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
# if there is no specific appropriate InterRepository, this will get
971
# the InterRepository base class, which raises an
972
# IncompatibleRepositories when asked to fetch.
973
inter = InterRepository.get(source, self)
974
return inter.fetch(revision_id=revision_id, pb=pb,
975
find_ghosts=find_ghosts)
977
def create_bundle(self, target, base, fileobj, format=None):
978
return serializer.write_bundle(self, target, base, fileobj, format)
980
def get_commit_builder(self, branch, parents, config, timestamp=None,
981
timezone=None, committer=None, revprops=None,
983
"""Obtain a CommitBuilder for this repository.
985
:param branch: Branch to commit to.
986
:param parents: Revision ids of the parents of the new revision.
987
:param config: Configuration to use.
988
:param timestamp: Optional timestamp recorded for commit.
989
:param timezone: Optional timezone for timestamp.
990
:param committer: Optional committer to set for commit.
991
:param revprops: Optional dictionary of revision properties.
992
:param revision_id: Optional revision id.
994
result = self._commit_builder_class(self, parents, config,
995
timestamp, timezone, committer, revprops, revision_id)
996
self.start_write_group()
1000
if (self.control_files._lock_count == 1 and
1001
self.control_files._lock_mode == 'w'):
1002
if self._write_group is not None:
1003
self.abort_write_group()
1004
self.control_files.unlock()
1005
raise errors.BzrError(
1006
'Must end write groups before releasing write locks.')
1007
self.control_files.unlock()
1008
for repo in self._fallback_repositories:
1012
def clone(self, a_bzrdir, revision_id=None):
1013
"""Clone this repository into a_bzrdir using the current format.
1015
Currently no check is made that the format of this repository and
1016
the bzrdir format are compatible. FIXME RBC 20060201.
1018
:return: The newly created destination repository.
1020
# TODO: deprecate after 0.16; cloning this with all its settings is
1021
# probably not very useful -- mbp 20070423
1022
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1023
self.copy_content_into(dest_repo, revision_id)
1026
def start_write_group(self):
1027
"""Start a write group in the repository.
1029
Write groups are used by repositories which do not have a 1:1 mapping
1030
between file ids and backend store to manage the insertion of data from
1031
both fetch and commit operations.
1033
A write lock is required around the start_write_group/commit_write_group
1034
for the support of lock-requiring repository formats.
1036
One can only insert data into a repository inside a write group.
1040
if not self.is_write_locked():
1041
raise errors.NotWriteLocked(self)
1042
if self._write_group:
1043
raise errors.BzrError('already in a write group')
1044
self._start_write_group()
1045
# so we can detect unlock/relock - the write group is now entered.
1046
self._write_group = self.get_transaction()
1048
def _start_write_group(self):
1049
"""Template method for per-repository write group startup.
1051
This is called before the write group is considered to be
1056
def sprout(self, to_bzrdir, revision_id=None):
1057
"""Create a descendent repository for new development.
1059
Unlike clone, this does not copy the settings of the repository.
1061
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1062
dest_repo.fetch(self, revision_id=revision_id)
1065
def _create_sprouting_repo(self, a_bzrdir, shared):
1066
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1067
# use target default format.
1068
dest_repo = a_bzrdir.create_repository()
1070
# Most control formats need the repository to be specifically
1071
# created, but on some old all-in-one formats it's not needed
1073
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1074
except errors.UninitializableFormat:
1075
dest_repo = a_bzrdir.open_repository()
1079
def has_revision(self, revision_id):
1080
"""True if this repository has a copy of the revision."""
1081
return revision_id in self.has_revisions((revision_id,))
1084
def has_revisions(self, revision_ids):
1085
"""Probe to find out the presence of multiple revisions.
1087
:param revision_ids: An iterable of revision_ids.
1088
:return: A set of the revision_ids that were present.
1090
parent_map = self.revisions.get_parent_map(
1091
[(rev_id,) for rev_id in revision_ids])
1093
if _mod_revision.NULL_REVISION in revision_ids:
1094
result.add(_mod_revision.NULL_REVISION)
1095
result.update([key[0] for key in parent_map])
1099
def get_revision(self, revision_id):
1100
"""Return the Revision object for a named revision."""
1101
return self.get_revisions([revision_id])[0]
1104
def get_revision_reconcile(self, revision_id):
1105
"""'reconcile' helper routine that allows access to a revision always.
1107
This variant of get_revision does not cross check the weave graph
1108
against the revision one as get_revision does: but it should only
1109
be used by reconcile, or reconcile-alike commands that are correcting
1110
or testing the revision graph.
1112
return self._get_revisions([revision_id])[0]
1115
def get_revisions(self, revision_ids):
1116
"""Get many revisions at once."""
1117
return self._get_revisions(revision_ids)
1120
def _get_revisions(self, revision_ids):
1121
"""Core work logic to get many revisions without sanity checks."""
1122
for rev_id in revision_ids:
1123
if not rev_id or not isinstance(rev_id, basestring):
1124
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1125
keys = [(key,) for key in revision_ids]
1126
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1128
for record in stream:
1129
if record.storage_kind == 'absent':
1130
raise errors.NoSuchRevision(self, record.key[0])
1131
text = record.get_bytes_as('fulltext')
1132
rev = self._serializer.read_revision_from_string(text)
1133
revs[record.key[0]] = rev
1134
return [revs[revid] for revid in revision_ids]
1137
def get_revision_xml(self, revision_id):
1138
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1139
# would have already do it.
1140
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1141
rev = self.get_revision(revision_id)
1142
rev_tmp = StringIO()
1143
# the current serializer..
1144
self._serializer.write_revision(rev, rev_tmp)
1146
return rev_tmp.getvalue()
1148
def get_deltas_for_revisions(self, revisions):
1149
"""Produce a generator of revision deltas.
1151
Note that the input is a sequence of REVISIONS, not revision_ids.
1152
Trees will be held in memory until the generator exits.
1153
Each delta is relative to the revision's lefthand predecessor.
1155
required_trees = set()
1156
for revision in revisions:
1157
required_trees.add(revision.revision_id)
1158
required_trees.update(revision.parent_ids[:1])
1159
trees = dict((t.get_revision_id(), t) for
1160
t in self.revision_trees(required_trees))
1161
for revision in revisions:
1162
if not revision.parent_ids:
1163
old_tree = self.revision_tree(None)
1165
old_tree = trees[revision.parent_ids[0]]
1166
yield trees[revision.revision_id].changes_from(old_tree)
1169
def get_revision_delta(self, revision_id):
1170
"""Return the delta for one revision.
