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
self._check_fallback_repository(repository)
534
self._fallback_repositories.append(repository)
535
self.texts.add_fallback_versioned_files(repository.texts)
536
self.inventories.add_fallback_versioned_files(repository.inventories)
537
self.revisions.add_fallback_versioned_files(repository.revisions)
538
self.signatures.add_fallback_versioned_files(repository.signatures)
540
def _check_fallback_repository(self, repository):
541
"""Check that this repository can fallback to repository safely.
543
Raise an error if not.
545
:param repository: A repository to fallback to.
547
return InterRepository._assert_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 = []
693
# What order should fetch operations request streams in?
694
# The default is topological, because it is the style that is 'most
695
# correct' for targets, even if it is the most expensive for source
696
# formats to generate.
697
self._fetch_order = 'topological'
698
# Does this repository use deltas that can be fetched as-deltas ?
699
# (E.g. knits, where the knit deltas can be transplanted intact.
700
# We default to False, which will ensure that enough data to get
701
# a full text out of any fetch stream will be grabbed.
702
self._fetch_uses_deltas = False
703
# Should fetch trigger a reconcile after the fetch? Only needed for
704
# some repository formats that can suffer internal inconsistencies.
705
self._fetch_reconcile = False
708
return '%s(%r)' % (self.__class__.__name__,
711
def has_same_location(self, other):
712
"""Returns a boolean indicating if this repository is at the same
713
location as another repository.
715
This might return False even when two repository objects are accessing
716
the same physical repository via different URLs.
718
if self.__class__ is not other.__class__:
720
return (self._transport.base == other._transport.base)
722
def is_in_write_group(self):
723
"""Return True if there is an open write group.
725
:seealso: start_write_group.
727
return self._write_group is not None
730
return self.control_files.is_locked()
732
def is_write_locked(self):
733
"""Return True if this object is write locked."""
734
return self.is_locked() and self.control_files._lock_mode == 'w'
736
def lock_write(self, token=None):
737
"""Lock this repository for writing.
739
This causes caching within the repository obejct to start accumlating
740
data during reads, and allows a 'write_group' to be obtained. Write
741
groups must be used for actual data insertion.
743
:param token: if this is already locked, then lock_write will fail
744
unless the token matches the existing lock.
745
:returns: a token if this instance supports tokens, otherwise None.
746
:raises TokenLockingNotSupported: when a token is given but this
747
instance doesn't support using token locks.
748
:raises MismatchedToken: if the specified token doesn't match the token
749
of the existing lock.
750
:seealso: start_write_group.
752
A token should be passed in if you know that you have locked the object
753
some other way, and need to synchronise this object's state with that
756
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
758
result = self.control_files.lock_write(token=token)
759
for repo in self._fallback_repositories:
760
# Writes don't affect fallback repos
766
self.control_files.lock_read()
767
for repo in self._fallback_repositories:
771
def get_physical_lock_status(self):
772
return self.control_files.get_physical_lock_status()
774
def leave_lock_in_place(self):
775
"""Tell this repository not to release the physical lock when this
778
If lock_write doesn't return a token, then this method is not supported.
780
self.control_files.leave_in_place()
782
def dont_leave_lock_in_place(self):
783
"""Tell this repository to release the physical lock when this
784
object is unlocked, even if it didn't originally acquire it.
786
If lock_write doesn't return a token, then this method is not supported.
788
self.control_files.dont_leave_in_place()
791
def gather_stats(self, revid=None, committers=None):
792
"""Gather statistics from a revision id.
794
:param revid: The revision id to gather statistics from, if None, then
795
no revision specific statistics are gathered.
796
:param committers: Optional parameter controlling whether to grab
797
a count of committers from the revision specific statistics.
798
:return: A dictionary of statistics. Currently this contains:
799
committers: The number of committers if requested.
800
firstrev: A tuple with timestamp, timezone for the penultimate left
801
most ancestor of revid, if revid is not the NULL_REVISION.
802
latestrev: A tuple with timestamp, timezone for revid, if revid is
803
not the NULL_REVISION.
804
revisions: The total revision count in the repository.
805
size: An estimate disk size of the repository in bytes.
808
if revid and committers:
809
result['committers'] = 0
810
if revid and revid != _mod_revision.NULL_REVISION:
812
all_committers = set()
813
revisions = self.get_ancestry(revid)
814
# pop the leading None
816
first_revision = None
818
# ignore the revisions in the middle - just grab first and last
819
revisions = revisions[0], revisions[-1]
820
for revision in self.get_revisions(revisions):
821
if not first_revision:
822
first_revision = revision
824
all_committers.add(revision.committer)
825
last_revision = revision
827
result['committers'] = len(all_committers)
828
result['firstrev'] = (first_revision.timestamp,
829
first_revision.timezone)
830
result['latestrev'] = (last_revision.timestamp,
831
last_revision.timezone)
833
# now gather global repository information
834
# XXX: This is available for many repos regardless of listability.
835
if self.bzrdir.root_transport.listable():
836
# XXX: do we want to __define len__() ?
837
# Maybe the versionedfiles object should provide a different
838
# method to get the number of keys.
839
result['revisions'] = len(self.revisions.keys())
843
def find_branches(self, using=False):
844
"""Find branches underneath this repository.
846
This will include branches inside other branches.
848
:param using: If True, list only branches using this repository.
850
if using and not self.is_shared():
852
return [self.bzrdir.open_branch()]
853
except errors.NotBranchError:
855
class Evaluator(object):
858
self.first_call = True
860
def __call__(self, bzrdir):
861
# On the first call, the parameter is always the bzrdir
862
# containing the current repo.
863
if not self.first_call:
865
repository = bzrdir.open_repository()
866
except errors.NoRepositoryPresent:
869
return False, (None, repository)
870
self.first_call = False
872
value = (bzrdir.open_branch(), None)
873
except errors.NotBranchError:
878
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
879
self.bzrdir.root_transport, evaluate=Evaluator()):
880
if branch is not None:
881
branches.append(branch)
882
if not using and repository is not None:
883
branches.extend(repository.find_branches())
887
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
888
"""Return the revision ids that other has that this does not.
890
These are returned in topological order.
892
revision_id: only return revision ids included by revision_id.
894
return InterRepository.get(other, self).search_missing_revision_ids(
895
revision_id, find_ghosts)
897
@deprecated_method(one_two)
899
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
900
"""Return the revision ids that other has that this does not.
902
These are returned in topological order.
904
revision_id: only return revision ids included by revision_id.
906
keys = self.search_missing_revision_ids(
907
other, revision_id, find_ghosts).get_keys()
910
parents = other.get_graph().get_parent_map(keys)
913
return tsort.topo_sort(parents)
917
"""Open the repository rooted at base.
919
For instance, if the repository is at URL/.bzr/repository,
920
Repository.open(URL) -> a Repository instance.
922
control = bzrdir.BzrDir.open(base)
923
return control.open_repository()
925
def copy_content_into(self, destination, revision_id=None):
926
"""Make a complete copy of the content in self into destination.
928
This is a destructive operation! Do not use it on existing
931
return InterRepository.get(self, destination).copy_content(revision_id)
933
def commit_write_group(self):
934
"""Commit the contents accrued within the current write group.
936
:seealso: start_write_group.
938
if self._write_group is not self.get_transaction():
939
# has an unlock or relock occured ?
940
raise errors.BzrError('mismatched lock context %r and '
942
(self.get_transaction(), self._write_group))
943
self._commit_write_group()
944
self._write_group = None
946
def _commit_write_group(self):
947
"""Template method for per-repository write group cleanup.
949
This is called before the write group is considered to be
950
finished and should ensure that all data handed to the repository
951
for writing during the write group is safely committed (to the
952
extent possible considering file system caching etc).
955
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
956
"""Fetch the content required to construct revision_id from source.
958
If revision_id is None all content is copied.
959
:param find_ghosts: Find and copy revisions in the source that are
960
ghosts in the target (and not reachable directly by walking out to
961
the first-present revision in target from revision_id).
963
# fast path same-url fetch operations
964
if self.has_same_location(source):
965
# check that last_revision is in 'from' and then return a
967
if (revision_id is not None and
968
not _mod_revision.is_null(revision_id)):
969
self.get_revision(revision_id)
971
# if there is no specific appropriate InterRepository, this will get
972
# the InterRepository base class, which raises an
973
# IncompatibleRepositories when asked to fetch.
974
inter = InterRepository.get(source, self)
975
return inter.fetch(revision_id=revision_id, pb=pb,
976
find_ghosts=find_ghosts)
978
def create_bundle(self, target, base, fileobj, format=None):
979
return serializer.write_bundle(self, target, base, fileobj, format)
981
def get_commit_builder(self, branch, parents, config, timestamp=None,
982
timezone=None, committer=None, revprops=None,
984
"""Obtain a CommitBuilder for this repository.
986
:param branch: Branch to commit to.
987
:param parents: Revision ids of the parents of the new revision.
988
:param config: Configuration to use.
989
:param timestamp: Optional timestamp recorded for commit.
990
:param timezone: Optional timezone for timestamp.
991
:param committer: Optional committer to set for commit.
992
:param revprops: Optional dictionary of revision properties.
993
:param revision_id: Optional revision id.
995
result = self._commit_builder_class(self, parents, config,
996
timestamp, timezone, committer, revprops, revision_id)
997
self.start_write_group()
1001
if (self.control_files._lock_count == 1 and
1002
self.control_files._lock_mode == 'w'):
1003
if self._write_group is not None:
1004
self.abort_write_group()
1005
self.control_files.unlock()
1006
raise errors.BzrError(
1007
'Must end write groups before releasing write locks.')
1008
self.control_files.unlock()
1009
for repo in self._fallback_repositories:
1013
def clone(self, a_bzrdir, revision_id=None):
1014
"""Clone this repository into a_bzrdir using the current format.
1016
Currently no check is made that the format of this repository and
1017
the bzrdir format are compatible. FIXME RBC 20060201.
1019
:return: The newly created destination repository.
