1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from bzrlib.lazy_import import lazy_import
18
lazy_import(globals(), """
37
revision as _mod_revision,
42
from bzrlib.bundle import serializer
43
from bzrlib.revisiontree import RevisionTree
44
from bzrlib.store.versioned import VersionedFileStore
45
from bzrlib.testament import Testament
48
from bzrlib import registry
49
from bzrlib.decorators import needs_read_lock, needs_write_lock
50
from bzrlib.inter import InterObject
51
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
52
from bzrlib.symbol_versioning import (
58
from bzrlib.trace import (
59
log_exception_quietly, note, mutter, mutter_callsite, warning)
62
# Old formats display a warning, but only once
63
_deprecation_warning_done = False
66
class CommitBuilder(object):
67
"""Provides an interface to build up a commit.
69
This allows describing a tree to be committed without needing to
70
know the internals of the format of the repository.
73
# all clients should supply tree roots.
74
record_root_entry = True
75
# the default CommitBuilder does not manage trees whose root is versioned.
76
_versioned_root = False
78
def __init__(self, repository, parents, config, timestamp=None,
79
timezone=None, committer=None, revprops=None,
81
"""Initiate a CommitBuilder.
83
:param repository: Repository to commit to.
84
:param parents: Revision ids of the parents of the new revision.
85
:param config: Configuration to use.
86
:param timestamp: Optional timestamp recorded for commit.
87
:param timezone: Optional timezone for timestamp.
88
:param committer: Optional committer to set for commit.
89
:param revprops: Optional dictionary of revision properties.
90
:param revision_id: Optional revision id.
95
self._committer = self._config.username()
97
self._committer = committer
99
self.new_inventory = Inventory(None)
100
self._new_revision_id = revision_id
101
self.parents = parents
102
self.repository = repository
105
if revprops is not None:
106
self._revprops.update(revprops)
108
if timestamp is None:
109
timestamp = time.time()
110
# Restrict resolution to 1ms
111
self._timestamp = round(timestamp, 3)
114
self._timezone = osutils.local_time_offset()
116
self._timezone = int(timezone)
118
self._generate_revision_if_needed()
119
self.__heads = graph.HeadsCache(repository.get_graph()).heads
121
def commit(self, message):
122
"""Make the actual commit.
124
:return: The revision id of the recorded revision.
126
rev = _mod_revision.Revision(
127
timestamp=self._timestamp,
128
timezone=self._timezone,
129
committer=self._committer,
131
inventory_sha1=self.inv_sha1,
132
revision_id=self._new_revision_id,
133
properties=self._revprops)
134
rev.parent_ids = self.parents
135
self.repository.add_revision(self._new_revision_id, rev,
136
self.new_inventory, self._config)
137
self.repository.commit_write_group()
138
return self._new_revision_id
141
"""Abort the commit that is being built.
143
self.repository.abort_write_group()
145
def revision_tree(self):
146
"""Return the tree that was just committed.
148
After calling commit() this can be called to get a RevisionTree
149
representing the newly committed tree. This is preferred to
150
calling Repository.revision_tree() because that may require
151
deserializing the inventory, while we already have a copy in
154
return RevisionTree(self.repository, self.new_inventory,
155
self._new_revision_id)
157
def finish_inventory(self):
158
"""Tell the builder that the inventory is finished."""
159
if self.new_inventory.root is None:
160
raise AssertionError('Root entry should be supplied to'
161
' record_entry_contents, as of bzr 0.10.')
162
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
163
self.new_inventory.revision_id = self._new_revision_id
164
self.inv_sha1 = self.repository.add_inventory(
165
self._new_revision_id,
170
def _gen_revision_id(self):
171
"""Return new revision-id."""
172
return generate_ids.gen_revision_id(self._config.username(),
175
def _generate_revision_if_needed(self):
176
"""Create a revision id if None was supplied.
178
If the repository can not support user-specified revision ids
179
they should override this function and raise CannotSetRevisionId
180
if _new_revision_id is not None.
182
:raises: CannotSetRevisionId
184
if self._new_revision_id is None:
185
self._new_revision_id = self._gen_revision_id()
186
self.random_revid = True
188
self.random_revid = False
190
def _heads(self, file_id, revision_ids):
191
"""Calculate the graph heads for revision_ids in the graph of file_id.
193
This can use either a per-file graph or a global revision graph as we
194
have an identity relationship between the two graphs.
196
return self.__heads(revision_ids)
198
def _check_root(self, ie, parent_invs, tree):
199
"""Helper for record_entry_contents.
201
:param ie: An entry being added.
202
:param parent_invs: The inventories of the parent revisions of the
204
:param tree: The tree that is being committed.
206
# In this revision format, root entries have no knit or weave When
207
# serializing out to disk and back in root.revision is always
209
ie.revision = self._new_revision_id
211
def _get_delta(self, ie, basis_inv, path):
212
"""Get a delta against the basis inventory for ie."""
213
if ie.file_id not in basis_inv:
215
return (None, path, ie.file_id, ie)
216
elif ie != basis_inv[ie.file_id]:
218
# TODO: avoid tis id2path call.
219
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
224
def record_entry_contents(self, ie, parent_invs, path, tree,
226
"""Record the content of ie from tree into the commit if needed.
228
Side effect: sets ie.revision when unchanged
230
:param ie: An inventory entry present in the commit.
231
:param parent_invs: The inventories of the parent revisions of the
233
:param path: The path the entry is at in the tree.
234
:param tree: The tree which contains this entry and should be used to
236
:param content_summary: Summary data from the tree about the paths
237
content - stat, length, exec, sha/link target. This is only
238
accessed when the entry has a revision of None - that is when it is
239
a candidate to commit.
240
:return: A tuple (change_delta, version_recorded, fs_hash).
241
change_delta is an inventory_delta change for this entry against
242
the basis tree of the commit, or None if no change occured against
244
version_recorded is True if a new version of the entry has been
245
recorded. For instance, committing a merge where a file was only
246
changed on the other side will return (delta, False).
247
fs_hash is either None, or the hash details for the path (currently
248
a tuple of the contents sha1 and the statvalue returned by
249
tree.get_file_with_stat()).
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)
290
return delta, False, None
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))
301
return None, False, None
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:
337
# Stat cache fingerprint feedback for the caller - None as we usually
338
# don't generate one.
341
if content_summary[2] is None:
342
raise ValueError("Files must not have executable = None")
344
if (# if the file length changed we have to store:
345
parent_entry.text_size != content_summary[1] or
346
# if the exec bit has changed we have to store:
347
parent_entry.executable != content_summary[2]):
349
elif parent_entry.text_sha1 == content_summary[3]:
350
# all meta and content is unchanged (using a hash cache
351
# hit to check the sha)
352
ie.revision = parent_entry.revision
353
ie.text_size = parent_entry.text_size
354
ie.text_sha1 = parent_entry.text_sha1
355
ie.executable = parent_entry.executable
356
return self._get_delta(ie, basis_inv, path), False, None
358
# Either there is only a hash change(no hash cache entry,
359
# or same size content change), or there is no change on
361
# Provide the parent's hash to the store layer, so that the
362
# content is unchanged we will not store a new node.
363
nostore_sha = parent_entry.text_sha1
365
# We want to record a new node regardless of the presence or
366
# absence of a content change in the file.
368
ie.executable = content_summary[2]
369
file_obj, stat_value = tree.get_file_with_stat(ie.file_id, path)
371
lines = file_obj.readlines()
375
ie.text_sha1, ie.text_size = self._add_text_to_weave(
376
ie.file_id, lines, heads, nostore_sha)
377
# Let the caller know we generated a stat fingerprint.
378
fingerprint = (ie.text_sha1, stat_value)
379
except errors.ExistingContent:
380
# Turns out that the file content was unchanged, and we were
381
# only going to store a new node if it was changed. Carry over
383
ie.revision = parent_entry.revision
384
ie.text_size = parent_entry.text_size
385
ie.text_sha1 = parent_entry.text_sha1
386
ie.executable = parent_entry.executable
387
return self._get_delta(ie, basis_inv, path), False, None
388
elif kind == 'directory':
390
# all data is meta here, nothing specific to directory, so
392
ie.revision = parent_entry.revision
393
return self._get_delta(ie, basis_inv, path), False, None
395
self._add_text_to_weave(ie.file_id, lines, heads, None)
396
elif kind == 'symlink':
397
current_link_target = content_summary[3]
399
# symlink target is not generic metadata, check if it has
401
if current_link_target != parent_entry.symlink_target:
404
# unchanged, carry over.
405
ie.revision = parent_entry.revision
406
ie.symlink_target = parent_entry.symlink_target
407
return self._get_delta(ie, basis_inv, path), False, None
408
ie.symlink_target = current_link_target
410
self._add_text_to_weave(ie.file_id, lines, heads, None)
411
elif kind == 'tree-reference':
413
if content_summary[3] != parent_entry.reference_revision:
416
# unchanged, carry over.
417
ie.reference_revision = parent_entry.reference_revision
418
ie.revision = parent_entry.revision
419
return self._get_delta(ie, basis_inv, path), False, None
420
ie.reference_revision = content_summary[3]
422
self._add_text_to_weave(ie.file_id, lines, heads, None)
424
raise NotImplementedError('unknown kind')
425
ie.revision = self._new_revision_id
426
return self._get_delta(ie, basis_inv, path), True, fingerprint
428
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
429
# Note: as we read the content directly from the tree, we know its not
430
# been turned into unicode or badly split - but a broken tree
431
# implementation could give us bad output from readlines() so this is
432
# not a guarantee of safety. What would be better is always checking
433
# the content during test suite execution. RBC 20070912
434
parent_keys = tuple((file_id, parent) for parent in parents)
435
return self.repository.texts.add_lines(
436
(file_id, self._new_revision_id), parent_keys, new_lines,
437
nostore_sha=nostore_sha, random_id=self.random_revid,
438
check_content=False)[0:2]
441
class RootCommitBuilder(CommitBuilder):
442
"""This commitbuilder actually records the root id"""
444
# the root entry gets versioned properly by this builder.
445
_versioned_root = True
447
def _check_root(self, ie, parent_invs, tree):
448
"""Helper for record_entry_contents.
450
:param ie: An entry being added.
451
:param parent_invs: The inventories of the parent revisions of the
453
:param tree: The tree that is being committed.
457
######################################################################
460
class Repository(object):
461
"""Repository holding history for one or more branches.
463
The repository holds and retrieves historical information including
464
revisions and file history. It's normally accessed only by the Branch,
465
which views a particular line of development through that history.
467
The Repository builds on top of some byte storage facilies (the revisions,
468
signatures, inventories and texts attributes) and a Transport, which
469
respectively provide byte storage and a means to access the (possibly
472
The byte storage facilities are addressed via tuples, which we refer to
473
as 'keys' throughout the code base. Revision_keys, inventory_keys and
474
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
475
(file_id, revision_id). We use this interface because it allows low
476
friction with the underlying code that implements disk indices, network
477
encoding and other parts of bzrlib.
479
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
480
the serialised revisions for the repository. This can be used to obtain
481
revision graph information or to access raw serialised revisions.
482
The result of trying to insert data into the repository via this store
483
is undefined: it should be considered read-only except for implementors
485
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
486
the serialised signatures for the repository. This can be used to
487
obtain access to raw serialised signatures. The result of trying to
488
insert data into the repository via this store is undefined: it should
489
be considered read-only except for implementors of repositories.
490
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
491
the serialised inventories for the repository. This can be used to
492
obtain unserialised inventories. The result of trying to insert data
493
into the repository via this store is undefined: it should be
494
considered read-only except for implementors of repositories.
495
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
496
texts of files and directories for the repository. This can be used to
497
obtain file texts or file graphs. Note that Repository.iter_file_bytes
498
is usually a better interface for accessing file texts.
499
The result of trying to insert data into the repository via this store
500
is undefined: it should be considered read-only except for implementors
502
:ivar _transport: Transport for file access to repository, typically
503
pointing to .bzr/repository.
506
# What class to use for a CommitBuilder. Often its simpler to change this
507
# in a Repository class subclass rather than to override
508
# get_commit_builder.
509
_commit_builder_class = CommitBuilder
510
# The search regex used by xml based repositories to determine what things
511
# where changed in a single commit.
512
_file_ids_altered_regex = lazy_regex.lazy_compile(
513
r'file_id="(?P<file_id>[^"]+)"'
514
r'.* revision="(?P<revision_id>[^"]+)"'
517
def abort_write_group(self, suppress_errors=False):
518
"""Commit the contents accrued within the current write group.
520
:param suppress_errors: if true, abort_write_group will catch and log
521
unexpected errors that happen during the abort, rather than
522
allowing them to propagate. Defaults to False.
524
:seealso: start_write_group.