1172
The delta is relative to the left-hand predecessor of the
1175
r = self.get_revision(revision_id)
1176
return list(self.get_deltas_for_revisions([r]))[0]
1179
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1180
signature = gpg_strategy.sign(plaintext)
1181
self.add_signature_text(revision_id, signature)
1184
def add_signature_text(self, revision_id, signature):
1185
self.signatures.add_lines((revision_id,), (),
1186
osutils.split_lines(signature))
1188
def find_text_key_references(self):
1189
"""Find the text key references within the repository.
1191
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1192
revision_ids. Each altered file-ids has the exact revision_ids that
1193
altered it listed explicitly.
1194
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1195
to whether they were referred to by the inventory of the
1196
revision_id that they contain. The inventory texts from all present
1197
revision ids are assessed to generate this report.
1199
revision_keys = self.revisions.keys()
1200
w = self.inventories
1201
pb = ui.ui_factory.nested_progress_bar()
1203
return self._find_text_key_references_from_xml_inventory_lines(
1204
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1208
def _find_text_key_references_from_xml_inventory_lines(self,
1210
"""Core routine for extracting references to texts from inventories.
1212
This performs the translation of xml lines to revision ids.
1214
:param line_iterator: An iterator of lines, origin_version_id
1215
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1216
to whether they were referred to by the inventory of the
1217
revision_id that they contain. Note that if that revision_id was
1218
not part of the line_iterator's output then False will be given -
1219
even though it may actually refer to that key.
1221
if not self._serializer.support_altered_by_hack:
1222
raise AssertionError(
1223
"_find_text_key_references_from_xml_inventory_lines only "
1224
"supported for branches which store inventory as unnested xml"
1225
", not on %r" % self)
1228
# this code needs to read every new line in every inventory for the
1229
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1230
# not present in one of those inventories is unnecessary but not
1231
# harmful because we are filtering by the revision id marker in the
1232
# inventory lines : we only select file ids altered in one of those
1233
# revisions. We don't need to see all lines in the inventory because
1234
# only those added in an inventory in rev X can contain a revision=X
1236
unescape_revid_cache = {}
1237
unescape_fileid_cache = {}
1239
# jam 20061218 In a big fetch, this handles hundreds of thousands
1240
# of lines, so it has had a lot of inlining and optimizing done.
1241
# Sorry that it is a little bit messy.
1242
# Move several functions to be local variables, since this is a long
1244
search = self._file_ids_altered_regex.search
1245
unescape = _unescape_xml
1246
setdefault = result.setdefault
1247
for line, line_key in line_iterator:
1248
match = search(line)
1251
# One call to match.group() returning multiple items is quite a
1252
# bit faster than 2 calls to match.group() each returning 1
1253
file_id, revision_id = match.group('file_id', 'revision_id')
1255
# Inlining the cache lookups helps a lot when you make 170,000
1256
# lines and 350k ids, versus 8.4 unique ids.
1257
# Using a cache helps in 2 ways:
1258
# 1) Avoids unnecessary decoding calls
1259
# 2) Re-uses cached strings, which helps in future set and
1261
# (2) is enough that removing encoding entirely along with
1262
# the cache (so we are using plain strings) results in no
1263
# performance improvement.
1265
revision_id = unescape_revid_cache[revision_id]
1267
unescaped = unescape(revision_id)
1268
unescape_revid_cache[revision_id] = unescaped
1269
revision_id = unescaped
1271
# Note that unconditionally unescaping means that we deserialise
1272
# every fileid, which for general 'pull' is not great, but we don't
1273
# really want to have some many fulltexts that this matters anyway.
1276
file_id = unescape_fileid_cache[file_id]
1278
unescaped = unescape(file_id)
1279
unescape_fileid_cache[file_id] = unescaped
1282
key = (file_id, revision_id)
1283
setdefault(key, False)
1284
if revision_id == line_key[-1]:
1288
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1290
"""Helper routine for fileids_altered_by_revision_ids.
1292
This performs the translation of xml lines to revision ids.
1294
:param line_iterator: An iterator of lines, origin_version_id
1295
:param revision_ids: The revision ids to filter for. This should be a
1296
set or other type which supports efficient __contains__ lookups, as
1297
the revision id from each parsed line will be looked up in the
1298
revision_ids filter.
1299
:return: a dictionary mapping altered file-ids to an iterable of
1300
revision_ids. Each altered file-ids has the exact revision_ids that
1301
altered it listed explicitly.
1304
setdefault = result.setdefault
1306
self._find_text_key_references_from_xml_inventory_lines(
1307
line_iterator).iterkeys():
1308
# once data is all ensured-consistent; then this is
1309
# if revision_id == version_id
1310
if key[-1:] in revision_ids:
1311
setdefault(key[0], set()).add(key[-1])
1314
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1315
"""Find the file ids and versions affected by revisions.
1317
:param revisions: an iterable containing revision ids.
1318
:param _inv_weave: The inventory weave from this repository or None.
1319
If None, the inventory weave will be opened automatically.
1320
:return: a dictionary mapping altered file-ids to an iterable of
1321
revision_ids. Each altered file-ids has the exact revision_ids that
1322
altered it listed explicitly.
1324
selected_keys = set((revid,) for revid in revision_ids)
1325
w = _inv_weave or self.inventories
1326
pb = ui.ui_factory.nested_progress_bar()
1328
return self._find_file_ids_from_xml_inventory_lines(
1329
w.iter_lines_added_or_present_in_keys(
1330
selected_keys, pb=pb),
1335
def iter_files_bytes(self, desired_files):
1336
"""Iterate through file versions.
1338
Files will not necessarily be returned in the order they occur in
1339
desired_files. No specific order is guaranteed.
1341
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1342
value supplied by the caller as part of desired_files. It should
1343
uniquely identify the file version in the caller's context. (Examples:
1344
an index number or a TreeTransform trans_id.)
1346
bytes_iterator is an iterable of bytestrings for the file. The
1347
kind of iterable and length of the bytestrings are unspecified, but for
1348
this implementation, it is a list of bytes produced by
1349
VersionedFile.get_record_stream().
