1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from cStringIO import StringIO
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
39
revision as _mod_revision,
45
from bzrlib.bundle import serializer
46
from bzrlib.revisiontree import RevisionTree
47
from bzrlib.store.versioned import VersionedFileStore
48
from bzrlib.store.text import TextStore
49
from bzrlib.testament import Testament
50
from bzrlib.util import bencode
53
from bzrlib.decorators import needs_read_lock, needs_write_lock
54
from bzrlib.inter import InterObject
55
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
56
from bzrlib.symbol_versioning import (
63
from bzrlib.trace import mutter, mutter_callsite, note, warning
66
# Old formats display a warning, but only once
67
_deprecation_warning_done = False
70
class CommitBuilder(object):
71
"""Provides an interface to build up a commit.
73
This allows describing a tree to be committed without needing to
74
know the internals of the format of the repository.
77
# all clients should supply tree roots.
78
record_root_entry = True
79
# the default CommitBuilder does not manage trees whose root is versioned.
80
_versioned_root = False
82
def __init__(self, repository, parents, config, timestamp=None,
83
timezone=None, committer=None, revprops=None,
85
"""Initiate a CommitBuilder.
87
:param repository: Repository to commit to.
88
:param parents: Revision ids of the parents of the new revision.
89
:param config: Configuration to use.
90
:param timestamp: Optional timestamp recorded for commit.
91
:param timezone: Optional timezone for timestamp.
92
:param committer: Optional committer to set for commit.
93
:param revprops: Optional dictionary of revision properties.
94
:param revision_id: Optional revision id.
99
self._committer = self._config.username()
101
self._committer = committer
103
self.new_inventory = Inventory(None)
104
self._new_revision_id = revision_id
105
self.parents = parents
106
self.repository = repository
109
if revprops is not None:
110
self._revprops.update(revprops)
112
if timestamp is None:
113
timestamp = time.time()
114
# Restrict resolution to 1ms
115
self._timestamp = round(timestamp, 3)
118
self._timezone = osutils.local_time_offset()
120
self._timezone = int(timezone)
122
self._generate_revision_if_needed()
123
self.__heads = graph.HeadsCache(repository.get_graph()).heads
125
def commit(self, message):
126
"""Make the actual commit.
128
:return: The revision id of the recorded revision.
130
rev = _mod_revision.Revision(
131
timestamp=self._timestamp,
132
timezone=self._timezone,
133
committer=self._committer,
135
inventory_sha1=self.inv_sha1,
136
revision_id=self._new_revision_id,
137
properties=self._revprops)
138
rev.parent_ids = self.parents
139
self.repository.add_revision(self._new_revision_id, rev,
140
self.new_inventory, self._config)
141
self.repository.commit_write_group()
142
return self._new_revision_id
145
"""Abort the commit that is being built.
147
self.repository.abort_write_group()
149
def revision_tree(self):
150
"""Return the tree that was just committed.
152
After calling commit() this can be called to get a RevisionTree
153
representing the newly committed tree. This is preferred to
154
calling Repository.revision_tree() because that may require
155
deserializing the inventory, while we already have a copy in
158
return RevisionTree(self.repository, self.new_inventory,
159
self._new_revision_id)
161
def finish_inventory(self):
162
"""Tell the builder that the inventory is finished."""
163
if self.new_inventory.root is None:
164
raise AssertionError('Root entry should be supplied to'
165
' record_entry_contents, as of bzr 0.10.')
166
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
167
self.new_inventory.revision_id = self._new_revision_id
168
self.inv_sha1 = self.repository.add_inventory(
169
self._new_revision_id,
174
def _gen_revision_id(self):
175
"""Return new revision-id."""
176
return generate_ids.gen_revision_id(self._config.username(),
179
def _generate_revision_if_needed(self):
180
"""Create a revision id if None was supplied.
182
If the repository can not support user-specified revision ids
183
they should override this function and raise CannotSetRevisionId
184
if _new_revision_id is not None.
186
:raises: CannotSetRevisionId
188
if self._new_revision_id is None:
189
self._new_revision_id = self._gen_revision_id()
190
self.random_revid = True
192
self.random_revid = False
194
def _heads(self, file_id, revision_ids):
195
"""Calculate the graph heads for revision_ids in the graph of file_id.
197
This can use either a per-file graph or a global revision graph as we
198
have an identity relationship between the two graphs.
200
return self.__heads(revision_ids)
202
def _check_root(self, ie, parent_invs, tree):
203
"""Helper for record_entry_contents.
205
:param ie: An entry being added.
206
:param parent_invs: The inventories of the parent revisions of the
208
:param tree: The tree that is being committed.
210
# In this revision format, root entries have no knit or weave When
211
# serializing out to disk and back in root.revision is always
213
ie.revision = self._new_revision_id
215
def _get_delta(self, ie, basis_inv, path):
216
"""Get a delta against the basis inventory for ie."""
217
if ie.file_id not in basis_inv:
219
return (None, path, ie.file_id, ie)
220
elif ie != basis_inv[ie.file_id]:
222
# TODO: avoid tis id2path call.
223
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
228
def record_entry_contents(self, ie, parent_invs, path, tree,
230
"""Record the content of ie from tree into the commit if needed.
232
Side effect: sets ie.revision when unchanged
234
:param ie: An inventory entry present in the commit.
235
:param parent_invs: The inventories of the parent revisions of the
237
:param path: The path the entry is at in the tree.
238
:param tree: The tree which contains this entry and should be used to
240
:param content_summary: Summary data from the tree about the paths
241
content - stat, length, exec, sha/link target. This is only
242
accessed when the entry has a revision of None - that is when it is
243
a candidate to commit.
244
:return: A tuple (change_delta, version_recorded). change_delta is
245
an inventory_delta change for this entry against the basis tree of
246
the commit, or None if no change occured against the basis tree.
247
version_recorded is True if a new version of the entry has been
248
recorded. For instance, committing a merge where a file was only
249
changed on the other side will return (delta, False).
251
if self.new_inventory.root is None:
252
if ie.parent_id is not None:
253
raise errors.RootMissing()
254
self._check_root(ie, parent_invs, tree)
255
if ie.revision is None:
256
kind = content_summary[0]
258
# ie is carried over from a prior commit
260
# XXX: repository specific check for nested tree support goes here - if
261
# the repo doesn't want nested trees we skip it ?
262
if (kind == 'tree-reference' and
263
not self.repository._format.supports_tree_reference):
264
# mismatch between commit builder logic and repository:
265
# this needs the entry creation pushed down into the builder.
266
raise NotImplementedError('Missing repository subtree support.')
267
self.new_inventory.add(ie)
269
# TODO: slow, take it out of the inner loop.
271
basis_inv = parent_invs[0]
273
basis_inv = Inventory(root_id=None)
275
# ie.revision is always None if the InventoryEntry is considered
276
# for committing. We may record the previous parents revision if the
277
# content is actually unchanged against a sole head.
278
if ie.revision is not None:
279
if not self._versioned_root and path == '':
280
# repositories that do not version the root set the root's
281
# revision to the new commit even when no change occurs, and
282
# this masks when a change may have occurred against the basis,
283
# so calculate if one happened.
284
if ie.file_id in basis_inv:
285
delta = (basis_inv.id2path(ie.file_id), path,
289
delta = (None, path, ie.file_id, ie)
292
# we don't need to commit this, because the caller already
293
# determined that an existing revision of this file is
295
return None, (ie.revision == self._new_revision_id)
296
# XXX: Friction: parent_candidates should return a list not a dict
297
# so that we don't have to walk the inventories again.
298
parent_candiate_entries = ie.parent_candidates(parent_invs)
299
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
301
for inv in parent_invs:
302
if ie.file_id in inv:
303
old_rev = inv[ie.file_id].revision
304
if old_rev in head_set:
305
heads.append(inv[ie.file_id].revision)
306
head_set.remove(inv[ie.file_id].revision)
309
# now we check to see if we need to write a new record to the
311
# We write a new entry unless there is one head to the ancestors, and
312
# the kind-derived content is unchanged.
314
# Cheapest check first: no ancestors, or more the one head in the
315
# ancestors, we write a new node.
319
# There is a single head, look it up for comparison
320
parent_entry = parent_candiate_entries[heads[0]]
321
# if the non-content specific data has changed, we'll be writing a
323
if (parent_entry.parent_id != ie.parent_id or
324
parent_entry.name != ie.name):
326
# now we need to do content specific checks:
328
# if the kind changed the content obviously has
329
if kind != parent_entry.kind:
332
if content_summary[2] is None:
333
raise ValueError("Files must not have executable = None")
335
if (# if the file length changed we have to store:
336
parent_entry.text_size != content_summary[1] or
337
# if the exec bit has changed we have to store:
338
parent_entry.executable != content_summary[2]):
340
elif parent_entry.text_sha1 == content_summary[3]:
341
# all meta and content is unchanged (using a hash cache
342
# hit to check the sha)
343
ie.revision = parent_entry.revision
344
ie.text_size = parent_entry.text_size
345
ie.text_sha1 = parent_entry.text_sha1
346
ie.executable = parent_entry.executable
347
return self._get_delta(ie, basis_inv, path), False
349
# Either there is only a hash change(no hash cache entry,
350
# or same size content change), or there is no change on
352
# Provide the parent's hash to the store layer, so that the
353
# content is unchanged we will not store a new node.
354
nostore_sha = parent_entry.text_sha1
356
# We want to record a new node regardless of the presence or
357
# absence of a content change in the file.
359
ie.executable = content_summary[2]
360
lines = tree.get_file(ie.file_id, path).readlines()
362
ie.text_sha1, ie.text_size = self._add_text_to_weave(
363
ie.file_id, lines, heads, nostore_sha)
364
except errors.ExistingContent:
365
# Turns out that the file content was unchanged, and we were
366
# only going to store a new node if it was changed. Carry over
368
ie.revision = parent_entry.revision
369
ie.text_size = parent_entry.text_size
370
ie.text_sha1 = parent_entry.text_sha1
371
ie.executable = parent_entry.executable
372
return self._get_delta(ie, basis_inv, path), False
373
elif kind == 'directory':
375
# all data is meta here, nothing specific to directory, so
377
ie.revision = parent_entry.revision
378
return self._get_delta(ie, basis_inv, path), False
380
self._add_text_to_weave(ie.file_id, lines, heads, None)
381
elif kind == 'symlink':
382
current_link_target = content_summary[3]
384
# symlink target is not generic metadata, check if it has
386
if current_link_target != parent_entry.symlink_target:
389
# unchanged, carry over.
390
ie.revision = parent_entry.revision
391
ie.symlink_target = parent_entry.symlink_target
392
return self._get_delta(ie, basis_inv, path), False
393
ie.symlink_target = current_link_target
395
self._add_text_to_weave(ie.file_id, lines, heads, None)
396
elif kind == 'tree-reference':
398
if content_summary[3] != parent_entry.reference_revision:
401
# unchanged, carry over.
402
ie.reference_revision = parent_entry.reference_revision
403
ie.revision = parent_entry.revision
404
return self._get_delta(ie, basis_inv, path), False
405
ie.reference_revision = content_summary[3]
407
self._add_text_to_weave(ie.file_id, lines, heads, None)
409
raise NotImplementedError('unknown kind')
410
ie.revision = self._new_revision_id
411
return self._get_delta(ie, basis_inv, path), True
413
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
414
# Note: as we read the content directly from the tree, we know its not
415
# been turned into unicode or badly split - but a broken tree
416
# implementation could give us bad output from readlines() so this is
417
# not a guarantee of safety. What would be better is always checking
418
# the content during test suite execution. RBC 20070912
419
parent_keys = tuple((file_id, parent) for parent in parents)
420
return self.repository.texts.add_lines(
421
(file_id, self._new_revision_id), parent_keys, new_lines,
422
nostore_sha=nostore_sha, random_id=self.random_revid,
423
check_content=False)[0:2]
426
class RootCommitBuilder(CommitBuilder):
427
"""This commitbuilder actually records the root id"""
429
# the root entry gets versioned properly by this builder.
430
_versioned_root = True
432
def _check_root(self, ie, parent_invs, tree):
433
"""Helper for record_entry_contents.
