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 cStringIO import StringIO
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
39
revision as _mod_revision,
45
from bzrlib.bundle import serializer
46
from bzrlib.revisiontree import RevisionTree
47
from bzrlib.store.versioned import VersionedFileStore
48
from bzrlib.store.text import TextStore
49
from bzrlib.testament import Testament
50
from bzrlib.util import bencode
53
from bzrlib.decorators import needs_read_lock, needs_write_lock
54
from bzrlib.inter import InterObject
55
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
56
from bzrlib.symbol_versioning import (
63
from bzrlib.trace import mutter, mutter_callsite, note, warning
66
# Old formats display a warning, but only once
67
_deprecation_warning_done = False
70
class CommitBuilder(object):
71
"""Provides an interface to build up a commit.
73
This allows describing a tree to be committed without needing to
74
know the internals of the format of the repository.
77
# all clients should supply tree roots.
78
record_root_entry = True
79
# the default CommitBuilder does not manage trees whose root is versioned.
80
_versioned_root = False
82
def __init__(self, repository, parents, config, timestamp=None,
83
timezone=None, committer=None, revprops=None,
85
"""Initiate a CommitBuilder.
87
:param repository: Repository to commit to.
88
:param parents: Revision ids of the parents of the new revision.
89
:param config: Configuration to use.
90
:param timestamp: Optional timestamp recorded for commit.
91
:param timezone: Optional timezone for timestamp.
92
:param committer: Optional committer to set for commit.
93
:param revprops: Optional dictionary of revision properties.
94
:param revision_id: Optional revision id.
99
self._committer = self._config.username()
101
self._committer = committer
103
self.new_inventory = Inventory(None)
104
self._new_revision_id = revision_id
105
self.parents = parents
106
self.repository = repository
109
if revprops is not None:
110
self._revprops.update(revprops)
112
if timestamp is None:
113
timestamp = time.time()
114
# Restrict resolution to 1ms
115
self._timestamp = round(timestamp, 3)
118
self._timezone = osutils.local_time_offset()
120
self._timezone = int(timezone)
122
self._generate_revision_if_needed()
123
self.__heads = graph.HeadsCache(repository.get_graph()).heads
125
def commit(self, message):
126
"""Make the actual commit.
128
:return: The revision id of the recorded revision.
130
rev = _mod_revision.Revision(
131
timestamp=self._timestamp,
132
timezone=self._timezone,
133
committer=self._committer,
135
inventory_sha1=self.inv_sha1,
136
revision_id=self._new_revision_id,
137
properties=self._revprops)
138
rev.parent_ids = self.parents
139
self.repository.add_revision(self._new_revision_id, rev,
140
self.new_inventory, self._config)
141
self.repository.commit_write_group()
142
return self._new_revision_id
145
"""Abort the commit that is being built.
147
self.repository.abort_write_group()
149
def revision_tree(self):
150
"""Return the tree that was just committed.
152
After calling commit() this can be called to get a RevisionTree
153
representing the newly committed tree. This is preferred to
154
calling Repository.revision_tree() because that may require
155
deserializing the inventory, while we already have a copy in
158
return RevisionTree(self.repository, self.new_inventory,
159
self._new_revision_id)
161
def finish_inventory(self):
162
"""Tell the builder that the inventory is finished."""
163
if self.new_inventory.root is None:
164
raise AssertionError('Root entry should be supplied to'
165
' record_entry_contents, as of bzr 0.10.')
166
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
167
self.new_inventory.revision_id = self._new_revision_id
168
self.inv_sha1 = self.repository.add_inventory(
169
self._new_revision_id,
174
def _gen_revision_id(self):
175
"""Return new revision-id."""
176
return generate_ids.gen_revision_id(self._config.username(),
179
def _generate_revision_if_needed(self):
180
"""Create a revision id if None was supplied.
182
If the repository can not support user-specified revision ids
183
they should override this function and raise CannotSetRevisionId
184
if _new_revision_id is not None.
186
:raises: CannotSetRevisionId
188
if self._new_revision_id is None:
189
self._new_revision_id = self._gen_revision_id()
190
self.random_revid = True
192
self.random_revid = False
194
def _heads(self, file_id, revision_ids):
195
"""Calculate the graph heads for revision_ids in the graph of file_id.
197
This can use either a per-file graph or a global revision graph as we
198
have an identity relationship between the two graphs.
200
return self.__heads(revision_ids)
202
def _check_root(self, ie, parent_invs, tree):
203
"""Helper for record_entry_contents.
205
:param ie: An entry being added.
206
:param parent_invs: The inventories of the parent revisions of the
208
:param tree: The tree that is being committed.
210
# In this revision format, root entries have no knit or weave When
211
# serializing out to disk and back in root.revision is always
213
ie.revision = self._new_revision_id
215
def _get_delta(self, ie, basis_inv, path):
216
"""Get a delta against the basis inventory for ie."""
217
if ie.file_id not in basis_inv:
219
return (None, path, ie.file_id, ie)
220
elif ie != basis_inv[ie.file_id]:
222
# TODO: avoid tis id2path call.
223
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
228
def record_entry_contents(self, ie, parent_invs, path, tree,
230
"""Record the content of ie from tree into the commit if needed.
232
Side effect: sets ie.revision when unchanged
234
:param ie: An inventory entry present in the commit.
235
:param parent_invs: The inventories of the parent revisions of the
237
:param path: The path the entry is at in the tree.
238
:param tree: The tree which contains this entry and should be used to
240
:param content_summary: Summary data from the tree about the paths
241
content - stat, length, exec, sha/link target. This is only
242
accessed when the entry has a revision of None - that is when it is
243
a candidate to commit.
244
:return: A tuple (change_delta, version_recorded). change_delta is
245
an inventory_delta change for this entry against the basis tree of
246
the commit, or None if no change occured against the basis tree.
247
version_recorded is True if a new version of the entry has been
248
recorded. For instance, committing a merge where a file was only
249
changed on the other side will return (delta, False).
251
if self.new_inventory.root is None:
252
if ie.parent_id is not None:
253
raise errors.RootMissing()
254
self._check_root(ie, parent_invs, tree)
255
if ie.revision is None:
256
kind = content_summary[0]
258
# ie is carried over from a prior commit
260
# XXX: repository specific check for nested tree support goes here - if
261
# the repo doesn't want nested trees we skip it ?
262
if (kind == 'tree-reference' and
263
not self.repository._format.supports_tree_reference):
264
# mismatch between commit builder logic and repository:
265
# this needs the entry creation pushed down into the builder.
266
raise NotImplementedError('Missing repository subtree support.')
267
self.new_inventory.add(ie)
269
# TODO: slow, take it out of the inner loop.
271
basis_inv = parent_invs[0]
273
basis_inv = Inventory(root_id=None)
275
# ie.revision is always None if the InventoryEntry is considered
276
# for committing. We may record the previous parents revision if the
277
# content is actually unchanged against a sole head.
278
if ie.revision is not None:
279
if not self._versioned_root and path == '':
280
# repositories that do not version the root set the root's
281
# revision to the new commit even when no change occurs, and
282
# this masks when a change may have occurred against the basis,
283
# so calculate if one happened.
284
if ie.file_id in basis_inv:
285
delta = (basis_inv.id2path(ie.file_id), path,
289
delta = (None, path, ie.file_id, ie)
292
# we don't need to commit this, because the caller already
293
# determined that an existing revision of this file is
295
return None, (ie.revision == self._new_revision_id)
296
# XXX: Friction: parent_candidates should return a list not a dict
297
# so that we don't have to walk the inventories again.
298
parent_candiate_entries = ie.parent_candidates(parent_invs)
299
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
301
for inv in parent_invs:
302
if ie.file_id in inv:
303
old_rev = inv[ie.file_id].revision
304
if old_rev in head_set:
305
heads.append(inv[ie.file_id].revision)
306
head_set.remove(inv[ie.file_id].revision)
309
# now we check to see if we need to write a new record to the
311
# We write a new entry unless there is one head to the ancestors, and
312
# the kind-derived content is unchanged.
314
# Cheapest check first: no ancestors, or more the one head in the
315
# ancestors, we write a new node.
319
# There is a single head, look it up for comparison
320
parent_entry = parent_candiate_entries[heads[0]]
321
# if the non-content specific data has changed, we'll be writing a
323
if (parent_entry.parent_id != ie.parent_id or
324
parent_entry.name != ie.name):
326
# now we need to do content specific checks:
328
# if the kind changed the content obviously has
329
if kind != parent_entry.kind:
332
if content_summary[2] is None:
333
raise ValueError("Files must not have executable = None")
335
if (# if the file length changed we have to store:
336
parent_entry.text_size != content_summary[1] or
337
# if the exec bit has changed we have to store:
338
parent_entry.executable != content_summary[2]):
340
elif parent_entry.text_sha1 == content_summary[3]:
341
# all meta and content is unchanged (using a hash cache
342
# hit to check the sha)
343
ie.revision = parent_entry.revision
344
ie.text_size = parent_entry.text_size
345
ie.text_sha1 = parent_entry.text_sha1
346
ie.executable = parent_entry.executable
347
return self._get_delta(ie, basis_inv, path), False
349
# Either there is only a hash change(no hash cache entry,
350
# or same size content change), or there is no change on
352
# Provide the parent's hash to the store layer, so that the
353
# content is unchanged we will not store a new node.
354
nostore_sha = parent_entry.text_sha1
356
# We want to record a new node regardless of the presence or
357
# absence of a content change in the file.
359
ie.executable = content_summary[2]
360
lines = tree.get_file(ie.file_id, path).readlines()
362
ie.text_sha1, ie.text_size = self._add_text_to_weave(
363
ie.file_id, lines, heads, nostore_sha)
364
except errors.ExistingContent:
365
# Turns out that the file content was unchanged, and we were
366
# only going to store a new node if it was changed. Carry over
368
ie.revision = parent_entry.revision
369
ie.text_size = parent_entry.text_size
370
ie.text_sha1 = parent_entry.text_sha1
371
ie.executable = parent_entry.executable
372
return self._get_delta(ie, basis_inv, path), False
373
elif kind == 'directory':
375
# all data is meta here, nothing specific to directory, so
377
ie.revision = parent_entry.revision
378
return self._get_delta(ie, basis_inv, path), False
380
self._add_text_to_weave(ie.file_id, lines, heads, None)
381
elif kind == 'symlink':
382
current_link_target = content_summary[3]
384
# symlink target is not generic metadata, check if it has
386
if current_link_target != parent_entry.symlink_target:
389
# unchanged, carry over.
390
ie.revision = parent_entry.revision
391
ie.symlink_target = parent_entry.symlink_target
392
return self._get_delta(ie, basis_inv, path), False
393
ie.symlink_target = current_link_target
395
self._add_text_to_weave(ie.file_id, lines, heads, None)
396
elif kind == 'tree-reference':
398
if content_summary[3] != parent_entry.reference_revision:
401
# unchanged, carry over.
402
ie.reference_revision = parent_entry.reference_revision
403
ie.revision = parent_entry.revision
404
return self._get_delta(ie, basis_inv, path), False
405
ie.reference_revision = content_summary[3]
407
self._add_text_to_weave(ie.file_id, lines, heads, None)
409
raise NotImplementedError('unknown kind')
410
ie.revision = self._new_revision_id
411
return self._get_delta(ie, basis_inv, path), True
413
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
414
# Note: as we read the content directly from the tree, we know its not
415
# been turned into unicode or badly split - but a broken tree
416
# implementation could give us bad output from readlines() so this is
417
# not a guarantee of safety. What would be better is always checking
418
# the content during test suite execution. RBC 20070912
419
parent_keys = tuple((file_id, parent) for parent in parents)
420
return self.repository.texts.add_lines(
421
(file_id, self._new_revision_id), parent_keys, new_lines,
422
nostore_sha=nostore_sha, random_id=self.random_revid,
423
check_content=False)[0:2]
426
class RootCommitBuilder(CommitBuilder):
427
"""This commitbuilder actually records the root id"""
429
# the root entry gets versioned properly by this builder.
430
_versioned_root = True
432
def _check_root(self, ie, parent_invs, tree):
433
"""Helper for record_entry_contents.
435
:param ie: An entry being added.
436
:param parent_invs: The inventories of the parent revisions of the
438
:param tree: The tree that is being committed.