1351
:param desired_files: a list of (file_id, revision_id, identifier)
1354
transaction = self.get_transaction()
1356
for file_id, revision_id, callable_data in desired_files:
1357
text_keys[(file_id, revision_id)] = callable_data
1358
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1359
if record.storage_kind == 'absent':
1360
raise errors.RevisionNotPresent(record.key, self)
1361
yield text_keys[record.key], record.get_bytes_as('fulltext')
1363
def _generate_text_key_index(self, text_key_references=None,
1365
"""Generate a new text key index for the repository.
1367
This is an expensive function that will take considerable time to run.
1369
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1370
list of parents, also text keys. When a given key has no parents,
1371
the parents list will be [NULL_REVISION].
1373
# All revisions, to find inventory parents.
1374
if ancestors is None:
1375
graph = self.get_graph()
1376
ancestors = graph.get_parent_map(self.all_revision_ids())
1377
if text_key_references is None:
1378
text_key_references = self.find_text_key_references()
1379
pb = ui.ui_factory.nested_progress_bar()
1381
return self._do_generate_text_key_index(ancestors,
1382
text_key_references, pb)
1386
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1387
"""Helper for _generate_text_key_index to avoid deep nesting."""
1388
revision_order = tsort.topo_sort(ancestors)
1389
invalid_keys = set()
1391
for revision_id in revision_order:
1392
revision_keys[revision_id] = set()
1393
text_count = len(text_key_references)
1394
# a cache of the text keys to allow reuse; costs a dict of all the
1395
# keys, but saves a 2-tuple for every child of a given key.
1397
for text_key, valid in text_key_references.iteritems():
1399
invalid_keys.add(text_key)
1401
revision_keys[text_key[1]].add(text_key)
1402
text_key_cache[text_key] = text_key
1403
del text_key_references
1405
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1406
NULL_REVISION = _mod_revision.NULL_REVISION
1407
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1408
# too small for large or very branchy trees. However, for 55K path
1409
# trees, it would be easy to use too much memory trivially. Ideally we
1410
# could gauge this by looking at available real memory etc, but this is
1411
# always a tricky proposition.
1412
inventory_cache = lru_cache.LRUCache(10)
1413
batch_size = 10 # should be ~150MB on a 55K path tree
1414
batch_count = len(revision_order) / batch_size + 1
1416
pb.update("Calculating text parents.", processed_texts, text_count)
1417
for offset in xrange(batch_count):
1418
to_query = revision_order[offset * batch_size:(offset + 1) *
1422
for rev_tree in self.revision_trees(to_query):
1423
revision_id = rev_tree.get_revision_id()
1424
parent_ids = ancestors[revision_id]
1425
for text_key in revision_keys[revision_id]:
1426
pb.update("Calculating text parents.", processed_texts)
1427
processed_texts += 1
1428
candidate_parents = []
1429
for parent_id in parent_ids:
1430
parent_text_key = (text_key[0], parent_id)
1432
check_parent = parent_text_key not in \
1433
revision_keys[parent_id]
1435
# the parent parent_id is a ghost:
1436
check_parent = False
1437
# truncate the derived graph against this ghost.
1438
parent_text_key = None
1440
# look at the parent commit details inventories to
1441
# determine possible candidates in the per file graph.
1444
inv = inventory_cache[parent_id]
1446
inv = self.revision_tree(parent_id).inventory
1447
inventory_cache[parent_id] = inv
1448
parent_entry = inv._byid.get(text_key[0], None)
1449
if parent_entry is not None:
1451
text_key[0], parent_entry.revision)
1453
parent_text_key = None
1454
if parent_text_key is not None:
1455
candidate_parents.append(
1456
text_key_cache[parent_text_key])
1457
parent_heads = text_graph.heads(candidate_parents)
1458
new_parents = list(parent_heads)
1459
new_parents.sort(key=lambda x:candidate_parents.index(x))
1460
if new_parents == []:
1461
new_parents = [NULL_REVISION]
1462
text_index[text_key] = new_parents
1464
for text_key in invalid_keys:
1465
text_index[text_key] = [NULL_REVISION]
1468
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1469
"""Get an iterable listing the keys of all the data introduced by a set
1472
The keys will be ordered so that the corresponding items can be safely
1473
fetched and inserted in that order.
1475
:returns: An iterable producing tuples of (knit-kind, file-id,
1476
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1477
'revisions'. file-id is None unless knit-kind is 'file'.
1479
# XXX: it's a bit weird to control the inventory weave caching in this
1480
# generator. Ideally the caching would be done in fetch.py I think. Or
1481
# maybe this generator should explicitly have the contract that it
1482
# should not be iterated until the previously yielded item has been
1484
inv_w = self.inventories
1486
# file ids that changed
1487
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1489
num_file_ids = len(file_ids)
1490
for file_id, altered_versions in file_ids.iteritems():
1491
if _files_pb is not None:
1492
_files_pb.update("fetch texts", count, num_file_ids)
1494
yield ("file", file_id, altered_versions)
1495
# We're done with the files_pb. Note that it finished by the caller,
1496
# just as it was created by the caller.
1500
yield ("inventory", None, revision_ids)
1503
revisions_with_signatures = set()
1504
for rev_id in revision_ids:
1506
self.get_signature_text(rev_id)
1507
except errors.NoSuchRevision:
1511
revisions_with_signatures.add(rev_id)
1512
yield ("signatures", None, revisions_with_signatures)
1515
yield ("revisions", None, revision_ids)
1518
def get_inventory(self, revision_id):
1519
"""Get Inventory object by revision id."""
1520
return self.iter_inventories([revision_id]).next()
1522
def iter_inventories(self, revision_ids):
1523
"""Get many inventories by revision_ids.
1525
This will buffer some or all of the texts used in constructing the
1526
inventories in memory, but will only parse a single inventory at a
1529
:return: An iterator of inventories.
1531
if ((None in revision_ids)
1532
or (_mod_revision.NULL_REVISION in revision_ids)):
1533
raise ValueError('cannot get null revision inventory')
1534
return self._iter_inventories(revision_ids)
1536
def _iter_inventories(self, revision_ids):
1537
"""single-document based inventory iteration."""