435
:param ie: An entry being added.
436
:param parent_invs: The inventories of the parent revisions of the
438
:param tree: The tree that is being committed.
442
######################################################################
445
class Repository(object):
446
"""Repository holding history for one or more branches.
448
The repository holds and retrieves historical information including
449
revisions and file history. It's normally accessed only by the Branch,
450
which views a particular line of development through that history.
452
The Repository builds on top of some byte storage facilies (the revisions,
453
signatures, inventories and texts attributes) and a Transport, which
454
respectively provide byte storage and a means to access the (possibly
457
The byte storage facilities are addressed via tuples, which we refer to
458
as 'keys' throughout the code base. Revision_keys, inventory_keys and
459
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
460
(file_id, revision_id). We use this interface because it allows low
461
friction with the underlying code that implements disk indices, network
462
encoding and other parts of bzrlib.
464
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
465
the serialised revisions for the repository. This can be used to obtain
466
revision graph information or to access raw serialised revisions.
467
The result of trying to insert data into the repository via this store
468
is undefined: it should be considered read-only except for implementors
470
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
471
the serialised signatures for the repository. This can be used to
472
obtain access to raw serialised signatures. The result of trying to
473
insert data into the repository via this store is undefined: it should
474
be considered read-only except for implementors of repositories.
475
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
476
the serialised inventories for the repository. This can be used to
477
obtain unserialised inventories. The result of trying to insert data
478
into the repository via this store is undefined: it should be
479
considered read-only except for implementors of repositories.
480
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
481
texts of files and directories for the repository. This can be used to
482
obtain file texts or file graphs. Note that Repository.iter_file_bytes
483
is usually a better interface for accessing file texts.
484
The result of trying to insert data into the repository via this store
485
is undefined: it should be considered read-only except for implementors
487
:ivar _transport: Transport for file access to repository, typically
488
pointing to .bzr/repository.
491
# What class to use for a CommitBuilder. Often its simpler to change this
492
# in a Repository class subclass rather than to override
493
# get_commit_builder.
494
_commit_builder_class = CommitBuilder
495
# The search regex used by xml based repositories to determine what things
496
# where changed in a single commit.
497
_file_ids_altered_regex = lazy_regex.lazy_compile(
498
r'file_id="(?P<file_id>[^"]+)"'
499
r'.* revision="(?P<revision_id>[^"]+)"'
502
def abort_write_group(self):
503
"""Commit the contents accrued within the current write group.
505
:seealso: start_write_group.
507
if self._write_group is not self.get_transaction():
508
# has an unlock or relock occured ?
509
raise errors.BzrError('mismatched lock context and write group.')
510
self._abort_write_group()
511
self._write_group = None
513
def _abort_write_group(self):
514
"""Template method for per-repository write group cleanup.
516
This is called during abort before the write group is considered to be
517
finished and should cleanup any internal state accrued during the write
518
group. There is no requirement that data handed to the repository be
519
*not* made available - this is not a rollback - but neither should any
520
attempt be made to ensure that data added is fully commited. Abort is
521
invoked when an error has occured so futher disk or network operations
522
may not be possible or may error and if possible should not be
526
def add_fallback_repository(self, repository):
527
"""Add a repository to use for looking up data not held locally.
529
:param repository: A repository.
531
if not self._format.supports_external_lookups:
532
raise errors.UnstackableRepositoryFormat(self._format, self.base)
533
if not self._add_fallback_repository_check(repository):
534
raise errors.IncompatibleRepositories(self, repository)
535
self._fallback_repositories.append(repository)
536
self.texts.add_fallback_versioned_files(repository.texts)
537
self.inventories.add_fallback_versioned_files(repository.inventories)
538
self.revisions.add_fallback_versioned_files(repository.revisions)
539
self.signatures.add_fallback_versioned_files(repository.signatures)
541
def _add_fallback_repository_check(self, repository):
542
"""Check that this repository can fallback to repository safely.
544
:param repository: A repository to fallback to.
545
:return: True if the repositories can stack ok.
547
return InterRepository._same_model(self, repository)
549
def add_inventory(self, revision_id, inv, parents):
550
"""Add the inventory inv to the repository as revision_id.
552
:param parents: The revision ids of the parents that revision_id
553
is known to have and are in the repository already.
555
:returns: The validator(which is a sha1 digest, though what is sha'd is
556
repository format specific) of the serialized inventory.
558
if not self.is_in_write_group():
559
raise AssertionError("%r not in write group" % (self,))
560
_mod_revision.check_not_reserved_id(revision_id)
561
if not (inv.revision_id is None or inv.revision_id == revision_id):
562
raise AssertionError(
563
"Mismatch between inventory revision"
564
" id and insertion revid (%r, %r)"
565
% (inv.revision_id, revision_id))
567
raise AssertionError()
568
inv_lines = self._serialise_inventory_to_lines(inv)
569
return self._inventory_add_lines(revision_id, parents,
570
inv_lines, check_content=False)
572
def _inventory_add_lines(self, revision_id, parents, lines,
574
"""Store lines in inv_vf and return the sha1 of the inventory."""
575
parents = [(parent,) for parent in parents]
576
return self.inventories.add_lines((revision_id,), parents, lines,
577
check_content=check_content)[0]
579
def add_revision(self, revision_id, rev, inv=None, config=None):
580
"""Add rev to the revision store as revision_id.
582
:param revision_id: the revision id to use.
583
:param rev: The revision object.
584
:param inv: The inventory for the revision. if None, it will be looked
585
up in the inventory storer
586
:param config: If None no digital signature will be created.
587
If supplied its signature_needed method will be used
588
to determine if a signature should be made.
590
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
592
_mod_revision.check_not_reserved_id(revision_id)
593
if config is not None and config.signature_needed():
595
inv = self.get_inventory(revision_id)
596
plaintext = Testament(rev, inv).as_short_text()
597
self.store_revision_signature(
598
gpg.GPGStrategy(config), plaintext, revision_id)
599
# check inventory present
600
if not self.inventories.get_parent_map([(revision_id,)]):
602
raise errors.WeaveRevisionNotPresent(revision_id,
605
# yes, this is not suitable for adding with ghosts.
606
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
610
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
611
self._add_revision(rev)
613
def _add_revision(self, revision):
614
text = self._serializer.write_revision_to_string(revision)
615
key = (revision.revision_id,)
616
parents = tuple((parent,) for parent in revision.parent_ids)
617
self.revisions.add_lines(key, parents, osutils.split_lines(text))
619
def all_revision_ids(self):
620
"""Returns a list of all the revision ids in the repository.
622
This is conceptually deprecated because code should generally work on
623
the graph reachable from a particular revision, and ignore any other
624
revisions that might be present. There is no direct replacement
627
if 'evil' in debug.debug_flags:
628
mutter_callsite(2, "all_revision_ids is linear with history.")
629
return self._all_revision_ids()
631
def _all_revision_ids(self):
632
"""Returns a list of all the revision ids in the repository.
634
These are in as much topological order as the underlying store can
637
raise NotImplementedError(self._all_revision_ids)
639
def break_lock(self):
640
"""Break a lock if one is present from another instance.
642
Uses the ui factory to ask for confirmation if the lock may be from
645
self.control_files.break_lock()
648
def _eliminate_revisions_not_present(self, revision_ids):
649
"""Check every revision id in revision_ids to see if we have it.
651
Returns a set of the present revisions.
654
graph = self.get_graph()
655
parent_map = graph.get_parent_map(revision_ids)
656
# The old API returned a list, should this actually be a set?
657
return parent_map.keys()
660
def create(a_bzrdir):
661
"""Construct the current default format repository in a_bzrdir."""
662
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
664
def __init__(self, _format, a_bzrdir, control_files):
665
"""instantiate a Repository.
667
:param _format: The format of the repository on disk.
668
:param a_bzrdir: The BzrDir of the repository.
670
In the future we will have a single api for all stores for
671
getting file texts, inventories and revisions, then
672
this construct will accept instances of those things.
674
super(Repository, self).__init__()
675
self._format = _format
676
# the following are part of the public API for Repository:
677
self.bzrdir = a_bzrdir
678
self.control_files = control_files
679
self._transport = control_files._transport
680
self.base = self._transport.base
682
self._reconcile_does_inventory_gc = True
683
self._reconcile_fixes_text_parents = False
684
self._reconcile_backsup_inventory = True
685
# not right yet - should be more semantically clear ?
687
# TODO: make sure to construct the right store classes, etc, depending
688
# on whether escaping is required.
689
self._warn_if_deprecated()
690
self._write_group = None
691
# Additional places to query for data.
692
self._fallback_repositories = []
693
# What order should fetch operations request streams in?
694
# The default is unsorted as that is the cheapest for an origin to
696
self._fetch_order = 'unsorted'
697
# Does this repository use deltas that can be fetched as-deltas ?
698
# (E.g. knits, where the knit deltas can be transplanted intact.
699
# We default to False, which will ensure that enough data to get
700
# a full text out of any fetch stream will be grabbed.
701
self._fetch_uses_deltas = False
702
# Should fetch trigger a reconcile after the fetch? Only needed for
703
# some repository formats that can suffer internal inconsistencies.
704
self._fetch_reconcile = False
707
return '%s(%r)' % (self.__class__.__name__,
710
def has_same_location(self, other):
711
"""Returns a boolean indicating if this repository is at the same
712
location as another repository.
714
This might return False even when two repository objects are accessing
715
the same physical repository via different URLs.
717
if self.__class__ is not other.__class__:
719
return (self._transport.base == other._transport.base)
721
def is_in_write_group(self):
722
"""Return True if there is an open write group.
724
:seealso: start_write_group.
726
return self._write_group is not None
729
return self.control_files.is_locked()
731
def is_write_locked(self):
732
"""Return True if this object is write locked."""
733
return self.is_locked() and self.control_files._lock_mode == 'w'
735
def lock_write(self, token=None):
736
"""Lock this repository for writing.