442
######################################################################
445
class Repository(object):
446
"""Repository holding history for one or more branches.
448
The repository holds and retrieves historical information including
449
revisions and file history. It's normally accessed only by the Branch,
450
which views a particular line of development through that history.
452
The Repository builds on top of some byte storage facilies (the revisions,
453
signatures, inventories and texts attributes) and a Transport, which
454
respectively provide byte storage and a means to access the (possibly
457
The byte storage facilities are addressed via tuples, which we refer to
458
as 'keys' throughout the code base. Revision_keys, inventory_keys and
459
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
460
(file_id, revision_id). We use this interface because it allows low
461
friction with the underlying code that implements disk indices, network
462
encoding and other parts of bzrlib.
464
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
465
the serialised revisions for the repository. This can be used to obtain
466
revision graph information or to access raw serialised revisions.
467
The result of trying to insert data into the repository via this store
468
is undefined: it should be considered read-only except for implementors
470
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
471
the serialised signatures for the repository. This can be used to
472
obtain access to raw serialised signatures. The result of trying to
473
insert data into the repository via this store is undefined: it should
474
be considered read-only except for implementors of repositories.
475
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
476
the serialised inventories for the repository. This can be used to
477
obtain unserialised inventories. The result of trying to insert data
478
into the repository via this store is undefined: it should be
479
considered read-only except for implementors of repositories.
480
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
481
texts of files and directories for the repository. This can be used to
482
obtain file texts or file graphs. Note that Repository.iter_file_bytes
483
is usually a better interface for accessing file texts.
484
The result of trying to insert data into the repository via this store
485
is undefined: it should be considered read-only except for implementors
487
:ivar _transport: Transport for file access to repository, typically
488
pointing to .bzr/repository.
491
# What class to use for a CommitBuilder. Often its simpler to change this
492
# in a Repository class subclass rather than to override
493
# get_commit_builder.
494
_commit_builder_class = CommitBuilder
495
# The search regex used by xml based repositories to determine what things
496
# where changed in a single commit.
497
_file_ids_altered_regex = lazy_regex.lazy_compile(
498
r'file_id="(?P<file_id>[^"]+)"'
499
r'.* revision="(?P<revision_id>[^"]+)"'
502
def abort_write_group(self):
503
"""Commit the contents accrued within the current write group.
505
:seealso: start_write_group.
507
if self._write_group is not self.get_transaction():
508
# has an unlock or relock occured ?
509
raise errors.BzrError('mismatched lock context and write group.')
510
self._abort_write_group()
511
self._write_group = None
513
def _abort_write_group(self):
514
"""Template method for per-repository write group cleanup.
516
This is called during abort before the write group is considered to be
517
finished and should cleanup any internal state accrued during the write
518
group. There is no requirement that data handed to the repository be
519
*not* made available - this is not a rollback - but neither should any
520
attempt be made to ensure that data added is fully commited. Abort is
521
invoked when an error has occured so futher disk or network operations
522
may not be possible or may error and if possible should not be
526
def add_fallback_repository(self, repository):
527
"""Add a repository to use for looking up data not held locally.
529
:param repository: A repository.
531
if not self._format.supports_external_lookups:
532
raise errors.UnstackableRepositoryFormat(self._format, self.base)
533
if not self._add_fallback_repository_check(repository):
534
raise errors.IncompatibleRepositories(self, repository)
535
self._fallback_repositories.append(repository)
536
self.texts.add_fallback_versioned_files(repository.texts)
537
self.inventories.add_fallback_versioned_files(repository.inventories)
538
self.revisions.add_fallback_versioned_files(repository.revisions)
539
self.signatures.add_fallback_versioned_files(repository.signatures)
541
def _add_fallback_repository_check(self, repository):
542
"""Check that this repository can fallback to repository safely.
544
:param repository: A repository to fallback to.
545
:return: True if the repositories can stack ok.
547
return InterRepository._same_model(self, repository)
549
def add_inventory(self, revision_id, inv, parents):
550
"""Add the inventory inv to the repository as revision_id.
552
:param parents: The revision ids of the parents that revision_id
553
is known to have and are in the repository already.
555
:returns: The validator(which is a sha1 digest, though what is sha'd is
556
repository format specific) of the serialized inventory.
558
if not self.is_in_write_group():
559
raise AssertionError("%r not in write group" % (self,))
560
_mod_revision.check_not_reserved_id(revision_id)
561
if not (inv.revision_id is None or inv.revision_id == revision_id):
562
raise AssertionError(
563
"Mismatch between inventory revision"
564
" id and insertion revid (%r, %r)"
565
% (inv.revision_id, revision_id))
567
raise AssertionError()
568
inv_lines = self._serialise_inventory_to_lines(inv)
569
return self._inventory_add_lines(revision_id, parents,
570
inv_lines, check_content=False)
572
def _inventory_add_lines(self, revision_id, parents, lines,
574
"""Store lines in inv_vf and return the sha1 of the inventory."""
575
parents = [(parent,) for parent in parents]
576
return self.inventories.add_lines((revision_id,), parents, lines,
577
check_content=check_content)[0]
579
def add_revision(self, revision_id, rev, inv=None, config=None):
580
"""Add rev to the revision store as revision_id.
582
:param revision_id: the revision id to use.
583
:param rev: The revision object.
584
:param inv: The inventory for the revision. if None, it will be looked
585
up in the inventory storer
586
:param config: If None no digital signature will be created.
587
If supplied its signature_needed method will be used
588
to determine if a signature should be made.
590
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
592
_mod_revision.check_not_reserved_id(revision_id)
593
if config is not None and config.signature_needed():
595
inv = self.get_inventory(revision_id)
596
plaintext = Testament(rev, inv).as_short_text()
597
self.store_revision_signature(
598
gpg.GPGStrategy(config), plaintext, revision_id)
599
# check inventory present
600
if not self.inventories.get_parent_map([(revision_id,)]):
602
raise errors.WeaveRevisionNotPresent(revision_id,
605
# yes, this is not suitable for adding with ghosts.
606
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
610
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
611
self._add_revision(rev)
613
def _add_revision(self, revision):
614
text = self._serializer.write_revision_to_string(revision)
615
key = (revision.revision_id,)
616
parents = tuple((parent,) for parent in revision.parent_ids)
617
self.revisions.add_lines(key, parents, osutils.split_lines(text))
619
def all_revision_ids(self):
620
"""Returns a list of all the revision ids in the repository.
622
This is conceptually deprecated because code should generally work on
623
the graph reachable from a particular revision, and ignore any other
624
revisions that might be present. There is no direct replacement
627
if 'evil' in debug.debug_flags:
628
mutter_callsite(2, "all_revision_ids is linear with history.")
629
return self._all_revision_ids()
631
def _all_revision_ids(self):
632
"""Returns a list of all the revision ids in the repository.
634
These are in as much topological order as the underlying store can
637
raise NotImplementedError(self._all_revision_ids)
639
def break_lock(self):
640
"""Break a lock if one is present from another instance.
642
Uses the ui factory to ask for confirmation if the lock may be from
645
self.control_files.break_lock()
648
def _eliminate_revisions_not_present(self, revision_ids):
649
"""Check every revision id in revision_ids to see if we have it.
651
Returns a set of the present revisions.
654
graph = self.get_graph()
655
parent_map = graph.get_parent_map(revision_ids)
656
# The old API returned a list, should this actually be a set?
657
return parent_map.keys()
660
def create(a_bzrdir):
661
"""Construct the current default format repository in a_bzrdir."""
662
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
664
def __init__(self, _format, a_bzrdir, control_files):
665
"""instantiate a Repository.
667
:param _format: The format of the repository on disk.
668
:param a_bzrdir: The BzrDir of the repository.
670
In the future we will have a single api for all stores for
671
getting file texts, inventories and revisions, then
672
this construct will accept instances of those things.
674
super(Repository, self).__init__()
675
self._format = _format
676
# the following are part of the public API for Repository:
677
self.bzrdir = a_bzrdir
678
self.control_files = control_files
679
self._transport = control_files._transport
680
self.base = self._transport.base
682
self._reconcile_does_inventory_gc = True
683
self._reconcile_fixes_text_parents = False
684
self._reconcile_backsup_inventory = True
685
# not right yet - should be more semantically clear ?
687
# TODO: make sure to construct the right store classes, etc, depending
688
# on whether escaping is required.
689
self._warn_if_deprecated()
690
self._write_group = None
691
# Additional places to query for data.
692
self._fallback_repositories = []
695
return '%s(%r)' % (self.__class__.__name__,
698
def has_same_location(self, other):
699
"""Returns a boolean indicating if this repository is at the same
700
location as another repository.
702
This might return False even when two repository objects are accessing
703
the same physical repository via different URLs.
705
if self.__class__ is not other.__class__:
707
return (self._transport.base == other._transport.base)
709
def is_in_write_group(self):
710
"""Return True if there is an open write group.
712
:seealso: start_write_group.
714
return self._write_group is not None
717
return self.control_files.is_locked()
719
def is_write_locked(self):
720
"""Return True if this object is write locked."""
721
return self.is_locked() and self.control_files._lock_mode == 'w'
723
def lock_write(self, token=None):
724
"""Lock this repository for writing.
726
This causes caching within the repository obejct to start accumlating
727
data during reads, and allows a 'write_group' to be obtained. Write
728
groups must be used for actual data insertion.
730
:param token: if this is already locked, then lock_write will fail
731
unless the token matches the existing lock.
732
:returns: a token if this instance supports tokens, otherwise None.
733
:raises TokenLockingNotSupported: when a token is given but this
734
instance doesn't support using token locks.
735
:raises MismatchedToken: if the specified token doesn't match the token
736
of the existing lock.
737
:seealso: start_write_group.
739
A token should be passed in if you know that you have locked the object
740
some other way, and need to synchronise this object's state with that
743
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
745
result = self.control_files.lock_write(token=token)
746
for repo in self._fallback_repositories:
747
# Writes don't affect fallback repos
753
self.control_files.lock_read()
754
for repo in self._fallback_repositories:
758
def get_physical_lock_status(self):
759
return self.control_files.get_physical_lock_status()
761
def leave_lock_in_place(self):
762
"""Tell this repository not to release the physical lock when this
765
If lock_write doesn't return a token, then this method is not supported.
767
self.control_files.leave_in_place()
769
def dont_leave_lock_in_place(self):
770
"""Tell this repository to release the physical lock when this
771
object is unlocked, even if it didn't originally acquire it.
773
If lock_write doesn't return a token, then this method is not supported.
775
self.control_files.dont_leave_in_place()
778
def gather_stats(self, revid=None, committers=None):
779
"""Gather statistics from a revision id.
781
:param revid: The revision id to gather statistics from, if None, then
782
no revision specific statistics are gathered.
783
:param committers: Optional parameter controlling whether to grab
784
a count of committers from the revision specific statistics.
785
:return: A dictionary of statistics. Currently this contains:
786
committers: The number of committers if requested.
787
firstrev: A tuple with timestamp, timezone for the penultimate left
788
most ancestor of revid, if revid is not the NULL_REVISION.
789
latestrev: A tuple with timestamp, timezone for revid, if revid is
790
not the NULL_REVISION.
791
revisions: The total revision count in the repository.
792
size: An estimate disk size of the repository in bytes.
795
if revid and committers:
796
result['committers'] = 0
797
if revid and revid != _mod_revision.NULL_REVISION:
799
all_committers = set()
800
revisions = self.get_ancestry(revid)
801
# pop the leading None
803
first_revision = None
805
# ignore the revisions in the middle - just grab first and last
806
revisions = revisions[0], revisions[-1]
807
for revision in self.get_revisions(revisions):
808
if not first_revision:
809
first_revision = revision
811
all_committers.add(revision.committer)
812
last_revision = revision
814
result['committers'] = len(all_committers)
815
result['firstrev'] = (first_revision.timestamp,
816
first_revision.timezone)
817
result['latestrev'] = (last_revision.timestamp,
818
last_revision.timezone)
820
# now gather global repository information
821
# XXX: This is available for many repos regardless of listability.
822
if self.bzrdir.root_transport.listable():
823
# XXX: do we want to __define len__() ?
824
# Maybe the versionedfiles object should provide a different
825
# method to get the number of keys.
826
result['revisions'] = len(self.revisions.keys())
830
def find_branches(self, using=False):
831
"""Find branches underneath this repository.
