
return f.count_copied, f.failed_revisions
2876
def copy_content(self, revision_id=None):
2877
"""Make a complete copy of the content in self into destination.
2879
This is a destructive operation! Do not use it on existing
2882
:param revision_id: Only copy the content needed to construct
2883
revision_id and its parents.
2886
self.target.set_make_working_trees(self.source.make_working_trees())
2887
except NotImplementedError:
2889
# but don't bother fetching if we have the needed data now.
2890
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2891
self.target.has_revision(revision_id)):
2893
self.target.fetch(self.source, revision_id=revision_id)
2896
class InterKnit1and2(InterKnitRepo):
2899
def _get_repo_format_to_test(self):
2903
def is_compatible(source, target):
2904
"""Be compatible with Knit1 source and Knit3 target"""
2906
from bzrlib.repofmt.knitrepo import (
2907
RepositoryFormatKnit1,
2908
RepositoryFormatKnit3,
2910
from bzrlib.repofmt.pack_repo import (
2911
RepositoryFormatKnitPack1,
2912
RepositoryFormatKnitPack3,
2913
RepositoryFormatKnitPack4,
2914
RepositoryFormatKnitPack5,
2915
RepositoryFormatKnitPack5RichRoot,
2916
RepositoryFormatPackDevelopment1,
2917
RepositoryFormatPackDevelopment1Subtree,
2920
RepositoryFormatKnit1, # no rr, no subtree
2921
RepositoryFormatKnitPack1, # no rr, no subtree
2922
RepositoryFormatPackDevelopment1, # no rr, no subtree
2923
RepositoryFormatKnitPack5, # no rr, no subtree
2926
RepositoryFormatKnit3, # rr, subtree
2927
RepositoryFormatKnitPack3, # rr, subtree
2928
RepositoryFormatKnitPack4, # rr, no subtree
2929
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
2930
RepositoryFormatPackDevelopment1Subtree, # rr, subtree
2932
for format in norichroot:
2933
if format.rich_root_data:
2934
raise AssertionError('Format %s is a rich-root format'
2935
' but is included in the non-rich-root list'
2937
for format in richroot:
2938
if not format.rich_root_data:
2939
raise AssertionError('Format %s is not a rich-root format'
2940
' but is included in the rich-root list'
2942
# TODO: One alternative is to just check format.rich_root_data,
2943
# instead of keeping membership lists. However, the formats
2944
# *also* have to use the same 'Knit' style of storage
2945
# (line-deltas, fulltexts, etc.)
2946
return (isinstance(source._format, norichroot) and
2947
isinstance(target._format, richroot))
2948
except AttributeError:
2952
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2953
"""See InterRepository.fetch()."""
2954
from bzrlib.fetch import Knit1to2Fetcher
2955
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2956
self.source, self.source._format, self.target,
2957
self.target._format)
2958
f = Knit1to2Fetcher(to_repository=self.target,
2959
from_repository=self.source,
2960
last_revision=revision_id,
2961
pb=pb, find_ghosts=find_ghosts)
2962
return f.count_copied, f.failed_revisions
2965
class InterDifferingSerializer(InterKnitRepo):
2968
def _get_repo_format_to_test(self):
2972
def is_compatible(source, target):
2973
"""Be compatible with Knit2 source and Knit3 target"""
2974
if source.supports_rich_root() != target.supports_rich_root():
2976
# Ideally, we'd support fetching if the source had no tree references
2977
# even if it supported them...
2978
if (getattr(source, '_format.supports_tree_reference', False) and
2979
not getattr(target, '_format.supports_tree_reference', False)):
2984
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2985
"""See InterRepository.fetch()."""