1021
# TODO: deprecate after 0.16; cloning this with all its settings is
1022
# probably not very useful -- mbp 20070423
1023
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1024
self.copy_content_into(dest_repo, revision_id)
1027
def start_write_group(self):
1028
"""Start a write group in the repository.
1030
Write groups are used by repositories which do not have a 1:1 mapping
1031
between file ids and backend store to manage the insertion of data from
1032
both fetch and commit operations.
1034
A write lock is required around the start_write_group/commit_write_group
1035
for the support of lock-requiring repository formats.
1037
One can only insert data into a repository inside a write group.
1041
if not self.is_write_locked():
1042
raise errors.NotWriteLocked(self)
1043
if self._write_group:
1044
raise errors.BzrError('already in a write group')
1045
self._start_write_group()
1046
# so we can detect unlock/relock - the write group is now entered.
1047
self._write_group = self.get_transaction()
1049
def _start_write_group(self):
1050
"""Template method for per-repository write group startup.
1052
This is called before the write group is considered to be
1057
def sprout(self, to_bzrdir, revision_id=None):
1058
"""Create a descendent repository for new development.
1060
Unlike clone, this does not copy the settings of the repository.
1062
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1063
dest_repo.fetch(self, revision_id=revision_id)
1066
def _create_sprouting_repo(self, a_bzrdir, shared):
1067
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1068
# use target default format.
1069
dest_repo = a_bzrdir.create_repository()
1071
# Most control formats need the repository to be specifically
1072
# created, but on some old all-in-one formats it's not needed
1074
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1075
except errors.UninitializableFormat:
1076
dest_repo = a_bzrdir.open_repository()
1080
def has_revision(self, revision_id):
1081
"""True if this repository has a copy of the revision."""
1082
return revision_id in self.has_revisions((revision_id,))
1085
def has_revisions(self, revision_ids):
1086
"""Probe to find out the presence of multiple revisions.
1088
:param revision_ids: An iterable of revision_ids.
1089
:return: A set of the revision_ids that were present.
1091
parent_map = self.revisions.get_parent_map(
1092
[(rev_id,) for rev_id in revision_ids])
1094
if _mod_revision.NULL_REVISION in revision_ids:
1095
result.add(_mod_revision.NULL_REVISION)
1096
result.update([key[0] for key in parent_map])
1100
def get_revision(self, revision_id):
1101
"""Return the Revision object for a named revision."""
1102
return self.get_revisions([revision_id])[0]
1105
def get_revision_reconcile(self, revision_id):
1106
"""'reconcile' helper routine that allows access to a revision always.
1108
This variant of get_revision does not cross check the weave graph
1109
against the revision one as get_revision does: but it should only
1110
be used by reconcile, or reconcile-alike commands that are correcting
1111
or testing the revision graph.
1113
return self._get_revisions([revision_id])[0]
1116
def get_revisions(self, revision_ids):
1117
"""Get many revisions at once."""
1118
return self._get_revisions(revision_ids)
1121
def _get_revisions(self, revision_ids):
1122
"""Core work logic to get many revisions without sanity checks."""
1123
for rev_id in revision_ids:
1124
if not rev_id or not isinstance(rev_id, basestring):
1125
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1126
keys = [(key,) for key in revision_ids]
1127
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1129
for record in stream:
1130
if record.storage_kind == 'absent':
1131
raise errors.NoSuchRevision(self, record.key[0])
1132
text = record.get_bytes_as('fulltext')
1133
rev = self._serializer.read_revision_from_string(text)
1134
revs[record.key[0]] = rev
1135
return [revs[revid] for revid in revision_ids]
1138
def get_revision_xml(self, revision_id):
1139
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1140
# would have already do it.
1141
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1142
rev = self.get_revision(revision_id)
1143
rev_tmp = StringIO()
1144
# the current serializer..
1145
self._serializer.write_revision(rev, rev_tmp)
1147
return rev_tmp.getvalue()
1149
def get_deltas_for_revisions(self, revisions):
1150
"""Produce a generator of revision deltas.
1152
Note that the input is a sequence of REVISIONS, not revision_ids.
1153
Trees will be held in memory until the generator exits.
1154
Each delta is relative to the revision's lefthand predecessor.
1156
required_trees = set()
1157
for revision in revisions:
1158
required_trees.add(revision.revision_id)
1159
required_trees.update(revision.parent_ids[:1])
1160
trees = dict((t.get_revision_id(), t) for
1161
t in self.revision_trees(required_trees))
1162
for revision in revisions:
1163
if not revision.parent_ids:
1164
old_tree = self.revision_tree(None)
1166
old_tree = trees[revision.parent_ids[0]]
1167
yield trees[revision.revision_id].changes_from(old_tree)
1170
def get_revision_delta(self, revision_id):
1171
"""Return the delta for one revision.
1173
The delta is relative to the left-hand predecessor of the
1176
r = self.get_revision(revision_id)
1177
return list(self.get_deltas_for_revisions([r]))[0]
1180
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1181
signature = gpg_strategy.sign(plaintext)
1182
self.add_signature_text(revision_id, signature)
1185
def add_signature_text(self, revision_id, signature):
1186
self.signatures.add_lines((revision_id,), (),
1187
osutils.split_lines(signature))
1189
def find_text_key_references(self):
1190
"""Find the text key references within the repository.
1192
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1193
revision_ids. Each altered file-ids has the exact revision_ids that
1194
altered it listed explicitly.
1195
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1196
to whether they were referred to by the inventory of the
1197
revision_id that they contain. The inventory texts from all present
1198
revision ids are assessed to generate this report.
1200
revision_keys = self.revisions.keys()
1201
w = self.inventories
1202
pb = ui.ui_factory.nested_progress_bar()
1204
return self._find_text_key_references_from_xml_inventory_lines(
1205
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1209
def _find_text_key_references_from_xml_inventory_lines(self,
1211
"""Core routine for extracting references to texts from inventories.
1213
This performs the translation of xml lines to revision ids.
1215
:param line_iterator: An iterator of lines, origin_version_id
1216
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1217
to whether they were referred to by the inventory of the
1218
revision_id that they contain. Note that if that revision_id was
1219
not part of the line_iterator's output then False will be given -
1220
even though it may actually refer to that key.
1222
if not self._serializer.support_altered_by_hack:
1223
raise AssertionError(
1224
"_find_text_key_references_from_xml_inventory_lines only "
1225
"supported for branches which store inventory as unnested xml"
1226
", not on %r" % self)
1229
# this code needs to read every new line in every inventory for the
1230
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1231
# not present in one of those inventories is unnecessary but not
1232
# harmful because we are filtering by the revision id marker in the
1233
# inventory lines : we only select file ids altered in one of those
1234
# revisions. We don't need to see all lines in the inventory because
1235
# only those added in an inventory in rev X can contain a revision=X
1237
unescape_revid_cache = {}
1238
unescape_fileid_cache = {}
1240
# jam 20061218 In a big fetch, this handles hundreds of thousands
1241
# of lines, so it has had a lot of inlining and optimizing done.
1242
# Sorry that it is a little bit messy.
1243
# Move several functions to be local variables, since this is a long
1245
search = self._file_ids_altered_regex.search
1246
unescape = _unescape_xml
1247
setdefault = result.setdefault
1248
for line, line_key in line_iterator:
1249
match = search(line)
1252
# One call to match.group() returning multiple items is quite a
1253
# bit faster than 2 calls to match.group() each returning 1
1254
file_id, revision_id = match.group('file_id', 'revision_id')
1256
# Inlining the cache lookups helps a lot when you make 170,000
1257
# lines and 350k ids, versus 8.4 unique ids.
1258
# Using a cache helps in 2 ways:
1259
# 1) Avoids unnecessary decoding calls
1260
# 2) Re-uses cached strings, which helps in future set and
1262
# (2) is enough that removing encoding entirely along with
1263
# the cache (so we are using plain strings) results in no
1264
# performance improvement.
1266
revision_id = unescape_revid_cache[revision_id]
1268
unescaped = unescape(revision_id)
1269
unescape_revid_cache[revision_id] = unescaped
1270
revision_id = unescaped
1272
# Note that unconditionally unescaping means that we deserialise
1273
# every fileid, which for general 'pull' is not great, but we don't
1274
# really want to have some many fulltexts that this matters anyway.
1277
file_id = unescape_fileid_cache[file_id]
1279
unescaped = unescape(file_id)
1280
unescape_fileid_cache[file_id] = unescaped
1283
key = (file_id, revision_id)
1284
setdefault(key, False)
1285
if revision_id == line_key[-1]:
1289
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1291
"""Helper routine for fileids_altered_by_revision_ids.
1293
This performs the translation of xml lines to revision ids.
1295
:param line_iterator: An iterator of lines, origin_version_id
1296
:param revision_ids: The revision ids to filter for. This should be a
1297
set or other type which supports efficient __contains__ lookups, as
1298
the revision id from each parsed line will be looked up in the
1299
revision_ids filter.
1300
:return: a dictionary mapping altered file-ids to an iterable of
1301
revision_ids. Each altered file-ids has the exact revision_ids that
1302
altered it listed explicitly.
1305
setdefault = result.setdefault
1307
self._find_text_key_references_from_xml_inventory_lines(
1308
line_iterator).iterkeys():
1309
# once data is all ensured-consistent; then this is
1310
# if revision_id == version_id
1311
if key[-1:] in revision_ids:
1312
setdefault(key[0], set()).add(key[-1])
1315
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1316
"""Find the file ids and versions affected by revisions.
1318
:param revisions: an iterable containing revision ids.
1319
:param _inv_weave: The inventory weave from this repository or None.
1320
If None, the inventory weave will be opened automatically.
1321
:return: a dictionary mapping altered file-ids to an iterable of
1322
revision_ids. Each altered file-ids has the exact revision_ids that
1323
altered it listed explicitly.
1325
selected_keys = set((revid,) for revid in revision_ids)
1326
w = _inv_weave or self.inventories
1327
pb = ui.ui_factory.nested_progress_bar()
1329
return self._find_file_ids_from_xml_inventory_lines(
1330
w.iter_lines_added_or_present_in_keys(
1331
selected_keys, pb=pb),
1336
def iter_files_bytes(self, desired_files):
1337
"""Iterate through file versions.