833
This will include branches inside other branches.
835
:param using: If True, list only branches using this repository.
837
if using and not self.is_shared():
839
return [self.bzrdir.open_branch()]
840
except errors.NotBranchError:
842
class Evaluator(object):
845
self.first_call = True
847
def __call__(self, bzrdir):
848
# On the first call, the parameter is always the bzrdir
849
# containing the current repo.
850
if not self.first_call:
852
repository = bzrdir.open_repository()
853
except errors.NoRepositoryPresent:
856
return False, (None, repository)
857
self.first_call = False
859
value = (bzrdir.open_branch(), None)
860
except errors.NotBranchError:
865
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
866
self.bzrdir.root_transport, evaluate=Evaluator()):
867
if branch is not None:
868
branches.append(branch)
869
if not using and repository is not None:
870
branches.extend(repository.find_branches())
874
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
875
"""Return the revision ids that other has that this does not.
877
These are returned in topological order.
879
revision_id: only return revision ids included by revision_id.
881
return InterRepository.get(other, self).search_missing_revision_ids(
882
revision_id, find_ghosts)
884
@deprecated_method(one_two)
886
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
887
"""Return the revision ids that other has that this does not.
889
These are returned in topological order.
891
revision_id: only return revision ids included by revision_id.
893
keys = self.search_missing_revision_ids(
894
other, revision_id, find_ghosts).get_keys()
897
parents = other.get_graph().get_parent_map(keys)
900
return tsort.topo_sort(parents)
904
"""Open the repository rooted at base.
906
For instance, if the repository is at URL/.bzr/repository,
907
Repository.open(URL) -> a Repository instance.
909
control = bzrdir.BzrDir.open(base)
910
return control.open_repository()
912
def copy_content_into(self, destination, revision_id=None):
913
"""Make a complete copy of the content in self into destination.
915
This is a destructive operation! Do not use it on existing
918
return InterRepository.get(self, destination).copy_content(revision_id)
920
def commit_write_group(self):
921
"""Commit the contents accrued within the current write group.
923
:seealso: start_write_group.
925
if self._write_group is not self.get_transaction():
926
# has an unlock or relock occured ?
927
raise errors.BzrError('mismatched lock context %r and '
929
(self.get_transaction(), self._write_group))
930
self._commit_write_group()
931
self._write_group = None
933
def _commit_write_group(self):
934
"""Template method for per-repository write group cleanup.
936
This is called before the write group is considered to be
937
finished and should ensure that all data handed to the repository
938
for writing during the write group is safely committed (to the
939
extent possible considering file system caching etc).
942
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
943
"""Fetch the content required to construct revision_id from source.
945
If revision_id is None all content is copied.
946
:param find_ghosts: Find and copy revisions in the source that are
947
ghosts in the target (and not reachable directly by walking out to
948
the first-present revision in target from revision_id).
950
# fast path same-url fetch operations
951
if self.has_same_location(source):
952
# check that last_revision is in 'from' and then return a
954
if (revision_id is not None and
955
not _mod_revision.is_null(revision_id)):
956
self.get_revision(revision_id)
958
inter = InterRepository.get(source, self)
960
return inter.fetch(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts)
961
except NotImplementedError:
962
raise errors.IncompatibleRepositories(source, self)
964
def create_bundle(self, target, base, fileobj, format=None):
965
return serializer.write_bundle(self, target, base, fileobj, format)
967
def get_commit_builder(self, branch, parents, config, timestamp=None,
968
timezone=None, committer=None, revprops=None,
970
"""Obtain a CommitBuilder for this repository.
972
:param branch: Branch to commit to.
973
:param parents: Revision ids of the parents of the new revision.
974
:param config: Configuration to use.
975
:param timestamp: Optional timestamp recorded for commit.
976
:param timezone: Optional timezone for timestamp.
977
:param committer: Optional committer to set for commit.
978
:param revprops: Optional dictionary of revision properties.
979
:param revision_id: Optional revision id.
981
result = self._commit_builder_class(self, parents, config,
982
timestamp, timezone, committer, revprops, revision_id)
983
self.start_write_group()
987
if (self.control_files._lock_count == 1 and
988
self.control_files._lock_mode == 'w'):
989
if self._write_group is not None:
990
self.abort_write_group()
991
self.control_files.unlock()
992
raise errors.BzrError(
993
'Must end write groups before releasing write locks.')
994
self.control_files.unlock()
995
for repo in self._fallback_repositories:
999
def clone(self, a_bzrdir, revision_id=None):
1000
"""Clone this repository into a_bzrdir using the current format.
1002
Currently no check is made that the format of this repository and
1003
the bzrdir format are compatible. FIXME RBC 20060201.
1005
:return: The newly created destination repository.
1007
# TODO: deprecate after 0.16; cloning this with all its settings is
1008
# probably not very useful -- mbp 20070423
1009
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1010
self.copy_content_into(dest_repo, revision_id)
1013
def start_write_group(self):
1014
"""Start a write group in the repository.
1016
Write groups are used by repositories which do not have a 1:1 mapping
1017
between file ids and backend store to manage the insertion of data from
1018
both fetch and commit operations.
1020
A write lock is required around the start_write_group/commit_write_group
1021
for the support of lock-requiring repository formats.
1023
One can only insert data into a repository inside a write group.
1027
if not self.is_write_locked():
1028
raise errors.NotWriteLocked(self)
1029
if self._write_group:
1030
raise errors.BzrError('already in a write group')
1031
self._start_write_group()
1032
# so we can detect unlock/relock - the write group is now entered.
1033
self._write_group = self.get_transaction()
1035
def _start_write_group(self):
1036
"""Template method for per-repository write group startup.
1038
This is called before the write group is considered to be
1043
def sprout(self, to_bzrdir, revision_id=None):
1044
"""Create a descendent repository for new development.
1046
Unlike clone, this does not copy the settings of the repository.
1048
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1049
dest_repo.fetch(self, revision_id=revision_id)
1052
def _create_sprouting_repo(self, a_bzrdir, shared):
1053
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1054
# use target default format.
1055
dest_repo = a_bzrdir.create_repository()
1057
# Most control formats need the repository to be specifically
1058
# created, but on some old all-in-one formats it's not needed
1060
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1061
except errors.UninitializableFormat:
1062
dest_repo = a_bzrdir.open_repository()
1066
def has_revision(self, revision_id):
1067
"""True if this repository has a copy of the revision."""
1068
return revision_id in self.has_revisions((revision_id,))
1071
def has_revisions(self, revision_ids):
1072
"""Probe to find out the presence of multiple revisions.
1074
:param revision_ids: An iterable of revision_ids.
1075
:return: A set of the revision_ids that were present.
1077
parent_map = self.revisions.get_parent_map(
1078
[(rev_id,) for rev_id in revision_ids])
1080
if _mod_revision.NULL_REVISION in revision_ids:
1081
result.add(_mod_revision.NULL_REVISION)
1082
result.update([key[0] for key in parent_map])
1086
def get_revision(self, revision_id):
1087
"""Return the Revision object for a named revision."""
1088
return self.get_revisions([revision_id])[0]
1091
def get_revision_reconcile(self, revision_id):
1092
"""'reconcile' helper routine that allows access to a revision always.
1094
This variant of get_revision does not cross check the weave graph
1095
against the revision one as get_revision does: but it should only
1096
be used by reconcile, or reconcile-alike commands that are correcting
1097
or testing the revision graph.
1099
return self._get_revisions([revision_id])[0]
1102
def get_revisions(self, revision_ids):
1103
"""Get many revisions at once."""
1104
return self._get_revisions(revision_ids)
1107
def _get_revisions(self, revision_ids):
1108
"""Core work logic to get many revisions without sanity checks."""
1109
for rev_id in revision_ids:
1110
if not rev_id or not isinstance(rev_id, basestring):
1111
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1112
keys = [(key,) for key in revision_ids]
1113
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1115
for record in stream:
1116
if record.storage_kind == 'absent':
1117
raise errors.NoSuchRevision(self, record.key[0])
1118
text = record.get_bytes_as('fulltext')
1119
rev = self._serializer.read_revision_from_string(text)
1120
revs[record.key[0]] = rev
1121
return [revs[revid] for revid in revision_ids]
1124
def get_revision_xml(self, revision_id):
1125
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1126
# would have already do it.
1127
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1128
rev = self.get_revision(revision_id)
1129
rev_tmp = StringIO()
1130
# the current serializer..
1131
self._serializer.write_revision(rev, rev_tmp)
1133
return rev_tmp.getvalue()
1135
def get_deltas_for_revisions(self, revisions):
1136
"""Produce a generator of revision deltas.
1138
Note that the input is a sequence of REVISIONS, not revision_ids.
1139
Trees will be held in memory until the generator exits.
1140
Each delta is relative to the revision's lefthand predecessor.
1142
required_trees = set()
1143
for revision in revisions:
1144
required_trees.add(revision.revision_id)
1145
required_trees.update(revision.parent_ids[:1])
1146
trees = dict((t.get_revision_id(), t) for
1147
t in self.revision_trees(required_trees))
1148
for revision in revisions:
1149
if not revision.parent_ids:
1150
old_tree = self.revision_tree(None)
1152
old_tree = trees[revision.parent_ids[0]]
1153
yield trees[revision.revision_id].changes_from(old_tree)
1156
def get_revision_delta(self, revision_id):
1157
"""Return the delta for one revision.
1159
The delta is relative to the left-hand predecessor of the
1162
r = self.get_revision(revision_id)
1163
return list(self.get_deltas_for_revisions([r]))[0]
1166
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1167
signature = gpg_strategy.sign(plaintext)
1168
self.add_signature_text(revision_id, signature)
1171
def add_signature_text(self, revision_id, signature):
1172
self.signatures.add_lines((revision_id,), (),
1173
osutils.split_lines(signature))
1175
def find_text_key_references(self):
1176
"""Find the text key references within the repository.
1178
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1179
revision_ids. Each altered file-ids has the exact revision_ids that
1180
altered it listed explicitly.
1181
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1182
to whether they were referred to by the inventory of the
1183
revision_id that they contain. The inventory texts from all present
1184
revision ids are assessed to generate this report.
1186
revision_keys = self.revisions.keys()
1187
w = self.inventories
1188
pb = ui.ui_factory.nested_progress_bar()
1190
return self._find_text_key_references_from_xml_inventory_lines(
1191
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1195
def _find_text_key_references_from_xml_inventory_lines(self,
1197
"""Core routine for extracting references to texts from inventories.
1199
This performs the translation of xml lines to revision ids.
1201
:param line_iterator: An iterator of lines, origin_version_id
1202
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1203
to whether they were referred to by the inventory of the
1204
revision_id that they contain. Note that if that revision_id was
1205
not part of the line_iterator's output then False will be given -
1206
even though it may actually refer to that key.
1208
if not self._serializer.support_altered_by_hack:
1209
raise AssertionError(
1210
"_find_text_key_references_from_xml_inventory_lines only "
1211
"supported for branches which store inventory as unnested xml"
1212
", not on %r" % self)
1215
# this code needs to read every new line in every inventory for the
1216
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1217
# not present in one of those inventories is unnecessary but not
1218
# harmful because we are filtering by the revision id marker in the
1219
# inventory lines : we only select file ids altered in one of those
1220
# revisions. We don't need to see all lines in the inventory because
1221
# only those added in an inventory in rev X can contain a revision=X
1223
unescape_revid_cache = {}
1224
unescape_fileid_cache = {}
1226
# jam 20061218 In a big fetch, this handles hundreds of thousands
1227
# of lines, so it has had a lot of inlining and optimizing done.
1228
# Sorry that it is a little bit messy.
1229
# Move several functions to be local variables, since this is a long
1231
search = self._file_ids_altered_regex.search
1232
unescape = _unescape_xml
1233
setdefault = result.setdefault
1234
for line, line_key in line_iterator:
1235
match = search(line)
1238
# One call to match.group() returning multiple items is quite a
1239
# bit faster than 2 calls to match.group() each returning 1
1240
file_id, revision_id = match.group('file_id', 'revision_id')
1242
# Inlining the cache lookups helps a lot when you make 170,000
1243
# lines and 350k ids, versus 8.4 unique ids.
1244
# Using a cache helps in 2 ways:
1245
# 1) Avoids unnecessary decoding calls
1246
# 2) Re-uses cached strings, which helps in future set and
1248
# (2) is enough that removing encoding entirely along with
1249
# the cache (so we are using plain strings) results in no
1250
# performance improvement.
1252
revision_id = unescape_revid_cache[revision_id]
1254
unescaped = unescape(revision_id)
1255
unescape_revid_cache[revision_id] = unescaped
1256
revision_id = unescaped
1258
# Note that unconditionally unescaping means that we deserialise
1259
# every fileid, which for general 'pull' is not great, but we don't
1260
# really want to have some many fulltexts that this matters anyway.
1263
file_id = unescape_fileid_cache[file_id]
1265
unescaped = unescape(file_id)
1266
unescape_fileid_cache[file_id] = unescaped
1269
key = (file_id, revision_id)
1270
setdefault(key, False)
1271
if revision_id == line_key[-1]:
1275
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1277
"""Helper routine for fileids_altered_by_revision_ids.