1538
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1539
yield self.deserialise_inventory(revision_id, text)
1541
def _iter_inventory_xmls(self, revision_ids):
1542
keys = [(revision_id,) for revision_id in revision_ids]
1543
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1545
for record in stream:
1546
if record.storage_kind != 'absent':
1547
texts[record.key] = record.get_bytes_as('fulltext')
1549
raise errors.NoSuchRevision(self, record.key)
1551
yield texts[key], key[-1]
1553
def deserialise_inventory(self, revision_id, xml):
1554
"""Transform the xml into an inventory object.
1556
:param revision_id: The expected revision id of the inventory.
1557
:param xml: A serialised inventory.
1559
result = self._serializer.read_inventory_from_string(xml, revision_id)
1560
if result.revision_id != revision_id:
1561
raise AssertionError('revision id mismatch %s != %s' % (
1562
result.revision_id, revision_id))
1565
def serialise_inventory(self, inv):
1566
return self._serializer.write_inventory_to_string(inv)
1568
def _serialise_inventory_to_lines(self, inv):
1569
return self._serializer.write_inventory_to_lines(inv)
1571
def get_serializer_format(self):
1572
return self._serializer.format_num
1575
def get_inventory_xml(self, revision_id):
1576
"""Get inventory XML as a file object."""
1577
texts = self._iter_inventory_xmls([revision_id])
1579
text, revision_id = texts.next()
1580
except StopIteration:
1581
raise errors.HistoryMissing(self, 'inventory', revision_id)
1585
def get_inventory_sha1(self, revision_id):
1586
"""Return the sha1 hash of the inventory entry
1588
return self.get_revision(revision_id).inventory_sha1
1590
def iter_reverse_revision_history(self, revision_id):
1591
"""Iterate backwards through revision ids in the lefthand history
1593
:param revision_id: The revision id to start with. All its lefthand
1594
ancestors will be traversed.
1596
graph = self.get_graph()
1597
next_id = revision_id
1599
if next_id in (None, _mod_revision.NULL_REVISION):
1602
# Note: The following line may raise KeyError in the event of
1603
# truncated history. We decided not to have a try:except:raise
1604
# RevisionNotPresent here until we see a use for it, because of the
1605
# cost in an inner loop that is by its very nature O(history).
1606
# Robert Collins 20080326
1607
parents = graph.get_parent_map([next_id])[next_id]
1608
if len(parents) == 0:
1611
next_id = parents[0]
1614
def get_revision_inventory(self, revision_id):
1615
"""Return inventory of a past revision."""
1616
# TODO: Unify this with get_inventory()
1617
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1618
# must be the same as its revision, so this is trivial.
1619
if revision_id is None:
1620
# This does not make sense: if there is no revision,
1621
# then it is the current tree inventory surely ?!
1622
# and thus get_root_id() is something that looks at the last
1623
# commit on the branch, and the get_root_id is an inventory check.
1624
raise NotImplementedError
1625
# return Inventory(self.get_root_id())
1627
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
2008
def is_shared(self):
2009
"""Return True if this repository is flagged as a shared repository."""
2010
return self._transport.has('shared-storage')
2013
def set_make_working_trees(self, new_value):
2014
"""Set the policy flag for making working trees when creating branches.
2016
This only applies to branches that use this repository.
2018
The default is 'True'.
2019
:param new_value: True to restore the default, False to disable making
2024
self._transport.delete('no-working-trees')
2025
except errors.NoSuchFile:
2028
self._transport.put_bytes('no-working-trees', '',
2029
mode=self.bzrdir._get_file_mode())
2031
def make_working_trees(self):
2032
"""Returns the policy for making working trees on new branches."""
2033
return not self._transport.has('no-working-trees')
2036
class MetaDirVersionedFileRepository(MetaDirRepository):
2037
"""Repositories in a meta-dir, that work via versioned file objects."""
2039
def __init__(self, _format, a_bzrdir, control_files):
2040
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2044
class RepositoryFormatRegistry(registry.Registry):
2045
"""Registry of RepositoryFormats."""
2047
def get(self, format_string):
2048
r = registry.Registry.get(self, format_string)
2054
format_registry = RepositoryFormatRegistry()
2055
"""Registry of formats, indexed by their identifying format string.
2057
This can contain either format instances themselves, or classes/factories that
2058
can be called to obtain one.
2062
#####################################################################
2063
# Repository Formats
2065
class RepositoryFormat(object):
2066
"""A repository format.
2068
Formats provide three things:
2069
* An initialization routine to construct repository data on disk.
2070
* a format string which is used when the BzrDir supports versioned
2072
* an open routine which returns a Repository instance.
2074
There is one and only one Format subclass for each on-disk format. But
2075
there can be one Repository subclass that is used for several different
2076
formats. The _format attribute on a Repository instance can be used to
2077
determine the disk format.
2079
Formats are placed in an dict by their format string for reference
2080
during opening. These should be subclasses of RepositoryFormat
2083
Once a format is deprecated, just deprecate the initialize and open
2084
methods on the format class. Do not deprecate the object, as the
2085
object will be created every system load.
2087
Common instance attributes:
2088
_matchingbzrdir - the bzrdir format that the repository format was
2089
originally written to work with. This can be used if manually
2090
constructing a bzrdir and repository, or more commonly for test suite
2094
# Set to True or False in derived classes. True indicates that the format
2095
# supports ghosts gracefully.
2096
supports_ghosts = None
2097
# Can this repository be given external locations to lookup additional
2098
# data. Set to True or False in derived classes.
2099
supports_external_lookups = None
2102
return "<%s>" % self.__class__.__name__
2104
def __eq__(self, other):
2105
# format objects are generally stateless
2106
return isinstance(other, self.__class__)
2108
def __ne__(self, other):
2109
return not self == other
2112
def find_format(klass, a_bzrdir):
2113
"""Return the format for the repository object in a_bzrdir.
2115
This is used by bzr native formats that have a "format" file in
2116
the repository. Other methods may be used by different types of
2120
transport = a_bzrdir.get_repository_transport(None)
2121
format_string = transport.get("format").read()
2122
return format_registry.get(format_string)
2123
except errors.NoSuchFile:
2124
raise errors.NoRepositoryPresent(a_bzrdir)
2126
raise errors.UnknownFormatError(format=format_string,
2130
def register_format(klass, format):
2131
format_registry.register(format.get_format_string(), format)
2134
def unregister_format(klass, format):
2135
format_registry.remove(format.get_format_string())
2138
def get_default_format(klass):
2139
"""Return the current default format."""