1339
Files will not necessarily be returned in the order they occur in
1340
desired_files. No specific order is guaranteed.
1342
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1343
value supplied by the caller as part of desired_files. It should
1344
uniquely identify the file version in the caller's context. (Examples:
1345
an index number or a TreeTransform trans_id.)
1347
bytes_iterator is an iterable of bytestrings for the file. The
1348
kind of iterable and length of the bytestrings are unspecified, but for
1349
this implementation, it is a list of bytes produced by
1350
VersionedFile.get_record_stream().
1352
:param desired_files: a list of (file_id, revision_id, identifier)
1355
transaction = self.get_transaction()
1357
for file_id, revision_id, callable_data in desired_files:
1358
text_keys[(file_id, revision_id)] = callable_data
1359
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1360
if record.storage_kind == 'absent':
1361
raise errors.RevisionNotPresent(record.key, self)
1362
yield text_keys[record.key], record.get_bytes_as('fulltext')
1364
def _generate_text_key_index(self, text_key_references=None,
1366
"""Generate a new text key index for the repository.
1368
This is an expensive function that will take considerable time to run.
1370
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1371
list of parents, also text keys. When a given key has no parents,
1372
the parents list will be [NULL_REVISION].
1374
# All revisions, to find inventory parents.
1375
if ancestors is None:
1376
graph = self.get_graph()
1377
ancestors = graph.get_parent_map(self.all_revision_ids())
1378
if text_key_references is None:
1379
text_key_references = self.find_text_key_references()
1380
pb = ui.ui_factory.nested_progress_bar()
1382
return self._do_generate_text_key_index(ancestors,
1383
text_key_references, pb)
1387
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1388
"""Helper for _generate_text_key_index to avoid deep nesting."""
1389
revision_order = tsort.topo_sort(ancestors)
1390
invalid_keys = set()
1392
for revision_id in revision_order:
1393
revision_keys[revision_id] = set()
1394
text_count = len(text_key_references)
1395
# a cache of the text keys to allow reuse; costs a dict of all the
1396
# keys, but saves a 2-tuple for every child of a given key.
1398
for text_key, valid in text_key_references.iteritems():
1400
invalid_keys.add(text_key)
1402
revision_keys[text_key[1]].add(text_key)
1403
text_key_cache[text_key] = text_key
1404
del text_key_references
1406
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1407
NULL_REVISION = _mod_revision.NULL_REVISION
1408
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1409
# too small for large or very branchy trees. However, for 55K path
1410
# trees, it would be easy to use too much memory trivially. Ideally we
1411
# could gauge this by looking at available real memory etc, but this is
1412
# always a tricky proposition.
1413
inventory_cache = lru_cache.LRUCache(10)
1414
batch_size = 10 # should be ~150MB on a 55K path tree
1415
batch_count = len(revision_order) / batch_size + 1
1417
pb.update("Calculating text parents.", processed_texts, text_count)
1418
for offset in xrange(batch_count):
1419
to_query = revision_order[offset * batch_size:(offset + 1) *
1423
for rev_tree in self.revision_trees(to_query):
1424
revision_id = rev_tree.get_revision_id()
1425
parent_ids = ancestors[revision_id]
1426
for text_key in revision_keys[revision_id]:
1427
pb.update("Calculating text parents.", processed_texts)
1428
processed_texts += 1
1429
candidate_parents = []
1430
for parent_id in parent_ids:
1431
parent_text_key = (text_key[0], parent_id)
1433
check_parent = parent_text_key not in \
1434
revision_keys[parent_id]
1436
# the parent parent_id is a ghost:
1437
check_parent = False
1438
# truncate the derived graph against this ghost.
1439
parent_text_key = None
1441
# look at the parent commit details inventories to
1442
# determine possible candidates in the per file graph.
1445
inv = inventory_cache[parent_id]
1447
inv = self.revision_tree(parent_id).inventory
1448
inventory_cache[parent_id] = inv
1449
parent_entry = inv._byid.get(text_key[0], None)
1450
if parent_entry is not None:
1452
text_key[0], parent_entry.revision)
1454
parent_text_key = None
1455
if parent_text_key is not None:
1456
candidate_parents.append(
1457
text_key_cache[parent_text_key])
1458
parent_heads = text_graph.heads(candidate_parents)
1459
new_parents = list(parent_heads)
1460
new_parents.sort(key=lambda x:candidate_parents.index(x))
1461
if new_parents == []:
1462
new_parents = [NULL_REVISION]
1463
text_index[text_key] = new_parents
1465
for text_key in invalid_keys:
1466
text_index[text_key] = [NULL_REVISION]
1469
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1470
"""Get an iterable listing the keys of all the data introduced by a set
1473
The keys will be ordered so that the corresponding items can be safely
1474
fetched and inserted in that order.
1476
:returns: An iterable producing tuples of (knit-kind, file-id,
1477
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1478
'revisions'. file-id is None unless knit-kind is 'file'.
1480
# XXX: it's a bit weird to control the inventory weave caching in this
1481
# generator. Ideally the caching would be done in fetch.py I think. Or
1482
# maybe this generator should explicitly have the contract that it
1483
# should not be iterated until the previously yielded item has been
1485
inv_w = self.inventories
1487
# file ids that changed
1488
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1490
num_file_ids = len(file_ids)
1491
for file_id, altered_versions in file_ids.iteritems():
1492
if _files_pb is not None:
1493
_files_pb.update("fetch texts", count, num_file_ids)
1495
yield ("file", file_id, altered_versions)
1496
# We're done with the files_pb. Note that it finished by the caller,
1497
# just as it was created by the caller.
1501
yield ("inventory", None, revision_ids)
1504
revisions_with_signatures = set()
1505
for rev_id in revision_ids:
1507
self.get_signature_text(rev_id)
1508
except errors.NoSuchRevision:
1512
revisions_with_signatures.add(rev_id)
1513
yield ("signatures", None, revisions_with_signatures)
1516
yield ("revisions", None, revision_ids)
1519
def get_inventory(self, revision_id):
1520
"""Get Inventory object by revision id."""
1521
return self.iter_inventories([revision_id]).next()
1523
def iter_inventories(self, revision_ids):
1524
"""Get many inventories by revision_ids.
1526
This will buffer some or all of the texts used in constructing the
1527
inventories in memory, but will only parse a single inventory at a
1530
:return: An iterator of inventories.
1532
if ((None in revision_ids)
1533
or (_mod_revision.NULL_REVISION in revision_ids)):
1534
raise ValueError('cannot get null revision inventory')
1535
return self._iter_inventories(revision_ids)
1537
def _iter_inventories(self, revision_ids):
1538
"""single-document based inventory iteration."""
1539
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1540
yield self.deserialise_inventory(revision_id, text)
1542
def _iter_inventory_xmls(self, revision_ids):
1543
keys = [(revision_id,) for revision_id in revision_ids]
1544
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1546
for record in stream:
1547
if record.storage_kind != 'absent':
1548
texts[record.key] = record.get_bytes_as('fulltext')
1550
raise errors.NoSuchRevision(self, record.key)
1552
yield texts[key], key[-1]
1554
def deserialise_inventory(self, revision_id, xml):
1555
"""Transform the xml into an inventory object.
1557
:param revision_id: The expected revision id of the inventory.
1558
:param xml: A serialised inventory.
1560
result = self._serializer.read_inventory_from_string(xml, revision_id)
1561
if result.revision_id != revision_id:
1562
raise AssertionError('revision id mismatch %s != %s' % (
1563
result.revision_id, revision_id))
1566
def serialise_inventory(self, inv):
1567
return self._serializer.write_inventory_to_string(inv)
1569
def _serialise_inventory_to_lines(self, inv):
1570
return self._serializer.write_inventory_to_lines(inv)
1572
def get_serializer_format(self):
1573
return self._serializer.format_num
1576
def get_inventory_xml(self, revision_id):
1577
"""Get inventory XML as a file object."""
1578
texts = self._iter_inventory_xmls([revision_id])
1580
text, revision_id = texts.next()
1581
except StopIteration:
1582
raise errors.HistoryMissing(self, 'inventory', revision_id)
1586
def get_inventory_sha1(self, revision_id):
1587
"""Return the sha1 hash of the inventory entry
1589
return self.get_revision(revision_id).inventory_sha1
1591
def iter_reverse_revision_history(self, revision_id):
1592
"""Iterate backwards through revision ids in the lefthand history
1594
:param revision_id: The revision id to start with. All its lefthand
1595
ancestors will be traversed.
1597
graph = self.get_graph()
1598
next_id = revision_id
1600
if next_id in (None, _mod_revision.NULL_REVISION):
1603
# Note: The following line may raise KeyError in the event of
1604
# truncated history. We decided not to have a try:except:raise
1605
# RevisionNotPresent here until we see a use for it, because of the
1606
# cost in an inner loop that is by its very nature O(history).
1607
# Robert Collins 20080326
1608
parents = graph.get_parent_map([next_id])[next_id]
1609
if len(parents) == 0:
1612
next_id = parents[0]
1615
def get_revision_inventory(self, revision_id):
1616
"""Return inventory of a past revision."""
1617
# TODO: Unify this with get_inventory()
1618
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1619
# must be the same as its revision, so this is trivial.
1620
if revision_id is None:
1621
# This does not make sense: if there is no revision,
1622
# then it is the current tree inventory surely ?!
1623
# and thus get_root_id() is something that looks at the last
1624
# commit on the branch, and the get_root_id is an inventory check.
1625
raise NotImplementedError
1626
# return Inventory(self.get_root_id())
1628
return self.get_inventory(revision_id)
1630
def is_shared(self):
1631
"""Return True if this repository is flagged as a shared repository."""
1632
raise NotImplementedError(self.is_shared)
1635
def reconcile(self, other=None, thorough=False):
1636
"""Reconcile this repository."""