526
if self._write_group is not self.get_transaction():
527
# has an unlock or relock occured ?
528
raise errors.BzrError('mismatched lock context and write group.')
530
self._abort_write_group()
531
except Exception, exc:
532
self._write_group = None
533
if not suppress_errors:
535
mutter('abort_write_group failed')
536
log_exception_quietly()
537
note('bzr: ERROR (ignored): %s', exc)
538
self._write_group = None
540
def _abort_write_group(self):
541
"""Template method for per-repository write group cleanup.
543
This is called during abort before the write group is considered to be
544
finished and should cleanup any internal state accrued during the write
545
group. There is no requirement that data handed to the repository be
546
*not* made available - this is not a rollback - but neither should any
547
attempt be made to ensure that data added is fully commited. Abort is
548
invoked when an error has occured so futher disk or network operations
549
may not be possible or may error and if possible should not be
553
def add_fallback_repository(self, repository):
554
"""Add a repository to use for looking up data not held locally.
556
:param repository: A repository.
558
if not self._format.supports_external_lookups:
559
raise errors.UnstackableRepositoryFormat(self._format, self.base)
560
self._check_fallback_repository(repository)
561
self._fallback_repositories.append(repository)
562
self.texts.add_fallback_versioned_files(repository.texts)
563
self.inventories.add_fallback_versioned_files(repository.inventories)
564
self.revisions.add_fallback_versioned_files(repository.revisions)
565
self.signatures.add_fallback_versioned_files(repository.signatures)
567
def _check_fallback_repository(self, repository):
568
"""Check that this repository can fallback to repository safely.
570
Raise an error if not.
572
:param repository: A repository to fallback to.
574
return InterRepository._assert_same_model(self, repository)
576
def add_inventory(self, revision_id, inv, parents):
577
"""Add the inventory inv to the repository as revision_id.
579
:param parents: The revision ids of the parents that revision_id
580
is known to have and are in the repository already.
582
:returns: The validator(which is a sha1 digest, though what is sha'd is
583
repository format specific) of the serialized inventory.
585
if not self.is_in_write_group():
586
raise AssertionError("%r not in write group" % (self,))
587
_mod_revision.check_not_reserved_id(revision_id)
588
if not (inv.revision_id is None or inv.revision_id == revision_id):
589
raise AssertionError(
590
"Mismatch between inventory revision"
591
" id and insertion revid (%r, %r)"
592
% (inv.revision_id, revision_id))
594
raise AssertionError()
595
inv_lines = self._serialise_inventory_to_lines(inv)
596
return self._inventory_add_lines(revision_id, parents,
597
inv_lines, check_content=False)
599
def _inventory_add_lines(self, revision_id, parents, lines,
601
"""Store lines in inv_vf and return the sha1 of the inventory."""
602
parents = [(parent,) for parent in parents]
603
return self.inventories.add_lines((revision_id,), parents, lines,
604
check_content=check_content)[0]
606
def add_revision(self, revision_id, rev, inv=None, config=None):
607
"""Add rev to the revision store as revision_id.
609
:param revision_id: the revision id to use.
610
:param rev: The revision object.
611
:param inv: The inventory for the revision. if None, it will be looked
612
up in the inventory storer
613
:param config: If None no digital signature will be created.
614
If supplied its signature_needed method will be used
615
to determine if a signature should be made.
617
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
619
_mod_revision.check_not_reserved_id(revision_id)
620
if config is not None and config.signature_needed():
622
inv = self.get_inventory(revision_id)
623
plaintext = Testament(rev, inv).as_short_text()
624
self.store_revision_signature(
625
gpg.GPGStrategy(config), plaintext, revision_id)
626
# check inventory present
627
if not self.inventories.get_parent_map([(revision_id,)]):
629
raise errors.WeaveRevisionNotPresent(revision_id,
632
# yes, this is not suitable for adding with ghosts.
633
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
637
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
638
self._add_revision(rev)
640
def _add_revision(self, revision):
641
text = self._serializer.write_revision_to_string(revision)
642
key = (revision.revision_id,)
643
parents = tuple((parent,) for parent in revision.parent_ids)
644
self.revisions.add_lines(key, parents, osutils.split_lines(text))
646
def all_revision_ids(self):
647
"""Returns a list of all the revision ids in the repository.
649
This is conceptually deprecated because code should generally work on
650
the graph reachable from a particular revision, and ignore any other
651
revisions that might be present. There is no direct replacement
654
if 'evil' in debug.debug_flags:
655
mutter_callsite(2, "all_revision_ids is linear with history.")
656
return self._all_revision_ids()
658
def _all_revision_ids(self):
659
"""Returns a list of all the revision ids in the repository.
661
These are in as much topological order as the underlying store can
664
raise NotImplementedError(self._all_revision_ids)
666
def break_lock(self):
667
"""Break a lock if one is present from another instance.
669
Uses the ui factory to ask for confirmation if the lock may be from
672
self.control_files.break_lock()
675
def _eliminate_revisions_not_present(self, revision_ids):
676
"""Check every revision id in revision_ids to see if we have it.
678
Returns a set of the present revisions.
681
graph = self.get_graph()
682
parent_map = graph.get_parent_map(revision_ids)
683
# The old API returned a list, should this actually be a set?
684
return parent_map.keys()
687
def create(a_bzrdir):
688
"""Construct the current default format repository in a_bzrdir."""
689
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
691
def __init__(self, _format, a_bzrdir, control_files):
692
"""instantiate a Repository.
694
:param _format: The format of the repository on disk.
695
:param a_bzrdir: The BzrDir of the repository.
697
In the future we will have a single api for all stores for
698
getting file texts, inventories and revisions, then
699
this construct will accept instances of those things.
701
super(Repository, self).__init__()
702
self._format = _format
703
# the following are part of the public API for Repository:
704
self.bzrdir = a_bzrdir
705
self.control_files = control_files
706
self._transport = control_files._transport
707
self.base = self._transport.base
709
self._reconcile_does_inventory_gc = True
710
self._reconcile_fixes_text_parents = False
711
self._reconcile_backsup_inventory = True
712
# not right yet - should be more semantically clear ?
714
# TODO: make sure to construct the right store classes, etc, depending
715
# on whether escaping is required.
716
self._warn_if_deprecated()
717
self._write_group = None
718
# Additional places to query for data.
719
self._fallback_repositories = []
720
# What order should fetch operations request streams in?
721
# The default is unordered as that is the cheapest for an origin to
723
self._fetch_order = 'unordered'
724
# Does this repository use deltas that can be fetched as-deltas ?
725
# (E.g. knits, where the knit deltas can be transplanted intact.
726
# We default to False, which will ensure that enough data to get
727
# a full text out of any fetch stream will be grabbed.
728
self._fetch_uses_deltas = False
729
# Should fetch trigger a reconcile after the fetch? Only needed for
730
# some repository formats that can suffer internal inconsistencies.
731
self._fetch_reconcile = False
734
return '%s(%r)' % (self.__class__.__name__,
737
def has_same_location(self, other):
738
"""Returns a boolean indicating if this repository is at the same
739
location as another repository.
741
This might return False even when two repository objects are accessing
742
the same physical repository via different URLs.
744
if self.__class__ is not other.__class__:
746
return (self._transport.base == other._transport.base)
748
def is_in_write_group(self):
749
"""Return True if there is an open write group.
751
:seealso: start_write_group.
753
return self._write_group is not None
756
return self.control_files.is_locked()
758
def is_write_locked(self):
759
"""Return True if this object is write locked."""
760
return self.is_locked() and self.control_files._lock_mode == 'w'
762
def lock_write(self, token=None):
763
"""Lock this repository for writing.
765
This causes caching within the repository obejct to start accumlating
766
data during reads, and allows a 'write_group' to be obtained. Write
767
groups must be used for actual data insertion.
769
:param token: if this is already locked, then lock_write will fail
770
unless the token matches the existing lock.
771
:returns: a token if this instance supports tokens, otherwise None.
772
:raises TokenLockingNotSupported: when a token is given but this
773
instance doesn't support using token locks.
774
:raises MismatchedToken: if the specified token doesn't match the token
775
of the existing lock.
776
:seealso: start_write_group.
778
A token should be passed in if you know that you have locked the object
779
some other way, and need to synchronise this object's state with that
782
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
784
result = self.control_files.lock_write(token=token)
785
for repo in self._fallback_repositories:
786
# Writes don't affect fallback repos
792
self.control_files.lock_read()
793
for repo in self._fallback_repositories:
797
def get_physical_lock_status(self):
798
return self.control_files.get_physical_lock_status()
800
def leave_lock_in_place(self):
801
"""Tell this repository not to release the physical lock when this
804
If lock_write doesn't return a token, then this method is not supported.
806
self.control_files.leave_in_place()
808
def dont_leave_lock_in_place(self):
809
"""Tell this repository to release the physical lock when this
810
object is unlocked, even if it didn't originally acquire it.
812
If lock_write doesn't return a token, then this method is not supported.
814
self.control_files.dont_leave_in_place()
817
def gather_stats(self, revid=None, committers=None):
818
"""Gather statistics from a revision id.
820
:param revid: The revision id to gather statistics from, if None, then
821
no revision specific statistics are gathered.
822
:param committers: Optional parameter controlling whether to grab
823
a count of committers from the revision specific statistics.
824
:return: A dictionary of statistics. Currently this contains:
825
committers: The number of committers if requested.
826
firstrev: A tuple with timestamp, timezone for the penultimate left
827
most ancestor of revid, if revid is not the NULL_REVISION.
828
latestrev: A tuple with timestamp, timezone for revid, if revid is
829
not the NULL_REVISION.
830
revisions: The total revision count in the repository.
831
size: An estimate disk size of the repository in bytes.
834
if revid and committers:
835
result['committers'] = 0
836
if revid and revid != _mod_revision.NULL_REVISION:
838
all_committers = set()
839
revisions = self.get_ancestry(revid)
840
# pop the leading None
842
first_revision = None
844
# ignore the revisions in the middle - just grab first and last
845
revisions = revisions[0], revisions[-1]
846
for revision in self.get_revisions(revisions):
847
if not first_revision:
848
first_revision = revision
850
all_committers.add(revision.committer)
851
last_revision = revision
853
result['committers'] = len(all_committers)
854
result['firstrev'] = (first_revision.timestamp,
855
first_revision.timezone)
856
result['latestrev'] = (last_revision.timestamp,
857
last_revision.timezone)
859
# now gather global repository information
860
# XXX: This is available for many repos regardless of listability.
861
if self.bzrdir.root_transport.listable():
862
# XXX: do we want to __define len__() ?
863
# Maybe the versionedfiles object should provide a different
864
# method to get the number of keys.
865
result['revisions'] = len(self.revisions.keys())
869
def find_branches(self, using=False):
870
"""Find branches underneath this repository.
872
This will include branches inside other branches.
874
:param using: If True, list only branches using this repository.
876
if using and not self.is_shared():
878
return [self.bzrdir.open_branch()]
879
except errors.NotBranchError:
881
class Evaluator(object):
884
self.first_call = True
886
def __call__(self, bzrdir):
887
# On the first call, the parameter is always the bzrdir
888
# containing the current repo.
889
if not self.first_call:
891
repository = bzrdir.open_repository()
892
except errors.NoRepositoryPresent:
895
return False, (None, repository)
896
self.first_call = False
898
value = (bzrdir.open_branch(), None)
899
except errors.NotBranchError:
904
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
905
self.bzrdir.root_transport, evaluate=Evaluator()):
906
if branch is not None:
907
branches.append(branch)
908
if not using and repository is not None:
909
branches.extend(repository.find_branches())
913
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
914
"""Return the revision ids that other has that this does not.
916
These are returned in topological order.
918
revision_id: only return revision ids included by revision_id.
920
return InterRepository.get(other, self).search_missing_revision_ids(
921
revision_id, find_ghosts)
923
@deprecated_method(one_two)
925
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
926
"""Return the revision ids that other has that this does not.
928
These are returned in topological order.
930
revision_id: only return revision ids included by revision_id.
932
keys = self.search_missing_revision_ids(
933
other, revision_id, find_ghosts).get_keys()
936
parents = other.get_graph().get_parent_map(keys)
939
return tsort.topo_sort(parents)
943
"""Open the repository rooted at base.
945
For instance, if the repository is at URL/.bzr/repository,
946
Repository.open(URL) -> a Repository instance.
948
control = bzrdir.BzrDir.open(base)
949
return control.open_repository()
951
def copy_content_into(self, destination, revision_id=None):
952
"""Make a complete copy of the content in self into destination.
954
This is a destructive operation! Do not use it on existing
957
return InterRepository.get(self, destination).copy_content(revision_id)
959
def commit_write_group(self):
960
"""Commit the contents accrued within the current write group.
962
:seealso: start_write_group.