2986
revision_ids = self.target.search_missing_revision_ids(self.source,
2987
revision_id, find_ghosts=find_ghosts).get_keys()
2988
revision_ids = tsort.topo_sort(
2989
self.source.get_graph().get_parent_map(revision_ids))
2990
def revisions_iterator():
2991
for current_revision_id in revision_ids:
2992
revision = self.source.get_revision(current_revision_id)
2993
tree = self.source.revision_tree(current_revision_id)
2995
signature = self.source.get_signature_text(
2996
current_revision_id)
2997
except errors.NoSuchRevision:
2999
yield revision, tree, signature
3001
my_pb = ui.ui_factory.nested_progress_bar()
3006
install_revisions(self.target, revisions_iterator(),
3007
len(revision_ids), pb)
3009
if my_pb is not None:
3011
return len(revision_ids), 0
3014
class InterOtherToRemote(InterRepository):
3016
def __init__(self, source, target):
3017
InterRepository.__init__(self, source, target)
3018
self._real_inter = None
3021
def is_compatible(source, target):
3022
if isinstance(target, remote.RemoteRepository):
3026
def _ensure_real_inter(self):
3027
if self._real_inter is None:
3028
self.target._ensure_real()
3029
real_target = self.target._real_repository
3030
self._real_inter = InterRepository.get(self.source, real_target)
3032
def copy_content(self, revision_id=None):
3033
self._ensure_real_inter()
3034
self._real_inter.copy_content(revision_id=revision_id)
3036
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3037
self._ensure_real_inter()
3038
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3039
find_ghosts=find_ghosts)
3042
def _get_repo_format_to_test(self):
3046
class InterRemoteToOther(InterRepository):
3048
def __init__(self, source, target):
3049
InterRepository.__init__(self, source, target)
3050
self._real_inter = None
3053
def is_compatible(source, target):
3054
if not isinstance(source, remote.RemoteRepository):
3056
# Is source's model compatible with target's model?
3057
source._ensure_real()
3058
real_source = source._real_repository
3059
if isinstance(real_source, remote.RemoteRepository):
3060
raise NotImplementedError(
3061
"We don't support remote repos backed by remote repos yet.")
3062
return InterRepository._same_model(real_source, target)
3064
def _ensure_real_inter(self):
3065
if self._real_inter is None:
3066
self.source._ensure_real()
3067
real_source = self.source._real_repository
3068
self._real_inter = InterRepository.get(real_source, self.target)
3070
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3071
self._ensure_real_inter()
3072
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3073
find_ghosts=find_ghosts)
3075
def copy_content(self, revision_id=None):
3076
self._ensure_real_inter()
3077
self._real_inter.copy_content(revision_id=revision_id)
3080
def _get_repo_format_to_test(self):
3085
InterRepository.register_optimiser(InterDifferingSerializer)
3086
InterRepository.register_optimiser(InterSameDataRepository)
3087
InterRepository.register_optimiser(InterWeaveRepo)
3088
InterRepository.register_optimiser(InterKnitRepo)
3089
InterRepository.register_optimiser(InterModel1and2)
3090
InterRepository.register_optimiser(InterKnit1and2)
3091
InterRepository.register_optimiser(InterPackRepo)
3092
InterRepository.register_optimiser(InterOtherToRemote)
3093
InterRepository.register_optimiser(InterRemoteToOther)
3096
1712
class CopyConverter(object):
3097
1713
"""A repository conversion tool which just performs a copy of the content.
3099
1715
This is slow but quite reliable.
3105
1721
:param target_format: The format the resulting repository should be.
3107
1723
self.target_format = target_format
3109
1725
def convert(self, repo, pb):
3110
1726
"""Perform the conversion of to_convert, giving feedback via pb.
3112
1728
:param to_convert: The disk object to convert.
3113
1729
:param pb: a progress bar to use for progress information.
1731
pb = ui.ui_factory.nested_progress_bar()
3118
1734
# this is only useful with metadir layouts - separated repo content.
3119
1735
# trigger an assertion if not such
3120
1736
repo._format.get_format_string()
3121
1737
self.repo_dir = repo.bzrdir
3122
self.step('Moving repository to repository.backup')
1738
pb.update(gettext('Moving repository to repository.backup'))
3123
1739
self.repo_dir.transport.move('repository', 'repository.backup')
3124
1740
backup_transport = self.repo_dir.transport.clone('repository.backup')
3125
1741
repo._format.check_conversion_target(self.target_format)
3126
1742
self.source_repo = repo._format.open(self.repo_dir,
3128
1744
_override_transport=backup_transport)
3129
self.step('Creating new repository')
1745
pb.update(gettext('Creating new repository'))
3130
1746
converted = self.target_format.initialize(self.repo_dir,
3131
1747
self.source_repo.is_shared())
3132
1748
converted.lock_write()
3134
self.step('Copying content into repository.')