1637
from bzrlib.reconcile import RepoReconciler
1638
reconciler = RepoReconciler(self, thorough=thorough)
1639
reconciler.reconcile()
1642
def _refresh_data(self):
1643
"""Helper called from lock_* to ensure coherency with disk.
1645
The default implementation does nothing; it is however possible
1646
for repositories to maintain loaded indices across multiple locks
1647
by checking inside their implementation of this method to see
1648
whether their indices are still valid. This depends of course on
1649
the disk format being validatable in this manner.
1653
def revision_tree(self, revision_id):
1654
"""Return Tree for a revision on this branch.
1656
`revision_id` may be None for the empty tree revision.
1658
# TODO: refactor this to use an existing revision object
1659
# so we don't need to read it in twice.
1660
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1661
return RevisionTree(self, Inventory(root_id=None),
1662
_mod_revision.NULL_REVISION)
1664
inv = self.get_revision_inventory(revision_id)
1665
return RevisionTree(self, inv, revision_id)
1667
def revision_trees(self, revision_ids):
1668
"""Return Tree for a revision on this branch.
1670
`revision_id` may not be None or 'null:'"""
1671
inventories = self.iter_inventories(revision_ids)
1672
for inv in inventories:
1673
yield RevisionTree(self, inv, inv.revision_id)
1676
def get_ancestry(self, revision_id, topo_sorted=True):
1677
"""Return a list of revision-ids integrated by a revision.
1679
The first element of the list is always None, indicating the origin
1680
revision. This might change when we have history horizons, or
1681
perhaps we should have a new API.
1683
This is topologically sorted.
1685
if _mod_revision.is_null(revision_id):
1687
if not self.has_revision(revision_id):
1688
raise errors.NoSuchRevision(self, revision_id)
1689
graph = self.get_graph()
1691
search = graph._make_breadth_first_searcher([revision_id])
1694
found, ghosts = search.next_with_ghosts()
1695
except StopIteration:
1698
if _mod_revision.NULL_REVISION in keys:
1699
keys.remove(_mod_revision.NULL_REVISION)
1701
parent_map = graph.get_parent_map(keys)
1702
keys = tsort.topo_sort(parent_map)
1703
return [None] + list(keys)
1706
"""Compress the data within the repository.
1708
This operation only makes sense for some repository types. For other
1709
types it should be a no-op that just returns.
1711
This stub method does not require a lock, but subclasses should use
1712
@needs_write_lock as this is a long running call its reasonable to
1713
implicitly lock for the user.
1717
@deprecated_method(one_six)
1718
def print_file(self, file, revision_id):
1719
"""Print `file` to stdout.
1721
FIXME RBC 20060125 as John Meinel points out this is a bad api
1722
- it writes to stdout, it assumes that that is valid etc. Fix
1723
by creating a new more flexible convenience function.
1725
tree = self.revision_tree(revision_id)
1726
# use inventory as it was in that revision
1727
file_id = tree.inventory.path2id(file)
1729
# TODO: jam 20060427 Write a test for this code path
1730
# it had a bug in it, and was raising the wrong
1732
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1733
tree.print_file(file_id)
1735
def get_transaction(self):
1736
return self.control_files.get_transaction()
1738
@deprecated_method(one_one)
1739
def get_parents(self, revision_ids):
1740
"""See StackedParentsProvider.get_parents"""
1741
parent_map = self.get_parent_map(revision_ids)
1742
return [parent_map.get(r, None) for r in revision_ids]
1744
def get_parent_map(self, revision_ids):
1745
"""See graph._StackedParentsProvider.get_parent_map"""
1746
# revisions index works in keys; this just works in revisions
1747
# therefore wrap and unwrap
1750
for revision_id in revision_ids:
1751
if revision_id == _mod_revision.NULL_REVISION:
1752
result[revision_id] = ()
1753
elif revision_id is None:
1754
raise ValueError('get_parent_map(None) is not valid')
1756
query_keys.append((revision_id ,))
1757
for ((revision_id,), parent_keys) in \
1758
self.revisions.get_parent_map(query_keys).iteritems():
1760
result[revision_id] = tuple(parent_revid
1761
for (parent_revid,) in parent_keys)
1763
result[revision_id] = (_mod_revision.NULL_REVISION,)
1766
def _make_parents_provider(self):
1769
def get_graph(self, other_repository=None):
1770
"""Return the graph walker for this repository format"""
1771
parents_provider = self._make_parents_provider()
1772
if (other_repository is not None and
1773
not self.has_same_location(other_repository)):
1774
parents_provider = graph._StackedParentsProvider(
1775
[parents_provider, other_repository._make_parents_provider()])
1776
return graph.Graph(parents_provider)
1778
def _get_versioned_file_checker(self):
1779
"""Return an object suitable for checking versioned files."""
1780
return _VersionedFileChecker(self)
1782
def revision_ids_to_search_result(self, result_set):
1783
"""Convert a set of revision ids to a graph SearchResult."""
1784
result_parents = set()
1785
for parents in self.get_graph().get_parent_map(
1786
result_set).itervalues():
1787
result_parents.update(parents)
1788
included_keys = result_set.intersection(result_parents)
1789
start_keys = result_set.difference(included_keys)
1790
exclude_keys = result_parents.difference(result_set)
1791
result = graph.SearchResult(start_keys, exclude_keys,
1792
len(result_set), result_set)
1796
def set_make_working_trees(self, new_value):
1797
"""Set the policy flag for making working trees when creating branches.
1799
This only applies to branches that use this repository.
1801
The default is 'True'.
1802
:param new_value: True to restore the default, False to disable making
1805
raise NotImplementedError(self.set_make_working_trees)
1807
def make_working_trees(self):
1808
"""Returns the policy for making working trees on new branches."""
1809
raise NotImplementedError(self.make_working_trees)
1812
def sign_revision(self, revision_id, gpg_strategy):
1813
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1814
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1817
def has_signature_for_revision_id(self, revision_id):
1818
"""Query for a revision signature for revision_id in the repository."""
1819
if not self.has_revision(revision_id):
1820
raise errors.NoSuchRevision(self, revision_id)
1821
sig_present = (1 == len(
1822
self.signatures.get_parent_map([(revision_id,)])))
1826
def get_signature_text(self, revision_id):
1827
"""Return the text for a signature."""
1828
stream = self.signatures.get_record_stream([(revision_id,)],
1830
record = stream.next()
1831
if record.storage_kind == 'absent':
1832
raise errors.NoSuchRevision(self, revision_id)
1833
return record.get_bytes_as('fulltext')
1836
def check(self, revision_ids=None):
1837
"""Check consistency of all history of given revision_ids.
1839
Different repository implementations should override _check().
1841
:param revision_ids: A non-empty list of revision_ids whose ancestry
1842
will be checked. Typically the last revision_id of a branch.
1844
return self._check(revision_ids)
1846
def _check(self, revision_ids):
1847
result = check.Check(self)
1851
def _warn_if_deprecated(self):
1852
global _deprecation_warning_done
1853
if _deprecation_warning_done:
1855
_deprecation_warning_done = True
1856
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1857
% (self._format, self.bzrdir.transport.base))
1859
def supports_rich_root(self):
1860
return self._format.rich_root_data
1862
def _check_ascii_revisionid(self, revision_id, method):
1863
"""Private helper for ascii-only repositories."""
1864
# weave repositories refuse to store revisionids that are non-ascii.
1865
if revision_id is not None:
1866
# weaves require ascii revision ids.
1867
if isinstance(revision_id, unicode):
1869
revision_id.encode('ascii')
1870
except UnicodeEncodeError:
1871
raise errors.NonAsciiRevisionId(method, self)
1874
revision_id.decode('ascii')
1875
except UnicodeDecodeError:
1876
raise errors.NonAsciiRevisionId(method, self)
1878
def revision_graph_can_have_wrong_parents(self):
1879
"""Is it possible for this repository to have a revision graph with
1882
If True, then this repository must also implement
1883
_find_inconsistent_revision_parents so that check and reconcile can
1884
check for inconsistencies before proceeding with other checks that may
1885
depend on the revision index being consistent.
1887
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1890
# remove these delegates a while after bzr 0.15
1891
def __make_delegated(name, from_module):
1892
def _deprecated_repository_forwarder():
1893
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1894
% (name, from_module),
1897
m = __import__(from_module, globals(), locals(), [name])
1899
return getattr(m, name)
1900
except AttributeError:
1901
raise AttributeError('module %s has no name %s'
1903
globals()[name] = _deprecated_repository_forwarder
1906
'AllInOneRepository',
1907
'WeaveMetaDirRepository',
1908
'PreSplitOutRepositoryFormat',
1909
'RepositoryFormat4',
1910
'RepositoryFormat5',
1911
'RepositoryFormat6',
1912
'RepositoryFormat7',
1914
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1918
'RepositoryFormatKnit',
1919
'RepositoryFormatKnit1',
1921
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1924
def install_revision(repository, rev, revision_tree):
1925
"""Install all revision data into a repository."""
1926
install_revisions(repository, [(rev, revision_tree, None)])
1929
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1930
"""Install all revision data into a repository.
1932
Accepts an iterable of revision, tree, signature tuples. The signature
1935
repository.start_write_group()
1937
for n, (revision, revision_tree, signature) in enumerate(iterable):
1938
_install_revision(repository, revision, revision_tree, signature)
1940
pb.update('Transferring revisions', n + 1, num_revisions)
1942
repository.abort_write_group()
1945
repository.commit_write_group()
1948
def _install_revision(repository, rev, revision_tree, signature):
1949
"""Install all revision data into a repository."""