964
if self._write_group is not self.get_transaction():
965
# has an unlock or relock occured ?
966
raise errors.BzrError('mismatched lock context %r and '
968
(self.get_transaction(), self._write_group))
969
self._commit_write_group()
970
self._write_group = None
972
def _commit_write_group(self):
973
"""Template method for per-repository write group cleanup.
975
This is called before the write group is considered to be
976
finished and should ensure that all data handed to the repository
977
for writing during the write group is safely committed (to the
978
extent possible considering file system caching etc).
981
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
982
"""Fetch the content required to construct revision_id from source.
984
If revision_id is None all content is copied.
985
:param find_ghosts: Find and copy revisions in the source that are
986
ghosts in the target (and not reachable directly by walking out to
987
the first-present revision in target from revision_id).
989
# fast path same-url fetch operations
990
if self.has_same_location(source):
991
# check that last_revision is in 'from' and then return a
993
if (revision_id is not None and
994
not _mod_revision.is_null(revision_id)):
995
self.get_revision(revision_id)
997
# if there is no specific appropriate InterRepository, this will get
998
# the InterRepository base class, which raises an
999
# IncompatibleRepositories when asked to fetch.
1000
inter = InterRepository.get(source, self)
1001
return inter.fetch(revision_id=revision_id, pb=pb,
1002
find_ghosts=find_ghosts)
1004
def create_bundle(self, target, base, fileobj, format=None):
1005
return serializer.write_bundle(self, target, base, fileobj, format)
1007
def get_commit_builder(self, branch, parents, config, timestamp=None,
1008
timezone=None, committer=None, revprops=None,
1010
"""Obtain a CommitBuilder for this repository.
1012
:param branch: Branch to commit to.
1013
:param parents: Revision ids of the parents of the new revision.
1014
:param config: Configuration to use.
1015
:param timestamp: Optional timestamp recorded for commit.
1016
:param timezone: Optional timezone for timestamp.
1017
:param committer: Optional committer to set for commit.
1018
:param revprops: Optional dictionary of revision properties.
1019
:param revision_id: Optional revision id.
1021
result = self._commit_builder_class(self, parents, config,
1022
timestamp, timezone, committer, revprops, revision_id)
1023
self.start_write_group()
1027
if (self.control_files._lock_count == 1 and
1028
self.control_files._lock_mode == 'w'):
1029
if self._write_group is not None:
1030
self.abort_write_group()
1031
self.control_files.unlock()
1032
raise errors.BzrError(
1033
'Must end write groups before releasing write locks.')
1034
self.control_files.unlock()
1035
for repo in self._fallback_repositories:
1039
def clone(self, a_bzrdir, revision_id=None):
1040
"""Clone this repository into a_bzrdir using the current format.
1042
Currently no check is made that the format of this repository and
1043
the bzrdir format are compatible. FIXME RBC 20060201.
1045
:return: The newly created destination repository.
1047
# TODO: deprecate after 0.16; cloning this with all its settings is
1048
# probably not very useful -- mbp 20070423
1049
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1050
self.copy_content_into(dest_repo, revision_id)
1053
def start_write_group(self):
1054
"""Start a write group in the repository.
1056
Write groups are used by repositories which do not have a 1:1 mapping
1057
between file ids and backend store to manage the insertion of data from
1058
both fetch and commit operations.
1060
A write lock is required around the start_write_group/commit_write_group
1061
for the support of lock-requiring repository formats.
1063
One can only insert data into a repository inside a write group.
1067
if not self.is_write_locked():
1068
raise errors.NotWriteLocked(self)
1069
if self._write_group:
1070
raise errors.BzrError('already in a write group')
1071
self._start_write_group()
1072
# so we can detect unlock/relock - the write group is now entered.
1073
self._write_group = self.get_transaction()
1075
def _start_write_group(self):
1076
"""Template method for per-repository write group startup.
1078
This is called before the write group is considered to be
1083
def sprout(self, to_bzrdir, revision_id=None):
1084
"""Create a descendent repository for new development.
1086
Unlike clone, this does not copy the settings of the repository.
1088
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1089
dest_repo.fetch(self, revision_id=revision_id)
1092
def _create_sprouting_repo(self, a_bzrdir, shared):
1093
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1094
# use target default format.
1095
dest_repo = a_bzrdir.create_repository()
1097
# Most control formats need the repository to be specifically
1098
# created, but on some old all-in-one formats it's not needed
1100
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1101
except errors.UninitializableFormat:
1102
dest_repo = a_bzrdir.open_repository()
1106
def has_revision(self, revision_id):
1107
"""True if this repository has a copy of the revision."""
1108
return revision_id in self.has_revisions((revision_id,))
1111
def has_revisions(self, revision_ids):
1112
"""Probe to find out the presence of multiple revisions.
1114
:param revision_ids: An iterable of revision_ids.
1115
:return: A set of the revision_ids that were present.
1117
parent_map = self.revisions.get_parent_map(
1118
[(rev_id,) for rev_id in revision_ids])
1120
if _mod_revision.NULL_REVISION in revision_ids:
1121
result.add(_mod_revision.NULL_REVISION)
1122
result.update([key[0] for key in parent_map])
1126
def get_revision(self, revision_id):
1127
"""Return the Revision object for a named revision."""
1128
return self.get_revisions([revision_id])[0]
1131
def get_revision_reconcile(self, revision_id):
1132
"""'reconcile' helper routine that allows access to a revision always.
1134
This variant of get_revision does not cross check the weave graph
1135
against the revision one as get_revision does: but it should only
1136
be used by reconcile, or reconcile-alike commands that are correcting
1137
or testing the revision graph.
1139
return self._get_revisions([revision_id])[0]
1142
def get_revisions(self, revision_ids):
1143
"""Get many revisions at once."""
1144
return self._get_revisions(revision_ids)
1147
def _get_revisions(self, revision_ids):
1148
"""Core work logic to get many revisions without sanity checks."""
1149
for rev_id in revision_ids:
1150
if not rev_id or not isinstance(rev_id, basestring):
1151
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1152
keys = [(key,) for key in revision_ids]
1153
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1155
for record in stream:
1156
if record.storage_kind == 'absent':
1157
raise errors.NoSuchRevision(self, record.key[0])
1158
text = record.get_bytes_as('fulltext')
1159
rev = self._serializer.read_revision_from_string(text)
1160
revs[record.key[0]] = rev
1161
return [revs[revid] for revid in revision_ids]
1164
def get_revision_xml(self, revision_id):
1165
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1166
# would have already do it.
1167
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1168
rev = self.get_revision(revision_id)
1169
rev_tmp = cStringIO.StringIO()
1170
# the current serializer..
1171
self._serializer.write_revision(rev, rev_tmp)
1173
return rev_tmp.getvalue()
1175
def get_deltas_for_revisions(self, revisions):
1176
"""Produce a generator of revision deltas.
1178
Note that the input is a sequence of REVISIONS, not revision_ids.
1179
Trees will be held in memory until the generator exits.
1180
Each delta is relative to the revision's lefthand predecessor.
1182
required_trees = set()
1183
for revision in revisions:
1184
required_trees.add(revision.revision_id)
1185
required_trees.update(revision.parent_ids[:1])
1186
trees = dict((t.get_revision_id(), t) for
1187
t in self.revision_trees(required_trees))
1188
for revision in revisions:
1189
if not revision.parent_ids:
1190
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
1192
old_tree = trees[revision.parent_ids[0]]
1193
yield trees[revision.revision_id].changes_from(old_tree)
1196
def get_revision_delta(self, revision_id):
1197
"""Return the delta for one revision.
1199
The delta is relative to the left-hand predecessor of the
1202
r = self.get_revision(revision_id)
1203
return list(self.get_deltas_for_revisions([r]))[0]
1206
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1207
signature = gpg_strategy.sign(plaintext)
1208
self.add_signature_text(revision_id, signature)
1211
def add_signature_text(self, revision_id, signature):
1212
self.signatures.add_lines((revision_id,), (),
1213
osutils.split_lines(signature))
1215
def find_text_key_references(self):
1216
"""Find the text key references within the repository.
1218
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1219
revision_ids. Each altered file-ids has the exact revision_ids that
1220
altered it listed explicitly.
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. The inventory texts from all present
1224
revision ids are assessed to generate this report.
1226
revision_keys = self.revisions.keys()
1227
w = self.inventories
1228
pb = ui.ui_factory.nested_progress_bar()
1230
return self._find_text_key_references_from_xml_inventory_lines(
1231
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1235
def _find_text_key_references_from_xml_inventory_lines(self,
1237
"""Core routine for extracting references to texts from inventories.
1239
This performs the translation of xml lines to revision ids.
1241
:param line_iterator: An iterator of lines, origin_version_id
1242
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1243
to whether they were referred to by the inventory of the
1244
revision_id that they contain. Note that if that revision_id was
1245
not part of the line_iterator's output then False will be given -
1246
even though it may actually refer to that key.
1248
if not self._serializer.support_altered_by_hack:
1249
raise AssertionError(
1250
"_find_text_key_references_from_xml_inventory_lines only "
1251
"supported for branches which store inventory as unnested xml"
1252
", not on %r" % self)
1255
# this code needs to read every new line in every inventory for the
1256
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1257
# not present in one of those inventories is unnecessary but not
1258
# harmful because we are filtering by the revision id marker in the
1259
# inventory lines : we only select file ids altered in one of those
1260
# revisions. We don't need to see all lines in the inventory because
1261
# only those added in an inventory in rev X can contain a revision=X
1263
unescape_revid_cache = {}
1264
unescape_fileid_cache = {}
1266
# jam 20061218 In a big fetch, this handles hundreds of thousands
1267
# of lines, so it has had a lot of inlining and optimizing done.
1268
# Sorry that it is a little bit messy.
1269
# Move several functions to be local variables, since this is a long
1271
search = self._file_ids_altered_regex.search
1272
unescape = _unescape_xml
1273
setdefault = result.setdefault
1274
for line, line_key in line_iterator:
1275
match = search(line)
1278
# One call to match.group() returning multiple items is quite a
1279
# bit faster than 2 calls to match.group() each returning 1
1280
file_id, revision_id = match.group('file_id', 'revision_id')
1282
# Inlining the cache lookups helps a lot when you make 170,000
1283
# lines and 350k ids, versus 8.4 unique ids.
1284
# Using a cache helps in 2 ways:
1285
# 1) Avoids unnecessary decoding calls
1286
# 2) Re-uses cached strings, which helps in future set and
1288
# (2) is enough that removing encoding entirely along with
1289
# the cache (so we are using plain strings) results in no
1290
# performance improvement.
1292
revision_id = unescape_revid_cache[revision_id]
1294
unescaped = unescape(revision_id)
1295
unescape_revid_cache[revision_id] = unescaped
1296
revision_id = unescaped
1298
# Note that unconditionally unescaping means that we deserialise
1299
# every fileid, which for general 'pull' is not great, but we don't
1300
# really want to have some many fulltexts that this matters anyway.
1303
file_id = unescape_fileid_cache[file_id]
1305
unescaped = unescape(file_id)
1306
unescape_fileid_cache[file_id] = unescaped
1309
key = (file_id, revision_id)
1310
setdefault(key, False)
1311
if revision_id == line_key[-1]:
1315
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1317
"""Helper routine for fileids_altered_by_revision_ids.
1319
This performs the translation of xml lines to revision ids.
1321
:param line_iterator: An iterator of lines, origin_version_id
1322
:param revision_ids: The revision ids to filter for. This should be a
1323
set or other type which supports efficient __contains__ lookups, as
1324
the revision id from each parsed line will be looked up in the
1325
revision_ids filter.
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.
1331
setdefault = result.setdefault
1333
self._find_text_key_references_from_xml_inventory_lines(
1334
line_iterator).iterkeys():
1335
# once data is all ensured-consistent; then this is
1336
# if revision_id == version_id
1337
if key[-1:] in revision_ids:
1338
setdefault(key[0], set()).add(key[-1])
1341
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1342
"""Find the file ids and versions affected by revisions.
1344
:param revisions: an iterable containing revision ids.
1345
:param _inv_weave: The inventory weave from this repository or None.
1346
If None, the inventory weave will be opened automatically.
1347
:return: a dictionary mapping altered file-ids to an iterable of
1348
revision_ids. Each altered file-ids has the exact revision_ids that
1349
altered it listed explicitly.
1351
selected_keys = set((revid,) for revid in revision_ids)
1352
w = _inv_weave or self.inventories
1353
pb = ui.ui_factory.nested_progress_bar()
1355
return self._find_file_ids_from_xml_inventory_lines(
1356
w.iter_lines_added_or_present_in_keys(
1357
selected_keys, pb=pb),
1362
def iter_files_bytes(self, desired_files):
1363
"""Iterate through file versions.
1365
Files will not necessarily be returned in the order they occur in
1366
desired_files. No specific order is guaranteed.