1279
This performs the translation of xml lines to revision ids.
1281
:param line_iterator: An iterator of lines, origin_version_id
1282
:param revision_ids: The revision ids to filter for. This should be a
1283
set or other type which supports efficient __contains__ lookups, as
1284
the revision id from each parsed line will be looked up in the
1285
revision_ids filter.
1286
:return: a dictionary mapping altered file-ids to an iterable of
1287
revision_ids. Each altered file-ids has the exact revision_ids that
1288
altered it listed explicitly.
1291
setdefault = result.setdefault
1293
self._find_text_key_references_from_xml_inventory_lines(
1294
line_iterator).iterkeys():
1295
# once data is all ensured-consistent; then this is
1296
# if revision_id == version_id
1297
if key[-1:] in revision_ids:
1298
setdefault(key[0], set()).add(key[-1])
1301
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1302
"""Find the file ids and versions affected by revisions.
1304
:param revisions: an iterable containing revision ids.
1305
:param _inv_weave: The inventory weave from this repository or None.
1306
If None, the inventory weave will be opened automatically.
1307
:return: a dictionary mapping altered file-ids to an iterable of
1308
revision_ids. Each altered file-ids has the exact revision_ids that
1309
altered it listed explicitly.
1311
selected_keys = set((revid,) for revid in revision_ids)
1312
w = _inv_weave or self.inventories
1313
pb = ui.ui_factory.nested_progress_bar()
1315
return self._find_file_ids_from_xml_inventory_lines(
1316
w.iter_lines_added_or_present_in_keys(
1317
selected_keys, pb=pb),
1322
def iter_files_bytes(self, desired_files):
1323
"""Iterate through file versions.
1325
Files will not necessarily be returned in the order they occur in
1326
desired_files. No specific order is guaranteed.
1328
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1329
value supplied by the caller as part of desired_files. It should
1330
uniquely identify the file version in the caller's context. (Examples:
1331
an index number or a TreeTransform trans_id.)
1333
bytes_iterator is an iterable of bytestrings for the file. The
1334
kind of iterable and length of the bytestrings are unspecified, but for
1335
this implementation, it is a list of bytes produced by
1336
VersionedFile.get_record_stream().
1338
:param desired_files: a list of (file_id, revision_id, identifier)
1341
transaction = self.get_transaction()
1343
for file_id, revision_id, callable_data in desired_files:
1344
text_keys[(file_id, revision_id)] = callable_data
1345
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1346
if record.storage_kind == 'absent':
1347
raise errors.RevisionNotPresent(record.key, self)
1348
yield text_keys[record.key], record.get_bytes_as('fulltext')
1350
def _generate_text_key_index(self, text_key_references=None,
1352
"""Generate a new text key index for the repository.
1354
This is an expensive function that will take considerable time to run.
1356
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1357
list of parents, also text keys. When a given key has no parents,
1358
the parents list will be [NULL_REVISION].
1360
# All revisions, to find inventory parents.
1361
if ancestors is None:
1362
graph = self.get_graph()
1363
ancestors = graph.get_parent_map(self.all_revision_ids())
1364
if text_key_references is None:
1365
text_key_references = self.find_text_key_references()
1366
pb = ui.ui_factory.nested_progress_bar()
1368
return self._do_generate_text_key_index(ancestors,
1369
text_key_references, pb)
1373
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1374
"""Helper for _generate_text_key_index to avoid deep nesting."""
1375
revision_order = tsort.topo_sort(ancestors)
1376
invalid_keys = set()
1378
for revision_id in revision_order:
1379
revision_keys[revision_id] = set()
1380
text_count = len(text_key_references)
1381
# a cache of the text keys to allow reuse; costs a dict of all the
1382
# keys, but saves a 2-tuple for every child of a given key.
1384
for text_key, valid in text_key_references.iteritems():
1386
invalid_keys.add(text_key)
1388
revision_keys[text_key[1]].add(text_key)
1389
text_key_cache[text_key] = text_key
1390
del text_key_references
1392
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1393
NULL_REVISION = _mod_revision.NULL_REVISION
1394
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1395
# too small for large or very branchy trees. However, for 55K path
1396
# trees, it would be easy to use too much memory trivially. Ideally we
1397
# could gauge this by looking at available real memory etc, but this is
1398
# always a tricky proposition.
1399
inventory_cache = lru_cache.LRUCache(10)
1400
batch_size = 10 # should be ~150MB on a 55K path tree
1401
batch_count = len(revision_order) / batch_size + 1
1403
pb.update("Calculating text parents.", processed_texts, text_count)
1404
for offset in xrange(batch_count):
1405
to_query = revision_order[offset * batch_size:(offset + 1) *
1409
for rev_tree in self.revision_trees(to_query):
1410
revision_id = rev_tree.get_revision_id()
1411
parent_ids = ancestors[revision_id]
1412
for text_key in revision_keys[revision_id]:
1413
pb.update("Calculating text parents.", processed_texts)
1414
processed_texts += 1
1415
candidate_parents = []
1416
for parent_id in parent_ids:
1417
parent_text_key = (text_key[0], parent_id)
1419
check_parent = parent_text_key not in \
1420
revision_keys[parent_id]
1422
# the parent parent_id is a ghost:
1423
check_parent = False
1424
# truncate the derived graph against this ghost.
1425
parent_text_key = None
1427
# look at the parent commit details inventories to
1428
# determine possible candidates in the per file graph.
1431
inv = inventory_cache[parent_id]
1433
inv = self.revision_tree(parent_id).inventory
1434
inventory_cache[parent_id] = inv
1435
parent_entry = inv._byid.get(text_key[0], None)
1436
if parent_entry is not None:
1438
text_key[0], parent_entry.revision)
1440
parent_text_key = None
1441
if parent_text_key is not None:
1442
candidate_parents.append(
1443
text_key_cache[parent_text_key])
1444
parent_heads = text_graph.heads(candidate_parents)
1445
new_parents = list(parent_heads)
1446
new_parents.sort(key=lambda x:candidate_parents.index(x))
1447
if new_parents == []:
1448
new_parents = [NULL_REVISION]
1449
text_index[text_key] = new_parents
1451
for text_key in invalid_keys:
1452
text_index[text_key] = [NULL_REVISION]
1455
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1456
"""Get an iterable listing the keys of all the data introduced by a set
1459
The keys will be ordered so that the corresponding items can be safely
1460
fetched and inserted in that order.
1462
:returns: An iterable producing tuples of (knit-kind, file-id,
1463
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1464
'revisions'. file-id is None unless knit-kind is 'file'.
1466
# XXX: it's a bit weird to control the inventory weave caching in this
1467
# generator. Ideally the caching would be done in fetch.py I think. Or
1468
# maybe this generator should explicitly have the contract that it
1469
# should not be iterated until the previously yielded item has been
1471
inv_w = self.inventories
1473
# file ids that changed
1474
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1476
num_file_ids = len(file_ids)
1477
for file_id, altered_versions in file_ids.iteritems():
1478
if _files_pb is not None:
1479
_files_pb.update("fetch texts", count, num_file_ids)
1481
yield ("file", file_id, altered_versions)
1482
# We're done with the files_pb. Note that it finished by the caller,
1483
# just as it was created by the caller.
1487
yield ("inventory", None, revision_ids)
1490
revisions_with_signatures = set()
1491
for rev_id in revision_ids:
1493
self.get_signature_text(rev_id)
1494
except errors.NoSuchRevision:
1498
revisions_with_signatures.add(rev_id)
1499
yield ("signatures", None, revisions_with_signatures)
1502
yield ("revisions", None, revision_ids)
1505
def get_inventory(self, revision_id):
1506
"""Get Inventory object by revision id."""
1507
return self.iter_inventories([revision_id]).next()
1509
def iter_inventories(self, revision_ids):
1510
"""Get many inventories by revision_ids.
1512
This will buffer some or all of the texts used in constructing the
1513
inventories in memory, but will only parse a single inventory at a
1516
:return: An iterator of inventories.
1518
if ((None in revision_ids)
1519
or (_mod_revision.NULL_REVISION in revision_ids)):
1520
raise ValueError('cannot get null revision inventory')
1521
return self._iter_inventories(revision_ids)
1523
def _iter_inventories(self, revision_ids):
1524
"""single-document based inventory iteration."""
1525
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1526
yield self.deserialise_inventory(revision_id, text)
1528
def _iter_inventory_xmls(self, revision_ids):
1529
keys = [(revision_id,) for revision_id in revision_ids]
1530
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1532
for record in stream:
1533
if record.storage_kind != 'absent':
1534
texts[record.key] = record.get_bytes_as('fulltext')
1536
raise errors.NoSuchRevision(self, record.key)
1538
yield texts[key], key[-1]
1540
def deserialise_inventory(self, revision_id, xml):
1541
"""Transform the xml into an inventory object.
1543
:param revision_id: The expected revision id of the inventory.
1544
:param xml: A serialised inventory.
1546
result = self._serializer.read_inventory_from_string(xml, revision_id)
1547
if result.revision_id != revision_id:
1548
raise AssertionError('revision id mismatch %s != %s' % (
1549
result.revision_id, revision_id))
1552
def serialise_inventory(self, inv):
1553
return self._serializer.write_inventory_to_string(inv)
1555
def _serialise_inventory_to_lines(self, inv):
1556
return self._serializer.write_inventory_to_lines(inv)
1558
def get_serializer_format(self):
1559
return self._serializer.format_num
1562
def get_inventory_xml(self, revision_id):
1563
"""Get inventory XML as a file object."""
1564
texts = self._iter_inventory_xmls([revision_id])
1566
text, revision_id = texts.next()
1567
except StopIteration:
1568
raise errors.HistoryMissing(self, 'inventory', revision_id)
1572
def get_inventory_sha1(self, revision_id):
1573
"""Return the sha1 hash of the inventory entry
1575
return self.get_revision(revision_id).inventory_sha1
1577
def iter_reverse_revision_history(self, revision_id):
1578
"""Iterate backwards through revision ids in the lefthand history
1580
:param revision_id: The revision id to start with. All its lefthand
1581
ancestors will be traversed.
1583
graph = self.get_graph()
1584
next_id = revision_id
1586
if next_id in (None, _mod_revision.NULL_REVISION):
1589
# Note: The following line may raise KeyError in the event of
1590
# truncated history. We decided not to have a try:except:raise
1591
# RevisionNotPresent here until we see a use for it, because of the
1592
# cost in an inner loop that is by its very nature O(history).
1593
# Robert Collins 20080326
1594
parents = graph.get_parent_map([next_id])[next_id]
1595
if len(parents) == 0:
1598
next_id = parents[0]
1601
def get_revision_inventory(self, revision_id):
1602
"""Return inventory of a past revision."""
1603
# TODO: Unify this with get_inventory()
1604
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1605
# must be the same as its revision, so this is trivial.
1606
if revision_id is None:
1607
# This does not make sense: if there is no revision,
1608
# then it is the current tree inventory surely ?!
1609
# and thus get_root_id() is something that looks at the last
1610
# commit on the branch, and the get_root_id is an inventory check.
1611
raise NotImplementedError
1612
# return Inventory(self.get_root_id())
1614
return self.get_inventory(revision_id)
1617
def is_shared(self):
1618
"""Return True if this repository is flagged as a shared repository."""