738
This causes caching within the repository obejct to start accumlating
739
data during reads, and allows a 'write_group' to be obtained. Write
740
groups must be used for actual data insertion.
742
:param token: if this is already locked, then lock_write will fail
743
unless the token matches the existing lock.
744
:returns: a token if this instance supports tokens, otherwise None.
745
:raises TokenLockingNotSupported: when a token is given but this
746
instance doesn't support using token locks.
747
:raises MismatchedToken: if the specified token doesn't match the token
748
of the existing lock.
749
:seealso: start_write_group.
751
A token should be passed in if you know that you have locked the object
752
some other way, and need to synchronise this object's state with that
755
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
757
result = self.control_files.lock_write(token=token)
758
for repo in self._fallback_repositories:
759
# Writes don't affect fallback repos
765
self.control_files.lock_read()
766
for repo in self._fallback_repositories:
770
def get_physical_lock_status(self):
771
return self.control_files.get_physical_lock_status()
773
def leave_lock_in_place(self):
774
"""Tell this repository not to release the physical lock when this
777
If lock_write doesn't return a token, then this method is not supported.
779
self.control_files.leave_in_place()
781
def dont_leave_lock_in_place(self):
782
"""Tell this repository to release the physical lock when this
783
object is unlocked, even if it didn't originally acquire it.
785
If lock_write doesn't return a token, then this method is not supported.
787
self.control_files.dont_leave_in_place()
790
def gather_stats(self, revid=None, committers=None):
791
"""Gather statistics from a revision id.
793
:param revid: The revision id to gather statistics from, if None, then
794
no revision specific statistics are gathered.
795
:param committers: Optional parameter controlling whether to grab
796
a count of committers from the revision specific statistics.
797
:return: A dictionary of statistics. Currently this contains:
798
committers: The number of committers if requested.
799
firstrev: A tuple with timestamp, timezone for the penultimate left
800
most ancestor of revid, if revid is not the NULL_REVISION.
801
latestrev: A tuple with timestamp, timezone for revid, if revid is
802
not the NULL_REVISION.
803
revisions: The total revision count in the repository.
804
size: An estimate disk size of the repository in bytes.
807
if revid and committers:
808
result['committers'] = 0
809
if revid and revid != _mod_revision.NULL_REVISION:
811
all_committers = set()
812
revisions = self.get_ancestry(revid)
813
# pop the leading None
815
first_revision = None
817
# ignore the revisions in the middle - just grab first and last
818
revisions = revisions[0], revisions[-1]
819
for revision in self.get_revisions(revisions):
820
if not first_revision:
821
first_revision = revision
823
all_committers.add(revision.committer)
824
last_revision = revision
826
result['committers'] = len(all_committers)
827
result['firstrev'] = (first_revision.timestamp,
828
first_revision.timezone)
829
result['latestrev'] = (last_revision.timestamp,
830
last_revision.timezone)
832
# now gather global repository information
833
# XXX: This is available for many repos regardless of listability.
834
if self.bzrdir.root_transport.listable():
835
# XXX: do we want to __define len__() ?
836
# Maybe the versionedfiles object should provide a different
837
# method to get the number of keys.
838
result['revisions'] = len(self.revisions.keys())
842
def find_branches(self, using=False):
843
"""Find branches underneath this repository.
845
This will include branches inside other branches.
847
:param using: If True, list only branches using this repository.
849
if using and not self.is_shared():
851
return [self.bzrdir.open_branch()]
852
except errors.NotBranchError:
854
class Evaluator(object):
857
self.first_call = True
859
def __call__(self, bzrdir):
860
# On the first call, the parameter is always the bzrdir
861
# containing the current repo.
862
if not self.first_call:
864
repository = bzrdir.open_repository()
865
except errors.NoRepositoryPresent:
868
return False, (None, repository)
869
self.first_call = False
871
value = (bzrdir.open_branch(), None)
872
except errors.NotBranchError:
877
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
878
self.bzrdir.root_transport, evaluate=Evaluator()):
879
if branch is not None:
880
branches.append(branch)
881
if not using and repository is not None:
882
branches.extend(repository.find_branches())
886
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
887
"""Return the revision ids that other has that this does not.
889
These are returned in topological order.
891
revision_id: only return revision ids included by revision_id.
893
return InterRepository.get(other, self).search_missing_revision_ids(
894
revision_id, find_ghosts)
896
@deprecated_method(one_two)
898
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
899
"""Return the revision ids that other has that this does not.
901
These are returned in topological order.
903
revision_id: only return revision ids included by revision_id.
905
keys = self.search_missing_revision_ids(
906
other, revision_id, find_ghosts).get_keys()
909
parents = other.get_graph().get_parent_map(keys)
912
return tsort.topo_sort(parents)
916
"""Open the repository rooted at base.
918
For instance, if the repository is at URL/.bzr/repository,
919
Repository.open(URL) -> a Repository instance.
921
control = bzrdir.BzrDir.open(base)
922
return control.open_repository()
924
def copy_content_into(self, destination, revision_id=None):
925
"""Make a complete copy of the content in self into destination.
927
This is a destructive operation! Do not use it on existing
930
return InterRepository.get(self, destination).copy_content(revision_id)
932
def commit_write_group(self):
933
"""Commit the contents accrued within the current write group.
935
:seealso: start_write_group.
937
if self._write_group is not self.get_transaction():
938
# has an unlock or relock occured ?
939
raise errors.BzrError('mismatched lock context %r and '
941
(self.get_transaction(), self._write_group))
942
self._commit_write_group()
943
self._write_group = None
945
def _commit_write_group(self):
946
"""Template method for per-repository write group cleanup.
948
This is called before the write group is considered to be
949
finished and should ensure that all data handed to the repository
950
for writing during the write group is safely committed (to the
951
extent possible considering file system caching etc).
954
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
955
"""Fetch the content required to construct revision_id from source.
957
If revision_id is None all content is copied.
958
:param find_ghosts: Find and copy revisions in the source that are
959
ghosts in the target (and not reachable directly by walking out to
960
the first-present revision in target from revision_id).
962
# fast path same-url fetch operations
963
if self.has_same_location(source):
964
# check that last_revision is in 'from' and then return a
966
if (revision_id is not None and
967
not _mod_revision.is_null(revision_id)):
968
self.get_revision(revision_id)
970
inter = InterRepository.get(source, self)
972
return inter.fetch(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts)
973
except NotImplementedError:
974
raise errors.IncompatibleRepositories(source, self)
976
def create_bundle(self, target, base, fileobj, format=None):
977
return serializer.write_bundle(self, target, base, fileobj, format)
979
def get_commit_builder(self, branch, parents, config, timestamp=None,
980
timezone=None, committer=None, revprops=None,
982
"""Obtain a CommitBuilder for this repository.
984
:param branch: Branch to commit to.
985
:param parents: Revision ids of the parents of the new revision.
986
:param config: Configuration to use.
987
:param timestamp: Optional timestamp recorded for commit.
988
:param timezone: Optional timezone for timestamp.
989
:param committer: Optional committer to set for commit.
990
:param revprops: Optional dictionary of revision properties.
991
:param revision_id: Optional revision id.
993
result = self._commit_builder_class(self, parents, config,
994
timestamp, timezone, committer, revprops, revision_id)
995
self.start_write_group()
999
if (self.control_files._lock_count == 1 and
1000
self.control_files._lock_mode == 'w'):
1001
if self._write_group is not None:
1002
self.abort_write_group()
1003
self.control_files.unlock()
1004
raise errors.BzrError(
1005
'Must end write groups before releasing write locks.')
1006
self.control_files.unlock()
1007
for repo in self._fallback_repositories:
1011
def clone(self, a_bzrdir, revision_id=None):
1012
"""Clone this repository into a_bzrdir using the current format.
1014
Currently no check is made that the format of this repository and
1015
the bzrdir format are compatible. FIXME RBC 20060201.
1017
:return: The newly created destination repository.
1019
# TODO: deprecate after 0.16; cloning this with all its settings is
1020
# probably not very useful -- mbp 20070423
1021
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1022
self.copy_content_into(dest_repo, revision_id)
1025
def start_write_group(self):
1026
"""Start a write group in the repository.
1028
Write groups are used by repositories which do not have a 1:1 mapping
1029
between file ids and backend store to manage the insertion of data from
1030
both fetch and commit operations.
1032
A write lock is required around the start_write_group/commit_write_group
1033
for the support of lock-requiring repository formats.
1035
One can only insert data into a repository inside a write group.
1039
if not self.is_write_locked():
1040
raise errors.NotWriteLocked(self)
1041
if self._write_group:
1042
raise errors.BzrError('already in a write group')
1043
self._start_write_group()
1044
# so we can detect unlock/relock - the write group is now entered.
1045
self._write_group = self.get_transaction()
1047
def _start_write_group(self):
1048
"""Template method for per-repository write group startup.
1050
This is called before the write group is considered to be
1055
def sprout(self, to_bzrdir, revision_id=None):
1056
"""Create a descendent repository for new development.
1058
Unlike clone, this does not copy the settings of the repository.
1060
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1061
dest_repo.fetch(self, revision_id=revision_id)
1064
def _create_sprouting_repo(self, a_bzrdir, shared):
1065
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1066
# use target default format.
1067
dest_repo = a_bzrdir.create_repository()
1069
# Most control formats need the repository to be specifically
1070
# created, but on some old all-in-one formats it's not needed
1072
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1073
except errors.UninitializableFormat:
1074
dest_repo = a_bzrdir.open_repository()
1078
def has_revision(self, revision_id):
1079
"""True if this repository has a copy of the revision."""
1080
return revision_id in self.has_revisions((revision_id,))
1083
def has_revisions(self, revision_ids):
1084
"""Probe to find out the presence of multiple revisions.
1086
:param revision_ids: An iterable of revision_ids.
1087
:return: A set of the revision_ids that were present.
1089
parent_map = self.revisions.get_parent_map(
1090
[(rev_id,) for rev_id in revision_ids])
1092
if _mod_revision.NULL_REVISION in revision_ids:
1093
result.add(_mod_revision.NULL_REVISION)
1094
result.update([key[0] for key in parent_map])
1098
def get_revision(self, revision_id):
1099
"""Return the Revision object for a named revision."""
1100
return self.get_revisions([revision_id])[0]
1103
def get_revision_reconcile(self, revision_id):
1104
"""'reconcile' helper routine that allows access to a revision always.
1106
This variant of get_revision does not cross check the weave graph
1107
against the revision one as get_revision does: but it should only
1108
be used by reconcile, or reconcile-alike commands that are correcting
1109
or testing the revision graph.
1111
return self._get_revisions([revision_id])[0]
1114
def get_revisions(self, revision_ids):
1115
"""Get many revisions at once."""
1116
return self._get_revisions(revision_ids)
1119
def _get_revisions(self, revision_ids):
1120
"""Core work logic to get many revisions without sanity checks."""