2140
from bzrlib import bzrdir
2141
return bzrdir.format_registry.make_bzrdir('default').repository_format
2143
def get_format_string(self):
2144
"""Return the ASCII format string that identifies this format.
2146
Note that in pre format ?? repositories the format string is
2147
not permitted nor written to disk.
2149
raise NotImplementedError(self.get_format_string)
2151
def get_format_description(self):
2152
"""Return the short description for this format."""
2153
raise NotImplementedError(self.get_format_description)
2155
# TODO: this shouldn't be in the base class, it's specific to things that
2156
# use weaves or knits -- mbp 20070207
2157
def _get_versioned_file_store(self,
2162
versionedfile_class=None,
2163
versionedfile_kwargs={},
2165
if versionedfile_class is None:
2166
versionedfile_class = self._versionedfile_class
2167
weave_transport = control_files._transport.clone(name)
2168
dir_mode = control_files._dir_mode
2169
file_mode = control_files._file_mode
2170
return VersionedFileStore(weave_transport, prefixed=prefixed,
2172
file_mode=file_mode,
2173
versionedfile_class=versionedfile_class,
2174
versionedfile_kwargs=versionedfile_kwargs,
2177
def initialize(self, a_bzrdir, shared=False):
2178
"""Initialize a repository of this format in a_bzrdir.
2180
:param a_bzrdir: The bzrdir to put the new repository in it.
2181
:param shared: The repository should be initialized as a sharable one.
2182
:returns: The new repository object.
2184
This may raise UninitializableFormat if shared repository are not
2185
compatible the a_bzrdir.
2187
raise NotImplementedError(self.initialize)
2189
def is_supported(self):
2190
"""Is this format supported?
2192
Supported formats must be initializable and openable.
2193
Unsupported formats may not support initialization or committing or
2194
some other features depending on the reason for not being supported.
2198
def check_conversion_target(self, target_format):
2199
raise NotImplementedError(self.check_conversion_target)
2201
def open(self, a_bzrdir, _found=False):
2202
"""Return an instance of this format for the bzrdir a_bzrdir.
2204
_found is a private parameter, do not use it.
2206
raise NotImplementedError(self.open)
2209
class MetaDirRepositoryFormat(RepositoryFormat):
2210
"""Common base class for the new repositories using the metadir layout."""
2212
rich_root_data = False
2213
supports_tree_reference = False
2214
supports_external_lookups = False
2215
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2218
super(MetaDirRepositoryFormat, self).__init__()
2220
def _create_control_files(self, a_bzrdir):
2221
"""Create the required files and the initial control_files object."""
2222
# FIXME: RBC 20060125 don't peek under the covers
2223
# NB: no need to escape relative paths that are url safe.
2224
repository_transport = a_bzrdir.get_repository_transport(self)
2225
control_files = lockable_files.LockableFiles(repository_transport,
2226
'lock', lockdir.LockDir)
2227
control_files.create_lock()
2228
return control_files
2230
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
2231
"""Upload the initial blank content."""
2232
control_files = self._create_control_files(a_bzrdir)
2233
control_files.lock_write()
2234
transport = control_files._transport
2236
utf8_files += [('shared-storage', '')]
2238
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2239
for (filename, content_stream) in files:
2240
transport.put_file(filename, content_stream,
2241
mode=a_bzrdir._get_file_mode())
2242
for (filename, content_bytes) in utf8_files:
2243
transport.put_bytes_non_atomic(filename, content_bytes,
2244
mode=a_bzrdir._get_file_mode())
2246
control_files.unlock()
2249
# formats which have no format string are not discoverable
2250
# and not independently creatable, so are not registered. They're
2251
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2252
# needed, it's constructed directly by the BzrDir. Non-native formats where
2253
# the repository is not separately opened are similar.
2255
format_registry.register_lazy(
2256
'Bazaar-NG Repository format 7',
2257
'bzrlib.repofmt.weaverepo',
2261
format_registry.register_lazy(
2262
'Bazaar-NG Knit Repository Format 1',
2263
'bzrlib.repofmt.knitrepo',
2264
'RepositoryFormatKnit1',
2267
format_registry.register_lazy(
2268
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2269
'bzrlib.repofmt.knitrepo',
2270
'RepositoryFormatKnit3',
2273
format_registry.register_lazy(
2274
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2275
'bzrlib.repofmt.knitrepo',
2276
'RepositoryFormatKnit4',
2279
# Pack-based formats. There is one format for pre-subtrees, and one for
2280
# post-subtrees to allow ease of testing.
2281
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2282
format_registry.register_lazy(
2283
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2284
'bzrlib.repofmt.pack_repo',
2285
'RepositoryFormatKnitPack1',
2287
format_registry.register_lazy(
2288
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2289
'bzrlib.repofmt.pack_repo',
2290
'RepositoryFormatKnitPack3',
2292
format_registry.register_lazy(
2293
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2294
'bzrlib.repofmt.pack_repo',
2295
'RepositoryFormatKnitPack4',
2297
format_registry.register_lazy(
2298
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2299
'bzrlib.repofmt.pack_repo',
2300
'RepositoryFormatKnitPack5',
2302
format_registry.register_lazy(
2303
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2304
'bzrlib.repofmt.pack_repo',
2305
'RepositoryFormatKnitPack5RichRoot',
2308
# Development formats.
2310
# development 0 - stub to introduce development versioning scheme.
2311
format_registry.register_lazy(
2312
"Bazaar development format 0 (needs bzr.dev from before 1.3)\n",
2313
'bzrlib.repofmt.pack_repo',
2314
'RepositoryFormatPackDevelopment0',
2316
format_registry.register_lazy(
2317
("Bazaar development format 0 with subtree support "
2318
"(needs bzr.dev from before 1.3)\n"),
2319
'bzrlib.repofmt.pack_repo',
2320
'RepositoryFormatPackDevelopment0Subtree',
2322
format_registry.register_lazy(
2323
"Bazaar development format 1 (needs bzr.dev from before 1.6)\n",
2324
'bzrlib.repofmt.pack_repo',
2325
'RepositoryFormatPackDevelopment1',
2327
format_registry.register_lazy(
2328
("Bazaar development format 1 with subtree support "
2329
"(needs bzr.dev from before 1.6)\n"),
2330
'bzrlib.repofmt.pack_repo',
2331
'RepositoryFormatPackDevelopment1Subtree',
2333
# 1.3->1.4 go below here
2336
class InterRepository(InterObject):
2337
"""This class represents operations taking place between two repositories.