1619
raise NotImplementedError(self.is_shared)
1622
def reconcile(self, other=None, thorough=False):
1623
"""Reconcile this repository."""
1624
from bzrlib.reconcile import RepoReconciler
1625
reconciler = RepoReconciler(self, thorough=thorough)
1626
reconciler.reconcile()
1629
def _refresh_data(self):
1630
"""Helper called from lock_* to ensure coherency with disk.
1632
The default implementation does nothing; it is however possible
1633
for repositories to maintain loaded indices across multiple locks
1634
by checking inside their implementation of this method to see
1635
whether their indices are still valid. This depends of course on
1636
the disk format being validatable in this manner.
1640
def revision_tree(self, revision_id):
1641
"""Return Tree for a revision on this branch.
1643
`revision_id` may be None for the empty tree revision.
1645
# TODO: refactor this to use an existing revision object
1646
# so we don't need to read it in twice.
1647
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1648
return RevisionTree(self, Inventory(root_id=None),
1649
_mod_revision.NULL_REVISION)
1651
inv = self.get_revision_inventory(revision_id)
1652
return RevisionTree(self, inv, revision_id)
1654
def revision_trees(self, revision_ids):
1655
"""Return Tree for a revision on this branch.
1657
`revision_id` may not be None or 'null:'"""
1658
inventories = self.iter_inventories(revision_ids)
1659
for inv in inventories:
1660
yield RevisionTree(self, inv, inv.revision_id)
1663
def get_ancestry(self, revision_id, topo_sorted=True):
1664
"""Return a list of revision-ids integrated by a revision.
1666
The first element of the list is always None, indicating the origin
1667
revision. This might change when we have history horizons, or
1668
perhaps we should have a new API.
1670
This is topologically sorted.
1672
if _mod_revision.is_null(revision_id):
1674
if not self.has_revision(revision_id):
1675
raise errors.NoSuchRevision(self, revision_id)
1676
graph = self.get_graph()
1678
search = graph._make_breadth_first_searcher([revision_id])
1681
found, ghosts = search.next_with_ghosts()
1682
except StopIteration:
1685
if _mod_revision.NULL_REVISION in keys:
1686
keys.remove(_mod_revision.NULL_REVISION)
1688
parent_map = graph.get_parent_map(keys)
1689
keys = tsort.topo_sort(parent_map)
1690
return [None] + list(keys)
1693
"""Compress the data within the repository.
1695
This operation only makes sense for some repository types. For other
1696
types it should be a no-op that just returns.
1698
This stub method does not require a lock, but subclasses should use
1699
@needs_write_lock as this is a long running call its reasonable to
1700
implicitly lock for the user.
1704
@deprecated_method(one_six)
1705
def print_file(self, file, revision_id):
1706
"""Print `file` to stdout.
1708
FIXME RBC 20060125 as John Meinel points out this is a bad api
1709
- it writes to stdout, it assumes that that is valid etc. Fix
1710
by creating a new more flexible convenience function.
1712
tree = self.revision_tree(revision_id)
1713
# use inventory as it was in that revision
1714
file_id = tree.inventory.path2id(file)
1716
# TODO: jam 20060427 Write a test for this code path
1717
# it had a bug in it, and was raising the wrong
1719
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1720
tree.print_file(file_id)
1722
def get_transaction(self):
1723
return self.control_files.get_transaction()
1725
@deprecated_method(one_one)
1726
def get_parents(self, revision_ids):
1727
"""See StackedParentsProvider.get_parents"""
1728
parent_map = self.get_parent_map(revision_ids)
1729
return [parent_map.get(r, None) for r in revision_ids]
1731
def get_parent_map(self, revision_ids):
1732
"""See graph._StackedParentsProvider.get_parent_map"""
1733
# revisions index works in keys; this just works in revisions
1734
# therefore wrap and unwrap
1737
for revision_id in revision_ids:
1738
if revision_id == _mod_revision.NULL_REVISION:
1739
result[revision_id] = ()
1740
elif revision_id is None:
1741
raise ValueError('get_parent_map(None) is not valid')
1743
query_keys.append((revision_id ,))
1744
for ((revision_id,), parent_keys) in \
1745
self.revisions.get_parent_map(query_keys).iteritems():
1747
result[revision_id] = tuple(parent_revid
1748
for (parent_revid,) in parent_keys)
1750
result[revision_id] = (_mod_revision.NULL_REVISION,)
1753
def _make_parents_provider(self):
1756
def get_graph(self, other_repository=None):
1757
"""Return the graph walker for this repository format"""
1758
parents_provider = self._make_parents_provider()
1759
if (other_repository is not None and
1760
not self.has_same_location(other_repository)):
1761
parents_provider = graph._StackedParentsProvider(
1762
[parents_provider, other_repository._make_parents_provider()])
1763
return graph.Graph(parents_provider)
1765
def _get_versioned_file_checker(self):
1766
"""Return an object suitable for checking versioned files."""
1767
return _VersionedFileChecker(self)
1769
def revision_ids_to_search_result(self, result_set):
1770
"""Convert a set of revision ids to a graph SearchResult."""
1771
result_parents = set()
1772
for parents in self.get_graph().get_parent_map(
1773
result_set).itervalues():
1774
result_parents.update(parents)
1775
included_keys = result_set.intersection(result_parents)
1776
start_keys = result_set.difference(included_keys)
1777
exclude_keys = result_parents.difference(result_set)
1778
result = graph.SearchResult(start_keys, exclude_keys,
1779
len(result_set), result_set)
1783
def set_make_working_trees(self, new_value):
1784
"""Set the policy flag for making working trees when creating branches.
1786
This only applies to branches that use this repository.
1788
The default is 'True'.
1789
:param new_value: True to restore the default, False to disable making
1792
raise NotImplementedError(self.set_make_working_trees)
1794
def make_working_trees(self):
1795
"""Returns the policy for making working trees on new branches."""
1796
raise NotImplementedError(self.make_working_trees)
1799
def sign_revision(self, revision_id, gpg_strategy):
1800
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1801
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1804
def has_signature_for_revision_id(self, revision_id):
1805
"""Query for a revision signature for revision_id in the repository."""
1806
if not self.has_revision(revision_id):
1807
raise errors.NoSuchRevision(self, revision_id)
1808
sig_present = (1 == len(
1809
self.signatures.get_parent_map([(revision_id,)])))
1813
def get_signature_text(self, revision_id):
1814
"""Return the text for a signature."""
1815
stream = self.signatures.get_record_stream([(revision_id,)],
1817
record = stream.next()
1818
if record.storage_kind == 'absent':
1819
raise errors.NoSuchRevision(self, revision_id)
1820
return record.get_bytes_as('fulltext')
1823
def check(self, revision_ids=None):
1824
"""Check consistency of all history of given revision_ids.
1826
Different repository implementations should override _check().
1828
:param revision_ids: A non-empty list of revision_ids whose ancestry
1829
will be checked. Typically the last revision_id of a branch.
1831
return self._check(revision_ids)
1833
def _check(self, revision_ids):
1834
result = check.Check(self)
1838
def _warn_if_deprecated(self):
1839
global _deprecation_warning_done
1840
if _deprecation_warning_done:
1842
_deprecation_warning_done = True
1843
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1844
% (self._format, self.bzrdir.transport.base))
1846
def supports_rich_root(self):
1847
return self._format.rich_root_data
1849
def _check_ascii_revisionid(self, revision_id, method):
1850
"""Private helper for ascii-only repositories."""
1851
# weave repositories refuse to store revisionids that are non-ascii.
1852
if revision_id is not None:
1853
# weaves require ascii revision ids.
1854
if isinstance(revision_id, unicode):
1856
revision_id.encode('ascii')
1857
except UnicodeEncodeError:
1858
raise errors.NonAsciiRevisionId(method, self)
1861
revision_id.decode('ascii')
1862
except UnicodeDecodeError:
1863
raise errors.NonAsciiRevisionId(method, self)
1865
def revision_graph_can_have_wrong_parents(self):
1866
"""Is it possible for this repository to have a revision graph with
1869
If True, then this repository must also implement
1870
_find_inconsistent_revision_parents so that check and reconcile can
1871
check for inconsistencies before proceeding with other checks that may
1872
depend on the revision index being consistent.
1874
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1877
# remove these delegates a while after bzr 0.15
1878
def __make_delegated(name, from_module):
1879
def _deprecated_repository_forwarder():
1880
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1881
% (name, from_module),
1884
m = __import__(from_module, globals(), locals(), [name])
1886
return getattr(m, name)
1887
except AttributeError:
1888
raise AttributeError('module %s has no name %s'
1890
globals()[name] = _deprecated_repository_forwarder
1893
'AllInOneRepository',
1894
'WeaveMetaDirRepository',
1895
'PreSplitOutRepositoryFormat',
1896
'RepositoryFormat4',
1897
'RepositoryFormat5',
1898
'RepositoryFormat6',
1899
'RepositoryFormat7',
1901
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1905
'RepositoryFormatKnit',
1906
'RepositoryFormatKnit1',
1908
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1911
def install_revision(repository, rev, revision_tree):
1912
"""Install all revision data into a repository."""
1913
install_revisions(repository, [(rev, revision_tree, None)])
1916
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1917
"""Install all revision data into a repository.
1919
Accepts an iterable of revision, tree, signature tuples. The signature
1922
repository.start_write_group()
1924
for n, (revision, revision_tree, signature) in enumerate(iterable):
1925
_install_revision(repository, revision, revision_tree, signature)
1927
pb.update('Transferring revisions', n + 1, num_revisions)
1929
repository.abort_write_group()
1932
repository.commit_write_group()
1935
def _install_revision(repository, rev, revision_tree, signature):
1936
"""Install all revision data into a repository."""
1937
present_parents = []
1939
for p_id in rev.parent_ids:
1940
if repository.has_revision(p_id):
1941
present_parents.append(p_id)
1942
parent_trees[p_id] = repository.revision_tree(p_id)
1944
parent_trees[p_id] = repository.revision_tree(None)
1946
inv = revision_tree.inventory
1947
entries = inv.iter_entries()
1948
# backwards compatibility hack: skip the root id.
1949
if not repository.supports_rich_root():
1950
path, root = entries.next()
1951
if root.revision != rev.revision_id:
1952
raise errors.IncompatibleRevision(repr(repository))
1954
for path, ie in entries:
1955
text_keys[(ie.file_id, ie.revision)] = ie
1956
text_parent_map = repository.texts.get_parent_map(text_keys)
1957
missing_texts = set(text_keys) - set(text_parent_map)
1958
# Add the texts that are not already present
1959
for text_key in missing_texts:
1960
ie = text_keys[text_key]
1962
# FIXME: TODO: The following loop overlaps/duplicates that done by
1963
# commit to determine parents. There is a latent/real bug here where
1964
# the parents inserted are not those commit would do - in particular
1965
# they are not filtered by heads(). RBC, AB
1966
for revision, tree in parent_trees.iteritems():
1967
if ie.file_id not in tree:
1969
parent_id = tree.inventory[ie.file_id].revision
1970
if parent_id in text_parents:
1972
text_parents.append((ie.file_id, parent_id))
1973
lines = revision_tree.get_file(ie.file_id).readlines()
1974
repository.texts.add_lines(text_key, text_parents, lines)
1976
# install the inventory
1977
repository.add_inventory(rev.revision_id, inv, present_parents)
1978
except errors.RevisionAlreadyPresent:
1980
if signature is not None:
1981
repository.add_signature_text(rev.revision_id, signature)
1982
repository.add_revision(rev.revision_id, rev, inv)
1985
class MetaDirRepository(Repository):
1986
"""Repositories in the new meta-dir layout.
1988
:ivar _transport: Transport for access to repository control files,
1989
typically pointing to .bzr/repository.
1992
def __init__(self, _format, a_bzrdir, control_files):
1993
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
1994
self._transport = control_files._transport
1997
def is_shared(self):
1998
"""Return True if this repository is flagged as a shared repository."""
1999
return self._transport.has('shared-storage')
2002
def set_make_working_trees(self, new_value):
2003
"""Set the policy flag for making working trees when creating branches.
2005
This only applies to branches that use this repository.
2007
The default is 'True'.