1121
for rev_id in revision_ids:
1122
if not rev_id or not isinstance(rev_id, basestring):
1123
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1124
keys = [(key,) for key in revision_ids]
1125
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1127
for record in stream:
1128
if record.storage_kind == 'absent':
1129
raise errors.NoSuchRevision(self, record.key[0])
1130
text = record.get_bytes_as('fulltext')
1131
rev = self._serializer.read_revision_from_string(text)
1132
revs[record.key[0]] = rev
1133
return [revs[revid] for revid in revision_ids]
1136
def get_revision_xml(self, revision_id):
1137
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1138
# would have already do it.
1139
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1140
rev = self.get_revision(revision_id)
1141
rev_tmp = StringIO()
1142
# the current serializer..
1143
self._serializer.write_revision(rev, rev_tmp)
1145
return rev_tmp.getvalue()
1147
def get_deltas_for_revisions(self, revisions):
1148
"""Produce a generator of revision deltas.
1150
Note that the input is a sequence of REVISIONS, not revision_ids.
1151
Trees will be held in memory until the generator exits.
1152
Each delta is relative to the revision's lefthand predecessor.
1154
required_trees = set()
1155
for revision in revisions:
1156
required_trees.add(revision.revision_id)
1157
required_trees.update(revision.parent_ids[:1])
1158
trees = dict((t.get_revision_id(), t) for
1159
t in self.revision_trees(required_trees))
1160
for revision in revisions:
1161
if not revision.parent_ids:
1162
old_tree = self.revision_tree(None)
1164
old_tree = trees[revision.parent_ids[0]]
1165
yield trees[revision.revision_id].changes_from(old_tree)
1168
def get_revision_delta(self, revision_id):
1169
"""Return the delta for one revision.
1171
The delta is relative to the left-hand predecessor of the
1174
r = self.get_revision(revision_id)
1175
return list(self.get_deltas_for_revisions([r]))[0]
1178
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1179
signature = gpg_strategy.sign(plaintext)
1180
self.add_signature_text(revision_id, signature)
1183
def add_signature_text(self, revision_id, signature):
1184
self.signatures.add_lines((revision_id,), (),
1185
osutils.split_lines(signature))
1187
def find_text_key_references(self):
1188
"""Find the text key references within the repository.
1190
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1191
revision_ids. Each altered file-ids has the exact revision_ids that
1192
altered it listed explicitly.
1193
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1194
to whether they were referred to by the inventory of the
1195
revision_id that they contain. The inventory texts from all present
1196
revision ids are assessed to generate this report.
1198
revision_keys = self.revisions.keys()
1199
w = self.inventories
1200
pb = ui.ui_factory.nested_progress_bar()
1202
return self._find_text_key_references_from_xml_inventory_lines(
1203
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1207
def _find_text_key_references_from_xml_inventory_lines(self,
1209
"""Core routine for extracting references to texts from inventories.
1211
This performs the translation of xml lines to revision ids.
1213
:param line_iterator: An iterator of lines, origin_version_id
1214
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1215
to whether they were referred to by the inventory of the
1216
revision_id that they contain. Note that if that revision_id was
1217
not part of the line_iterator's output then False will be given -
1218
even though it may actually refer to that key.
1220
if not self._serializer.support_altered_by_hack:
1221
raise AssertionError(
1222
"_find_text_key_references_from_xml_inventory_lines only "
1223
"supported for branches which store inventory as unnested xml"
1224
", not on %r" % self)
1227
# this code needs to read every new line in every inventory for the
1228
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1229
# not present in one of those inventories is unnecessary but not
1230
# harmful because we are filtering by the revision id marker in the
1231
# inventory lines : we only select file ids altered in one of those
1232
# revisions. We don't need to see all lines in the inventory because
1233
# only those added in an inventory in rev X can contain a revision=X
1235
unescape_revid_cache = {}
1236
unescape_fileid_cache = {}
1238
# jam 20061218 In a big fetch, this handles hundreds of thousands
1239
# of lines, so it has had a lot of inlining and optimizing done.
1240
# Sorry that it is a little bit messy.
1241
# Move several functions to be local variables, since this is a long
1243
search = self._file_ids_altered_regex.search
1244
unescape = _unescape_xml
1245
setdefault = result.setdefault
1246
for line, line_key in line_iterator:
1247
match = search(line)
1250
# One call to match.group() returning multiple items is quite a
1251
# bit faster than 2 calls to match.group() each returning 1
1252
file_id, revision_id = match.group('file_id', 'revision_id')
1254
# Inlining the cache lookups helps a lot when you make 170,000
1255
# lines and 350k ids, versus 8.4 unique ids.
1256
# Using a cache helps in 2 ways:
1257
# 1) Avoids unnecessary decoding calls
1258
# 2) Re-uses cached strings, which helps in future set and
1260
# (2) is enough that removing encoding entirely along with
1261
# the cache (so we are using plain strings) results in no
1262
# performance improvement.
1264
revision_id = unescape_revid_cache[revision_id]
1266
unescaped = unescape(revision_id)
1267
unescape_revid_cache[revision_id] = unescaped
1268
revision_id = unescaped
1270
# Note that unconditionally unescaping means that we deserialise
1271
# every fileid, which for general 'pull' is not great, but we don't
1272
# really want to have some many fulltexts that this matters anyway.
1275
file_id = unescape_fileid_cache[file_id]
1277
unescaped = unescape(file_id)
1278
unescape_fileid_cache[file_id] = unescaped
1281
key = (file_id, revision_id)
1282
setdefault(key, False)
1283
if revision_id == line_key[-1]:
1287
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1289
"""Helper routine for fileids_altered_by_revision_ids.
1291
This performs the translation of xml lines to revision ids.
1293
:param line_iterator: An iterator of lines, origin_version_id
1294
:param revision_ids: The revision ids to filter for. This should be a
1295
set or other type which supports efficient __contains__ lookups, as
1296
the revision id from each parsed line will be looked up in the
1297
revision_ids filter.
1298
:return: a dictionary mapping altered file-ids to an iterable of
1299
revision_ids. Each altered file-ids has the exact revision_ids that
1300
altered it listed explicitly.
1303
setdefault = result.setdefault
1305
self._find_text_key_references_from_xml_inventory_lines(
1306
line_iterator).iterkeys():
1307
# once data is all ensured-consistent; then this is
1308
# if revision_id == version_id
1309
if key[-1:] in revision_ids:
1310
setdefault(key[0], set()).add(key[-1])
1313
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1314
"""Find the file ids and versions affected by revisions.
1316
:param revisions: an iterable containing revision ids.
1317
:param _inv_weave: The inventory weave from this repository or None.
1318
If None, the inventory weave will be opened automatically.
1319
:return: a dictionary mapping altered file-ids to an iterable of
1320
revision_ids. Each altered file-ids has the exact revision_ids that
1321
altered it listed explicitly.
1323
selected_keys = set((revid,) for revid in revision_ids)
1324
w = _inv_weave or self.inventories
1325
pb = ui.ui_factory.nested_progress_bar()
1327
return self._find_file_ids_from_xml_inventory_lines(
1328
w.iter_lines_added_or_present_in_keys(
1329
selected_keys, pb=pb),
1334
def iter_files_bytes(self, desired_files):
1335
"""Iterate through file versions.
1337
Files will not necessarily be returned in the order they occur in
1338
desired_files. No specific order is guaranteed.
1340
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1341
value supplied by the caller as part of desired_files. It should
1342
uniquely identify the file version in the caller's context. (Examples:
1343
an index number or a TreeTransform trans_id.)
1345
bytes_iterator is an iterable of bytestrings for the file. The
1346
kind of iterable and length of the bytestrings are unspecified, but for
1347
this implementation, it is a list of bytes produced by
1348
VersionedFile.get_record_stream().
1350
:param desired_files: a list of (file_id, revision_id, identifier)
1353
transaction = self.get_transaction()
1355
for file_id, revision_id, callable_data in desired_files:
1356
text_keys[(file_id, revision_id)] = callable_data
1357
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1358
if record.storage_kind == 'absent':
1359
raise errors.RevisionNotPresent(record.key, self)
1360
yield text_keys[record.key], record.get_bytes_as('fulltext')
1362
def _generate_text_key_index(self, text_key_references=None,
1364
"""Generate a new text key index for the repository.
1366
This is an expensive function that will take considerable time to run.
1368
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1369
list of parents, also text keys. When a given key has no parents,
1370
the parents list will be [NULL_REVISION].
1372
# All revisions, to find inventory parents.
1373
if ancestors is None:
1374
graph = self.get_graph()
1375
ancestors = graph.get_parent_map(self.all_revision_ids())
1376
if text_key_references is None:
1377
text_key_references = self.find_text_key_references()
1378
pb = ui.ui_factory.nested_progress_bar()
1380
return self._do_generate_text_key_index(ancestors,
1381
text_key_references, pb)
1385
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1386
"""Helper for _generate_text_key_index to avoid deep nesting."""
1387
revision_order = tsort.topo_sort(ancestors)
1388
invalid_keys = set()
1390
for revision_id in revision_order:
1391
revision_keys[revision_id] = set()
1392
text_count = len(text_key_references)
1393
# a cache of the text keys to allow reuse; costs a dict of all the
1394
# keys, but saves a 2-tuple for every child of a given key.
1396
for text_key, valid in text_key_references.iteritems():
1398
invalid_keys.add(text_key)
1400
revision_keys[text_key[1]].add(text_key)
1401
text_key_cache[text_key] = text_key
1402
del text_key_references
1404
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1405
NULL_REVISION = _mod_revision.NULL_REVISION
1406
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1407
# too small for large or very branchy trees. However, for 55K path
1408
# trees, it would be easy to use too much memory trivially. Ideally we
1409
# could gauge this by looking at available real memory etc, but this is
1410
# always a tricky proposition.
1411
inventory_cache = lru_cache.LRUCache(10)
1412
batch_size = 10 # should be ~150MB on a 55K path tree
1413
batch_count = len(revision_order) / batch_size + 1
1415
pb.update("Calculating text parents.", processed_texts, text_count)
1416
for offset in xrange(batch_count):
1417
to_query = revision_order[offset * batch_size:(offset + 1) *
1421
for rev_tree in self.revision_trees(to_query):
1422
revision_id = rev_tree.get_revision_id()
1423
parent_ids = ancestors[revision_id]
1424
for text_key in revision_keys[revision_id]:
1425
pb.update("Calculating text parents.", processed_texts)
1426
processed_texts += 1
1427
candidate_parents = []
1428
for parent_id in parent_ids:
1429
parent_text_key = (text_key[0], parent_id)
1431
check_parent = parent_text_key not in \
1432
revision_keys[parent_id]
1434
# the parent parent_id is a ghost:
1435
check_parent = False
1436
# truncate the derived graph against this ghost.
1437
parent_text_key = None
1439
# look at the parent commit details inventories to
1440
# determine possible candidates in the per file graph.
1443
inv = inventory_cache[parent_id]
1445
inv = self.revision_tree(parent_id).inventory
1446
inventory_cache[parent_id] = inv
1447
parent_entry = inv._byid.get(text_key[0], None)
1448
if parent_entry is not None:
1450
text_key[0], parent_entry.revision)
1452
parent_text_key = None
1453
if parent_text_key is not None:
1454
candidate_parents.append(
1455
text_key_cache[parent_text_key])
1456
parent_heads = text_graph.heads(candidate_parents)
1457
new_parents = list(parent_heads)
1458
new_parents.sort(key=lambda x:candidate_parents.index(x))
1459
if new_parents == []:
1460
new_parents = [NULL_REVISION]
1461
text_index[text_key] = new_parents
1463
for text_key in invalid_keys:
1464
text_index[text_key] = [NULL_REVISION]
1467
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1468
"""Get an iterable listing the keys of all the data introduced by a set
1471
The keys will be ordered so that the corresponding items can be safely
1472
fetched and inserted in that order.