1368
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1369
value supplied by the caller as part of desired_files. It should
1370
uniquely identify the file version in the caller's context. (Examples:
1371
an index number or a TreeTransform trans_id.)
1373
bytes_iterator is an iterable of bytestrings for the file. The
1374
kind of iterable and length of the bytestrings are unspecified, but for
1375
this implementation, it is a list of bytes produced by
1376
VersionedFile.get_record_stream().
1378
:param desired_files: a list of (file_id, revision_id, identifier)
1381
transaction = self.get_transaction()
1383
for file_id, revision_id, callable_data in desired_files:
1384
text_keys[(file_id, revision_id)] = callable_data
1385
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1386
if record.storage_kind == 'absent':
1387
raise errors.RevisionNotPresent(record.key, self)
1388
yield text_keys[record.key], record.get_bytes_as('fulltext')
1390
def _generate_text_key_index(self, text_key_references=None,
1392
"""Generate a new text key index for the repository.
1394
This is an expensive function that will take considerable time to run.
1396
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1397
list of parents, also text keys. When a given key has no parents,
1398
the parents list will be [NULL_REVISION].
1400
# All revisions, to find inventory parents.
1401
if ancestors is None:
1402
graph = self.get_graph()
1403
ancestors = graph.get_parent_map(self.all_revision_ids())
1404
if text_key_references is None:
1405
text_key_references = self.find_text_key_references()
1406
pb = ui.ui_factory.nested_progress_bar()
1408
return self._do_generate_text_key_index(ancestors,
1409
text_key_references, pb)
1413
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1414
"""Helper for _generate_text_key_index to avoid deep nesting."""
1415
revision_order = tsort.topo_sort(ancestors)
1416
invalid_keys = set()
1418
for revision_id in revision_order:
1419
revision_keys[revision_id] = set()
1420
text_count = len(text_key_references)
1421
# a cache of the text keys to allow reuse; costs a dict of all the
1422
# keys, but saves a 2-tuple for every child of a given key.
1424
for text_key, valid in text_key_references.iteritems():
1426
invalid_keys.add(text_key)
1428
revision_keys[text_key[1]].add(text_key)
1429
text_key_cache[text_key] = text_key
1430
del text_key_references
1432
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1433
NULL_REVISION = _mod_revision.NULL_REVISION
1434
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1435
# too small for large or very branchy trees. However, for 55K path
1436
# trees, it would be easy to use too much memory trivially. Ideally we
1437
# could gauge this by looking at available real memory etc, but this is
1438
# always a tricky proposition.
1439
inventory_cache = lru_cache.LRUCache(10)
1440
batch_size = 10 # should be ~150MB on a 55K path tree
1441
batch_count = len(revision_order) / batch_size + 1
1443
pb.update("Calculating text parents.", processed_texts, text_count)
1444
for offset in xrange(batch_count):
1445
to_query = revision_order[offset * batch_size:(offset + 1) *
1449
for rev_tree in self.revision_trees(to_query):
1450
revision_id = rev_tree.get_revision_id()
1451
parent_ids = ancestors[revision_id]
1452
for text_key in revision_keys[revision_id]:
1453
pb.update("Calculating text parents.", processed_texts)
1454
processed_texts += 1
1455
candidate_parents = []
1456
for parent_id in parent_ids:
1457
parent_text_key = (text_key[0], parent_id)
1459
check_parent = parent_text_key not in \
1460
revision_keys[parent_id]
1462
# the parent parent_id is a ghost:
1463
check_parent = False
1464
# truncate the derived graph against this ghost.
1465
parent_text_key = None
1467
# look at the parent commit details inventories to
1468
# determine possible candidates in the per file graph.
1471
inv = inventory_cache[parent_id]
1473
inv = self.revision_tree(parent_id).inventory
1474
inventory_cache[parent_id] = inv
1475
parent_entry = inv._byid.get(text_key[0], None)
1476
if parent_entry is not None:
1478
text_key[0], parent_entry.revision)
1480
parent_text_key = None
1481
if parent_text_key is not None:
1482
candidate_parents.append(
1483
text_key_cache[parent_text_key])
1484
parent_heads = text_graph.heads(candidate_parents)
1485
new_parents = list(parent_heads)
1486
new_parents.sort(key=lambda x:candidate_parents.index(x))
1487
if new_parents == []:
1488
new_parents = [NULL_REVISION]
1489
text_index[text_key] = new_parents
1491
for text_key in invalid_keys:
1492
text_index[text_key] = [NULL_REVISION]
1495
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1496
"""Get an iterable listing the keys of all the data introduced by a set
1499
The keys will be ordered so that the corresponding items can be safely
1500
fetched and inserted in that order.
1502
:returns: An iterable producing tuples of (knit-kind, file-id,
1503
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1504
'revisions'. file-id is None unless knit-kind is 'file'.
1506
# XXX: it's a bit weird to control the inventory weave caching in this
1507
# generator. Ideally the caching would be done in fetch.py I think. Or
1508
# maybe this generator should explicitly have the contract that it
1509
# should not be iterated until the previously yielded item has been
1511
inv_w = self.inventories
1513
# file ids that changed
1514
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1516
num_file_ids = len(file_ids)
1517
for file_id, altered_versions in file_ids.iteritems():
1518
if _files_pb is not None:
1519
_files_pb.update("fetch texts", count, num_file_ids)
1521
yield ("file", file_id, altered_versions)
1522
# We're done with the files_pb. Note that it finished by the caller,
1523
# just as it was created by the caller.
1527
yield ("inventory", None, revision_ids)
1530
revisions_with_signatures = set()
1531
for rev_id in revision_ids:
1533
self.get_signature_text(rev_id)
1534
except errors.NoSuchRevision:
1538
revisions_with_signatures.add(rev_id)
1539
yield ("signatures", None, revisions_with_signatures)
1542
yield ("revisions", None, revision_ids)
1545
def get_inventory(self, revision_id):
1546
"""Get Inventory object by revision id."""
1547
return self.iter_inventories([revision_id]).next()
1549
def iter_inventories(self, revision_ids):
1550
"""Get many inventories by revision_ids.
1552
This will buffer some or all of the texts used in constructing the
1553
inventories in memory, but will only parse a single inventory at a
1556
:return: An iterator of inventories.
1558
if ((None in revision_ids)
1559
or (_mod_revision.NULL_REVISION in revision_ids)):
1560
raise ValueError('cannot get null revision inventory')
1561
return self._iter_inventories(revision_ids)
1563
def _iter_inventories(self, revision_ids):
1564
"""single-document based inventory iteration."""
1565
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1566
yield self.deserialise_inventory(revision_id, text)
1568
def _iter_inventory_xmls(self, revision_ids):
1569
keys = [(revision_id,) for revision_id in revision_ids]
1570
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1572
for record in stream:
1573
if record.storage_kind != 'absent':
1574
texts[record.key] = record.get_bytes_as('fulltext')
1576
raise errors.NoSuchRevision(self, record.key)
1578
yield texts[key], key[-1]
1580
def deserialise_inventory(self, revision_id, xml):
1581
"""Transform the xml into an inventory object.
1583
:param revision_id: The expected revision id of the inventory.
1584
:param xml: A serialised inventory.
1586
result = self._serializer.read_inventory_from_string(xml, revision_id)
1587
if result.revision_id != revision_id:
1588
raise AssertionError('revision id mismatch %s != %s' % (
1589
result.revision_id, revision_id))
1592
def serialise_inventory(self, inv):
1593
return self._serializer.write_inventory_to_string(inv)
1595
def _serialise_inventory_to_lines(self, inv):
1596
return self._serializer.write_inventory_to_lines(inv)
1598
def get_serializer_format(self):
1599
return self._serializer.format_num
1602
def get_inventory_xml(self, revision_id):
1603
"""Get inventory XML as a file object."""
1604
texts = self._iter_inventory_xmls([revision_id])
1606
text, revision_id = texts.next()
1607
except StopIteration:
1608
raise errors.HistoryMissing(self, 'inventory', revision_id)
1612
def get_inventory_sha1(self, revision_id):
1613
"""Return the sha1 hash of the inventory entry
1615
return self.get_revision(revision_id).inventory_sha1
1617
def iter_reverse_revision_history(self, revision_id):
1618
"""Iterate backwards through revision ids in the lefthand history
1620
:param revision_id: The revision id to start with. All its lefthand
1621
ancestors will be traversed.
1623
graph = self.get_graph()
1624
next_id = revision_id
1626
if next_id in (None, _mod_revision.NULL_REVISION):
1629
# Note: The following line may raise KeyError in the event of
1630
# truncated history. We decided not to have a try:except:raise
1631
# RevisionNotPresent here until we see a use for it, because of the
1632
# cost in an inner loop that is by its very nature O(history).
1633
# Robert Collins 20080326
1634
parents = graph.get_parent_map([next_id])[next_id]
1635
if len(parents) == 0:
1638
next_id = parents[0]
1641
def get_revision_inventory(self, revision_id):
1642
"""Return inventory of a past revision."""
1643
# TODO: Unify this with get_inventory()
1644
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1645
# must be the same as its revision, so this is trivial.
1646
if revision_id is None:
1647
# This does not make sense: if there is no revision,
1648
# then it is the current tree inventory surely ?!
1649
# and thus get_root_id() is something that looks at the last
1650
# commit on the branch, and the get_root_id is an inventory check.
1651
raise NotImplementedError
1652
# return Inventory(self.get_root_id())
1654
return self.get_inventory(revision_id)
1656
def is_shared(self):
1657
"""Return True if this repository is flagged as a shared repository."""
1658
raise NotImplementedError(self.is_shared)
1661
def reconcile(self, other=None, thorough=False):
1662
"""Reconcile this repository."""
1663
from bzrlib.reconcile import RepoReconciler
1664
reconciler = RepoReconciler(self, thorough=thorough)
1665
reconciler.reconcile()
1668
def _refresh_data(self):
1669
"""Helper called from lock_* to ensure coherency with disk.
1671
The default implementation does nothing; it is however possible
1672
for repositories to maintain loaded indices across multiple locks
1673
by checking inside their implementation of this method to see
1674
whether their indices are still valid. This depends of course on
1675
the disk format being validatable in this manner.
1679
def revision_tree(self, revision_id):
1680
"""Return Tree for a revision on this branch.
1682
`revision_id` may be NULL_REVISION for the empty tree revision.
1684
revision_id = _mod_revision.ensure_null(revision_id)
1685
# TODO: refactor this to use an existing revision object
1686
# so we don't need to read it in twice.
1687
if revision_id == _mod_revision.NULL_REVISION:
1688
return RevisionTree(self, Inventory(root_id=None),
1689
_mod_revision.NULL_REVISION)
1691
inv = self.get_revision_inventory(revision_id)
1692
return RevisionTree(self, inv, revision_id)
1694
def revision_trees(self, revision_ids):
1695
"""Return Tree for a revision on this branch.
1697
`revision_id` may not be None or 'null:'"""
1698
inventories = self.iter_inventories(revision_ids)
1699
for inv in inventories:
1700
yield RevisionTree(self, inv, inv.revision_id)
1703
def get_ancestry(self, revision_id, topo_sorted=True):
1704
"""Return a list of revision-ids integrated by a revision.
1706
The first element of the list is always None, indicating the origin
1707
revision. This might change when we have history horizons, or
1708
perhaps we should have a new API.
1710
This is topologically sorted.
1712
if _mod_revision.is_null(revision_id):
1714
if not self.has_revision(revision_id):
1715
raise errors.NoSuchRevision(self, revision_id)
1716
graph = self.get_graph()
1718
search = graph._make_breadth_first_searcher([revision_id])
1721
found, ghosts = search.next_with_ghosts()
1722
except StopIteration:
1725
if _mod_revision.NULL_REVISION in keys:
1726
keys.remove(_mod_revision.NULL_REVISION)
1728
parent_map = graph.get_parent_map(keys)
1729
keys = tsort.topo_sort(parent_map)
1730
return [None] + list(keys)
1733
"""Compress the data within the repository.
1735
This operation only makes sense for some repository types. For other
1736
types it should be a no-op that just returns.
1738
This stub method does not require a lock, but subclasses should use
1739
@needs_write_lock as this is a long running call its reasonable to
1740
implicitly lock for the user.
1744
@deprecated_method(one_six)
1745
def print_file(self, file, revision_id):
1746
"""Print `file` to stdout.
1748
FIXME RBC 20060125 as John Meinel points out this is a bad api
1749
- it writes to stdout, it assumes that that is valid etc. Fix
1750
by creating a new more flexible convenience function.