1950
present_parents = []
1952
for p_id in rev.parent_ids:
1953
if repository.has_revision(p_id):
1954
present_parents.append(p_id)
1955
parent_trees[p_id] = repository.revision_tree(p_id)
1957
parent_trees[p_id] = repository.revision_tree(None)
1959
inv = revision_tree.inventory
1960
entries = inv.iter_entries()
1961
# backwards compatibility hack: skip the root id.
1962
if not repository.supports_rich_root():
1963
path, root = entries.next()
1964
if root.revision != rev.revision_id:
1965
raise errors.IncompatibleRevision(repr(repository))
1967
for path, ie in entries:
1968
text_keys[(ie.file_id, ie.revision)] = ie
1969
text_parent_map = repository.texts.get_parent_map(text_keys)
1970
missing_texts = set(text_keys) - set(text_parent_map)
1971
# Add the texts that are not already present
1972
for text_key in missing_texts:
1973
ie = text_keys[text_key]
1975
# FIXME: TODO: The following loop overlaps/duplicates that done by
1976
# commit to determine parents. There is a latent/real bug here where
1977
# the parents inserted are not those commit would do - in particular
1978
# they are not filtered by heads(). RBC, AB
1979
for revision, tree in parent_trees.iteritems():
1980
if ie.file_id not in tree:
1982
parent_id = tree.inventory[ie.file_id].revision
1983
if parent_id in text_parents:
1985
text_parents.append((ie.file_id, parent_id))
1986
lines = revision_tree.get_file(ie.file_id).readlines()
1987
repository.texts.add_lines(text_key, text_parents, lines)
1989
# install the inventory
1990
repository.add_inventory(rev.revision_id, inv, present_parents)
1991
except errors.RevisionAlreadyPresent:
1993
if signature is not None:
1994
repository.add_signature_text(rev.revision_id, signature)
1995
repository.add_revision(rev.revision_id, rev, inv)
1998
class MetaDirRepository(Repository):
1999
"""Repositories in the new meta-dir layout.
2001
:ivar _transport: Transport for access to repository control files,
2002
typically pointing to .bzr/repository.
2005
def __init__(self, _format, a_bzrdir, control_files):
2006
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2007
self._transport = control_files._transport
2009
def is_shared(self):
2010
"""Return True if this repository is flagged as a shared repository."""
2011
return self._transport.has('shared-storage')
2014
def set_make_working_trees(self, new_value):
2015
"""Set the policy flag for making working trees when creating branches.
2017
This only applies to branches that use this repository.
2019
The default is 'True'.
2020
:param new_value: True to restore the default, False to disable making
2025
self._transport.delete('no-working-trees')
2026
except errors.NoSuchFile:
2029
self._transport.put_bytes('no-working-trees', '',
2030
mode=self.bzrdir._get_file_mode())
2032
def make_working_trees(self):
2033
"""Returns the policy for making working trees on new branches."""
2034
return not self._transport.has('no-working-trees')
2037
class MetaDirVersionedFileRepository(MetaDirRepository):
2038
"""Repositories in a meta-dir, that work via versioned file objects."""
2040
def __init__(self, _format, a_bzrdir, control_files):
2041
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2045
class RepositoryFormatRegistry(registry.Registry):
2046
"""Registry of RepositoryFormats."""
2048
def get(self, format_string):
2049
r = registry.Registry.get(self, format_string)
2055
format_registry = RepositoryFormatRegistry()
2056
"""Registry of formats, indexed by their identifying format string.
2058
This can contain either format instances themselves, or classes/factories that
2059
can be called to obtain one.
2063
#####################################################################
2064
# Repository Formats
2066
class RepositoryFormat(object):
2067
"""A repository format.
2069
Formats provide three things:
2070
* An initialization routine to construct repository data on disk.
2071
* a format string which is used when the BzrDir supports versioned
2073
* an open routine which returns a Repository instance.
2075
There is one and only one Format subclass for each on-disk format. But
2076
there can be one Repository subclass that is used for several different
2077
formats. The _format attribute on a Repository instance can be used to
2078
determine the disk format.
2080
Formats are placed in an dict by their format string for reference
2081
during opening. These should be subclasses of RepositoryFormat
2084
Once a format is deprecated, just deprecate the initialize and open
2085
methods on the format class. Do not deprecate the object, as the
2086
object will be created every system load.
2088
Common instance attributes:
2089
_matchingbzrdir - the bzrdir format that the repository format was
2090
originally written to work with. This can be used if manually
2091
constructing a bzrdir and repository, or more commonly for test suite
2095
# Set to True or False in derived classes. True indicates that the format
2096
# supports ghosts gracefully.
2097
supports_ghosts = None
2098
# Can this repository be given external locations to lookup additional
2099
# data. Set to True or False in derived classes.
2100
supports_external_lookups = None
2103
return "<%s>" % self.__class__.__name__
2105
def __eq__(self, other):
2106
# format objects are generally stateless
2107
return isinstance(other, self.__class__)
2109
def __ne__(self, other):
2110
return not self == other
2113
def find_format(klass, a_bzrdir):
2114
"""Return the format for the repository object in a_bzrdir.
2116
This is used by bzr native formats that have a "format" file in
2117
the repository. Other methods may be used by different types of
2121
transport = a_bzrdir.get_repository_transport(None)
2122
format_string = transport.get("format").read()
2123
return format_registry.get(format_string)
2124
except errors.NoSuchFile:
2125
raise errors.NoRepositoryPresent(a_bzrdir)
2127
raise errors.UnknownFormatError(format=format_string,
2131
def register_format(klass, format):
2132
format_registry.register(format.get_format_string(), format)
2135
def unregister_format(klass, format):
2136
format_registry.remove(format.get_format_string())
2139
def get_default_format(klass):
2140
"""Return the current default format."""
2141
from bzrlib import bzrdir
2142
return bzrdir.format_registry.make_bzrdir('default').repository_format
2144
def get_format_string(self):
2145
"""Return the ASCII format string that identifies this format.
2147
Note that in pre format ?? repositories the format string is
2148
not permitted nor written to disk.
2150
raise NotImplementedError(self.get_format_string)
2152
def get_format_description(self):
2153
"""Return the short description for this format."""
2154
raise NotImplementedError(self.get_format_description)
2156
# TODO: this shouldn't be in the base class, it's specific to things that
2157
# use weaves or knits -- mbp 20070207
2158
def _get_versioned_file_store(self,
2163
versionedfile_class=None,
2164
versionedfile_kwargs={},
2166
if versionedfile_class is None:
2167
versionedfile_class = self._versionedfile_class
2168
weave_transport = control_files._transport.clone(name)
2169
dir_mode = control_files._dir_mode
2170
file_mode = control_files._file_mode
2171
return VersionedFileStore(weave_transport, prefixed=prefixed,
2173
file_mode=file_mode,
2174
versionedfile_class=versionedfile_class,
2175
versionedfile_kwargs=versionedfile_kwargs,
2178
def initialize(self, a_bzrdir, shared=False):
2179
"""Initialize a repository of this format in a_bzrdir.
2181
:param a_bzrdir: The bzrdir to put the new repository in it.
2182
:param shared: The repository should be initialized as a sharable one.
2183
:returns: The new repository object.
2185
This may raise UninitializableFormat if shared repository are not
2186
compatible the a_bzrdir.
2188
raise NotImplementedError(self.initialize)
2190
def is_supported(self):
2191
"""Is this format supported?
2193
Supported formats must be initializable and openable.
2194
Unsupported formats may not support initialization or committing or
2195
some other features depending on the reason for not being supported.
2199
def check_conversion_target(self, target_format):
2200
raise NotImplementedError(self.check_conversion_target)
2202
def open(self, a_bzrdir, _found=False):
2203
"""Return an instance of this format for the bzrdir a_bzrdir.
2205
_found is a private parameter, do not use it.
2207
raise NotImplementedError(self.open)
2210
class MetaDirRepositoryFormat(RepositoryFormat):
2211
"""Common base class for the new repositories using the metadir layout."""
2213
rich_root_data = False
2214
supports_tree_reference = False
2215
supports_external_lookups = False
2216
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2219
super(MetaDirRepositoryFormat, self).__init__()
2221
def _create_control_files(self, a_bzrdir):
2222
"""Create the required files and the initial control_files object."""
2223
# FIXME: RBC 20060125 don't peek under the covers
2224
# NB: no need to escape relative paths that are url safe.
2225
repository_transport = a_bzrdir.get_repository_transport(self)
2226
control_files = lockable_files.LockableFiles(repository_transport,
2227
'lock', lockdir.LockDir)
2228
control_files.create_lock()
2229
return control_files
2231
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
2232
"""Upload the initial blank content."""
2233
control_files = self._create_control_files(a_bzrdir)
2234
control_files.lock_write()
2235
transport = control_files._transport
2237
utf8_files += [('shared-storage', '')]
2239
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2240
for (filename, content_stream) in files:
2241
transport.put_file(filename, content_stream,
2242
mode=a_bzrdir._get_file_mode())
2243
for (filename, content_bytes) in utf8_files:
2244
transport.put_bytes_non_atomic(filename, content_bytes,
2245
mode=a_bzrdir._get_file_mode())
2247
control_files.unlock()