2008
:param new_value: True to restore the default, False to disable making
2013
self._transport.delete('no-working-trees')
2014
except errors.NoSuchFile:
2017
self._transport.put_bytes('no-working-trees', '',
2018
mode=self.bzrdir._get_file_mode())
2020
def make_working_trees(self):
2021
"""Returns the policy for making working trees on new branches."""
2022
return not self._transport.has('no-working-trees')
2025
class MetaDirVersionedFileRepository(MetaDirRepository):
2026
"""Repositories in a meta-dir, that work via versioned file objects."""
2028
def __init__(self, _format, a_bzrdir, control_files):
2029
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2033
class RepositoryFormatRegistry(registry.Registry):
2034
"""Registry of RepositoryFormats."""
2036
def get(self, format_string):
2037
r = registry.Registry.get(self, format_string)
2043
format_registry = RepositoryFormatRegistry()
2044
"""Registry of formats, indexed by their identifying format string.
2046
This can contain either format instances themselves, or classes/factories that
2047
can be called to obtain one.
2051
#####################################################################
2052
# Repository Formats
2054
class RepositoryFormat(object):
2055
"""A repository format.
2057
Formats provide three things:
2058
* An initialization routine to construct repository data on disk.
2059
* a format string which is used when the BzrDir supports versioned
2061
* an open routine which returns a Repository instance.
2063
There is one and only one Format subclass for each on-disk format. But
2064
there can be one Repository subclass that is used for several different
2065
formats. The _format attribute on a Repository instance can be used to
2066
determine the disk format.
2068
Formats are placed in an dict by their format string for reference
2069
during opening. These should be subclasses of RepositoryFormat
2072
Once a format is deprecated, just deprecate the initialize and open
2073
methods on the format class. Do not deprecate the object, as the
2074
object will be created every system load.
2076
Common instance attributes:
2077
_matchingbzrdir - the bzrdir format that the repository format was
2078
originally written to work with. This can be used if manually
2079
constructing a bzrdir and repository, or more commonly for test suite
2083
# Set to True or False in derived classes. True indicates that the format
2084
# supports ghosts gracefully.
2085
supports_ghosts = None
2086
# Can this repository be given external locations to lookup additional
2087
# data. Set to True or False in derived classes.
2088
supports_external_lookups = None
2091
return "<%s>" % self.__class__.__name__
2093
def __eq__(self, other):
2094
# format objects are generally stateless
2095
return isinstance(other, self.__class__)
2097
def __ne__(self, other):
2098
return not self == other
2101
def find_format(klass, a_bzrdir):
2102
"""Return the format for the repository object in a_bzrdir.
2104
This is used by bzr native formats that have a "format" file in
2105
the repository. Other methods may be used by different types of
2109
transport = a_bzrdir.get_repository_transport(None)
2110
format_string = transport.get("format").read()
2111
return format_registry.get(format_string)
2112
except errors.NoSuchFile:
2113
raise errors.NoRepositoryPresent(a_bzrdir)
2115
raise errors.UnknownFormatError(format=format_string,
2119
def register_format(klass, format):
2120
format_registry.register(format.get_format_string(), format)
2123
def unregister_format(klass, format):
2124
format_registry.remove(format.get_format_string())
2127
def get_default_format(klass):
2128
"""Return the current default format."""
2129
from bzrlib import bzrdir
2130
return bzrdir.format_registry.make_bzrdir('default').repository_format
2132
def get_format_string(self):
2133
"""Return the ASCII format string that identifies this format.
2135
Note that in pre format ?? repositories the format string is
2136
not permitted nor written to disk.
2138
raise NotImplementedError(self.get_format_string)
2140
def get_format_description(self):
2141
"""Return the short description for this format."""
2142
raise NotImplementedError(self.get_format_description)
2144
# TODO: this shouldn't be in the base class, it's specific to things that
2145
# use weaves or knits -- mbp 20070207
2146
def _get_versioned_file_store(self,
2151
versionedfile_class=None,
2152
versionedfile_kwargs={},
2154
if versionedfile_class is None:
2155
versionedfile_class = self._versionedfile_class
2156
weave_transport = control_files._transport.clone(name)
2157
dir_mode = control_files._dir_mode
2158
file_mode = control_files._file_mode
2159
return VersionedFileStore(weave_transport, prefixed=prefixed,
2161
file_mode=file_mode,
2162
versionedfile_class=versionedfile_class,
2163
versionedfile_kwargs=versionedfile_kwargs,
2166
def initialize(self, a_bzrdir, shared=False):
2167
"""Initialize a repository of this format in a_bzrdir.
2169
:param a_bzrdir: The bzrdir to put the new repository in it.
2170
:param shared: The repository should be initialized as a sharable one.
2171
:returns: The new repository object.
2173
This may raise UninitializableFormat if shared repository are not
2174
compatible the a_bzrdir.
2176
raise NotImplementedError(self.initialize)
2178
def is_supported(self):
2179
"""Is this format supported?
2181
Supported formats must be initializable and openable.
2182
Unsupported formats may not support initialization or committing or
2183
some other features depending on the reason for not being supported.
2187
def check_conversion_target(self, target_format):
2188
raise NotImplementedError(self.check_conversion_target)
2190
def open(self, a_bzrdir, _found=False):
2191
"""Return an instance of this format for the bzrdir a_bzrdir.
2193
_found is a private parameter, do not use it.
2195
raise NotImplementedError(self.open)
2198
class MetaDirRepositoryFormat(RepositoryFormat):
2199
"""Common base class for the new repositories using the metadir layout."""
2201
rich_root_data = False
2202
supports_tree_reference = False
2203
supports_external_lookups = False
2204
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2207
super(MetaDirRepositoryFormat, self).__init__()
2209
def _create_control_files(self, a_bzrdir):
2210
"""Create the required files and the initial control_files object."""
2211
# FIXME: RBC 20060125 don't peek under the covers
2212
# NB: no need to escape relative paths that are url safe.
2213
repository_transport = a_bzrdir.get_repository_transport(self)
2214
control_files = lockable_files.LockableFiles(repository_transport,
2215
'lock', lockdir.LockDir)
2216
control_files.create_lock()
2217
return control_files
2219
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
2220
"""Upload the initial blank content."""
2221
control_files = self._create_control_files(a_bzrdir)
2222
control_files.lock_write()
2223
transport = control_files._transport
2225
utf8_files += [('shared-storage', '')]
2227
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2228
for (filename, content_stream) in files:
2229
transport.put_file(filename, content_stream,
2230
mode=a_bzrdir._get_file_mode())
2231
for (filename, content_bytes) in utf8_files:
2232
transport.put_bytes_non_atomic(filename, content_bytes,
2233
mode=a_bzrdir._get_file_mode())
2235
control_files.unlock()
2238
# formats which have no format string are not discoverable
2239
# and not independently creatable, so are not registered. They're
2240
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2241
# needed, it's constructed directly by the BzrDir. Non-native formats where
2242
# the repository is not separately opened are similar.
2244
format_registry.register_lazy(
2245
'Bazaar-NG Repository format 7',
2246
'bzrlib.repofmt.weaverepo',
2250
format_registry.register_lazy(
2251
'Bazaar-NG Knit Repository Format 1',
2252
'bzrlib.repofmt.knitrepo',
2253
'RepositoryFormatKnit1',
2256
format_registry.register_lazy(
2257
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2258
'bzrlib.repofmt.knitrepo',
2259
'RepositoryFormatKnit3',
2262
format_registry.register_lazy(
2263
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2264
'bzrlib.repofmt.knitrepo',
2265
'RepositoryFormatKnit4',
2268
# Pack-based formats. There is one format for pre-subtrees, and one for
2269
# post-subtrees to allow ease of testing.
2270
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2271
format_registry.register_lazy(
2272
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2273
'bzrlib.repofmt.pack_repo',
2274
'RepositoryFormatKnitPack1',
2276
format_registry.register_lazy(
2277
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2278
'bzrlib.repofmt.pack_repo',
2279
'RepositoryFormatKnitPack3',
2281
format_registry.register_lazy(
2282
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2283
'bzrlib.repofmt.pack_repo',
2284
'RepositoryFormatKnitPack4',
2286
format_registry.register_lazy(
2287
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2288
'bzrlib.repofmt.pack_repo',
2289
'RepositoryFormatKnitPack5',
2291
format_registry.register_lazy(
2292
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2293
'bzrlib.repofmt.pack_repo',
2294
'RepositoryFormatKnitPack5RichRoot',
2297
# Development formats.
2299
# development 0 - stub to introduce development versioning scheme.
2300
format_registry.register_lazy(
2301
"Bazaar development format 0 (needs bzr.dev from before 1.3)\n",
2302
'bzrlib.repofmt.pack_repo',
2303
'RepositoryFormatPackDevelopment0',
2305
format_registry.register_lazy(
2306
("Bazaar development format 0 with subtree support "
2307
"(needs bzr.dev from before 1.3)\n"),
2308
'bzrlib.repofmt.pack_repo',
2309
'RepositoryFormatPackDevelopment0Subtree',
2311
format_registry.register_lazy(
2312
"Bazaar development format 1 (needs bzr.dev from before 1.6)\n",
2313
'bzrlib.repofmt.pack_repo',
2314
'RepositoryFormatPackDevelopment1',
2316
format_registry.register_lazy(
2317
("Bazaar development format 1 with subtree support "
2318
"(needs bzr.dev from before 1.6)\n"),
2319
'bzrlib.repofmt.pack_repo',
2320
'RepositoryFormatPackDevelopment1Subtree',
2322
# 1.3->1.4 go below here
2325
class InterRepository(InterObject):
2326
"""This class represents operations taking place between two repositories.
2328
Its instances have methods like copy_content and fetch, and contain
2329
references to the source and target repositories these operations can be
2332
Often we will provide convenience methods on 'repository' which carry out
2333
operations with another repository - they will always forward to
2334
InterRepository.get(other).method_name(parameters).
2338
"""The available optimised InterRepository types."""
2340
def copy_content(self, revision_id=None):
2341
raise NotImplementedError(self.copy_content)
2343
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2344
"""Fetch the content required to construct revision_id.
2346
The content is copied from self.source to self.target.
2348
:param revision_id: if None all content is copied, if NULL_REVISION no
2350
:param pb: optional progress bar to use for progress reports. If not
2351
provided a default one will be created.
2353
Returns the copied revision count and the failed revisions in a tuple:
2356
raise NotImplementedError(self.fetch)
2358
def _walk_to_common_revisions(self, revision_ids):
2359
"""Walk out from revision_ids in source to revisions target has.
2361
:param revision_ids: The start point for the search.
2362
:return: A set of revision ids.
2364
target_graph = self.target.get_graph()
2365
revision_ids = frozenset(revision_ids)
2366
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2367
return graph.SearchResult(revision_ids, set(), 0, set())
2368
missing_revs = set()
2369
source_graph = self.source.get_graph()
2370
# ensure we don't pay silly lookup costs.
2371
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2372
null_set = frozenset([_mod_revision.NULL_REVISION])
2375
next_revs, ghosts = searcher.next_with_ghosts()
2376
except StopIteration:
2378
if revision_ids.intersection(ghosts):
2379
absent_ids = set(revision_ids.intersection(ghosts))
2380
# If all absent_ids are present in target, no error is needed.
2381
absent_ids.difference_update(
2382
set(target_graph.get_parent_map(absent_ids)))
2384
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2385
# we don't care about other ghosts as we can't fetch them and
2386
# haven't been asked to.
2387
next_revs = set(next_revs)
2388
# we always have NULL_REVISION present.
2389
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2390
missing_revs.update(next_revs - have_revs)
2391
searcher.stop_searching_any(have_revs)
2392
return searcher.get_result()
2394
@deprecated_method(one_two)
2396
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2397
"""Return the revision ids that source has that target does not.
2399
These are returned in topological order.
2401
:param revision_id: only return revision ids included by this
2403
:param find_ghosts: If True find missing revisions in deep history
2404
rather than just finding the surface difference.
2406
return list(self.search_missing_revision_ids(
2407
revision_id, find_ghosts).get_keys())
2410
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2411
"""Return the revision ids that source has that target does not.
2413
:param revision_id: only return revision ids included by this
2415
:param find_ghosts: If True find missing revisions in deep history
2416
rather than just finding the surface difference.