1474
:returns: An iterable producing tuples of (knit-kind, file-id,
1475
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1476
'revisions'. file-id is None unless knit-kind is 'file'.
1478
# XXX: it's a bit weird to control the inventory weave caching in this
1479
# generator. Ideally the caching would be done in fetch.py I think. Or
1480
# maybe this generator should explicitly have the contract that it
1481
# should not be iterated until the previously yielded item has been
1483
inv_w = self.inventories
1485
# file ids that changed
1486
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1488
num_file_ids = len(file_ids)
1489
for file_id, altered_versions in file_ids.iteritems():
1490
if _files_pb is not None:
1491
_files_pb.update("fetch texts", count, num_file_ids)
1493
yield ("file", file_id, altered_versions)
1494
# We're done with the files_pb. Note that it finished by the caller,
1495
# just as it was created by the caller.
1499
yield ("inventory", None, revision_ids)
1502
revisions_with_signatures = set()
1503
for rev_id in revision_ids:
1505
self.get_signature_text(rev_id)
1506
except errors.NoSuchRevision:
1510
revisions_with_signatures.add(rev_id)
1511
yield ("signatures", None, revisions_with_signatures)
1514
yield ("revisions", None, revision_ids)
1517
def get_inventory(self, revision_id):
1518
"""Get Inventory object by revision id."""
1519
return self.iter_inventories([revision_id]).next()
1521
def iter_inventories(self, revision_ids):
1522
"""Get many inventories by revision_ids.
1524
This will buffer some or all of the texts used in constructing the
1525
inventories in memory, but will only parse a single inventory at a
1528
:return: An iterator of inventories.
1530
if ((None in revision_ids)
1531
or (_mod_revision.NULL_REVISION in revision_ids)):
1532
raise ValueError('cannot get null revision inventory')
1533
return self._iter_inventories(revision_ids)
1535
def _iter_inventories(self, revision_ids):
1536
"""single-document based inventory iteration."""
1537
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1538
yield self.deserialise_inventory(revision_id, text)
1540
def _iter_inventory_xmls(self, revision_ids):
1541
keys = [(revision_id,) for revision_id in revision_ids]
1542
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1544
for record in stream:
1545
if record.storage_kind != 'absent':
1546
texts[record.key] = record.get_bytes_as('fulltext')
1548
raise errors.NoSuchRevision(self, record.key)
1550
yield texts[key], key[-1]
1552
def deserialise_inventory(self, revision_id, xml):
1553
"""Transform the xml into an inventory object.
1555
:param revision_id: The expected revision id of the inventory.
1556
:param xml: A serialised inventory.
1558
result = self._serializer.read_inventory_from_string(xml, revision_id)
1559
if result.revision_id != revision_id:
1560
raise AssertionError('revision id mismatch %s != %s' % (
1561
result.revision_id, revision_id))
1564
def serialise_inventory(self, inv):
1565
return self._serializer.write_inventory_to_string(inv)
1567
def _serialise_inventory_to_lines(self, inv):
1568
return self._serializer.write_inventory_to_lines(inv)
1570
def get_serializer_format(self):
1571
return self._serializer.format_num
1574
def get_inventory_xml(self, revision_id):
1575
"""Get inventory XML as a file object."""
1576
texts = self._iter_inventory_xmls([revision_id])
1578
text, revision_id = texts.next()
1579
except StopIteration:
1580
raise errors.HistoryMissing(self, 'inventory', revision_id)
1584
def get_inventory_sha1(self, revision_id):
1585
"""Return the sha1 hash of the inventory entry
1587
return self.get_revision(revision_id).inventory_sha1
1589
def iter_reverse_revision_history(self, revision_id):
1590
"""Iterate backwards through revision ids in the lefthand history
1592
:param revision_id: The revision id to start with. All its lefthand
1593
ancestors will be traversed.
1595
graph = self.get_graph()
1596
next_id = revision_id
1598
if next_id in (None, _mod_revision.NULL_REVISION):
1601
# Note: The following line may raise KeyError in the event of
1602
# truncated history. We decided not to have a try:except:raise
1603
# RevisionNotPresent here until we see a use for it, because of the
1604
# cost in an inner loop that is by its very nature O(history).
1605
# Robert Collins 20080326
1606
parents = graph.get_parent_map([next_id])[next_id]
1607
if len(parents) == 0:
1610
next_id = parents[0]
1613
def get_revision_inventory(self, revision_id):
1614
"""Return inventory of a past revision."""
1615
# TODO: Unify this with get_inventory()
1616
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1617
# must be the same as its revision, so this is trivial.
1618
if revision_id is None:
1619
# This does not make sense: if there is no revision,
1620
# then it is the current tree inventory surely ?!
1621
# and thus get_root_id() is something that looks at the last
1622
# commit on the branch, and the get_root_id is an inventory check.
1623
raise NotImplementedError
1624
# return Inventory(self.get_root_id())
1626
return self.get_inventory(revision_id)
1629
def is_shared(self):
1630
"""Return True if this repository is flagged as a shared repository."""
1631
raise NotImplementedError(self.is_shared)
1634
def reconcile(self, other=None, thorough=False):
1635
"""Reconcile this repository."""
1636
from bzrlib.reconcile import RepoReconciler
1637
reconciler = RepoReconciler(self, thorough=thorough)
1638
reconciler.reconcile()
1641
def _refresh_data(self):
1642
"""Helper called from lock_* to ensure coherency with disk.
1644
The default implementation does nothing; it is however possible
1645
for repositories to maintain loaded indices across multiple locks
1646
by checking inside their implementation of this method to see
1647
whether their indices are still valid. This depends of course on
1648
the disk format being validatable in this manner.
1652
def revision_tree(self, revision_id):
1653
"""Return Tree for a revision on this branch.
1655
`revision_id` may be None for the empty tree revision.
1657
# TODO: refactor this to use an existing revision object
1658
# so we don't need to read it in twice.
1659
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1660
return RevisionTree(self, Inventory(root_id=None),
1661
_mod_revision.NULL_REVISION)
1663
inv = self.get_revision_inventory(revision_id)
1664
return RevisionTree(self, inv, revision_id)
1666
def revision_trees(self, revision_ids):
1667
"""Return Tree for a revision on this branch.
1669
`revision_id` may not be None or 'null:'"""
1670
inventories = self.iter_inventories(revision_ids)
1671
for inv in inventories:
1672
yield RevisionTree(self, inv, inv.revision_id)
1675
def get_ancestry(self, revision_id, topo_sorted=True):
1676
"""Return a list of revision-ids integrated by a revision.
1678
The first element of the list is always None, indicating the origin
1679
revision. This might change when we have history horizons, or
1680
perhaps we should have a new API.
1682
This is topologically sorted.
1684
if _mod_revision.is_null(revision_id):
1686
if not self.has_revision(revision_id):
1687
raise errors.NoSuchRevision(self, revision_id)
1688
graph = self.get_graph()
1690
search = graph._make_breadth_first_searcher([revision_id])
1693
found, ghosts = search.next_with_ghosts()
1694
except StopIteration:
1697
if _mod_revision.NULL_REVISION in keys:
1698
keys.remove(_mod_revision.NULL_REVISION)
1700
parent_map = graph.get_parent_map(keys)
1701
keys = tsort.topo_sort(parent_map)
1702
return [None] + list(keys)
1705
"""Compress the data within the repository.
1707
This operation only makes sense for some repository types. For other
1708
types it should be a no-op that just returns.
1710
This stub method does not require a lock, but subclasses should use
1711
@needs_write_lock as this is a long running call its reasonable to
1712
implicitly lock for the user.
1716
@deprecated_method(one_six)
1717
def print_file(self, file, revision_id):
1718
"""Print `file` to stdout.
1720
FIXME RBC 20060125 as John Meinel points out this is a bad api
1721
- it writes to stdout, it assumes that that is valid etc. Fix
1722
by creating a new more flexible convenience function.
1724
tree = self.revision_tree(revision_id)
1725
# use inventory as it was in that revision
1726
file_id = tree.inventory.path2id(file)
1728
# TODO: jam 20060427 Write a test for this code path
1729
# it had a bug in it, and was raising the wrong
1731
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1732
tree.print_file(file_id)
1734
def get_transaction(self):
1735
return self.control_files.get_transaction()
1737
@deprecated_method(one_one)
1738
def get_parents(self, revision_ids):
1739
"""See StackedParentsProvider.get_parents"""
1740
parent_map = self.get_parent_map(revision_ids)
1741
return [parent_map.get(r, None) for r in revision_ids]
1743
def get_parent_map(self, revision_ids):
1744
"""See graph._StackedParentsProvider.get_parent_map"""
1745
# revisions index works in keys; this just works in revisions
1746
# therefore wrap and unwrap
1749
for revision_id in revision_ids:
1750
if revision_id == _mod_revision.NULL_REVISION:
1751
result[revision_id] = ()
1752
elif revision_id is None:
1753
raise ValueError('get_parent_map(None) is not valid')
1755
query_keys.append((revision_id ,))
1756
for ((revision_id,), parent_keys) in \
1757
self.revisions.get_parent_map(query_keys).iteritems():
1759
result[revision_id] = tuple(parent_revid
1760
for (parent_revid,) in parent_keys)
1762
result[revision_id] = (_mod_revision.NULL_REVISION,)
1765
def _make_parents_provider(self):
1768
def get_graph(self, other_repository=None):
1769
"""Return the graph walker for this repository format"""
1770
parents_provider = self._make_parents_provider()
1771
if (other_repository is not None and
1772
not self.has_same_location(other_repository)):
1773
parents_provider = graph._StackedParentsProvider(
1774
[parents_provider, other_repository._make_parents_provider()])
1775
return graph.Graph(parents_provider)
1777
def _get_versioned_file_checker(self):
1778
"""Return an object suitable for checking versioned files."""
1779
return _VersionedFileChecker(self)
1781
def revision_ids_to_search_result(self, result_set):
1782
"""Convert a set of revision ids to a graph SearchResult."""
1783
result_parents = set()
1784
for parents in self.get_graph().get_parent_map(
1785
result_set).itervalues():
1786
result_parents.update(parents)
1787
included_keys = result_set.intersection(result_parents)
1788
start_keys = result_set.difference(included_keys)
1789
exclude_keys = result_parents.difference(result_set)
1790
result = graph.SearchResult(start_keys, exclude_keys,
1791
len(result_set), result_set)
1795
def set_make_working_trees(self, new_value):
1796
"""Set the policy flag for making working trees when creating branches.