2250
# formats which have no format string are not discoverable
2251
# and not independently creatable, so are not registered. They're
2252
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2253
# needed, it's constructed directly by the BzrDir. Non-native formats where
2254
# the repository is not separately opened are similar.
2256
format_registry.register_lazy(
2257
'Bazaar-NG Repository format 7',
2258
'bzrlib.repofmt.weaverepo',
2262
format_registry.register_lazy(
2263
'Bazaar-NG Knit Repository Format 1',
2264
'bzrlib.repofmt.knitrepo',
2265
'RepositoryFormatKnit1',
2268
format_registry.register_lazy(
2269
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2270
'bzrlib.repofmt.knitrepo',
2271
'RepositoryFormatKnit3',
2274
format_registry.register_lazy(
2275
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2276
'bzrlib.repofmt.knitrepo',
2277
'RepositoryFormatKnit4',
2280
# Pack-based formats. There is one format for pre-subtrees, and one for
2281
# post-subtrees to allow ease of testing.
2282
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2283
format_registry.register_lazy(
2284
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2285
'bzrlib.repofmt.pack_repo',
2286
'RepositoryFormatKnitPack1',
2288
format_registry.register_lazy(
2289
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2290
'bzrlib.repofmt.pack_repo',
2291
'RepositoryFormatKnitPack3',
2293
format_registry.register_lazy(
2294
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2295
'bzrlib.repofmt.pack_repo',
2296
'RepositoryFormatKnitPack4',
2298
format_registry.register_lazy(
2299
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2300
'bzrlib.repofmt.pack_repo',
2301
'RepositoryFormatKnitPack5',
2303
format_registry.register_lazy(
2304
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2305
'bzrlib.repofmt.pack_repo',
2306
'RepositoryFormatKnitPack5RichRoot',
2309
# Development formats.
2311
# development 0 - stub to introduce development versioning scheme.
2312
format_registry.register_lazy(
2313
"Bazaar development format 0 (needs bzr.dev from before 1.3)\n",
2314
'bzrlib.repofmt.pack_repo',
2315
'RepositoryFormatPackDevelopment0',
2317
format_registry.register_lazy(
2318
("Bazaar development format 0 with subtree support "
2319
"(needs bzr.dev from before 1.3)\n"),
2320
'bzrlib.repofmt.pack_repo',
2321
'RepositoryFormatPackDevelopment0Subtree',
2323
format_registry.register_lazy(
2324
"Bazaar development format 1 (needs bzr.dev from before 1.6)\n",
2325
'bzrlib.repofmt.pack_repo',
2326
'RepositoryFormatPackDevelopment1',
2328
format_registry.register_lazy(
2329
("Bazaar development format 1 with subtree support "
2330
"(needs bzr.dev from before 1.6)\n"),
2331
'bzrlib.repofmt.pack_repo',
2332
'RepositoryFormatPackDevelopment1Subtree',
2334
# 1.3->1.4 go below here
2337
class InterRepository(InterObject):
2338
"""This class represents operations taking place between two repositories.
2340
Its instances have methods like copy_content and fetch, and contain
2341
references to the source and target repositories these operations can be
2344
Often we will provide convenience methods on 'repository' which carry out
2345
operations with another repository - they will always forward to
2346
InterRepository.get(other).method_name(parameters).
2350
"""The available optimised InterRepository types."""
2352
def copy_content(self, revision_id=None):
2353
raise NotImplementedError(self.copy_content)
2355
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2356
"""Fetch the content required to construct revision_id.
2358
The content is copied from self.source to self.target.
2360
:param revision_id: if None all content is copied, if NULL_REVISION no
2362
:param pb: optional progress bar to use for progress reports. If not
2363
provided a default one will be created.
2365
:returns: (copied_revision_count, failures).
2367
# Normally we should find a specific InterRepository subclass to do
2368
# the fetch; if nothing else then at least InterSameDataRepository.
2369
# If none of them is suitable it looks like fetching is not possible;
2370
# we try to give a good message why. _assert_same_model will probably
2371
# give a helpful message; otherwise a generic one.
2372
self._assert_same_model(self.source, self.target)
2373
raise errors.IncompatibleRepositories(self.source, self.target,
2374
"no suitableInterRepository found")
2376
def _walk_to_common_revisions(self, revision_ids):
2377
"""Walk out from revision_ids in source to revisions target has.
2379
:param revision_ids: The start point for the search.
2380
:return: A set of revision ids.
2382
target_graph = self.target.get_graph()
2383
revision_ids = frozenset(revision_ids)
2384
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2385
return graph.SearchResult(revision_ids, set(), 0, set())
2386
missing_revs = set()
2387
source_graph = self.source.get_graph()
2388
# ensure we don't pay silly lookup costs.
2389
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2390
null_set = frozenset([_mod_revision.NULL_REVISION])
2393
next_revs, ghosts = searcher.next_with_ghosts()
2394
except StopIteration:
2396
if revision_ids.intersection(ghosts):
2397
absent_ids = set(revision_ids.intersection(ghosts))
2398
# If all absent_ids are present in target, no error is needed.
2399
absent_ids.difference_update(
2400
set(target_graph.get_parent_map(absent_ids)))
2402
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2403
# we don't care about other ghosts as we can't fetch them and
2404
# haven't been asked to.
2405
next_revs = set(next_revs)
2406
# we always have NULL_REVISION present.
2407
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2408
missing_revs.update(next_revs - have_revs)
2409
searcher.stop_searching_any(have_revs)
2410
return searcher.get_result()
2412
@deprecated_method(one_two)
2414
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2415
"""Return the revision ids that source has that target does not.
2417
These are returned in topological order.
2419
:param revision_id: only return revision ids included by this
2421
:param find_ghosts: If True find missing revisions in deep history
2422
rather than just finding the surface difference.
2424
return list(self.search_missing_revision_ids(
2425
revision_id, find_ghosts).get_keys())
2428
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2429
"""Return the revision ids that source has that target does not.
2431
:param revision_id: only return revision ids included by this
2433
:param find_ghosts: If True find missing revisions in deep history
2434
rather than just finding the surface difference.
2435
:return: A bzrlib.graph.SearchResult.
2437
# stop searching at found target revisions.
2438
if not find_ghosts and revision_id is not None:
2439
return self._walk_to_common_revisions([revision_id])
2440
# generic, possibly worst case, slow code path.
2441
target_ids = set(self.target.all_revision_ids())
2442
if revision_id is not None:
2443
source_ids = self.source.get_ancestry(revision_id)
2444
if source_ids[0] is not None:
2445
raise AssertionError()
2448
source_ids = self.source.all_revision_ids()
2449
result_set = set(source_ids).difference(target_ids)
2450
return self.source.revision_ids_to_search_result(result_set)
2453
def _same_model(source, target):
2454
"""True if source and target have the same data representation.
2456
Note: this is always called on the base class; overriding it in a
2457
subclass will have no effect.
2460
InterRepository._assert_same_model(source, target)
2462
except errors.IncompatibleRepositories, e:
2466
def _assert_same_model(source, target):
2467
"""Raise an exception if two repositories do not use the same model.
2469
if source.supports_rich_root() != target.supports_rich_root():
2470
raise errors.IncompatibleRepositories(source, target,
2471
"different rich-root support")
2472
if source._serializer != target._serializer:
2473
raise errors.IncompatibleRepositories(source, target,
2474
"different serializers")
2477
class InterSameDataRepository(InterRepository):
2478
"""Code for converting between repositories that represent the same data.
2480
Data format and model must match for this to work.
2484
def _get_repo_format_to_test(self):
2485
"""Repository format for testing with.
2487
InterSameData can pull from subtree to subtree and from non-subtree to
2488
non-subtree, so we test this with the richest repository format.
2490
from bzrlib.repofmt import knitrepo
2491
return knitrepo.RepositoryFormatKnit3()
2494
def is_compatible(source, target):
2495
return InterRepository._same_model(source, target)
2498
def copy_content(self, revision_id=None):
2499
"""Make a complete copy of the content in self into destination.
2501
This copies both the repository's revision data, and configuration information
2502
such as the make_working_trees setting.
2504
This is a destructive operation! Do not use it on existing
2507
:param revision_id: Only copy the content needed to construct
2508
revision_id and its parents.
2511
self.target.set_make_working_trees(self.source.make_working_trees())
2512
except NotImplementedError:
2514
# but don't bother fetching if we have the needed data now.
2515
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2516
self.target.has_revision(revision_id)):
2518
self.target.fetch(self.source, revision_id=revision_id)
2521
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2522
"""See InterRepository.fetch()."""
2523
from bzrlib.fetch import RepoFetcher
2524
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2525
self.source, self.source._format, self.target,
2526
self.target._format)
2527
f = RepoFetcher(to_repository=self.target,
2528
from_repository=self.source,
2529
last_revision=revision_id,
2530
pb=pb, find_ghosts=find_ghosts)
2531
return f.count_copied, f.failed_revisions
2534
class InterWeaveRepo(InterSameDataRepository):
2535
"""Optimised code paths between Weave based repositories.
2537
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2538
implemented lazy inter-object optimisation.
2542
def _get_repo_format_to_test(self):
2543
from bzrlib.repofmt import weaverepo
2544
return weaverepo.RepositoryFormat7()
2547
def is_compatible(source, target):
2548
"""Be compatible with known Weave formats.
2550
We don't test for the stores being of specific types because that
2551
could lead to confusing results, and there is no need to be
2554
from bzrlib.repofmt.weaverepo import (
2560
return (isinstance(source._format, (RepositoryFormat5,
2562
RepositoryFormat7)) and
2563
isinstance(target._format, (RepositoryFormat5,
2565
RepositoryFormat7)))
2566
except AttributeError:
2570
def copy_content(self, revision_id=None):
2571
"""See InterRepository.copy_content()."""
2572
# weave specific optimised path:
2574
self.target.set_make_working_trees(self.source.make_working_trees())
2575
except (errors.RepositoryUpgradeRequired, NotImplemented):
2577
# FIXME do not peek!
2578
if self.source._transport.listable():
2579
pb = ui.ui_factory.nested_progress_bar()
2581
self.target.texts.insert_record_stream(
2582
self.source.texts.get_record_stream(
2583
self.source.texts.keys(), 'topological', False))
2584
pb.update('copying inventory', 0, 1)
2585
self.target.inventories.insert_record_stream(
2586
self.source.inventories.get_record_stream(
2587
self.source.inventories.keys(), 'topological', False))
2588
self.target.signatures.insert_record_stream(
2589
self.source.signatures.get_record_stream(
2590
self.source.signatures.keys(),
2592
self.target.revisions.insert_record_stream(
2593
self.source.revisions.get_record_stream(
2594
self.source.revisions.keys(),
2595
'topological', True))
2599
self.target.fetch(self.source, revision_id=revision_id)
2602
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2603
"""See InterRepository.fetch()."""
2604
from bzrlib.fetch import RepoFetcher
2605
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2606
self.source, self.source._format, self.target, self.target._format)
2607
f = RepoFetcher(to_repository=self.target,
2608
from_repository=self.source,
2609
last_revision=revision_id,
2610
pb=pb, find_ghosts=find_ghosts)
2611
return f.count_copied, f.failed_revisions
2614
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2615
"""See InterRepository.missing_revision_ids()."""