1752
tree = self.revision_tree(revision_id)
1753
# use inventory as it was in that revision
1754
file_id = tree.inventory.path2id(file)
1756
# TODO: jam 20060427 Write a test for this code path
1757
# it had a bug in it, and was raising the wrong
1759
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1760
tree.print_file(file_id)
1762
def get_transaction(self):
1763
return self.control_files.get_transaction()
1765
@deprecated_method(one_one)
1766
def get_parents(self, revision_ids):
1767
"""See StackedParentsProvider.get_parents"""
1768
parent_map = self.get_parent_map(revision_ids)
1769
return [parent_map.get(r, None) for r in revision_ids]
1771
def get_parent_map(self, revision_ids):
1772
"""See graph._StackedParentsProvider.get_parent_map"""
1773
# revisions index works in keys; this just works in revisions
1774
# therefore wrap and unwrap
1777
for revision_id in revision_ids:
1778
if revision_id == _mod_revision.NULL_REVISION:
1779
result[revision_id] = ()
1780
elif revision_id is None:
1781
raise ValueError('get_parent_map(None) is not valid')
1783
query_keys.append((revision_id ,))
1784
for ((revision_id,), parent_keys) in \
1785
self.revisions.get_parent_map(query_keys).iteritems():
1787
result[revision_id] = tuple(parent_revid
1788
for (parent_revid,) in parent_keys)
1790
result[revision_id] = (_mod_revision.NULL_REVISION,)
1793
def _make_parents_provider(self):
1796
def get_graph(self, other_repository=None):
1797
"""Return the graph walker for this repository format"""
1798
parents_provider = self._make_parents_provider()
1799
if (other_repository is not None and
1800
not self.has_same_location(other_repository)):
1801
parents_provider = graph._StackedParentsProvider(
1802
[parents_provider, other_repository._make_parents_provider()])
1803
return graph.Graph(parents_provider)
1805
def _get_versioned_file_checker(self):
1806
"""Return an object suitable for checking versioned files."""
1807
return _VersionedFileChecker(self)
1809
def revision_ids_to_search_result(self, result_set):
1810
"""Convert a set of revision ids to a graph SearchResult."""
1811
result_parents = set()
1812
for parents in self.get_graph().get_parent_map(
1813
result_set).itervalues():
1814
result_parents.update(parents)
1815
included_keys = result_set.intersection(result_parents)
1816
start_keys = result_set.difference(included_keys)
1817
exclude_keys = result_parents.difference(result_set)
1818
result = graph.SearchResult(start_keys, exclude_keys,
1819
len(result_set), result_set)
1823
def set_make_working_trees(self, new_value):
1824
"""Set the policy flag for making working trees when creating branches.
1826
This only applies to branches that use this repository.
1828
The default is 'True'.
1829
:param new_value: True to restore the default, False to disable making
1832
raise NotImplementedError(self.set_make_working_trees)
1834
def make_working_trees(self):
1835
"""Returns the policy for making working trees on new branches."""
1836
raise NotImplementedError(self.make_working_trees)
1839
def sign_revision(self, revision_id, gpg_strategy):
1840
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1841
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1844
def has_signature_for_revision_id(self, revision_id):
1845
"""Query for a revision signature for revision_id in the repository."""
1846
if not self.has_revision(revision_id):
1847
raise errors.NoSuchRevision(self, revision_id)
1848
sig_present = (1 == len(
1849
self.signatures.get_parent_map([(revision_id,)])))
1853
def get_signature_text(self, revision_id):
1854
"""Return the text for a signature."""
1855
stream = self.signatures.get_record_stream([(revision_id,)],
1857
record = stream.next()
1858
if record.storage_kind == 'absent':
1859
raise errors.NoSuchRevision(self, revision_id)
1860
return record.get_bytes_as('fulltext')
1863
def check(self, revision_ids=None):
1864
"""Check consistency of all history of given revision_ids.
1866
Different repository implementations should override _check().
1868
:param revision_ids: A non-empty list of revision_ids whose ancestry
1869
will be checked. Typically the last revision_id of a branch.
1871
return self._check(revision_ids)
1873
def _check(self, revision_ids):
1874
result = check.Check(self)
1878
def _warn_if_deprecated(self):
1879
global _deprecation_warning_done
1880
if _deprecation_warning_done:
1882
_deprecation_warning_done = True
1883
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1884
% (self._format, self.bzrdir.transport.base))
1886
def supports_rich_root(self):
1887
return self._format.rich_root_data
1889
def _check_ascii_revisionid(self, revision_id, method):
1890
"""Private helper for ascii-only repositories."""
1891
# weave repositories refuse to store revisionids that are non-ascii.
1892
if revision_id is not None:
1893
# weaves require ascii revision ids.
1894
if isinstance(revision_id, unicode):
1896
revision_id.encode('ascii')
1897
except UnicodeEncodeError:
1898
raise errors.NonAsciiRevisionId(method, self)
1901
revision_id.decode('ascii')
1902
except UnicodeDecodeError:
1903
raise errors.NonAsciiRevisionId(method, self)
1905
def revision_graph_can_have_wrong_parents(self):
1906
"""Is it possible for this repository to have a revision graph with
1909
If True, then this repository must also implement
1910
_find_inconsistent_revision_parents so that check and reconcile can
1911
check for inconsistencies before proceeding with other checks that may
1912
depend on the revision index being consistent.
1914
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1917
# remove these delegates a while after bzr 0.15
1918
def __make_delegated(name, from_module):
1919
def _deprecated_repository_forwarder():
1920
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1921
% (name, from_module),
1924
m = __import__(from_module, globals(), locals(), [name])
1926
return getattr(m, name)
1927
except AttributeError:
1928
raise AttributeError('module %s has no name %s'
1930
globals()[name] = _deprecated_repository_forwarder
1933
'AllInOneRepository',
1934
'WeaveMetaDirRepository',
1935
'PreSplitOutRepositoryFormat',
1936
'RepositoryFormat4',
1937
'RepositoryFormat5',
1938
'RepositoryFormat6',
1939
'RepositoryFormat7',
1941
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1945
'RepositoryFormatKnit',
1946
'RepositoryFormatKnit1',
1948
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1951
def install_revision(repository, rev, revision_tree):
1952
"""Install all revision data into a repository."""
1953
install_revisions(repository, [(rev, revision_tree, None)])
1956
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1957
"""Install all revision data into a repository.
1959
Accepts an iterable of revision, tree, signature tuples. The signature
1962
repository.start_write_group()
1964
for n, (revision, revision_tree, signature) in enumerate(iterable):
1965
_install_revision(repository, revision, revision_tree, signature)
1967
pb.update('Transferring revisions', n + 1, num_revisions)
1969
repository.abort_write_group()
1972
repository.commit_write_group()
1975
def _install_revision(repository, rev, revision_tree, signature):
1976
"""Install all revision data into a repository."""
1977
present_parents = []
1979
for p_id in rev.parent_ids:
1980
if repository.has_revision(p_id):
1981
present_parents.append(p_id)
1982
parent_trees[p_id] = repository.revision_tree(p_id)
1984
parent_trees[p_id] = repository.revision_tree(
1985
_mod_revision.NULL_REVISION)
1987
inv = revision_tree.inventory
1988
entries = inv.iter_entries()
1989
# backwards compatibility hack: skip the root id.
1990
if not repository.supports_rich_root():
1991
path, root = entries.next()
1992
if root.revision != rev.revision_id:
1993
raise errors.IncompatibleRevision(repr(repository))
1995
for path, ie in entries:
1996
text_keys[(ie.file_id, ie.revision)] = ie
1997
text_parent_map = repository.texts.get_parent_map(text_keys)
1998
missing_texts = set(text_keys) - set(text_parent_map)
1999
# Add the texts that are not already present
2000
for text_key in missing_texts:
2001
ie = text_keys[text_key]
2003
# FIXME: TODO: The following loop overlaps/duplicates that done by
2004
# commit to determine parents. There is a latent/real bug here where
2005
# the parents inserted are not those commit would do - in particular
2006
# they are not filtered by heads(). RBC, AB
2007
for revision, tree in parent_trees.iteritems():
2008
if ie.file_id not in tree:
2010
parent_id = tree.inventory[ie.file_id].revision
2011
if parent_id in text_parents:
2013
text_parents.append((ie.file_id, parent_id))
2014
lines = revision_tree.get_file(ie.file_id).readlines()
2015
repository.texts.add_lines(text_key, text_parents, lines)
2017
# install the inventory
2018
repository.add_inventory(rev.revision_id, inv, present_parents)
2019
except errors.RevisionAlreadyPresent:
2021
if signature is not None:
2022
repository.add_signature_text(rev.revision_id, signature)
2023
repository.add_revision(rev.revision_id, rev, inv)
2026
class MetaDirRepository(Repository):
2027
"""Repositories in the new meta-dir layout.
2029
:ivar _transport: Transport for access to repository control files,
2030
typically pointing to .bzr/repository.
2033
def __init__(self, _format, a_bzrdir, control_files):
2034
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2035
self._transport = control_files._transport
2037
def is_shared(self):
2038
"""Return True if this repository is flagged as a shared repository."""
2039
return self._transport.has('shared-storage')
2042
def set_make_working_trees(self, new_value):
2043
"""Set the policy flag for making working trees when creating branches.
2045
This only applies to branches that use this repository.
2047
The default is 'True'.
2048
:param new_value: True to restore the default, False to disable making
2053
self._transport.delete('no-working-trees')
2054
except errors.NoSuchFile:
2057
self._transport.put_bytes('no-working-trees', '',
2058
mode=self.bzrdir._get_file_mode())
2060
def make_working_trees(self):
2061
"""Returns the policy for making working trees on new branches."""
2062
return not self._transport.has('no-working-trees')
2065
class MetaDirVersionedFileRepository(MetaDirRepository):
2066
"""Repositories in a meta-dir, that work via versioned file objects."""
2068
def __init__(self, _format, a_bzrdir, control_files):
2069
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2073
class RepositoryFormatRegistry(registry.Registry):
2074
"""Registry of RepositoryFormats."""
2076
def get(self, format_string):
2077
r = registry.Registry.get(self, format_string)
2083
format_registry = RepositoryFormatRegistry()
2084
"""Registry of formats, indexed by their identifying format string.
2086
This can contain either format instances themselves, or classes/factories that
2087
can be called to obtain one.
2091
#####################################################################
2092
# Repository Formats
2094
class RepositoryFormat(object):
2095
"""A repository format.
2097
Formats provide three things:
2098
* An initialization routine to construct repository data on disk.
2099
* a format string which is used when the BzrDir supports versioned
2101
* an open routine which returns a Repository instance.
2103
There is one and only one Format subclass for each on-disk format. But
2104
there can be one Repository subclass that is used for several different
2105
formats. The _format attribute on a Repository instance can be used to
2106
determine the disk format.
2108
Formats are placed in an dict by their format string for reference
2109
during opening. These should be subclasses of RepositoryFormat
2112
Once a format is deprecated, just deprecate the initialize and open
2113
methods on the format class. Do not deprecate the object, as the
2114
object will be created every system load.
2116
Common instance attributes:
2117
_matchingbzrdir - the bzrdir format that the repository format was
2118
originally written to work with. This can be used if manually
2119
constructing a bzrdir and repository, or more commonly for test suite
2123
# Set to True or False in derived classes. True indicates that the format
2124
# supports ghosts gracefully.
2125
supports_ghosts = None
2126
# Can this repository be given external locations to lookup additional
2127
# data. Set to True or False in derived classes.
2128
supports_external_lookups = None
2131
return "<%s>" % self.__class__.__name__
2133
def __eq__(self, other):
2134
# format objects are generally stateless
2135
return isinstance(other, self.__class__)
2137
def __ne__(self, other):
2138
return not self == other
2141
def find_format(klass, a_bzrdir):
2142
"""Return the format for the repository object in a_bzrdir.
2144
This is used by bzr native formats that have a "format" file in
2145
the repository. Other methods may be used by different types of
2149
transport = a_bzrdir.get_repository_transport(None)
2150
format_string = transport.get("format").read()
2151
return format_registry.get(format_string)
2152
except errors.NoSuchFile:
2153
raise errors.NoRepositoryPresent(a_bzrdir)
2155
raise errors.UnknownFormatError(format=format_string,
2159
def register_format(klass, format):
2160
format_registry.register(format.get_format_string(), format)
2163
def unregister_format(klass, format):
2164
format_registry.remove(format.get_format_string())
2167
def get_default_format(klass):
2168
"""Return the current default format."""
2169
from bzrlib import bzrdir
2170
return bzrdir.format_registry.make_bzrdir('default').repository_format
2172
def get_format_string(self):
2173
"""Return the ASCII format string that identifies this format.
2175
Note that in pre format ?? repositories the format string is
2176
not permitted nor written to disk.
2178
raise NotImplementedError(self.get_format_string)
2180
def get_format_description(self):
2181
"""Return the short description for this format."""