1798
This only applies to branches that use this repository.
1800
The default is 'True'.
1801
:param new_value: True to restore the default, False to disable making
1804
raise NotImplementedError(self.set_make_working_trees)
1806
def make_working_trees(self):
1807
"""Returns the policy for making working trees on new branches."""
1808
raise NotImplementedError(self.make_working_trees)
1811
def sign_revision(self, revision_id, gpg_strategy):
1812
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1813
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1816
def has_signature_for_revision_id(self, revision_id):
1817
"""Query for a revision signature for revision_id in the repository."""
1818
if not self.has_revision(revision_id):
1819
raise errors.NoSuchRevision(self, revision_id)
1820
sig_present = (1 == len(
1821
self.signatures.get_parent_map([(revision_id,)])))
1825
def get_signature_text(self, revision_id):
1826
"""Return the text for a signature."""
1827
stream = self.signatures.get_record_stream([(revision_id,)],
1829
record = stream.next()
1830
if record.storage_kind == 'absent':
1831
raise errors.NoSuchRevision(self, revision_id)
1832
return record.get_bytes_as('fulltext')
1835
def check(self, revision_ids=None):
1836
"""Check consistency of all history of given revision_ids.
1838
Different repository implementations should override _check().
1840
:param revision_ids: A non-empty list of revision_ids whose ancestry
1841
will be checked. Typically the last revision_id of a branch.
1843
return self._check(revision_ids)
1845
def _check(self, revision_ids):
1846
result = check.Check(self)
1850
def _warn_if_deprecated(self):
1851
global _deprecation_warning_done
1852
if _deprecation_warning_done:
1854
_deprecation_warning_done = True
1855
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1856
% (self._format, self.bzrdir.transport.base))
1858
def supports_rich_root(self):
1859
return self._format.rich_root_data
1861
def _check_ascii_revisionid(self, revision_id, method):
1862
"""Private helper for ascii-only repositories."""
1863
# weave repositories refuse to store revisionids that are non-ascii.
1864
if revision_id is not None:
1865
# weaves require ascii revision ids.
1866
if isinstance(revision_id, unicode):
1868
revision_id.encode('ascii')
1869
except UnicodeEncodeError:
1870
raise errors.NonAsciiRevisionId(method, self)
1873
revision_id.decode('ascii')
1874
except UnicodeDecodeError:
1875
raise errors.NonAsciiRevisionId(method, self)
1877
def revision_graph_can_have_wrong_parents(self):
1878
"""Is it possible for this repository to have a revision graph with
1881
If True, then this repository must also implement
1882
_find_inconsistent_revision_parents so that check and reconcile can
1883
check for inconsistencies before proceeding with other checks that may
1884
depend on the revision index being consistent.
1886
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1889
# remove these delegates a while after bzr 0.15
1890
def __make_delegated(name, from_module):
1891
def _deprecated_repository_forwarder():
1892
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1893
% (name, from_module),
1896
m = __import__(from_module, globals(), locals(), [name])
1898
return getattr(m, name)
1899
except AttributeError:
1900
raise AttributeError('module %s has no name %s'
1902
globals()[name] = _deprecated_repository_forwarder
1905
'AllInOneRepository',
1906
'WeaveMetaDirRepository',
1907
'PreSplitOutRepositoryFormat',
1908
'RepositoryFormat4',
1909
'RepositoryFormat5',
1910
'RepositoryFormat6',
1911
'RepositoryFormat7',
1913
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1917
'RepositoryFormatKnit',
1918
'RepositoryFormatKnit1',
1920
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1923
def install_revision(repository, rev, revision_tree):
1924
"""Install all revision data into a repository."""
1925
install_revisions(repository, [(rev, revision_tree, None)])
1928
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1929
"""Install all revision data into a repository.
1931
Accepts an iterable of revision, tree, signature tuples. The signature
1934
repository.start_write_group()
1936
for n, (revision, revision_tree, signature) in enumerate(iterable):
1937
_install_revision(repository, revision, revision_tree, signature)
1939
pb.update('Transferring revisions', n + 1, num_revisions)
1941
repository.abort_write_group()
1944
repository.commit_write_group()
1947
def _install_revision(repository, rev, revision_tree, signature):
1948
"""Install all revision data into a repository."""
1949
present_parents = []
1951
for p_id in rev.parent_ids:
1952
if repository.has_revision(p_id):
1953
present_parents.append(p_id)
1954
parent_trees[p_id] = repository.revision_tree(p_id)
1956
parent_trees[p_id] = repository.revision_tree(None)
1958
inv = revision_tree.inventory
1959
entries = inv.iter_entries()
1960
# backwards compatibility hack: skip the root id.
1961
if not repository.supports_rich_root():
1962
path, root = entries.next()
1963
if root.revision != rev.revision_id:
1964
raise errors.IncompatibleRevision(repr(repository))
1966
for path, ie in entries:
1967
text_keys[(ie.file_id, ie.revision)] = ie
1968
text_parent_map = repository.texts.get_parent_map(text_keys)
1969
missing_texts = set(text_keys) - set(text_parent_map)
1970
# Add the texts that are not already present
1971
for text_key in missing_texts:
1972
ie = text_keys[text_key]
1974
# FIXME: TODO: The following loop overlaps/duplicates that done by
1975
# commit to determine parents. There is a latent/real bug here where
1976
# the parents inserted are not those commit would do - in particular
1977
# they are not filtered by heads(). RBC, AB
1978
for revision, tree in parent_trees.iteritems():
1979
if ie.file_id not in tree:
1981
parent_id = tree.inventory[ie.file_id].revision
1982
if parent_id in text_parents:
1984
text_parents.append((ie.file_id, parent_id))
1985
lines = revision_tree.get_file(ie.file_id).readlines()
1986
repository.texts.add_lines(text_key, text_parents, lines)
1988
# install the inventory
1989
repository.add_inventory(rev.revision_id, inv, present_parents)
1990
except errors.RevisionAlreadyPresent:
1992
if signature is not None:
1993
repository.add_signature_text(rev.revision_id, signature)
1994
repository.add_revision(rev.revision_id, rev, inv)
1997
class MetaDirRepository(Repository):
1998
"""Repositories in the new meta-dir layout.
2000
:ivar _transport: Transport for access to repository control files,
2001
typically pointing to .bzr/repository.
2004
def __init__(self, _format, a_bzrdir, control_files):
2005
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2006
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 the copied revision count and the failed revisions in a tuple:
2368
raise NotImplementedError(self.fetch)
2370
def _walk_to_common_revisions(self, revision_ids):
2371
"""Walk out from revision_ids in source to revisions target has.
2373
:param revision_ids: The start point for the search.
2374
:return: A set of revision ids.
2376
target_graph = self.target.get_graph()
2377
revision_ids = frozenset(revision_ids)
2378
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2379
return graph.SearchResult(revision_ids, set(), 0, set())
2380
missing_revs = set()
2381
source_graph = self.source.get_graph()
2382
# ensure we don't pay silly lookup costs.
2383
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2384
null_set = frozenset([_mod_revision.NULL_REVISION])
2387
next_revs, ghosts = searcher.next_with_ghosts()
2388
except StopIteration:
2390
if revision_ids.intersection(ghosts):
2391
absent_ids = set(revision_ids.intersection(ghosts))
2392
# If all absent_ids are present in target, no error is needed.
2393
absent_ids.difference_update(
2394
set(target_graph.get_parent_map(absent_ids)))
2396
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2397
# we don't care about other ghosts as we can't fetch them and
2398
# haven't been asked to.
2399
next_revs = set(next_revs)
2400
# we always have NULL_REVISION present.
2401
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2402
missing_revs.update(next_revs - have_revs)
2403
searcher.stop_searching_any(have_revs)
2404
return searcher.get_result()
2406
@deprecated_method(one_two)
2408
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2409
"""Return the revision ids that source has that target does not.
2411
These are returned in topological order.
2413
:param revision_id: only return revision ids included by this
2415
:param find_ghosts: If True find missing revisions in deep history
2416
rather than just finding the surface difference.
2418
return list(self.search_missing_revision_ids(
2419
revision_id, find_ghosts).get_keys())
2422
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2423
"""Return the revision ids that source has that target does not.
2425
:param revision_id: only return revision ids included by this
2427
:param find_ghosts: If True find missing revisions in deep history
2428
rather than just finding the surface difference.
2429
:return: A bzrlib.graph.SearchResult.
2431
# stop searching at found target revisions.
2432
if not find_ghosts and revision_id is not None:
2433
return self._walk_to_common_revisions([revision_id])
2434
# generic, possibly worst case, slow code path.
2435
target_ids = set(self.target.all_revision_ids())
2436
if revision_id is not None:
2437
source_ids = self.source.get_ancestry(revision_id)
2438
if source_ids[0] is not None:
2439
raise AssertionError()
2442
source_ids = self.source.all_revision_ids()
2443
result_set = set(source_ids).difference(target_ids)
2444
return self.source.revision_ids_to_search_result(result_set)
2447
def _same_model(source, target):
2448
"""True if source and target have the same data representation."""
2449
if source.supports_rich_root() != target.supports_rich_root():
2451
if source._serializer != target._serializer:
2456
class InterSameDataRepository(InterRepository):
2457
"""Code for converting between repositories that represent the same data.
2459
Data format and model must match for this to work.
2463
def _get_repo_format_to_test(self):
2464
"""Repository format for testing with.
2466
InterSameData can pull from subtree to subtree and from non-subtree to
2467
non-subtree, so we test this with the richest repository format.
2469
from bzrlib.repofmt import knitrepo
2470
return knitrepo.RepositoryFormatKnit3()
2473
def is_compatible(source, target):
2474
return InterRepository._same_model(source, target)
2477
def copy_content(self, revision_id=None):
2478
"""Make a complete copy of the content in self into destination.
2480
This copies both the repository's revision data, and configuration information
2481
such as the make_working_trees setting.
2483
This is a destructive operation! Do not use it on existing
2486
:param revision_id: Only copy the content needed to construct
2487
revision_id and its parents.
2490
self.target.set_make_working_trees(self.source.make_working_trees())
2491
except NotImplementedError:
2493
# but don't bother fetching if we have the needed data now.
2494
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2495
self.target.has_revision(revision_id)):
2497
self.target.fetch(self.source, revision_id=revision_id)
2500
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2501
"""See InterRepository.fetch()."""
2502
from bzrlib.fetch import RepoFetcher
2503
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2504
self.source, self.source._format, self.target,
2505
self.target._format)
2506
f = RepoFetcher(to_repository=self.target,
2507
from_repository=self.source,
2508
last_revision=revision_id,
2509
pb=pb, find_ghosts=find_ghosts)
2510
return f.count_copied, f.failed_revisions
2513
class InterWeaveRepo(InterSameDataRepository):
2514
"""Optimised code paths between Weave based repositories.