2616
# we want all revisions to satisfy revision_id in source.
2617
# but we don't want to stat every file here and there.
2618
# we want then, all revisions other needs to satisfy revision_id
2619
# checked, but not those that we have locally.
2620
# so the first thing is to get a subset of the revisions to
2621
# satisfy revision_id in source, and then eliminate those that
2622
# we do already have.
2623
# this is slow on high latency connection to self, but as as this
2624
# disk format scales terribly for push anyway due to rewriting
2625
# inventory.weave, this is considered acceptable.
2627
if revision_id is not None:
2628
source_ids = self.source.get_ancestry(revision_id)
2629
if source_ids[0] is not None:
2630
raise AssertionError()
2633
source_ids = self.source._all_possible_ids()
2634
source_ids_set = set(source_ids)
2635
# source_ids is the worst possible case we may need to pull.
2636
# now we want to filter source_ids against what we actually
2637
# have in target, but don't try to check for existence where we know
2638
# we do not have a revision as that would be pointless.
2639
target_ids = set(self.target._all_possible_ids())
2640
possibly_present_revisions = target_ids.intersection(source_ids_set)
2641
actually_present_revisions = set(
2642
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2643
required_revisions = source_ids_set.difference(actually_present_revisions)
2644
if revision_id is not None:
2645
# we used get_ancestry to determine source_ids then we are assured all
2646
# revisions referenced are present as they are installed in topological order.
2647
# and the tip revision was validated by get_ancestry.
2648
result_set = required_revisions
2650
# if we just grabbed the possibly available ids, then
2651
# we only have an estimate of whats available and need to validate
2652
# that against the revision records.
2654
self.source._eliminate_revisions_not_present(required_revisions))
2655
return self.source.revision_ids_to_search_result(result_set)
2658
class InterKnitRepo(InterSameDataRepository):
2659
"""Optimised code paths between Knit based repositories."""
2662
def _get_repo_format_to_test(self):
2663
from bzrlib.repofmt import knitrepo
2664
return knitrepo.RepositoryFormatKnit1()
2667
def is_compatible(source, target):
2668
"""Be compatible with known Knit formats.
2670
We don't test for the stores being of specific types because that
2671
could lead to confusing results, and there is no need to be
2674
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2676
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2677
isinstance(target._format, RepositoryFormatKnit))
2678
except AttributeError:
2680
return are_knits and InterRepository._same_model(source, target)
2683
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2684
"""See InterRepository.fetch()."""
2685
from bzrlib.fetch import RepoFetcher
2686
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2687
self.source, self.source._format, self.target, self.target._format)
2688
f = RepoFetcher(to_repository=self.target,
2689
from_repository=self.source,
2690
last_revision=revision_id,
2691
pb=pb, find_ghosts=find_ghosts)
2692
return f.count_copied, f.failed_revisions
2695
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2696
"""See InterRepository.missing_revision_ids()."""
2697
if revision_id is not None:
2698
source_ids = self.source.get_ancestry(revision_id)
2699
if source_ids[0] is not None:
2700
raise AssertionError()
2703
source_ids = self.source.all_revision_ids()
2704
source_ids_set = set(source_ids)
2705
# source_ids is the worst possible case we may need to pull.
2706
# now we want to filter source_ids against what we actually
2707
# have in target, but don't try to check for existence where we know
2708
# we do not have a revision as that would be pointless.
2709
target_ids = set(self.target.all_revision_ids())
2710
possibly_present_revisions = target_ids.intersection(source_ids_set)
2711
actually_present_revisions = set(
2712
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2713
required_revisions = source_ids_set.difference(actually_present_revisions)
2714
if revision_id is not None:
2715
# we used get_ancestry to determine source_ids then we are assured all
2716
# revisions referenced are present as they are installed in topological order.
2717
# and the tip revision was validated by get_ancestry.
2718
result_set = required_revisions
2720
# if we just grabbed the possibly available ids, then
2721
# we only have an estimate of whats available and need to validate
2722
# that against the revision records.
2724
self.source._eliminate_revisions_not_present(required_revisions))
2725
return self.source.revision_ids_to_search_result(result_set)
2728
class InterPackRepo(InterSameDataRepository):
2729
"""Optimised code paths between Pack based repositories."""
2732
def _get_repo_format_to_test(self):
2733
from bzrlib.repofmt import pack_repo
2734
return pack_repo.RepositoryFormatKnitPack1()
2737
def is_compatible(source, target):
2738
"""Be compatible with known Pack formats.
2740
We don't test for the stores being of specific types because that
2741
could lead to confusing results, and there is no need to be
2744
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2746
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2747
isinstance(target._format, RepositoryFormatPack))
2748
except AttributeError:
2750
return are_packs and InterRepository._same_model(source, target)
2753
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2754
"""See InterRepository.fetch()."""
2755
if (len(self.source._fallback_repositories) > 0 or
2756
len(self.target._fallback_repositories) > 0):
2757
# The pack layer is not aware of fallback repositories, so when
2758
# fetching from a stacked repository or into a stacked repository
2759
# we use the generic fetch logic which uses the VersionedFiles
2760
# attributes on repository.
2761
from bzrlib.fetch import RepoFetcher
2762
fetcher = RepoFetcher(self.target, self.source, revision_id,
2764
return fetcher.count_copied, fetcher.failed_revisions
2765
from bzrlib.repofmt.pack_repo import Packer
2766
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2767
self.source, self.source._format, self.target, self.target._format)
2768
self.count_copied = 0
2769
if revision_id is None:
2771
# everything to do - use pack logic
2772
# to fetch from all packs to one without
2773
# inventory parsing etc, IFF nothing to be copied is in the target.
2775
source_revision_ids = frozenset(self.source.all_revision_ids())
2776
revision_ids = source_revision_ids - \
2777
frozenset(self.target.get_parent_map(source_revision_ids))
2778
revision_keys = [(revid,) for revid in revision_ids]
2779
index = self.target._pack_collection.revision_index.combined_index
2780
present_revision_ids = set(item[1][0] for item in
2781
index.iter_entries(revision_keys))
2782
revision_ids = set(revision_ids) - present_revision_ids
2783
# implementing the TODO will involve:
2784
# - detecting when all of a pack is selected
2785
# - avoiding as much as possible pre-selection, so the
2786
# more-core routines such as create_pack_from_packs can filter in
2787
# a just-in-time fashion. (though having a HEADS list on a
2788
# repository might make this a lot easier, because we could
2789
# sensibly detect 'new revisions' without doing a full index scan.
2790
elif _mod_revision.is_null(revision_id):
2795
revision_ids = self.search_missing_revision_ids(revision_id,
2796
find_ghosts=find_ghosts).get_keys()
2797
except errors.NoSuchRevision:
2798
raise errors.InstallFailed([revision_id])
2799
if len(revision_ids) == 0:
2801
packs = self.source._pack_collection.all_packs()
2802
pack = Packer(self.target._pack_collection, packs, '.fetch',
2803
revision_ids).pack()
2804
if pack is not None:
2805
self.target._pack_collection._save_pack_names()
2806
# Trigger an autopack. This may duplicate effort as we've just done
2807
# a pack creation, but for now it is simpler to think about as
2808
# 'upload data, then repack if needed'.
2809
self.target._pack_collection.autopack()
2810
return (pack.get_revision_count(), [])
2815
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2816
"""See InterRepository.missing_revision_ids().
2818
:param find_ghosts: Find ghosts throughout the ancestry of
2821
if not find_ghosts and revision_id is not None:
2822
return self._walk_to_common_revisions([revision_id])
2823
elif revision_id is not None:
2824
# Find ghosts: search for revisions pointing from one repository to
2825
# the other, and vice versa, anywhere in the history of revision_id.
2826
graph = self.target.get_graph(other_repository=self.source)
2827
searcher = graph._make_breadth_first_searcher([revision_id])
2831
next_revs, ghosts = searcher.next_with_ghosts()
2832
except StopIteration:
2834
if revision_id in ghosts:
2835
raise errors.NoSuchRevision(self.source, revision_id)
2836
found_ids.update(next_revs)
2837
found_ids.update(ghosts)
2838
found_ids = frozenset(found_ids)
2839
# Double query here: should be able to avoid this by changing the
2840
# graph api further.
2841
result_set = found_ids - frozenset(
2842
self.target.get_parent_map(found_ids))
2844
source_ids = self.source.all_revision_ids()
2845
# source_ids is the worst possible case we may need to pull.
2846
# now we want to filter source_ids against what we actually
2847
# have in target, but don't try to check for existence where we know
2848
# we do not have a revision as that would be pointless.
2849
target_ids = set(self.target.all_revision_ids())
2850
result_set = set(source_ids).difference(target_ids)
2851
return self.source.revision_ids_to_search_result(result_set)
2854
class InterModel1and2(InterRepository):
2857
def _get_repo_format_to_test(self):
2861
def is_compatible(source, target):
2862
if not source.supports_rich_root() and target.supports_rich_root():
2868
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2869
"""See InterRepository.fetch()."""
2870
from bzrlib.fetch import Model1toKnit2Fetcher
2871
f = Model1toKnit2Fetcher(to_repository=self.target,
2872
from_repository=self.source,
2873
last_revision=revision_id,
2874
pb=pb, find_ghosts=find_ghosts)
2875
return f.count_copied, f.failed_revisions
2878
def copy_content(self, revision_id=None):
2879
"""Make a complete copy of the content in self into destination.
2881
This is a destructive operation! Do not use it on existing
2884
:param revision_id: Only copy the content needed to construct
2885
revision_id and its parents.