2182
raise NotImplementedError(self.get_format_description)
2184
# TODO: this shouldn't be in the base class, it's specific to things that
2185
# use weaves or knits -- mbp 20070207
2186
def _get_versioned_file_store(self,
2191
versionedfile_class=None,
2192
versionedfile_kwargs={},
2194
if versionedfile_class is None:
2195
versionedfile_class = self._versionedfile_class
2196
weave_transport = control_files._transport.clone(name)
2197
dir_mode = control_files._dir_mode
2198
file_mode = control_files._file_mode
2199
return VersionedFileStore(weave_transport, prefixed=prefixed,
2201
file_mode=file_mode,
2202
versionedfile_class=versionedfile_class,
2203
versionedfile_kwargs=versionedfile_kwargs,
2206
def initialize(self, a_bzrdir, shared=False):
2207
"""Initialize a repository of this format in a_bzrdir.
2209
:param a_bzrdir: The bzrdir to put the new repository in it.
2210
:param shared: The repository should be initialized as a sharable one.
2211
:returns: The new repository object.
2213
This may raise UninitializableFormat if shared repository are not
2214
compatible the a_bzrdir.
2216
raise NotImplementedError(self.initialize)
2218
def is_supported(self):
2219
"""Is this format supported?
2221
Supported formats must be initializable and openable.
2222
Unsupported formats may not support initialization or committing or
2223
some other features depending on the reason for not being supported.
2227
def check_conversion_target(self, target_format):
2228
raise NotImplementedError(self.check_conversion_target)
2230
def open(self, a_bzrdir, _found=False):
2231
"""Return an instance of this format for the bzrdir a_bzrdir.
2233
_found is a private parameter, do not use it.
2235
raise NotImplementedError(self.open)
2238
class MetaDirRepositoryFormat(RepositoryFormat):
2239
"""Common base class for the new repositories using the metadir layout."""
2241
rich_root_data = False
2242
supports_tree_reference = False
2243
supports_external_lookups = False
2244
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2247
super(MetaDirRepositoryFormat, self).__init__()
2249
def _create_control_files(self, a_bzrdir):
2250
"""Create the required files and the initial control_files object."""
2251
# FIXME: RBC 20060125 don't peek under the covers
2252
# NB: no need to escape relative paths that are url safe.
2253
repository_transport = a_bzrdir.get_repository_transport(self)
2254
control_files = lockable_files.LockableFiles(repository_transport,
2255
'lock', lockdir.LockDir)
2256
control_files.create_lock()
2257
return control_files
2259
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
2260
"""Upload the initial blank content."""
2261
control_files = self._create_control_files(a_bzrdir)
2262
control_files.lock_write()
2263
transport = control_files._transport
2265
utf8_files += [('shared-storage', '')]
2267
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2268
for (filename, content_stream) in files:
2269
transport.put_file(filename, content_stream,
2270
mode=a_bzrdir._get_file_mode())
2271
for (filename, content_bytes) in utf8_files:
2272
transport.put_bytes_non_atomic(filename, content_bytes,
2273
mode=a_bzrdir._get_file_mode())
2275
control_files.unlock()
2278
# formats which have no format string are not discoverable
2279
# and not independently creatable, so are not registered. They're
2280
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2281
# needed, it's constructed directly by the BzrDir. Non-native formats where
2282
# the repository is not separately opened are similar.
2284
format_registry.register_lazy(
2285
'Bazaar-NG Repository format 7',
2286
'bzrlib.repofmt.weaverepo',
2290
format_registry.register_lazy(
2291
'Bazaar-NG Knit Repository Format 1',
2292
'bzrlib.repofmt.knitrepo',
2293
'RepositoryFormatKnit1',
2296
format_registry.register_lazy(
2297
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2298
'bzrlib.repofmt.knitrepo',
2299
'RepositoryFormatKnit3',
2302
format_registry.register_lazy(
2303
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2304
'bzrlib.repofmt.knitrepo',
2305
'RepositoryFormatKnit4',
2308
# Pack-based formats. There is one format for pre-subtrees, and one for
2309
# post-subtrees to allow ease of testing.
2310
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2311
format_registry.register_lazy(
2312
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2313
'bzrlib.repofmt.pack_repo',
2314
'RepositoryFormatKnitPack1',
2316
format_registry.register_lazy(
2317
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2318
'bzrlib.repofmt.pack_repo',
2319
'RepositoryFormatKnitPack3',
2321
format_registry.register_lazy(
2322
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2323
'bzrlib.repofmt.pack_repo',
2324
'RepositoryFormatKnitPack4',
2326
format_registry.register_lazy(
2327
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2328
'bzrlib.repofmt.pack_repo',
2329
'RepositoryFormatKnitPack5',
2331
format_registry.register_lazy(
2332
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2333
'bzrlib.repofmt.pack_repo',
2334
'RepositoryFormatKnitPack5RichRoot',
2336
format_registry.register_lazy(
2337
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2338
'bzrlib.repofmt.pack_repo',
2339
'RepositoryFormatKnitPack5RichRootBroken',
2341
format_registry.register_lazy(
2342
'Bazaar RepositoryFormatKnitPack6 (bzr 1.9)\n',
2343
'bzrlib.repofmt.pack_repo',
2344
'RepositoryFormatKnitPack6',
2346
format_registry.register_lazy(
2347
'Bazaar RepositoryFormatKnitPack6RichRoot (bzr 1.9)\n',
2348
'bzrlib.repofmt.pack_repo',
2349
'RepositoryFormatKnitPack6RichRoot',
2352
# Development formats.
2353
# 1.7->1.8 go below here
2354
format_registry.register_lazy(
2355
"Bazaar development format 2 (needs bzr.dev from before 1.8)\n",
2356
'bzrlib.repofmt.pack_repo',
2357
'RepositoryFormatPackDevelopment2',
2359
format_registry.register_lazy(
2360
("Bazaar development format 2 with subtree support "
2361
"(needs bzr.dev from before 1.8)\n"),
2362
'bzrlib.repofmt.pack_repo',
2363
'RepositoryFormatPackDevelopment2Subtree',
2367
class InterRepository(InterObject):
2368
"""This class represents operations taking place between two repositories.
2370
Its instances have methods like copy_content and fetch, and contain
2371
references to the source and target repositories these operations can be
2374
Often we will provide convenience methods on 'repository' which carry out
2375
operations with another repository - they will always forward to
2376
InterRepository.get(other).method_name(parameters).
2379
_walk_to_common_revisions_batch_size = 1
2381
"""The available optimised InterRepository types."""
2383
def __init__(self, source, target):
2384
InterObject.__init__(self, source, target)
2385
# These two attributes may be overridden by e.g. InterOtherToRemote to
2386
# provide a faster implementation.
2387
self.target_get_graph = self.target.get_graph
2388
self.target_get_parent_map = self.target.get_parent_map
2390
def copy_content(self, revision_id=None):
2391
raise NotImplementedError(self.copy_content)
2393
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2394
"""Fetch the content required to construct revision_id.
2396
The content is copied from self.source to self.target.
2398
:param revision_id: if None all content is copied, if NULL_REVISION no
2400
:param pb: optional progress bar to use for progress reports. If not
2401
provided a default one will be created.
2403
:returns: (copied_revision_count, failures).
2405
# Normally we should find a specific InterRepository subclass to do
2406
# the fetch; if nothing else then at least InterSameDataRepository.
2407
# If none of them is suitable it looks like fetching is not possible;
2408
# we try to give a good message why. _assert_same_model will probably
2409
# give a helpful message; otherwise a generic one.
2410
self._assert_same_model(self.source, self.target)
2411
raise errors.IncompatibleRepositories(self.source, self.target,
2412
"no suitableInterRepository found")
2414
def _walk_to_common_revisions(self, revision_ids):
2415
"""Walk out from revision_ids in source to revisions target has.
2417
:param revision_ids: The start point for the search.
2418
:return: A set of revision ids.
2420
target_graph = self.target_get_graph()
2421
revision_ids = frozenset(revision_ids)
2422
# Fast path for the case where all the revisions are already in the
2424
# (Although this does incur an extra round trip for the
2425
# fairly common case where the target doesn't already have the revision
2427
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2428
return graph.SearchResult(revision_ids, set(), 0, set())
2429
missing_revs = set()
2430
source_graph = self.source.get_graph()
2431
# ensure we don't pay silly lookup costs.
2432
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2433
null_set = frozenset([_mod_revision.NULL_REVISION])
2434
searcher_exhausted = False
2438
# Iterate the searcher until we have enough next_revs
2439
while len(next_revs) < self._walk_to_common_revisions_batch_size:
2441
next_revs_part, ghosts_part = searcher.next_with_ghosts()
2442
next_revs.update(next_revs_part)
2443
ghosts.update(ghosts_part)
2444
except StopIteration:
2445
searcher_exhausted = True
2447
# If there are ghosts in the source graph, and the caller asked for
2448
# them, make sure that they are present in the target.
2449
# We don't care about other ghosts as we can't fetch them and
2450
# haven't been asked to.
2451
ghosts_to_check = set(revision_ids.intersection(ghosts))
2452
revs_to_get = set(next_revs).union(ghosts_to_check)
2454
have_revs = set(target_graph.get_parent_map(revs_to_get))
2455
# we always have NULL_REVISION present.
2456
have_revs = have_revs.union(null_set)
2457
# Check if the target is missing any ghosts we need.
2458
ghosts_to_check.difference_update(have_revs)
2460
# One of the caller's revision_ids is a ghost in both the
2461
# source and the target.
2462
raise errors.NoSuchRevision(
2463
self.source, ghosts_to_check.pop())
2464
missing_revs.update(next_revs - have_revs)
2465
# Because we may have walked past the original stop point, make
2466
# sure everything is stopped
2467
stop_revs = searcher.find_seen_ancestors(have_revs)
2468
searcher.stop_searching_any(stop_revs)
2469
if searcher_exhausted:
2471
return searcher.get_result()
2473
@deprecated_method(one_two)
2475
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2476
"""Return the revision ids that source has that target does not.
2478
These are returned in topological order.
2480
:param revision_id: only return revision ids included by this
2482
:param find_ghosts: If True find missing revisions in deep history
2483
rather than just finding the surface difference.
2485
return list(self.search_missing_revision_ids(
2486
revision_id, find_ghosts).get_keys())
2489
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2490
"""Return the revision ids that source has that target does not.
2492
:param revision_id: only return revision ids included by this
2494
:param find_ghosts: If True find missing revisions in deep history
2495
rather than just finding the surface difference.
2496
:return: A bzrlib.graph.SearchResult.
2498
# stop searching at found target revisions.
2499
if not find_ghosts and revision_id is not None:
2500
return self._walk_to_common_revisions([revision_id])
2501
# generic, possibly worst case, slow code path.
2502
target_ids = set(self.target.all_revision_ids())
2503
if revision_id is not None:
2504
source_ids = self.source.get_ancestry(revision_id)
2505
if source_ids[0] is not None:
2506
raise AssertionError()
2509
source_ids = self.source.all_revision_ids()
2510
result_set = set(source_ids).difference(target_ids)
2511
return self.source.revision_ids_to_search_result(result_set)
2514
def _same_model(source, target):
2515
"""True if source and target have the same data representation.
2517
Note: this is always called on the base class; overriding it in a
2518
subclass will have no effect.
2521
InterRepository._assert_same_model(source, target)
2523
except errors.IncompatibleRepositories, e:
2527
def _assert_same_model(source, target):
2528
"""Raise an exception if two repositories do not use the same model.
2530
if source.supports_rich_root() != target.supports_rich_root():
2531
raise errors.IncompatibleRepositories(source, target,
2532
"different rich-root support")
2533
if source._serializer != target._serializer:
2534
raise errors.IncompatibleRepositories(source, target,
2535
"different serializers")
2538
class InterSameDataRepository(InterRepository):
2539
"""Code for converting between repositories that represent the same data.
2541
Data format and model must match for this to work.
2545
def _get_repo_format_to_test(self):
2546
"""Repository format for testing with.
2548
InterSameData can pull from subtree to subtree and from non-subtree to
2549
non-subtree, so we test this with the richest repository format.
2551
from bzrlib.repofmt import knitrepo
2552
return knitrepo.RepositoryFormatKnit3()
2555
def is_compatible(source, target):
2556
return InterRepository._same_model(source, target)
2559
def copy_content(self, revision_id=None):
2560
"""Make a complete copy of the content in self into destination.
2562
This copies both the repository's revision data, and configuration information
2563
such as the make_working_trees setting.
2565
This is a destructive operation! Do not use it on existing
2568
:param revision_id: Only copy the content needed to construct
2569
revision_id and its parents.
2572
self.target.set_make_working_trees(self.source.make_working_trees())
2573
except NotImplementedError:
2575
# but don't bother fetching if we have the needed data now.
2576
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2577
self.target.has_revision(revision_id)):
2579
self.target.fetch(self.source, revision_id=revision_id)
2582
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2583
"""See InterRepository.fetch()."""