2516
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2517
implemented lazy inter-object optimisation.
2521
def _get_repo_format_to_test(self):
2522
from bzrlib.repofmt import weaverepo
2523
return weaverepo.RepositoryFormat7()
2526
def is_compatible(source, target):
2527
"""Be compatible with known Weave formats.
2529
We don't test for the stores being of specific types because that
2530
could lead to confusing results, and there is no need to be
2533
from bzrlib.repofmt.weaverepo import (
2539
return (isinstance(source._format, (RepositoryFormat5,
2541
RepositoryFormat7)) and
2542
isinstance(target._format, (RepositoryFormat5,
2544
RepositoryFormat7)))
2545
except AttributeError:
2549
def copy_content(self, revision_id=None):
2550
"""See InterRepository.copy_content()."""
2551
# weave specific optimised path:
2553
self.target.set_make_working_trees(self.source.make_working_trees())
2554
except (errors.RepositoryUpgradeRequired, NotImplemented):
2556
# FIXME do not peek!
2557
if self.source._transport.listable():
2558
pb = ui.ui_factory.nested_progress_bar()
2560
self.target.texts.insert_record_stream(
2561
self.source.texts.get_record_stream(
2562
self.source.texts.keys(), 'topological', False))
2563
pb.update('copying inventory', 0, 1)
2564
self.target.inventories.insert_record_stream(
2565
self.source.inventories.get_record_stream(
2566
self.source.inventories.keys(), 'topological', False))
2567
self.target.signatures.insert_record_stream(
2568
self.source.signatures.get_record_stream(
2569
self.source.signatures.keys(),
2571
self.target.revisions.insert_record_stream(
2572
self.source.revisions.get_record_stream(
2573
self.source.revisions.keys(),
2574
'topological', True))
2578
self.target.fetch(self.source, revision_id=revision_id)
2581
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2582
"""See InterRepository.fetch()."""
2583
from bzrlib.fetch import RepoFetcher
2584
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2585
self.source, self.source._format, self.target, self.target._format)
2586
f = RepoFetcher(to_repository=self.target,
2587
from_repository=self.source,
2588
last_revision=revision_id,
2589
pb=pb, find_ghosts=find_ghosts)
2590
return f.count_copied, f.failed_revisions
2593
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2594
"""See InterRepository.missing_revision_ids()."""
2595
# we want all revisions to satisfy revision_id in source.
2596
# but we don't want to stat every file here and there.
2597
# we want then, all revisions other needs to satisfy revision_id
2598
# checked, but not those that we have locally.
2599
# so the first thing is to get a subset of the revisions to
2600
# satisfy revision_id in source, and then eliminate those that
2601
# we do already have.
2602
# this is slow on high latency connection to self, but as as this
2603
# disk format scales terribly for push anyway due to rewriting
2604
# inventory.weave, this is considered acceptable.
2606
if revision_id is not None:
2607
source_ids = self.source.get_ancestry(revision_id)
2608
if source_ids[0] is not None:
2609
raise AssertionError()
2612
source_ids = self.source._all_possible_ids()
2613
source_ids_set = set(source_ids)
2614
# source_ids is the worst possible case we may need to pull.
2615
# now we want to filter source_ids against what we actually
2616
# have in target, but don't try to check for existence where we know
2617
# we do not have a revision as that would be pointless.
2618
target_ids = set(self.target._all_possible_ids())
2619
possibly_present_revisions = target_ids.intersection(source_ids_set)
2620
actually_present_revisions = set(
2621
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2622
required_revisions = source_ids_set.difference(actually_present_revisions)
2623
if revision_id is not None:
2624
# we used get_ancestry to determine source_ids then we are assured all
2625
# revisions referenced are present as they are installed in topological order.
2626
# and the tip revision was validated by get_ancestry.
2627
result_set = required_revisions
2629
# if we just grabbed the possibly available ids, then
2630
# we only have an estimate of whats available and need to validate
2631
# that against the revision records.
2633
self.source._eliminate_revisions_not_present(required_revisions))
2634
return self.source.revision_ids_to_search_result(result_set)
2637
class InterKnitRepo(InterSameDataRepository):
2638
"""Optimised code paths between Knit based repositories."""
2641
def _get_repo_format_to_test(self):
2642
from bzrlib.repofmt import knitrepo
2643
return knitrepo.RepositoryFormatKnit1()
2646
def is_compatible(source, target):
2647
"""Be compatible with known Knit formats.
2649
We don't test for the stores being of specific types because that
2650
could lead to confusing results, and there is no need to be
2653
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2655
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2656
isinstance(target._format, RepositoryFormatKnit))
2657
except AttributeError:
2659
return are_knits and InterRepository._same_model(source, target)
2662
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2663
"""See InterRepository.fetch()."""
2664
from bzrlib.fetch import RepoFetcher
2665
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2666
self.source, self.source._format, self.target, self.target._format)
2667
f = RepoFetcher(to_repository=self.target,
2668
from_repository=self.source,
2669
last_revision=revision_id,
2670
pb=pb, find_ghosts=find_ghosts)
2671
return f.count_copied, f.failed_revisions
2674
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2675
"""See InterRepository.missing_revision_ids()."""
2676
if revision_id is not None:
2677
source_ids = self.source.get_ancestry(revision_id)
2678
if source_ids[0] is not None:
2679
raise AssertionError()
2682
source_ids = self.source.all_revision_ids()
2683
source_ids_set = set(source_ids)
2684
# source_ids is the worst possible case we may need to pull.
2685
# now we want to filter source_ids against what we actually
2686
# have in target, but don't try to check for existence where we know
2687
# we do not have a revision as that would be pointless.
2688
target_ids = set(self.target.all_revision_ids())
2689
possibly_present_revisions = target_ids.intersection(source_ids_set)
2690
actually_present_revisions = set(
2691
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2692
required_revisions = source_ids_set.difference(actually_present_revisions)
2693
if revision_id is not None:
2694
# we used get_ancestry to determine source_ids then we are assured all
2695
# revisions referenced are present as they are installed in topological order.
2696
# and the tip revision was validated by get_ancestry.
2697
result_set = required_revisions
2699
# if we just grabbed the possibly available ids, then
2700
# we only have an estimate of whats available and need to validate
2701
# that against the revision records.
2703
self.source._eliminate_revisions_not_present(required_revisions))
2704
return self.source.revision_ids_to_search_result(result_set)
2707
class InterPackRepo(InterSameDataRepository):
2708
"""Optimised code paths between Pack based repositories."""
2711
def _get_repo_format_to_test(self):
2712
from bzrlib.repofmt import pack_repo
2713
return pack_repo.RepositoryFormatKnitPack1()
2716
def is_compatible(source, target):
2717
"""Be compatible with known Pack formats.
2719
We don't test for the stores being of specific types because that
2720
could lead to confusing results, and there is no need to be
2723
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2725
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2726
isinstance(target._format, RepositoryFormatPack))
2727
except AttributeError:
2729
return are_packs and InterRepository._same_model(source, target)
2732
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2733
"""See InterRepository.fetch()."""
2734
if len(self.source._fallback_repositories) > 0:
2735
from bzrlib.fetch import RepoFetcher
2736
fetcher = RepoFetcher(self.target, self.source, revision_id,
2738
return fetcher.count_copied, fetcher.failed_revisions
2739
from bzrlib.repofmt.pack_repo import Packer
2740
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2741
self.source, self.source._format, self.target, self.target._format)
2742
self.count_copied = 0
2743
if revision_id is None:
2745
# everything to do - use pack logic
2746
# to fetch from all packs to one without
2747
# inventory parsing etc, IFF nothing to be copied is in the target.
2749
source_revision_ids = frozenset(self.source.all_revision_ids())
2750
revision_ids = source_revision_ids - \
2751
frozenset(self.target.get_parent_map(source_revision_ids))
2752
revision_keys = [(revid,) for revid in revision_ids]
2753
index = self.target._pack_collection.revision_index.combined_index
2754
present_revision_ids = set(item[1][0] for item in
2755
index.iter_entries(revision_keys))
2756
revision_ids = set(revision_ids) - present_revision_ids
2757
# implementing the TODO will involve:
2758
# - detecting when all of a pack is selected
2759
# - avoiding as much as possible pre-selection, so the
2760
# more-core routines such as create_pack_from_packs can filter in
2761
# a just-in-time fashion. (though having a HEADS list on a
2762
# repository might make this a lot easier, because we could
2763
# sensibly detect 'new revisions' without doing a full index scan.
2764
elif _mod_revision.is_null(revision_id):
2769
revision_ids = self.search_missing_revision_ids(revision_id,
2770
find_ghosts=find_ghosts).get_keys()
2771
except errors.NoSuchRevision:
2772
raise errors.InstallFailed([revision_id])
2773
if len(revision_ids) == 0:
2775
packs = self.source._pack_collection.all_packs()
2776
pack = Packer(self.target._pack_collection, packs, '.fetch',
2777
revision_ids).pack()
2778
if pack is not None:
2779
self.target._pack_collection._save_pack_names()
2780
# Trigger an autopack. This may duplicate effort as we've just done
2781
# a pack creation, but for now it is simpler to think about as
2782
# 'upload data, then repack if needed'.
2783
self.target._pack_collection.autopack()
2784
return (pack.get_revision_count(), [])
2789
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2790
"""See InterRepository.missing_revision_ids().
2792
:param find_ghosts: Find ghosts throughout the ancestry of
2795
if not find_ghosts and revision_id is not None:
2796
return self._walk_to_common_revisions([revision_id])
2797
elif revision_id is not None:
2798
# Find ghosts: search for revisions pointing from one repository to
2799
# the other, and vice versa, anywhere in the history of revision_id.
2800
graph = self.target.get_graph(other_repository=self.source)
2801
searcher = graph._make_breadth_first_searcher([revision_id])
2805
next_revs, ghosts = searcher.next_with_ghosts()
2806
except StopIteration:
2808
if revision_id in ghosts:
2809
raise errors.NoSuchRevision(self.source, revision_id)
2810
found_ids.update(next_revs)
2811
found_ids.update(ghosts)
2812
found_ids = frozenset(found_ids)
2813
# Double query here: should be able to avoid this by changing the
2814
# graph api further.
2815
result_set = found_ids - frozenset(
2816
self.target.get_parent_map(found_ids))
2818
source_ids = self.source.all_revision_ids()
2819
# source_ids is the worst possible case we may need to pull.
2820
# now we want to filter source_ids against what we actually
2821
# have in target, but don't try to check for existence where we know
2822
# we do not have a revision as that would be pointless.
2823
target_ids = set(self.target.all_revision_ids())
2824
result_set = set(source_ids).difference(target_ids)
2825
return self.source.revision_ids_to_search_result(result_set)
2828
class InterModel1and2(InterRepository):
2831
def _get_repo_format_to_test(self):
2835
def is_compatible(source, target):
2836
if not source.supports_rich_root() and target.supports_rich_root():
2842
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2843
"""See InterRepository.fetch()."""