2417
:return: A bzrlib.graph.SearchResult.
2419
# stop searching at found target revisions.
2420
if not find_ghosts and revision_id is not None:
2421
return self._walk_to_common_revisions([revision_id])
2422
# generic, possibly worst case, slow code path.
2423
target_ids = set(self.target.all_revision_ids())
2424
if revision_id is not None:
2425
source_ids = self.source.get_ancestry(revision_id)
2426
if source_ids[0] is not None:
2427
raise AssertionError()
2430
source_ids = self.source.all_revision_ids()
2431
result_set = set(source_ids).difference(target_ids)
2432
return self.source.revision_ids_to_search_result(result_set)
2435
def _same_model(source, target):
2436
"""True if source and target have the same data representation."""
2437
if source.supports_rich_root() != target.supports_rich_root():
2439
if source._serializer != target._serializer:
2444
class InterSameDataRepository(InterRepository):
2445
"""Code for converting between repositories that represent the same data.
2447
Data format and model must match for this to work.
2451
def _get_repo_format_to_test(self):
2452
"""Repository format for testing with.
2454
InterSameData can pull from subtree to subtree and from non-subtree to
2455
non-subtree, so we test this with the richest repository format.
2457
from bzrlib.repofmt import knitrepo
2458
return knitrepo.RepositoryFormatKnit3()
2461
def is_compatible(source, target):
2462
return InterRepository._same_model(source, target)
2465
def copy_content(self, revision_id=None):
2466
"""Make a complete copy of the content in self into destination.
2468
This copies both the repository's revision data, and configuration information
2469
such as the make_working_trees setting.
2471
This is a destructive operation! Do not use it on existing
2474
:param revision_id: Only copy the content needed to construct
2475
revision_id and its parents.
2478
self.target.set_make_working_trees(self.source.make_working_trees())
2479
except NotImplementedError:
2481
# but don't bother fetching if we have the needed data now.
2482
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2483
self.target.has_revision(revision_id)):
2485
self.target.fetch(self.source, revision_id=revision_id)
2488
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2489
"""See InterRepository.fetch()."""
2490
from bzrlib.fetch import GenericRepoFetcher
2491
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2492
self.source, self.source._format, self.target,
2493
self.target._format)
2494
f = GenericRepoFetcher(to_repository=self.target,
2495
from_repository=self.source,
2496
last_revision=revision_id,
2497
pb=pb, find_ghosts=find_ghosts)
2498
return f.count_copied, f.failed_revisions
2501
class InterWeaveRepo(InterSameDataRepository):
2502
"""Optimised code paths between Weave based repositories.
2504
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2505
implemented lazy inter-object optimisation.
2509
def _get_repo_format_to_test(self):
2510
from bzrlib.repofmt import weaverepo
2511
return weaverepo.RepositoryFormat7()
2514
def is_compatible(source, target):
2515
"""Be compatible with known Weave formats.
2517
We don't test for the stores being of specific types because that
2518
could lead to confusing results, and there is no need to be
2521
from bzrlib.repofmt.weaverepo import (
2527
return (isinstance(source._format, (RepositoryFormat5,
2529
RepositoryFormat7)) and
2530
isinstance(target._format, (RepositoryFormat5,
2532
RepositoryFormat7)))
2533
except AttributeError:
2537
def copy_content(self, revision_id=None):
2538
"""See InterRepository.copy_content()."""
2539
# weave specific optimised path:
2541
self.target.set_make_working_trees(self.source.make_working_trees())
2542
except (errors.RepositoryUpgradeRequired, NotImplemented):
2544
# FIXME do not peek!
2545
if self.source._transport.listable():
2546
pb = ui.ui_factory.nested_progress_bar()
2548
self.target.texts.insert_record_stream(
2549
self.source.texts.get_record_stream(
2550
self.source.texts.keys(), 'topological', False))
2551
pb.update('copying inventory', 0, 1)
2552
self.target.inventories.insert_record_stream(
2553
self.source.inventories.get_record_stream(
2554
self.source.inventories.keys(), 'topological', False))
2555
self.target.signatures.insert_record_stream(
2556
self.source.signatures.get_record_stream(
2557
self.source.signatures.keys(),
2559
self.target.revisions.insert_record_stream(
2560
self.source.revisions.get_record_stream(
2561
self.source.revisions.keys(),
2562
'topological', True))
2566
self.target.fetch(self.source, revision_id=revision_id)
2569
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2570
"""See InterRepository.fetch()."""
2571
from bzrlib.fetch import GenericRepoFetcher
2572
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2573
self.source, self.source._format, self.target, self.target._format)
2574
f = GenericRepoFetcher(to_repository=self.target,
2575
from_repository=self.source,
2576
last_revision=revision_id,
2577
pb=pb, find_ghosts=find_ghosts)
2578
return f.count_copied, f.failed_revisions
2581
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2582
"""See InterRepository.missing_revision_ids()."""
2583
# we want all revisions to satisfy revision_id in source.
2584
# but we don't want to stat every file here and there.
2585
# we want then, all revisions other needs to satisfy revision_id
2586
# checked, but not those that we have locally.
2587
# so the first thing is to get a subset of the revisions to
2588
# satisfy revision_id in source, and then eliminate those that
2589
# we do already have.
2590
# this is slow on high latency connection to self, but as as this
2591
# disk format scales terribly for push anyway due to rewriting
2592
# inventory.weave, this is considered acceptable.
2594
if revision_id is not None:
2595
source_ids = self.source.get_ancestry(revision_id)
2596
if source_ids[0] is not None:
2597
raise AssertionError()
2600
source_ids = self.source._all_possible_ids()
2601
source_ids_set = set(source_ids)
2602
# source_ids is the worst possible case we may need to pull.
2603
# now we want to filter source_ids against what we actually
2604
# have in target, but don't try to check for existence where we know
2605
# we do not have a revision as that would be pointless.
2606
target_ids = set(self.target._all_possible_ids())
2607
possibly_present_revisions = target_ids.intersection(source_ids_set)
2608
actually_present_revisions = set(
2609
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2610
required_revisions = source_ids_set.difference(actually_present_revisions)
2611
if revision_id is not None:
2612
# we used get_ancestry to determine source_ids then we are assured all
2613
# revisions referenced are present as they are installed in topological order.
2614
# and the tip revision was validated by get_ancestry.
2615
result_set = required_revisions
2617
# if we just grabbed the possibly available ids, then
2618
# we only have an estimate of whats available and need to validate
2619
# that against the revision records.
2621
self.source._eliminate_revisions_not_present(required_revisions))
2622
return self.source.revision_ids_to_search_result(result_set)
2625
class InterKnitRepo(InterSameDataRepository):
2626
"""Optimised code paths between Knit based repositories."""
2629
def _get_repo_format_to_test(self):
2630
from bzrlib.repofmt import knitrepo
2631
return knitrepo.RepositoryFormatKnit1()
2634
def is_compatible(source, target):
2635
"""Be compatible with known Knit formats.
2637
We don't test for the stores being of specific types because that
2638
could lead to confusing results, and there is no need to be
2641
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2643
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2644
isinstance(target._format, RepositoryFormatKnit))
2645
except AttributeError:
2647
return are_knits and InterRepository._same_model(source, target)
2650
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2651
"""See InterRepository.fetch()."""
2652
from bzrlib.fetch import KnitRepoFetcher
2653
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2654
self.source, self.source._format, self.target, self.target._format)
2655
f = KnitRepoFetcher(to_repository=self.target,
2656
from_repository=self.source,
2657
last_revision=revision_id,
2658
pb=pb, find_ghosts=find_ghosts)
2659
return f.count_copied, f.failed_revisions
2662
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2663
"""See InterRepository.missing_revision_ids()."""
2664
if revision_id is not None:
2665
source_ids = self.source.get_ancestry(revision_id)
2666
if source_ids[0] is not None:
2667
raise AssertionError()
2670
source_ids = self.source.all_revision_ids()
2671
source_ids_set = set(source_ids)
2672
# source_ids is the worst possible case we may need to pull.
2673
# now we want to filter source_ids against what we actually
2674
# have in target, but don't try to check for existence where we know
2675
# we do not have a revision as that would be pointless.
2676
target_ids = set(self.target.all_revision_ids())
2677
possibly_present_revisions = target_ids.intersection(source_ids_set)
2678
actually_present_revisions = set(
2679
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2680
required_revisions = source_ids_set.difference(actually_present_revisions)
2681
if revision_id is not None:
2682
# we used get_ancestry to determine source_ids then we are assured all
2683
# revisions referenced are present as they are installed in topological order.
2684
# and the tip revision was validated by get_ancestry.
2685
result_set = required_revisions
2687
# if we just grabbed the possibly available ids, then
2688
# we only have an estimate of whats available and need to validate
2689
# that against the revision records.
2691
self.source._eliminate_revisions_not_present(required_revisions))
2692
return self.source.revision_ids_to_search_result(result_set)
2695
class InterPackRepo(InterSameDataRepository):
2696
"""Optimised code paths between Pack based repositories."""
2699
def _get_repo_format_to_test(self):
2700
from bzrlib.repofmt import pack_repo
2701
return pack_repo.RepositoryFormatKnitPack1()
2704
def is_compatible(source, target):
2705
"""Be compatible with known Pack formats.
2707
We don't test for the stores being of specific types because that
2708
could lead to confusing results, and there is no need to be
2711
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2713
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2714
isinstance(target._format, RepositoryFormatPack))
2715
except AttributeError:
2717
return are_packs and InterRepository._same_model(source, target)
2720
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2721
"""See InterRepository.fetch()."""
2722
if len(self.source._fallback_repositories) > 0:
2723
from bzrlib.fetch import KnitRepoFetcher
2724
fetcher = KnitRepoFetcher(self.target, self.source, revision_id,
2726
return fetcher.count_copied, fetcher.failed_revisions
2727
from bzrlib.repofmt.pack_repo import Packer
2728
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2729
self.source, self.source._format, self.target, self.target._format)
2730
self.count_copied = 0
2731
if revision_id is None:
2733
# everything to do - use pack logic
2734
# to fetch from all packs to one without
2735
# inventory parsing etc, IFF nothing to be copied is in the target.
2737
source_revision_ids = frozenset(self.source.all_revision_ids())
2738
revision_ids = source_revision_ids - \
2739
frozenset(self.target.get_parent_map(source_revision_ids))
2740
revision_keys = [(revid,) for revid in revision_ids]
2741
index = self.target._pack_collection.revision_index.combined_index
2742
present_revision_ids = set(item[1][0] for item in
2743
index.iter_entries(revision_keys))
2744
revision_ids = set(revision_ids) - present_revision_ids
2745
# implementing the TODO will involve:
2746
# - detecting when all of a pack is selected
2747
# - avoiding as much as possible pre-selection, so the
2748
# more-core routines such as create_pack_from_packs can filter in
2749
# a just-in-time fashion. (though having a HEADS list on a
2750
# repository might make this a lot easier, because we could
2751
# sensibly detect 'new revisions' without doing a full index scan.
2752
elif _mod_revision.is_null(revision_id):
2757
revision_ids = self.search_missing_revision_ids(revision_id,
2758
find_ghosts=find_ghosts).get_keys()
2759
except errors.NoSuchRevision:
2760
raise errors.InstallFailed([revision_id])
2761
if len(revision_ids) == 0:
2763
packs = self.source._pack_collection.all_packs()
2764
pack = Packer(self.target._pack_collection, packs, '.fetch',
2765
revision_ids).pack()
2766
if pack is not None:
2767
self.target._pack_collection._save_pack_names()
2768
# Trigger an autopack. This may duplicate effort as we've just done
2769
# a pack creation, but for now it is simpler to think about as
2770
# 'upload data, then repack if needed'.
2771
self.target._pack_collection.autopack()
2772
return (pack.get_revision_count(), [])
2777
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2778
"""See InterRepository.missing_revision_ids().