2888
self.target.set_make_working_trees(self.source.make_working_trees())
2889
except NotImplementedError:
2891
# but don't bother fetching if we have the needed data now.
2892
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2893
self.target.has_revision(revision_id)):
2895
self.target.fetch(self.source, revision_id=revision_id)
2898
class InterKnit1and2(InterKnitRepo):
2901
def _get_repo_format_to_test(self):
2905
def is_compatible(source, target):
2906
"""Be compatible with Knit1 source and Knit3 target"""
2907
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
2909
from bzrlib.repofmt.knitrepo import (RepositoryFormatKnit1,
2910
RepositoryFormatKnit3)
2911
from bzrlib.repofmt.pack_repo import (
2912
RepositoryFormatKnitPack1,
2913
RepositoryFormatKnitPack3,
2914
RepositoryFormatPackDevelopment0,
2915
RepositoryFormatPackDevelopment0Subtree,
2918
RepositoryFormatKnit1,
2919
RepositoryFormatKnitPack1,
2920
RepositoryFormatPackDevelopment0,
2923
RepositoryFormatKnit3,
2924
RepositoryFormatKnitPack3,
2925
RepositoryFormatPackDevelopment0Subtree,
2927
return (isinstance(source._format, nosubtrees) and
2928
isinstance(target._format, subtrees))
2929
except AttributeError:
2933
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2934
"""See InterRepository.fetch()."""
2935
from bzrlib.fetch import Knit1to2Fetcher
2936
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2937
self.source, self.source._format, self.target,
2938
self.target._format)
2939
f = Knit1to2Fetcher(to_repository=self.target,
2940
from_repository=self.source,
2941
last_revision=revision_id,
2942
pb=pb, find_ghosts=find_ghosts)
2943
return f.count_copied, f.failed_revisions
2946
class InterDifferingSerializer(InterKnitRepo):
2949
def _get_repo_format_to_test(self):
2953
def is_compatible(source, target):
2954
"""Be compatible with Knit2 source and Knit3 target"""
2955
if source.supports_rich_root() != target.supports_rich_root():
2957
# Ideally, we'd support fetching if the source had no tree references
2958
# even if it supported them...
2959
if (getattr(source, '_format.supports_tree_reference', False) and
2960
not getattr(target, '_format.supports_tree_reference', False)):
2965
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2966
"""See InterRepository.fetch()."""
2967
revision_ids = self.target.search_missing_revision_ids(self.source,
2968
revision_id, find_ghosts=find_ghosts).get_keys()
2969
revision_ids = tsort.topo_sort(
2970
self.source.get_graph().get_parent_map(revision_ids))
2971
def revisions_iterator():
2972
for current_revision_id in revision_ids:
2973
revision = self.source.get_revision(current_revision_id)
2974
tree = self.source.revision_tree(current_revision_id)
2976
signature = self.source.get_signature_text(
2977
current_revision_id)
2978
except errors.NoSuchRevision:
2980
yield revision, tree, signature
2982
my_pb = ui.ui_factory.nested_progress_bar()
2987
install_revisions(self.target, revisions_iterator(),
2988
len(revision_ids), pb)
2990
if my_pb is not None:
2992
return len(revision_ids), 0
2995
class InterOtherToRemote(InterRepository):
2997
def __init__(self, source, target):
2998
InterRepository.__init__(self, source, target)
2999
self._real_inter = None
3002
def is_compatible(source, target):
3003
if isinstance(target, remote.RemoteRepository):
3007
def _ensure_real_inter(self):
3008
if self._real_inter is None:
3009
self.target._ensure_real()
3010
real_target = self.target._real_repository
3011
self._real_inter = InterRepository.get(self.source, real_target)
3013
def copy_content(self, revision_id=None):
3014
self._ensure_real_inter()
3015
self._real_inter.copy_content(revision_id=revision_id)
3017
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3018
self._ensure_real_inter()
3019
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3020
find_ghosts=find_ghosts)
3023
def _get_repo_format_to_test(self):
3027
class InterRemoteToOther(InterRepository):
3029
def __init__(self, source, target):
3030
InterRepository.__init__(self, source, target)
3031
self._real_inter = None
3034
def is_compatible(source, target):
3035
if not isinstance(source, remote.RemoteRepository):
3037
# Is source's model compatible with target's model?
3038
source._ensure_real()
3039
real_source = source._real_repository
3040
if isinstance(real_source, remote.RemoteRepository):
3041
raise NotImplementedError(
3042
"We don't support remote repos backed by remote repos yet.")
3043
return InterRepository._same_model(real_source, target)
3045
def _ensure_real_inter(self):
3046
if self._real_inter is None:
3047
self.source._ensure_real()
3048
real_source = self.source._real_repository
3049
self._real_inter = InterRepository.get(real_source, self.target)
3051
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3052
self._ensure_real_inter()
3053
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3054
find_ghosts=find_ghosts)
3056
def copy_content(self, revision_id=None):
3057
self._ensure_real_inter()
3058
self._real_inter.copy_content(revision_id=revision_id)
3061
def _get_repo_format_to_test(self):
3066
InterRepository.register_optimiser(InterDifferingSerializer)
3067
InterRepository.register_optimiser(InterSameDataRepository)
3068
InterRepository.register_optimiser(InterWeaveRepo)
3069
InterRepository.register_optimiser(InterKnitRepo)
3070
InterRepository.register_optimiser(InterModel1and2)
3071
InterRepository.register_optimiser(InterKnit1and2)
3072
InterRepository.register_optimiser(InterPackRepo)
3073
InterRepository.register_optimiser(InterOtherToRemote)
3074
InterRepository.register_optimiser(InterRemoteToOther)
3077
class CopyConverter(object):
3078
"""A repository conversion tool which just performs a copy of the content.
3080
This is slow but quite reliable.
3083
def __init__(self, target_format):
3084
"""Create a CopyConverter.
3086
:param target_format: The format the resulting repository should be.
3088
self.target_format = target_format
3090
def convert(self, repo, pb):
3091
"""Perform the conversion of to_convert, giving feedback via pb.
3093
:param to_convert: The disk object to convert.
3094
:param pb: a progress bar to use for progress information.
3099
# this is only useful with metadir layouts - separated repo content.
3100
# trigger an assertion if not such
3101
repo._format.get_format_string()
3102
self.repo_dir = repo.bzrdir
3103
self.step('Moving repository to repository.backup')
3104
self.repo_dir.transport.move('repository', 'repository.backup')
3105
backup_transport = self.repo_dir.transport.clone('repository.backup')
3106
repo._format.check_conversion_target(self.target_format)
3107
self.source_repo = repo._format.open(self.repo_dir,
3109
_override_transport=backup_transport)
3110
self.step('Creating new repository')
3111
converted = self.target_format.initialize(self.repo_dir,
3112
self.source_repo.is_shared())
3113
converted.lock_write()
3115
self.step('Copying content into repository.')
3116
self.source_repo.copy_content_into(converted)
3119
self.step('Deleting old repository content.')
3120
self.repo_dir.transport.delete_tree('repository.backup')
3121
self.pb.note('repository converted')
3123
def step(self, message):
3124
"""Update the pb by a step."""
3126
self.pb.update(message, self.count, self.total)
3138
def _unescaper(match, _map=_unescape_map):
3139
code = match.group(1)
3143
if not code.startswith('#'):
3145
return unichr(int(code[1:])).encode('utf8')
3151
def _unescape_xml(data):
3152
"""Unescape predefined XML entities in a string of data."""
3154
if _unescape_re is None:
3155
_unescape_re = re.compile('\&([^;]*);')
3156
return _unescape_re.sub(_unescaper, data)
3159
class _VersionedFileChecker(object):
3161
def __init__(self, repository):
3162
self.repository = repository
3163
self.text_index = self.repository._generate_text_key_index()
3165
def calculate_file_version_parents(self, text_key):
3166
"""Calculate the correct parents for a file version according to
3169
parent_keys = self.text_index[text_key]
3170
if parent_keys == [_mod_revision.NULL_REVISION]:
3172
return tuple(parent_keys)
3174
def check_file_version_parents(self, texts, progress_bar=None):
3175
"""Check the parents stored in a versioned file are correct.
3177
It also detects file versions that are not referenced by their
3178
corresponding revision's inventory.
3180
:returns: A tuple of (wrong_parents, dangling_file_versions).
3181
wrong_parents is a dict mapping {revision_id: (stored_parents,
3182
correct_parents)} for each revision_id where the stored parents
3183
are not correct. dangling_file_versions is a set of (file_id,
3184
revision_id) tuples for versions that are present in this versioned
3185
file, but not used by the corresponding inventory.
3188
self.file_ids = set([file_id for file_id, _ in
3189
self.text_index.iterkeys()])
3190
# text keys is now grouped by file_id
3191
n_weaves = len(self.file_ids)
3192
files_in_revisions = {}
3193
revisions_of_files = {}
3194
n_versions = len(self.text_index)
3195
progress_bar.update('loading text store', 0, n_versions)
3196
parent_map = self.repository.texts.get_parent_map(self.text_index)
3197
# On unlistable transports this could well be empty/error...
3198
text_keys = self.repository.texts.keys()
3199
unused_keys = frozenset(text_keys) - set(self.text_index)
3200
for num, key in enumerate(self.text_index.iterkeys()):
3201
if progress_bar is not None:
3202
progress_bar.update('checking text graph', num, n_versions)
3203
correct_parents = self.calculate_file_version_parents(key)
3205
knit_parents = parent_map[key]
3206
except errors.RevisionNotPresent:
3209
if correct_parents != knit_parents:
3210
wrong_parents[key] = (knit_parents, correct_parents)
3211
return wrong_parents, unused_keys
3214
def _old_get_graph(repository, revision_id):
3215
"""DO NOT USE. That is all. I'm serious."""
3216
graph = repository.get_graph()
3217
revision_graph = dict(((key, value) for key, value in
3218
graph.iter_ancestry([revision_id]) if value is not None))
3219
return _strip_NULL_ghosts(revision_graph)
3222
def _strip_NULL_ghosts(revision_graph):
3223
"""Also don't use this. more compatibility code for unmigrated clients."""
3224
# Filter ghosts, and null:
3225
if _mod_revision.NULL_REVISION in revision_graph:
3226
del revision_graph[_mod_revision.NULL_REVISION]
3227
for key, parents in revision_graph.items():
3228
revision_graph[key] = tuple(parent for parent in parents if parent
3230
return revision_graph
added
removed
bzrlib/repository.py