2584
from bzrlib.fetch import RepoFetcher
2585
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2586
self.source, self.source._format, self.target,
2587
self.target._format)
2588
f = RepoFetcher(to_repository=self.target,
2589
from_repository=self.source,
2590
last_revision=revision_id,
2591
pb=pb, find_ghosts=find_ghosts)
2592
return f.count_copied, f.failed_revisions
2595
class InterWeaveRepo(InterSameDataRepository):
2596
"""Optimised code paths between Weave based repositories.
2598
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2599
implemented lazy inter-object optimisation.
2603
def _get_repo_format_to_test(self):
2604
from bzrlib.repofmt import weaverepo
2605
return weaverepo.RepositoryFormat7()
2608
def is_compatible(source, target):
2609
"""Be compatible with known Weave formats.
2611
We don't test for the stores being of specific types because that
2612
could lead to confusing results, and there is no need to be
2615
from bzrlib.repofmt.weaverepo import (
2621
return (isinstance(source._format, (RepositoryFormat5,
2623
RepositoryFormat7)) and
2624
isinstance(target._format, (RepositoryFormat5,
2626
RepositoryFormat7)))
2627
except AttributeError:
2631
def copy_content(self, revision_id=None):
2632
"""See InterRepository.copy_content()."""
2633
# weave specific optimised path:
2635
self.target.set_make_working_trees(self.source.make_working_trees())
2636
except (errors.RepositoryUpgradeRequired, NotImplemented):
2638
# FIXME do not peek!
2639
if self.source._transport.listable():
2640
pb = ui.ui_factory.nested_progress_bar()
2642
self.target.texts.insert_record_stream(
2643
self.source.texts.get_record_stream(
2644
self.source.texts.keys(), 'topological', False))
2645
pb.update('copying inventory', 0, 1)
2646
self.target.inventories.insert_record_stream(
2647
self.source.inventories.get_record_stream(
2648
self.source.inventories.keys(), 'topological', False))
2649
self.target.signatures.insert_record_stream(
2650
self.source.signatures.get_record_stream(
2651
self.source.signatures.keys(),
2653
self.target.revisions.insert_record_stream(
2654
self.source.revisions.get_record_stream(
2655
self.source.revisions.keys(),
2656
'topological', True))
2660
self.target.fetch(self.source, revision_id=revision_id)
2663
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2664
"""See InterRepository.fetch()."""
2665
from bzrlib.fetch import RepoFetcher
2666
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2667
self.source, self.source._format, self.target, self.target._format)
2668
f = RepoFetcher(to_repository=self.target,
2669
from_repository=self.source,
2670
last_revision=revision_id,
2671
pb=pb, find_ghosts=find_ghosts)
2672
return f.count_copied, f.failed_revisions
2675
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2676
"""See InterRepository.missing_revision_ids()."""
2677
# we want all revisions to satisfy revision_id in source.
2678
# but we don't want to stat every file here and there.
2679
# we want then, all revisions other needs to satisfy revision_id
2680
# checked, but not those that we have locally.
2681
# so the first thing is to get a subset of the revisions to
2682
# satisfy revision_id in source, and then eliminate those that
2683
# we do already have.
2684
# this is slow on high latency connection to self, but as as this
2685
# disk format scales terribly for push anyway due to rewriting
2686
# inventory.weave, this is considered acceptable.
2688
if revision_id is not None:
2689
source_ids = self.source.get_ancestry(revision_id)
2690
if source_ids[0] is not None:
2691
raise AssertionError()
2694
source_ids = self.source._all_possible_ids()
2695
source_ids_set = set(source_ids)
2696
# source_ids is the worst possible case we may need to pull.
2697
# now we want to filter source_ids against what we actually
2698
# have in target, but don't try to check for existence where we know
2699
# we do not have a revision as that would be pointless.
2700
target_ids = set(self.target._all_possible_ids())
2701
possibly_present_revisions = target_ids.intersection(source_ids_set)
2702
actually_present_revisions = set(
2703
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2704
required_revisions = source_ids_set.difference(actually_present_revisions)
2705
if revision_id is not None:
2706
# we used get_ancestry to determine source_ids then we are assured all
2707
# revisions referenced are present as they are installed in topological order.
2708
# and the tip revision was validated by get_ancestry.
2709
result_set = required_revisions
2711
# if we just grabbed the possibly available ids, then
2712
# we only have an estimate of whats available and need to validate
2713
# that against the revision records.
2715
self.source._eliminate_revisions_not_present(required_revisions))
2716
return self.source.revision_ids_to_search_result(result_set)
2719
class InterKnitRepo(InterSameDataRepository):
2720
"""Optimised code paths between Knit based repositories."""
2723
def _get_repo_format_to_test(self):
2724
from bzrlib.repofmt import knitrepo
2725
return knitrepo.RepositoryFormatKnit1()
2728
def is_compatible(source, target):
2729
"""Be compatible with known Knit formats.
2731
We don't test for the stores being of specific types because that
2732
could lead to confusing results, and there is no need to be
2735
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2737
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2738
isinstance(target._format, RepositoryFormatKnit))
2739
except AttributeError:
2741
return are_knits and InterRepository._same_model(source, target)
2744
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2745
"""See InterRepository.fetch()."""
2746
from bzrlib.fetch import RepoFetcher
2747
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2748
self.source, self.source._format, self.target, self.target._format)
2749
f = RepoFetcher(to_repository=self.target,
2750
from_repository=self.source,
2751
last_revision=revision_id,
2752
pb=pb, find_ghosts=find_ghosts)
2753
return f.count_copied, f.failed_revisions
2756
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2757
"""See InterRepository.missing_revision_ids()."""
2758
if revision_id is not None:
2759
source_ids = self.source.get_ancestry(revision_id)
2760
if source_ids[0] is not None:
2761
raise AssertionError()
2764
source_ids = self.source.all_revision_ids()
2765
source_ids_set = set(source_ids)
2766
# source_ids is the worst possible case we may need to pull.
2767
# now we want to filter source_ids against what we actually
2768
# have in target, but don't try to check for existence where we know
2769
# we do not have a revision as that would be pointless.
2770
target_ids = set(self.target.all_revision_ids())
2771
possibly_present_revisions = target_ids.intersection(source_ids_set)
2772
actually_present_revisions = set(
2773
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2774
required_revisions = source_ids_set.difference(actually_present_revisions)
2775
if revision_id is not None:
2776
# we used get_ancestry to determine source_ids then we are assured all
2777
# revisions referenced are present as they are installed in topological order.
2778
# and the tip revision was validated by get_ancestry.
2779
result_set = required_revisions
2781
# if we just grabbed the possibly available ids, then
2782
# we only have an estimate of whats available and need to validate
2783
# that against the revision records.
2785
self.source._eliminate_revisions_not_present(required_revisions))
2786
return self.source.revision_ids_to_search_result(result_set)
2789
class InterPackRepo(InterSameDataRepository):
2790
"""Optimised code paths between Pack based repositories."""
2793
def _get_repo_format_to_test(self):
2794
from bzrlib.repofmt import pack_repo
2795
return pack_repo.RepositoryFormatKnitPack1()
2798
def is_compatible(source, target):
2799
"""Be compatible with known Pack formats.
2801
We don't test for the stores being of specific types because that
2802
could lead to confusing results, and there is no need to be
2805
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2807
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2808
isinstance(target._format, RepositoryFormatPack))
2809
except AttributeError:
2811
return are_packs and InterRepository._same_model(source, target)
2814
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2815
"""See InterRepository.fetch()."""
2816
if (len(self.source._fallback_repositories) > 0 or
2817
len(self.target._fallback_repositories) > 0):
2818
# The pack layer is not aware of fallback repositories, so when
2819
# fetching from a stacked repository or into a stacked repository
2820
# we use the generic fetch logic which uses the VersionedFiles
2821
# attributes on repository.
2823
# XXX: Andrew suggests removing the check on the target
2825
from bzrlib.fetch import RepoFetcher
2826
fetcher = RepoFetcher(self.target, self.source, revision_id,
2828
return fetcher.count_copied, fetcher.failed_revisions
2829
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2830
self.source, self.source._format, self.target, self.target._format)
2831
self.count_copied = 0
2832
if revision_id is None:
2834
# everything to do - use pack logic
2835
# to fetch from all packs to one without
2836
# inventory parsing etc, IFF nothing to be copied is in the target.
2838
source_revision_ids = frozenset(self.source.all_revision_ids())
2839
revision_ids = source_revision_ids - \
2840
frozenset(self.target_get_parent_map(source_revision_ids))
2841
revision_keys = [(revid,) for revid in revision_ids]
2842
target_pack_collection = self._get_target_pack_collection()
2843
index = target_pack_collection.revision_index.combined_index
2844
present_revision_ids = set(item[1][0] for item in
2845
index.iter_entries(revision_keys))
2846
revision_ids = set(revision_ids) - present_revision_ids
2847
# implementing the TODO will involve:
2848
# - detecting when all of a pack is selected
2849
# - avoiding as much as possible pre-selection, so the
2850
# more-core routines such as create_pack_from_packs can filter in
2851
# a just-in-time fashion. (though having a HEADS list on a
2852
# repository might make this a lot easier, because we could
2853
# sensibly detect 'new revisions' without doing a full index scan.
2854
elif _mod_revision.is_null(revision_id):
2859
revision_ids = self.search_missing_revision_ids(revision_id,
2860
find_ghosts=find_ghosts).get_keys()
2861
except errors.NoSuchRevision:
2862
raise errors.InstallFailed([revision_id])
2863
if len(revision_ids) == 0:
2865
return self._pack(self.source, self.target, revision_ids)
2867
def _pack(self, source, target, revision_ids):
2868
from bzrlib.repofmt.pack_repo import Packer
2869
target_pack_collection = self._get_target_pack_collection()
2870
packs = source._pack_collection.all_packs()
2871
pack = Packer(target_pack_collection, packs, '.fetch',
2872
revision_ids).pack()
2873
if pack is not None:
2874
target_pack_collection._save_pack_names()
2875
copied_revs = pack.get_revision_count()
2876
# Trigger an autopack. This may duplicate effort as we've just done
2877
# a pack creation, but for now it is simpler to think about as
2878
# 'upload data, then repack if needed'.
2880
return (copied_revs, [])
2884
def _autopack(self):
2885
self.target._pack_collection.autopack()
2887
def _get_target_pack_collection(self):
2888
return self.target._pack_collection
2891
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2892
"""See InterRepository.missing_revision_ids().
2894
:param find_ghosts: Find ghosts throughout the ancestry of
2897
if not find_ghosts and revision_id is not None:
2898
return self._walk_to_common_revisions([revision_id])
2899
elif revision_id is not None:
2900
# Find ghosts: search for revisions pointing from one repository to
2901
# the other, and vice versa, anywhere in the history of revision_id.
2902
graph = self.target_get_graph(other_repository=self.source)
2903
searcher = graph._make_breadth_first_searcher([revision_id])
2907
next_revs, ghosts = searcher.next_with_ghosts()
2908
except StopIteration:
2910
if revision_id in ghosts:
2911
raise errors.NoSuchRevision(self.source, revision_id)
2912
found_ids.update(next_revs)
2913
found_ids.update(ghosts)
2914
found_ids = frozenset(found_ids)
2915
# Double query here: should be able to avoid this by changing the
2916
# graph api further.
2917
result_set = found_ids - frozenset(
2918
self.target_get_parent_map(found_ids))
2920
source_ids = self.source.all_revision_ids()
2921
# source_ids is the worst possible case we may need to pull.
2922
# now we want to filter source_ids against what we actually
2923
# have in target, but don't try to check for existence where we know
2924
# we do not have a revision as that would be pointless.
2925
target_ids = set(self.target.all_revision_ids())
2926
result_set = set(source_ids).difference(target_ids)
2927
return self.source.revision_ids_to_search_result(result_set)
2930
class InterModel1and2(InterRepository):
2933
def _get_repo_format_to_test(self):
2937
def is_compatible(source, target):
2938
if not source.supports_rich_root() and target.supports_rich_root():
2944
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2945
"""See InterRepository.fetch()."""
2946
from bzrlib.fetch import Model1toKnit2Fetcher
2947
f = Model1toKnit2Fetcher(to_repository=self.target,
2948
from_repository=self.source,
2949
last_revision=revision_id,
2950
pb=pb, find_ghosts=find_ghosts)
2951
return f.count_copied, f.failed_revisions
2954
def copy_content(self, revision_id=None):
2955
"""Make a complete copy of the content in self into destination.
2957
This is a destructive operation! Do not use it on existing
2960
:param revision_id: Only copy the content needed to construct
2961
revision_id and its parents.