2780
:param find_ghosts: Find ghosts throughout the ancestry of
2783
if not find_ghosts and revision_id is not None:
2784
return self._walk_to_common_revisions([revision_id])
2785
elif revision_id is not None:
2786
# Find ghosts: search for revisions pointing from one repository to
2787
# the other, and vice versa, anywhere in the history of revision_id.
2788
graph = self.target.get_graph(other_repository=self.source)
2789
searcher = graph._make_breadth_first_searcher([revision_id])
2793
next_revs, ghosts = searcher.next_with_ghosts()
2794
except StopIteration:
2796
if revision_id in ghosts:
2797
raise errors.NoSuchRevision(self.source, revision_id)
2798
found_ids.update(next_revs)
2799
found_ids.update(ghosts)
2800
found_ids = frozenset(found_ids)
2801
# Double query here: should be able to avoid this by changing the
2802
# graph api further.
2803
result_set = found_ids - frozenset(
2804
self.target.get_parent_map(found_ids))
2806
source_ids = self.source.all_revision_ids()
2807
# source_ids is the worst possible case we may need to pull.
2808
# now we want to filter source_ids against what we actually
2809
# have in target, but don't try to check for existence where we know
2810
# we do not have a revision as that would be pointless.
2811
target_ids = set(self.target.all_revision_ids())
2812
result_set = set(source_ids).difference(target_ids)
2813
return self.source.revision_ids_to_search_result(result_set)
2816
class InterModel1and2(InterRepository):
2819
def _get_repo_format_to_test(self):
2823
def is_compatible(source, target):
2824
if not source.supports_rich_root() and target.supports_rich_root():
2830
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2831
"""See InterRepository.fetch()."""
2832
from bzrlib.fetch import Model1toKnit2Fetcher
2833
f = Model1toKnit2Fetcher(to_repository=self.target,
2834
from_repository=self.source,
2835
last_revision=revision_id,
2836
pb=pb, find_ghosts=find_ghosts)
2837
return f.count_copied, f.failed_revisions
2840
def copy_content(self, revision_id=None):
2841
"""Make a complete copy of the content in self into destination.
2843
This is a destructive operation! Do not use it on existing
2846
:param revision_id: Only copy the content needed to construct
2847
revision_id and its parents.
2850
self.target.set_make_working_trees(self.source.make_working_trees())
2851
except NotImplementedError:
2853
# but don't bother fetching if we have the needed data now.
2854
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2855
self.target.has_revision(revision_id)):
2857
self.target.fetch(self.source, revision_id=revision_id)
2860
class InterKnit1and2(InterKnitRepo):
2863
def _get_repo_format_to_test(self):
2867
def is_compatible(source, target):
2868
"""Be compatible with Knit1 source and Knit3 target"""
2869
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
2871
from bzrlib.repofmt.knitrepo import (RepositoryFormatKnit1,
2872
RepositoryFormatKnit3)
2873
from bzrlib.repofmt.pack_repo import (
2874
RepositoryFormatKnitPack1,
2875
RepositoryFormatKnitPack3,
2876
RepositoryFormatPackDevelopment0,
2877
RepositoryFormatPackDevelopment0Subtree,
2880
RepositoryFormatKnit1,
2881
RepositoryFormatKnitPack1,
2882
RepositoryFormatPackDevelopment0,
2885
RepositoryFormatKnit3,
2886
RepositoryFormatKnitPack3,
2887
RepositoryFormatPackDevelopment0Subtree,
2889
return (isinstance(source._format, nosubtrees) and
2890
isinstance(target._format, subtrees))
2891
except AttributeError:
2895
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2896
"""See InterRepository.fetch()."""
2897
from bzrlib.fetch import Knit1to2Fetcher
2898
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2899
self.source, self.source._format, self.target,
2900
self.target._format)
2901
f = Knit1to2Fetcher(to_repository=self.target,
2902
from_repository=self.source,
2903
last_revision=revision_id,
2904
pb=pb, find_ghosts=find_ghosts)
2905
return f.count_copied, f.failed_revisions
2908
class InterDifferingSerializer(InterKnitRepo):
2911
def _get_repo_format_to_test(self):
2915
def is_compatible(source, target):
2916
"""Be compatible with Knit2 source and Knit3 target"""
2917
if source.supports_rich_root() != target.supports_rich_root():
2919
# Ideally, we'd support fetching if the source had no tree references
2920
# even if it supported them...
2921
if (getattr(source, '_format.supports_tree_reference', False) and
2922
not getattr(target, '_format.supports_tree_reference', False)):
2927
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2928
"""See InterRepository.fetch()."""
2929
revision_ids = self.target.search_missing_revision_ids(self.source,
2930
revision_id, find_ghosts=find_ghosts).get_keys()
2931
revision_ids = tsort.topo_sort(
2932
self.source.get_graph().get_parent_map(revision_ids))
2933
def revisions_iterator():
2934
for current_revision_id in revision_ids:
2935
revision = self.source.get_revision(current_revision_id)
2936
tree = self.source.revision_tree(current_revision_id)
2938
signature = self.source.get_signature_text(
2939
current_revision_id)
2940
except errors.NoSuchRevision:
2942
yield revision, tree, signature
2944
my_pb = ui.ui_factory.nested_progress_bar()
2949
install_revisions(self.target, revisions_iterator(),
2950
len(revision_ids), pb)
2952
if my_pb is not None:
2954
return len(revision_ids), 0
2957
class InterOtherToRemote(InterRepository):
2959
def __init__(self, source, target):
2960
InterRepository.__init__(self, source, target)
2961
self._real_inter = None
2964
def is_compatible(source, target):
2965
if isinstance(target, remote.RemoteRepository):
2969
def _ensure_real_inter(self):
2970
if self._real_inter is None:
2971
self.target._ensure_real()
2972
real_target = self.target._real_repository
2973
self._real_inter = InterRepository.get(self.source, real_target)
2975
def copy_content(self, revision_id=None):
2976
self._ensure_real_inter()
2977
self._real_inter.copy_content(revision_id=revision_id)
2979
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2980
self._ensure_real_inter()
2981
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
2982
find_ghosts=find_ghosts)
2985
def _get_repo_format_to_test(self):
2989
class InterRemoteToOther(InterRepository):
2991
def __init__(self, source, target):
2992
InterRepository.__init__(self, source, target)
2993
self._real_inter = None
2996
def is_compatible(source, target):
2997
if not isinstance(source, remote.RemoteRepository):
2999
# Is source's model compatible with target's model?
3000
source._ensure_real()
3001
real_source = source._real_repository
3002
if isinstance(real_source, remote.RemoteRepository):
3003
raise NotImplementedError(
3004
"We don't support remote repos backed by remote repos yet.")
3005
return InterRepository._same_model(real_source, target)
3007
def _ensure_real_inter(self):
3008
if self._real_inter is None:
3009
self.source._ensure_real()
3010
real_source = self.source._real_repository
3011
self._real_inter = InterRepository.get(real_source, self.target)
3013
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3014
self._ensure_real_inter()
3015
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3016
find_ghosts=find_ghosts)
3018
def copy_content(self, revision_id=None):
3019
self._ensure_real_inter()
3020
self._real_inter.copy_content(revision_id=revision_id)
3023
def _get_repo_format_to_test(self):
3028
InterRepository.register_optimiser(InterDifferingSerializer)
3029
InterRepository.register_optimiser(InterSameDataRepository)
3030
InterRepository.register_optimiser(InterWeaveRepo)
3031
InterRepository.register_optimiser(InterKnitRepo)
3032
InterRepository.register_optimiser(InterModel1and2)
3033
InterRepository.register_optimiser(InterKnit1and2)
3034
InterRepository.register_optimiser(InterPackRepo)
3035
InterRepository.register_optimiser(InterOtherToRemote)
3036
InterRepository.register_optimiser(InterRemoteToOther)
3039
class CopyConverter(object):
3040
"""A repository conversion tool which just performs a copy of the content.
3042
This is slow but quite reliable.
3045
def __init__(self, target_format):
3046
"""Create a CopyConverter.
3048
:param target_format: The format the resulting repository should be.
3050
self.target_format = target_format
3052
def convert(self, repo, pb):
3053
"""Perform the conversion of to_convert, giving feedback via pb.
3055
:param to_convert: The disk object to convert.
3056
:param pb: a progress bar to use for progress information.
3061
# this is only useful with metadir layouts - separated repo content.
3062
# trigger an assertion if not such
3063
repo._format.get_format_string()
3064
self.repo_dir = repo.bzrdir
3065
self.step('Moving repository to repository.backup')
3066
self.repo_dir.transport.move('repository', 'repository.backup')
3067
backup_transport = self.repo_dir.transport.clone('repository.backup')
3068
repo._format.check_conversion_target(self.target_format)
3069
self.source_repo = repo._format.open(self.repo_dir,
3071
_override_transport=backup_transport)
3072
self.step('Creating new repository')
3073
converted = self.target_format.initialize(self.repo_dir,
3074
self.source_repo.is_shared())
3075
converted.lock_write()
3077
self.step('Copying content into repository.')
3078
self.source_repo.copy_content_into(converted)
3081
self.step('Deleting old repository content.')
3082
self.repo_dir.transport.delete_tree('repository.backup')
3083
self.pb.note('repository converted')
3085
def step(self, message):
3086
"""Update the pb by a step."""
3088
self.pb.update(message, self.count, self.total)
3100
def _unescaper(match, _map=_unescape_map):
3101
code = match.group(1)
3105
if not code.startswith('#'):
3107
return unichr(int(code[1:])).encode('utf8')
3113
def _unescape_xml(data):
3114
"""Unescape predefined XML entities in a string of data."""
3116
if _unescape_re is None:
3117
_unescape_re = re.compile('\&([^;]*);')
3118
return _unescape_re.sub(_unescaper, data)
3121
class _VersionedFileChecker(object):
3123
def __init__(self, repository):
3124
self.repository = repository
3125
self.text_index = self.repository._generate_text_key_index()
3127
def calculate_file_version_parents(self, text_key):
3128
"""Calculate the correct parents for a file version according to
3131
parent_keys = self.text_index[text_key]
3132
if parent_keys == [_mod_revision.NULL_REVISION]:
3134
return tuple(parent_keys)
3136
def check_file_version_parents(self, texts, progress_bar=None):
3137
"""Check the parents stored in a versioned file are correct.
3139
It also detects file versions that are not referenced by their
3140
corresponding revision's inventory.
3142
:returns: A tuple of (wrong_parents, dangling_file_versions).
3143
wrong_parents is a dict mapping {revision_id: (stored_parents,
3144
correct_parents)} for each revision_id where the stored parents
3145
are not correct. dangling_file_versions is a set of (file_id,
3146
revision_id) tuples for versions that are present in this versioned
3147
file, but not used by the corresponding inventory.
3150
self.file_ids = set([file_id for file_id, _ in
3151
self.text_index.iterkeys()])
3152
# text keys is now grouped by file_id
3153
n_weaves = len(self.file_ids)
3154
files_in_revisions = {}
3155
revisions_of_files = {}
3156
n_versions = len(self.text_index)
3157
progress_bar.update('loading text store', 0, n_versions)
3158
parent_map = self.repository.texts.get_parent_map(self.text_index)
3159
# On unlistable transports this could well be empty/error...
3160
text_keys = self.repository.texts.keys()
3161
unused_keys = frozenset(text_keys) - set(self.text_index)
3162
for num, key in enumerate(self.text_index.iterkeys()):
3163
if progress_bar is not None:
3164
progress_bar.update('checking text graph', num, n_versions)
3165
correct_parents = self.calculate_file_version_parents(key)
3167
knit_parents = parent_map[key]
3168
except errors.RevisionNotPresent:
3171
if correct_parents != knit_parents:
3172
wrong_parents[key] = (knit_parents, correct_parents)
3173
return wrong_parents, unused_keys
3176
def _old_get_graph(repository, revision_id):
3177
"""DO NOT USE. That is all. I'm serious."""
3178
graph = repository.get_graph()
3179
revision_graph = dict(((key, value) for key, value in
3180
graph.iter_ancestry([revision_id]) if value is not None))
3181
return _strip_NULL_ghosts(revision_graph)
3184
def _strip_NULL_ghosts(revision_graph):
3185
"""Also don't use this. more compatibility code for unmigrated clients."""
3186
# Filter ghosts, and null:
3187
if _mod_revision.NULL_REVISION in revision_graph:
3188
del revision_graph[_mod_revision.NULL_REVISION]
3189
for key, parents in revision_graph.items():
3190
revision_graph[key] = tuple(parent for parent in parents if parent
3192
return revision_graph
added
removed
bzrlib/repository.py