1750
pb.update(gettext('Copying content'))
3135
1751
self.source_repo.copy_content_into(converted)
3137
1753
converted.unlock()
3138
self.step('Deleting old repository content.')
1754
pb.update(gettext('Deleting old repository content'))
3139
1755
self.repo_dir.transport.delete_tree('repository.backup')
3140
self.pb.note('repository converted')
3142
def step(self, message):
3143
"""Update the pb by a step."""
3145
self.pb.update(message, self.count, self.total)
3157
def _unescaper(match, _map=_unescape_map):
3158
code = match.group(1)
3162
if not code.startswith('#'):
3164
return unichr(int(code[1:])).encode('utf8')
3170
def _unescape_xml(data):
3171
"""Unescape predefined XML entities in a string of data."""
3173
if _unescape_re is None:
3174
_unescape_re = re.compile('\&([^;]*);')
3175
return _unescape_re.sub(_unescaper, data)
3178
class _VersionedFileChecker(object):
3180
def __init__(self, repository):
3181
self.repository = repository
3182
self.text_index = self.repository._generate_text_key_index()
3184
def calculate_file_version_parents(self, text_key):
3185
"""Calculate the correct parents for a file version according to
3188
parent_keys = self.text_index[text_key]
3189
if parent_keys == [_mod_revision.NULL_REVISION]:
3191
return tuple(parent_keys)
3193
def check_file_version_parents(self, texts, progress_bar=None):
3194
"""Check the parents stored in a versioned file are correct.
3196
It also detects file versions that are not referenced by their
3197
corresponding revision's inventory.
3199
:returns: A tuple of (wrong_parents, dangling_file_versions).
3200
wrong_parents is a dict mapping {revision_id: (stored_parents,
3201
correct_parents)} for each revision_id where the stored parents
3202
are not correct. dangling_file_versions is a set of (file_id,
3203
revision_id) tuples for versions that are present in this versioned
3204
file, but not used by the corresponding inventory.
3207
self.file_ids = set([file_id for file_id, _ in
3208
self.text_index.iterkeys()])
3209
# text keys is now grouped by file_id
3210
n_weaves = len(self.file_ids)
3211
files_in_revisions = {}
3212
revisions_of_files = {}
3213
n_versions = len(self.text_index)
3214
progress_bar.update('loading text store', 0, n_versions)
3215
parent_map = self.repository.texts.get_parent_map(self.text_index)
3216
# On unlistable transports this could well be empty/error...
3217
text_keys = self.repository.texts.keys()
3218
unused_keys = frozenset(text_keys) - set(self.text_index)
3219
for num, key in enumerate(self.text_index.iterkeys()):
3220
if progress_bar is not None:
3221
progress_bar.update('checking text graph', num, n_versions)
3222
correct_parents = self.calculate_file_version_parents(key)
3224
knit_parents = parent_map[key]
3225
except errors.RevisionNotPresent:
3228
if correct_parents != knit_parents:
3229
wrong_parents[key] = (knit_parents, correct_parents)
3230
return wrong_parents, unused_keys
3233
def _old_get_graph(repository, revision_id):
3234
"""DO NOT USE. That is all. I'm serious."""
3235
graph = repository.get_graph()
3236
revision_graph = dict(((key, value) for key, value in
3237
graph.iter_ancestry([revision_id]) if value is not None))
3238
return _strip_NULL_ghosts(revision_graph)
1756
ui.ui_factory.note(gettext('repository converted'))
3241
1760
def _strip_NULL_ghosts(revision_graph):
added
removed
bzrlib/repository.py