2964
self.target.set_make_working_trees(self.source.make_working_trees())
2965
except NotImplementedError:
2967
# but don't bother fetching if we have the needed data now.
2968
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2969
self.target.has_revision(revision_id)):
2971
self.target.fetch(self.source, revision_id=revision_id)
2974
class InterKnit1and2(InterKnitRepo):
2977
def _get_repo_format_to_test(self):
2981
def is_compatible(source, target):
2982
"""Be compatible with Knit1 source and Knit3 target"""
2984
from bzrlib.repofmt.knitrepo import (
2985
RepositoryFormatKnit1,
2986
RepositoryFormatKnit3,
2988
from bzrlib.repofmt.pack_repo import (
2989
RepositoryFormatKnitPack1,
2990
RepositoryFormatKnitPack3,
2991
RepositoryFormatKnitPack4,
2992
RepositoryFormatKnitPack5,
2993
RepositoryFormatKnitPack5RichRoot,
2994
RepositoryFormatKnitPack6,
2995
RepositoryFormatKnitPack6RichRoot,
2996
RepositoryFormatPackDevelopment2,
2997
RepositoryFormatPackDevelopment2Subtree,
3000
RepositoryFormatKnit1, # no rr, no subtree
3001
RepositoryFormatKnitPack1, # no rr, no subtree
3002
RepositoryFormatPackDevelopment2, # no rr, no subtree
3003
RepositoryFormatKnitPack5, # no rr, no subtree
3004
RepositoryFormatKnitPack6, # no rr, no subtree
3007
RepositoryFormatKnit3, # rr, subtree
3008
RepositoryFormatKnitPack3, # rr, subtree
3009
RepositoryFormatKnitPack4, # rr, no subtree
3010
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
3011
RepositoryFormatKnitPack6RichRoot,# rr, no subtree
3012
RepositoryFormatPackDevelopment2Subtree, # rr, subtree
3014
for format in norichroot:
3015
if format.rich_root_data:
3016
raise AssertionError('Format %s is a rich-root format'
3017
' but is included in the non-rich-root list'
3019
for format in richroot:
3020
if not format.rich_root_data:
3021
raise AssertionError('Format %s is not a rich-root format'
3022
' but is included in the rich-root list'
3024
# TODO: One alternative is to just check format.rich_root_data,
3025
# instead of keeping membership lists. However, the formats
3026
# *also* have to use the same 'Knit' style of storage
3027
# (line-deltas, fulltexts, etc.)
3028
return (isinstance(source._format, norichroot) and
3029
isinstance(target._format, richroot))
3030
except AttributeError:
3034
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3035
"""See InterRepository.fetch()."""
3036
from bzrlib.fetch import Knit1to2Fetcher
3037
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
3038
self.source, self.source._format, self.target,
3039
self.target._format)
3040
f = Knit1to2Fetcher(to_repository=self.target,
3041
from_repository=self.source,
3042
last_revision=revision_id,
3043
pb=pb, find_ghosts=find_ghosts)
3044
return f.count_copied, f.failed_revisions
3047
class InterDifferingSerializer(InterKnitRepo):
3050
def _get_repo_format_to_test(self):
3054
def is_compatible(source, target):
3055
"""Be compatible with Knit2 source and Knit3 target"""
3056
if source.supports_rich_root() != target.supports_rich_root():
3058
# Ideally, we'd support fetching if the source had no tree references
3059
# even if it supported them...
3060
if (getattr(source, '_format.supports_tree_reference', False) and
3061
not getattr(target, '_format.supports_tree_reference', False)):
3066
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3067
"""See InterRepository.fetch()."""
3068
revision_ids = self.target.search_missing_revision_ids(self.source,
3069
revision_id, find_ghosts=find_ghosts).get_keys()
3070
revision_ids = tsort.topo_sort(
3071
self.source.get_graph().get_parent_map(revision_ids))
3072
def revisions_iterator():
3073
for current_revision_id in revision_ids:
3074
revision = self.source.get_revision(current_revision_id)
3075
tree = self.source.revision_tree(current_revision_id)
3077
signature = self.source.get_signature_text(
3078
current_revision_id)
3079
except errors.NoSuchRevision:
3081
yield revision, tree, signature
3083
my_pb = ui.ui_factory.nested_progress_bar()
3088
install_revisions(self.target, revisions_iterator(),
3089
len(revision_ids), pb)
3091
if my_pb is not None:
3093
return len(revision_ids), 0
3096
class InterOtherToRemote(InterRepository):
3097
"""An InterRepository that simply delegates to the 'real' InterRepository
3098
calculated for (source, target._real_repository).
3101
_walk_to_common_revisions_batch_size = 50
3103
def __init__(self, source, target):
3104
InterRepository.__init__(self, source, target)
3105
self._real_inter = None
3108
def is_compatible(source, target):
3109
if isinstance(target, remote.RemoteRepository):
3113
def _ensure_real_inter(self):
3114
if self._real_inter is None:
3115
self.target._ensure_real()
3116
real_target = self.target._real_repository
3117
self._real_inter = InterRepository.get(self.source, real_target)
3118
# Make _real_inter use the RemoteRepository for get_parent_map
3119
self._real_inter.target_get_graph = self.target.get_graph
3120
self._real_inter.target_get_parent_map = self.target.get_parent_map
3122
def copy_content(self, revision_id=None):
3123
self._ensure_real_inter()
3124
self._real_inter.copy_content(revision_id=revision_id)
3126
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3127
self._ensure_real_inter()
3128
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3129
find_ghosts=find_ghosts)
3132
def _get_repo_format_to_test(self):
3136
class InterRemoteToOther(InterRepository):
3138
def __init__(self, source, target):
3139
InterRepository.__init__(self, source, target)
3140
self._real_inter = None
3143
def is_compatible(source, target):
3144
if not isinstance(source, remote.RemoteRepository):
3146
# Is source's model compatible with target's model?
3147
source._ensure_real()
3148
real_source = source._real_repository
3149
if isinstance(real_source, remote.RemoteRepository):
3150
raise NotImplementedError(
3151
"We don't support remote repos backed by remote repos yet.")
3152
return InterRepository._same_model(real_source, target)
3154
def _ensure_real_inter(self):
3155
if self._real_inter is None:
3156
self.source._ensure_real()
3157
real_source = self.source._real_repository
3158
self._real_inter = InterRepository.get(real_source, self.target)
3160
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3161
self._ensure_real_inter()
3162
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3163
find_ghosts=find_ghosts)
3165
def copy_content(self, revision_id=None):
3166
self._ensure_real_inter()
3167
self._real_inter.copy_content(revision_id=revision_id)
3170
def _get_repo_format_to_test(self):
3175
class InterPackToRemotePack(InterPackRepo):
3176
"""A specialisation of InterPackRepo for a target that is a
3179
This will use the get_parent_map RPC rather than plain readvs, and also
3180
uses an RPC for autopacking.
3183
_walk_to_common_revisions_batch_size = 50
3186
def is_compatible(source, target):
3187
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
3188
if isinstance(source._format, RepositoryFormatPack):
3189
if isinstance(target, remote.RemoteRepository):
3190
target._ensure_real()
3191
if isinstance(target._real_repository._format,
3192
RepositoryFormatPack):
3193
if InterRepository._same_model(source, target):
3197
def _autopack(self):
3198
self.target.autopack()
3200
def _get_target_pack_collection(self):
3201
return self.target._real_repository._pack_collection
3204
def _get_repo_format_to_test(self):
3208
InterRepository.register_optimiser(InterDifferingSerializer)
3209
InterRepository.register_optimiser(InterSameDataRepository)
3210
InterRepository.register_optimiser(InterWeaveRepo)
3211
InterRepository.register_optimiser(InterKnitRepo)
3212
InterRepository.register_optimiser(InterModel1and2)
3213
InterRepository.register_optimiser(InterKnit1and2)
3214
InterRepository.register_optimiser(InterPackRepo)
3215
InterRepository.register_optimiser(InterOtherToRemote)
3216
InterRepository.register_optimiser(InterRemoteToOther)
3217
InterRepository.register_optimiser(InterPackToRemotePack)
3220
class CopyConverter(object):
3221
"""A repository conversion tool which just performs a copy of the content.
3223
This is slow but quite reliable.
3226
def __init__(self, target_format):
3227
"""Create a CopyConverter.
3229
:param target_format: The format the resulting repository should be.
3231
self.target_format = target_format
3233
def convert(self, repo, pb):
3234
"""Perform the conversion of to_convert, giving feedback via pb.
3236
:param to_convert: The disk object to convert.
3237
:param pb: a progress bar to use for progress information.
3242
# this is only useful with metadir layouts - separated repo content.
3243
# trigger an assertion if not such
3244
repo._format.get_format_string()
3245
self.repo_dir = repo.bzrdir
3246
self.step('Moving repository to repository.backup')
3247
self.repo_dir.transport.move('repository', 'repository.backup')
3248
backup_transport = self.repo_dir.transport.clone('repository.backup')
3249
repo._format.check_conversion_target(self.target_format)
3250
self.source_repo = repo._format.open(self.repo_dir,
3252
_override_transport=backup_transport)
3253
self.step('Creating new repository')
3254
converted = self.target_format.initialize(self.repo_dir,
3255
self.source_repo.is_shared())
3256
converted.lock_write()
3258
self.step('Copying content into repository.')
3259
self.source_repo.copy_content_into(converted)
3262
self.step('Deleting old repository content.')
3263
self.repo_dir.transport.delete_tree('repository.backup')
3264
self.pb.note('repository converted')
3266
def step(self, message):
3267
"""Update the pb by a step."""
3269
self.pb.update(message, self.count, self.total)
3281
def _unescaper(match, _map=_unescape_map):
3282
code = match.group(1)
3286
if not code.startswith('#'):
3288
return unichr(int(code[1:])).encode('utf8')
3294
def _unescape_xml(data):
3295
"""Unescape predefined XML entities in a string of data."""
3297
if _unescape_re is None:
3298
_unescape_re = re.compile('\&([^;]*);')
3299
return _unescape_re.sub(_unescaper, data)
3302
class _VersionedFileChecker(object):
3304
def __init__(self, repository):
3305
self.repository = repository
3306
self.text_index = self.repository._generate_text_key_index()
3308
def calculate_file_version_parents(self, text_key):
3309
"""Calculate the correct parents for a file version according to
3312
parent_keys = self.text_index[text_key]
3313
if parent_keys == [_mod_revision.NULL_REVISION]:
3315
return tuple(parent_keys)
3317
def check_file_version_parents(self, texts, progress_bar=None):
3318
"""Check the parents stored in a versioned file are correct.
3320
It also detects file versions that are not referenced by their
3321
corresponding revision's inventory.
3323
:returns: A tuple of (wrong_parents, dangling_file_versions).
3324
wrong_parents is a dict mapping {revision_id: (stored_parents,
3325
correct_parents)} for each revision_id where the stored parents
3326
are not correct. dangling_file_versions is a set of (file_id,
3327
revision_id) tuples for versions that are present in this versioned
3328
file, but not used by the corresponding inventory.
3331
self.file_ids = set([file_id for file_id, _ in
3332
self.text_index.iterkeys()])
3333
# text keys is now grouped by file_id
3334
n_weaves = len(self.file_ids)
3335
files_in_revisions = {}
3336
revisions_of_files = {}
3337
n_versions = len(self.text_index)
3338
progress_bar.update('loading text store', 0, n_versions)
3339
parent_map = self.repository.texts.get_parent_map(self.text_index)
3340
# On unlistable transports this could well be empty/error...
3341
text_keys = self.repository.texts.keys()
3342
unused_keys = frozenset(text_keys) - set(self.text_index)
3343
for num, key in enumerate(self.text_index.iterkeys()):
3344
if progress_bar is not None:
3345
progress_bar.update('checking text graph', num, n_versions)
3346
correct_parents = self.calculate_file_version_parents(key)
3348
knit_parents = parent_map[key]
3349
except errors.RevisionNotPresent:
3352
if correct_parents != knit_parents:
3353
wrong_parents[key] = (knit_parents, correct_parents)
3354
return wrong_parents, unused_keys
3357
def _old_get_graph(repository, revision_id):
3358
"""DO NOT USE. That is all. I'm serious."""
3359
graph = repository.get_graph()
3360
revision_graph = dict(((key, value) for key, value in
3361
graph.iter_ancestry([revision_id]) if value is not None))
3362
return _strip_NULL_ghosts(revision_graph)
3365
def _strip_NULL_ghosts(revision_graph):
3366
"""Also don't use this. more compatibility code for unmigrated clients."""
3367
# Filter ghosts, and null:
3368
if _mod_revision.NULL_REVISION in revision_graph:
3369
del revision_graph[_mod_revision.NULL_REVISION]
3370
for key, parents in revision_graph.items():
3371
revision_graph[key] = tuple(parent for parent in parents if parent
3373
return revision_graph
added
removed
bzrlib/repository.py