2339
Its instances have methods like copy_content and fetch, and contain
2340
references to the source and target repositories these operations can be
2343
Often we will provide convenience methods on 'repository' which carry out
2344
operations with another repository - they will always forward to
2345
InterRepository.get(other).method_name(parameters).
2349
"""The available optimised InterRepository types."""
2351
def copy_content(self, revision_id=None):
2352
raise NotImplementedError(self.copy_content)
2354
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2355
"""Fetch the content required to construct revision_id.
2357
The content is copied from self.source to self.target.
2359
:param revision_id: if None all content is copied, if NULL_REVISION no
2361
:param pb: optional progress bar to use for progress reports. If not
2362
provided a default one will be created.
2364
:returns: (copied_revision_count, failures).
2366
# Normally we should find a specific InterRepository subclass to do
2367
# the fetch; if nothing else then at least InterSameDataRepository.
2368
# If none of them is suitable it looks like fetching is not possible;
2369
# we try to give a good message why. _assert_same_model will probably
2370
# give a helpful message; otherwise a generic one.
2371
self._assert_same_model(self.source, self.target)
2372
raise errors.IncompatibleRepositories(self.source, self.target,
2373
"no suitableInterRepository found")
2375
def _walk_to_common_revisions(self, revision_ids):
2376
"""Walk out from revision_ids in source to revisions target has.
2378
:param revision_ids: The start point for the search.
2379
:return: A set of revision ids.
2381
target_graph = self.target.get_graph()
2382
revision_ids = frozenset(revision_ids)
2383
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2384
return graph.SearchResult(revision_ids, set(), 0, set())
2385
missing_revs = set()
2386
source_graph = self.source.get_graph()
2387
# ensure we don't pay silly lookup costs.
2388
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2389
null_set = frozenset([_mod_revision.NULL_REVISION])
2392
next_revs, ghosts = searcher.next_with_ghosts()
2393
except StopIteration:
2395
if revision_ids.intersection(ghosts):
2396
absent_ids = set(revision_ids.intersection(ghosts))
2397
# If all absent_ids are present in target, no error is needed.
2398
absent_ids.difference_update(
2399
set(target_graph.get_parent_map(absent_ids)))
2401
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2402
# we don't care about other ghosts as we can't fetch them and
2403
# haven't been asked to.
2404
next_revs = set(next_revs)
2405
# we always have NULL_REVISION present.
2406
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2407
missing_revs.update(next_revs - have_revs)
2408
searcher.stop_searching_any(have_revs)
2409
return searcher.get_result()
2411
@deprecated_method(one_two)
2413
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2414
"""Return the revision ids that source has that target does not.
2416
These are returned in topological order.
2418
:param revision_id: only return revision ids included by this
2420
:param find_ghosts: If True find missing revisions in deep history
2421
rather than just finding the surface difference.
2423
return list(self.search_missing_revision_ids(
2424
revision_id, find_ghosts).get_keys())
2427
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2428
"""Return the revision ids that source has that target does not.
2430
:param revision_id: only return revision ids included by this
2432
:param find_ghosts: If True find missing revisions in deep history
2433
rather than just finding the surface difference.
2434
:return: A bzrlib.graph.SearchResult.
2436
# stop searching at found target revisions.
2437
if not find_ghosts and revision_id is not None:
2438
return self._walk_to_common_revisions([revision_id])
2439
# generic, possibly worst case, slow code path.
2440
target_ids = set(self.target.all_revision_ids())
2441
if revision_id is not None:
2442
source_ids = self.source.get_ancestry(revision_id)
2443
if source_ids[0] is not None:
2444
raise AssertionError()
2447
source_ids = self.source.all_revision_ids()
2448
result_set = set(source_ids).difference(target_ids)
2449
return self.source.revision_ids_to_search_result(result_set)
2452
def _same_model(source, target):
2453
"""True if source and target have the same data representation.
2455
Note: this is always called on the base class; overriding it in a
2456
subclass will have no effect.
2459
InterRepository._assert_same_model(source, target)
2461
except errors.IncompatibleRepositories, e:
2465
def _assert_same_model(source, target):
2466
"""Raise an exception if two repositories do not use the same model.
2468
if source.supports_rich_root() != target.supports_rich_root():
2469
raise errors.IncompatibleRepositories(source, target,
2470
"different rich-root support")
2471
if source._serializer != target._serializer:
2472
raise errors.IncompatibleRepositories(source, target,
2473
"different serializers")
2476
class InterSameDataRepository(InterRepository):
2477
"""Code for converting between repositories that represent the same data.
2479
Data format and model must match for this to work.
2483
def _get_repo_format_to_test(self):
2484
"""Repository format for testing with.
2486
InterSameData can pull from subtree to subtree and from non-subtree to
2487
non-subtree, so we test this with the richest repository format.
2489
from bzrlib.repofmt import knitrepo
2490
return knitrepo.RepositoryFormatKnit3()
2493
def is_compatible(source, target):
2494
return InterRepository._same_model(source, target)
2497
def copy_content(self, revision_id=None):
2498
"""Make a complete copy of the content in self into destination.
2500
This copies both the repository's revision data, and configuration information
2501
such as the make_working_trees setting.
2503
This is a destructive operation! Do not use it on existing
2506
:param revision_id: Only copy the content needed to construct
2507
revision_id and its parents.
2510
self.target.set_make_working_trees(self.source.make_working_trees())
2511
except NotImplementedError:
2513
# but don't bother fetching if we have the needed data now.
2514
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2515
self.target.has_revision(revision_id)):
2517
self.target.fetch(self.source, revision_id=revision_id)
2520
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2521
"""See InterRepository.fetch()."""