2844
from bzrlib.fetch import Model1toKnit2Fetcher
2845
f = Model1toKnit2Fetcher(to_repository=self.target,
2846
from_repository=self.source,
2847
last_revision=revision_id,
2848
pb=pb, find_ghosts=find_ghosts)
2849
return f.count_copied, f.failed_revisions
2852
def copy_content(self, revision_id=None):
2853
"""Make a complete copy of the content in self into destination.
2855
This is a destructive operation! Do not use it on existing
2858
:param revision_id: Only copy the content needed to construct
2859
revision_id and its parents.
2862
self.target.set_make_working_trees(self.source.make_working_trees())
2863
except NotImplementedError:
2865
# but don't bother fetching if we have the needed data now.
2866
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2867
self.target.has_revision(revision_id)):
2869
self.target.fetch(self.source, revision_id=revision_id)
2872
class InterKnit1and2(InterKnitRepo):
2875
def _get_repo_format_to_test(self):
2879
def is_compatible(source, target):
2880
"""Be compatible with Knit1 source and Knit3 target"""
2881
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
2883
from bzrlib.repofmt.knitrepo import (RepositoryFormatKnit1,
2884
RepositoryFormatKnit3)
2885
from bzrlib.repofmt.pack_repo import (
2886
RepositoryFormatKnitPack1,
2887
RepositoryFormatKnitPack3,
2888
RepositoryFormatPackDevelopment0,
2889
RepositoryFormatPackDevelopment0Subtree,
2892
RepositoryFormatKnit1,
2893
RepositoryFormatKnitPack1,
2894
RepositoryFormatPackDevelopment0,
2897
RepositoryFormatKnit3,
2898
RepositoryFormatKnitPack3,
2899
RepositoryFormatPackDevelopment0Subtree,
2901
return (isinstance(source._format, nosubtrees) and
2902
isinstance(target._format, subtrees))
2903
except AttributeError:
2907
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2908
"""See InterRepository.fetch()."""
2909
from bzrlib.fetch import Knit1to2Fetcher
2910
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2911
self.source, self.source._format, self.target,
2912
self.target._format)
2913
f = Knit1to2Fetcher(to_repository=self.target,
2914
from_repository=self.source,
2915
last_revision=revision_id,
2916
pb=pb, find_ghosts=find_ghosts)
2917
return f.count_copied, f.failed_revisions
2920
class InterDifferingSerializer(InterKnitRepo):
2923
def _get_repo_format_to_test(self):
2927
def is_compatible(source, target):
2928
"""Be compatible with Knit2 source and Knit3 target"""
2929
if source.supports_rich_root() != target.supports_rich_root():
2931
# Ideally, we'd support fetching if the source had no tree references
2932
# even if it supported them...
2933
if (getattr(source, '_format.supports_tree_reference', False) and
2934
not getattr(target, '_format.supports_tree_reference', False)):
2939
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2940
"""See InterRepository.fetch()."""
2941
revision_ids = self.target.search_missing_revision_ids(self.source,
2942
revision_id, find_ghosts=find_ghosts).get_keys()
2943
revision_ids = tsort.topo_sort(
2944
self.source.get_graph().get_parent_map(revision_ids))
2945
def revisions_iterator():
2946
for current_revision_id in revision_ids:
2947
revision = self.source.get_revision(current_revision_id)
2948
tree = self.source.revision_tree(current_revision_id)
2950
signature = self.source.get_signature_text(
2951
current_revision_id)
2952
except errors.NoSuchRevision:
2954
yield revision, tree, signature
2956
my_pb = ui.ui_factory.nested_progress_bar()
2961
install_revisions(self.target, revisions_iterator(),
2962
len(revision_ids), pb)
2964
if my_pb is not None:
2966
return len(revision_ids), 0
2969
class InterOtherToRemote(InterRepository):
2971
def __init__(self, source, target):
2972
InterRepository.__init__(self, source, target)
2973
self._real_inter = None
2976
def is_compatible(source, target):
2977
if isinstance(target, remote.RemoteRepository):
2981
def _ensure_real_inter(self):
2982
if self._real_inter is None:
2983
self.target._ensure_real()
2984
real_target = self.target._real_repository
2985
self._real_inter = InterRepository.get(self.source, real_target)
2987
def copy_content(self, revision_id=None):
2988
self._ensure_real_inter()
2989
self._real_inter.copy_content(revision_id=revision_id)
2991
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2992
self._ensure_real_inter()
2993
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
2994
find_ghosts=find_ghosts)
2997
def _get_repo_format_to_test(self):
3001
class InterRemoteToOther(InterRepository):
3003
def __init__(self, source, target):
3004
InterRepository.__init__(self, source, target)
3005
self._real_inter = None
3008
def is_compatible(source, target):
3009
if not isinstance(source, remote.RemoteRepository):
3011
# Is source's model compatible with target's model?
3012
source._ensure_real()
3013
real_source = source._real_repository
3014
if isinstance(real_source, remote.RemoteRepository):
3015
raise NotImplementedError(
3016
"We don't support remote repos backed by remote repos yet.")
3017
return InterRepository._same_model(real_source, target)
3019
def _ensure_real_inter(self):
3020
if self._real_inter is None:
3021
self.source._ensure_real()
3022
real_source = self.source._real_repository
3023
self._real_inter = InterRepository.get(real_source, self.target)
3025
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3026
self._ensure_real_inter()
3027
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3028
find_ghosts=find_ghosts)
3030
def copy_content(self, revision_id=None):
3031
self._ensure_real_inter()
3032
self._real_inter.copy_content(revision_id=revision_id)
3035
def _get_repo_format_to_test(self):
3040
InterRepository.register_optimiser(InterDifferingSerializer)
3041
InterRepository.register_optimiser(InterSameDataRepository)
3042
InterRepository.register_optimiser(InterWeaveRepo)
3043
InterRepository.register_optimiser(InterKnitRepo)
3044
InterRepository.register_optimiser(InterModel1and2)
3045
InterRepository.register_optimiser(InterKnit1and2)
3046
InterRepository.register_optimiser(InterPackRepo)
3047
InterRepository.register_optimiser(InterOtherToRemote)
3048
InterRepository.register_optimiser(InterRemoteToOther)
3051
class CopyConverter(object):
3052
"""A repository conversion tool which just performs a copy of the content.
3054
This is slow but quite reliable.
3057
def __init__(self, target_format):
3058
"""Create a CopyConverter.
3060
:param target_format: The format the resulting repository should be.
3062
self.target_format = target_format
3064
def convert(self, repo, pb):
3065
"""Perform the conversion of to_convert, giving feedback via pb.
3067
:param to_convert: The disk object to convert.
3068
:param pb: a progress bar to use for progress information.
3073
# this is only useful with metadir layouts - separated repo content.
3074
# trigger an assertion if not such
3075
repo._format.get_format_string()
3076
self.repo_dir = repo.bzrdir
3077
self.step('Moving repository to repository.backup')
3078
self.repo_dir.transport.move('repository', 'repository.backup')
3079
backup_transport = self.repo_dir.transport.clone('repository.backup')
3080
repo._format.check_conversion_target(self.target_format)
3081
self.source_repo = repo._format.open(self.repo_dir,
3083
_override_transport=backup_transport)
3084
self.step('Creating new repository')
3085
converted = self.target_format.initialize(self.repo_dir,
3086
self.source_repo.is_shared())
3087
converted.lock_write()
3089
self.step('Copying content into repository.')
3090
self.source_repo.copy_content_into(converted)
3093
self.step('Deleting old repository content.')
3094
self.repo_dir.transport.delete_tree('repository.backup')
3095
self.pb.note('repository converted')
3097
def step(self, message):
3098
"""Update the pb by a step."""
3100
self.pb.update(message, self.count, self.total)
3112
def _unescaper(match, _map=_unescape_map):
3113
code = match.group(1)
3117
if not code.startswith('#'):
3119
return unichr(int(code[1:])).encode('utf8')
3125
def _unescape_xml(data):
3126
"""Unescape predefined XML entities in a string of data."""
3128
if _unescape_re is None:
3129
_unescape_re = re.compile('\&([^;]*);')
3130
return _unescape_re.sub(_unescaper, data)
3133
class _VersionedFileChecker(object):
3135
def __init__(self, repository):
3136
self.repository = repository
3137
self.text_index = self.repository._generate_text_key_index()
3139
def calculate_file_version_parents(self, text_key):
3140
"""Calculate the correct parents for a file version according to
3143
parent_keys = self.text_index[text_key]
3144
if parent_keys == [_mod_revision.NULL_REVISION]:
3146
return tuple(parent_keys)
3148
def check_file_version_parents(self, texts, progress_bar=None):
3149
"""Check the parents stored in a versioned file are correct.
3151
It also detects file versions that are not referenced by their
3152
corresponding revision's inventory.
3154
:returns: A tuple of (wrong_parents, dangling_file_versions).
3155
wrong_parents is a dict mapping {revision_id: (stored_parents,
3156
correct_parents)} for each revision_id where the stored parents
3157
are not correct. dangling_file_versions is a set of (file_id,
3158
revision_id) tuples for versions that are present in this versioned
3159
file, but not used by the corresponding inventory.
3162
self.file_ids = set([file_id for file_id, _ in
3163
self.text_index.iterkeys()])
3164
# text keys is now grouped by file_id
3165
n_weaves = len(self.file_ids)
3166
files_in_revisions = {}
3167
revisions_of_files = {}
3168
n_versions = len(self.text_index)
3169
progress_bar.update('loading text store', 0, n_versions)
3170
parent_map = self.repository.texts.get_parent_map(self.text_index)
3171
# On unlistable transports this could well be empty/error...
3172
text_keys = self.repository.texts.keys()
3173
unused_keys = frozenset(text_keys) - set(self.text_index)
3174
for num, key in enumerate(self.text_index.iterkeys()):
3175
if progress_bar is not None:
3176
progress_bar.update('checking text graph', num, n_versions)
3177
correct_parents = self.calculate_file_version_parents(key)
3179
knit_parents = parent_map[key]
3180
except errors.RevisionNotPresent:
3183
if correct_parents != knit_parents:
3184
wrong_parents[key] = (knit_parents, correct_parents)
3185
return wrong_parents, unused_keys
3188
def _old_get_graph(repository, revision_id):
3189
"""DO NOT USE. That is all. I'm serious."""
3190
graph = repository.get_graph()
3191
revision_graph = dict(((key, value) for key, value in
3192
graph.iter_ancestry([revision_id]) if value is not None))
3193
return _strip_NULL_ghosts(revision_graph)
3196
def _strip_NULL_ghosts(revision_graph):
3197
"""Also don't use this. more compatibility code for unmigrated clients."""
3198
# Filter ghosts, and null:
3199
if _mod_revision.NULL_REVISION in revision_graph:
3200
del revision_graph[_mod_revision.NULL_REVISION]
3201
for key, parents in revision_graph.items():
3202
revision_graph[key] = tuple(parent for parent in parents if parent
3204
return revision_graph
added
removed
bzrlib/repository.py