2522
from bzrlib.fetch import RepoFetcher
2523
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2524
self.source, self.source._format, self.target,
2525
self.target._format)
2526
f = RepoFetcher(to_repository=self.target,
2527
from_repository=self.source,
2528
last_revision=revision_id,
2529
pb=pb, find_ghosts=find_ghosts)
2530
return f.count_copied, f.failed_revisions
2533
class InterWeaveRepo(InterSameDataRepository):
2534
"""Optimised code paths between Weave based repositories.
2536
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2537
implemented lazy inter-object optimisation.
2541
def _get_repo_format_to_test(self):
2542
from bzrlib.repofmt import weaverepo
2543
return weaverepo.RepositoryFormat7()
2546
def is_compatible(source, target):
2547
"""Be compatible with known Weave formats.
2549
We don't test for the stores being of specific types because that
2550
could lead to confusing results, and there is no need to be
2553
from bzrlib.repofmt.weaverepo import (
2559
return (isinstance(source._format, (RepositoryFormat5,
2561
RepositoryFormat7)) and
2562
isinstance(target._format, (RepositoryFormat5,
2564
RepositoryFormat7)))
2565
except AttributeError:
2569
def copy_content(self, revision_id=None):
2570
"""See InterRepository.copy_content()."""
2571
# weave specific optimised path:
2573
self.target.set_make_working_trees(self.source.make_working_trees())
2574
except (errors.RepositoryUpgradeRequired, NotImplemented):
2576
# FIXME do not peek!
2577
if self.source._transport.listable():
2578
pb = ui.ui_factory.nested_progress_bar()
2580
self.target.texts.insert_record_stream(
2581
self.source.texts.get_record_stream(
2582
self.source.texts.keys(), 'topological', False))
2583
pb.update('copying inventory', 0, 1)
2584
self.target.inventories.insert_record_stream(
2585
self.source.inventories.get_record_stream(
2586
self.source.inventories.keys(), 'topological', False))
2587
self.target.signatures.insert_record_stream(
2588
self.source.signatures.get_record_stream(
2589
self.source.signatures.keys(),
2591
self.target.revisions.insert_record_stream(
2592
self.source.revisions.get_record_stream(
2593
self.source.revisions.keys(),
2594
'topological', True))
2598
self.target.fetch(self.source, revision_id=revision_id)
2601
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2602
"""See InterRepository.fetch()."""
2603
from bzrlib.fetch import RepoFetcher
2604
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2605
self.source, self.source._format, self.target, self.target._format)
2606
f = RepoFetcher(to_repository=self.target,
2607
from_repository=self.source,
2608
last_revision=revision_id,
2609
pb=pb, find_ghosts=find_ghosts)
2610
return f.count_copied, f.failed_revisions
2613
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2614
"""See InterRepository.missing_revision_ids()."""
2615
# we want all revisions to satisfy revision_id in source.
2616
# but we don't want to stat every file here and there.
2617
# we want then, all revisions other needs to satisfy revision_id
2618
# checked, but not those that we have locally.
2619
# so the first thing is to get a subset of the revisions to
2620
# satisfy revision_id in source, and then eliminate those that
2621
# we do already have.
2622
# this is slow on high latency connection to self, but as as this
2623
# disk format scales terribly for push anyway due to rewriting
2624
# inventory.weave, this is considered acceptable.
2626
if revision_id is not None:
2627
source_ids = self.source.get_ancestry(revision_id)
2628
if source_ids[0] is not None:
2629
raise AssertionError()
2632
source_ids = self.source._all_possible_ids()
2633
source_ids_set = set(source_ids)
2634
# source_ids is the worst possible case we may need to pull.
2635
# now we want to filter source_ids against what we actually
2636
# have in target, but don't try to check for existence where we know
2637
# we do not have a revision as that would be pointless.
2638
target_ids = set(self.target._all_possible_ids())
2639
possibly_present_revisions = target_ids.intersection(source_ids_set)
2640
actually_present_revisions = set(
2641
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2642
required_revisions = source_ids_set.difference(actually_present_revisions)
2643
if revision_id is not None:
2644
# we used get_ancestry to determine source_ids then we are assured all
2645
# revisions referenced are present as they are installed in topological order.
2646
# and the tip revision was validated by get_ancestry.
2647
result_set = required_revisions
2649
# if we just grabbed the possibly available ids, then
2650
# we only have an estimate of whats available and need to validate
2651
# that against the revision records.
2653
self.source._eliminate_revisions_not_present(required_revisions))
2654
return self.source.revision_ids_to_search_result(result_set)
2657
class InterKnitRepo(InterSameDataRepository):
2658
"""Optimised code paths between Knit based repositories."""
2661
def _get_repo_format_to_test(self):
2662
from bzrlib.repofmt import knitrepo
2663
return knitrepo.RepositoryFormatKnit1()
2666
def is_compatible(source, target):
2667
"""Be compatible with known Knit formats.
2669
We don't test for the stores being of specific types because that
2670
could lead to confusing results, and there is no need to be
2673
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2675
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2676
isinstance(target._format, RepositoryFormatKnit))
2677
except AttributeError:
2679
return are_knits and InterRepository._same_model(source, target)
2682
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2683
"""See InterRepository.fetch()."""
2684
from bzrlib.fetch import RepoFetcher
2685
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2686
self.source, self.source._format, self.target, self.target._format)
2687
f = RepoFetcher(to_repository=self.target,
2688
from_repository=self.source,
2689
last_revision=revision_id,
2690
pb=pb, find_ghosts=find_ghosts)
2691
return f.count_copied, f.failed_revisions
2694
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2695
"""See InterRepository.missing_revision_ids()."""
2696
if revision_id is not None:
2697
source_ids = self.source.get_ancestry(revision_id)
2698
if source_ids[0] is not None:
2699
raise AssertionError()
2702
source_ids = self.source.all_revision_ids()
2703
source_ids_set = set(source_ids)
2704
# source_ids is the worst possible case we may need to pull.
2705
# now we want to filter source_ids against what we actually
2706
# have in target, but don't try to check for existence where we know
2707
# we do not have a revision as that would be pointless.
2708
target_ids = set(self.target.all_revision_ids())
2709
possibly_present_revisions = target_ids.intersection(source_ids_set)
2710
actually_present_revisions = set(
2711
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2712
required_revisions = source_ids_set.difference(actually_present_revisions)
2713
if revision_id is not None:
2714
# we used get_ancestry to determine source_ids then we are assured all
2715
# revisions referenced are present as they are installed in topological order.
2716
# and the tip revision was validated by get_ancestry.
2717
result_set = required_revisions
2719
# if we just grabbed the possibly available ids, then
2720
# we only have an estimate of whats available and need to validate
2721
# that against the revision records.
2723
self.source._eliminate_revisions_not_present(required_revisions))
2724
return self.source.revision_ids_to_search_result(result_set)
2727
class InterPackRepo(InterSameDataRepository):
2728
"""Optimised code paths between Pack based repositories."""
2731
def _get_repo_format_to_test(self):
2732
from bzrlib.repofmt import pack_repo
2733
return pack_repo.RepositoryFormatKnitPack1()
2736
def is_compatible(source, target):
2737
"""Be compatible with known Pack formats.
2739
We don't test for the stores being of specific types because that
2740
could lead to confusing results, and there is no need to be
2743
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2745
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2746
isinstance(target._format, RepositoryFormatPack))
2747
except AttributeError:
2749
return are_packs and InterRepository._same_model(source, target)
2752
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2753
"""See InterRepository.fetch()."""
2754
if len(self.source._fallback_repositories) > 0:
2755
from bzrlib.fetch import RepoFetcher
2756
fetcher = RepoFetcher(self.target, self.source, revision_id,
2758
return fetcher.count_copied, fetcher.failed_revisions
2759
from bzrlib.repofmt.pack_repo import Packer
2760
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2761
self.source, self.source._format, self.target, self.target._format)
2762
self.count_copied = 0
2763
if revision_id is None:
2765
# everything to do - use pack logic
2766
# to fetch from all packs to one without
2767
# inventory parsing etc, IFF nothing to be copied is in the target.
2769
source_revision_ids = frozenset(self.source.all_revision_ids())
2770
revision_ids = source_revision_ids - \
2771
frozenset(self.target.get_parent_map(source_revision_ids))
2772
revision_keys = [(revid,) for revid in revision_ids]
2773
index = self.target._pack_collection.revision_index.combined_index
2774
present_revision_ids = set(item[1][0] for item in
2775
index.iter_entries(revision_keys))
2776
revision_ids = set(revision_ids) - present_revision_ids
2777
# implementing the TODO will involve:
2778
# - detecting when all of a pack is selected
2779
# - avoiding as much as possible pre-selection, so the
2780
# more-core routines such as create_pack_from_packs can filter in
2781
# a just-in-time fashion. (though having a HEADS list on a
2782
# repository might make this a lot easier, because we could
2783
# sensibly detect 'new revisions' without doing a full index scan.
2784
elif _mod_revision.is_null(revision_id):
2789
revision_ids = self.search_missing_revision_ids(revision_id,
2790
find_ghosts=find_ghosts).get_keys()
2791
except errors.NoSuchRevision:
2792
raise errors.InstallFailed([revision_id])
2793
if len(revision_ids) == 0:
2795
packs = self.source._pack_collection.all_packs()
2796
pack = Packer(self.target._pack_collection, packs, '.fetch',
2797
revision_ids).pack()
2798
if pack is not None:
2799
self.target._pack_collection._save_pack_names()
2800
# Trigger an autopack. This may duplicate effort as we've just done
2801
# a pack creation, but for now it is simpler to think about as
2802
# 'upload data, then repack if needed'.
2803
self.target._pack_collection.autopack()
2804
return (pack.get_revision_count(), [])
2809
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2810
"""See InterRepository.missing_revision_ids().
2812
:param find_ghosts: Find ghosts throughout the ancestry of
2815
if not find_ghosts and revision_id is not None:
2816
return self._walk_to_common_revisions([revision_id])
2817
elif revision_id is not None:
2818
# Find ghosts: search for revisions pointing from one repository to
2819
# the other, and vice versa, anywhere in the history of revision_id.
2820
graph = self.target.get_graph(other_repository=self.source)
2821
searcher = graph._make_breadth_first_searcher([revision_id])
2825
next_revs, ghosts = searcher.next_with_ghosts()
2826
except StopIteration:
2828
if revision_id in ghosts:
2829
raise errors.NoSuchRevision(self.source, revision_id)
2830
found_ids.update(next_revs)
2831
found_ids.update(ghosts)
2832
found_ids = frozenset(found_ids)
2833
# Double query here: should be able to avoid this by changing the
2834
# graph api further.
2835
result_set = found_ids - frozenset(
2836
self.target.get_parent_map(found_ids))
2838
source_ids = self.source.all_revision_ids()
2839
# source_ids is the worst possible case we may need to pull.
2840
# now we want to filter source_ids against what we actually
2841
# have in target, but don't try to check for existence where we know
2842
# we do not have a revision as that would be pointless.
2843
target_ids = set(self.target.all_revision_ids())
2844
result_set = set(source_ids).difference(target_ids)
2845
return self.source.revision_ids_to_search_result(result_set)
2848
class InterModel1and2(InterRepository):
2851
def _get_repo_format_to_test(self):
2855
def is_compatible(source, target):
2856
if not source.supports_rich_root() and target.supports_rich_root():
2862
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2863
"""See InterRepository.fetch()."""
2864
from bzrlib.fetch import Model1toKnit2Fetcher
2865
f = Model1toKnit2Fetcher(to_repository=self.target,
2866
from_repository=self.source,
2867
last_revision=revision_id,
2868
pb=pb, find_ghosts=find_ghosts)
2869
return f.count_copied, f.failed_revisions
2872
def copy_content(self, revision_id=None):
2873
"""Make a complete copy of the content in self into destination.
2875
This is a destructive operation! Do not use it on existing
2878
:param revision_id: Only copy the content needed to construct
2879
revision_id and its parents.
2882
self.target.set_make_working_trees(self.source.make_working_trees())
2883
except NotImplementedError:
2885
# but don't bother fetching if we have the needed data now.
2886
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2887
self.target.has_revision(revision_id)):
2889
self.target.fetch(self.source, revision_id=revision_id)
2892
class InterKnit1and2(InterKnitRepo):
2895
def _get_repo_format_to_test(self):
2899
def is_compatible(source, target):
2900
"""Be compatible with Knit1 source and Knit3 target"""
2901
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
2903
from bzrlib.repofmt.knitrepo import (RepositoryFormatKnit1,
2904
RepositoryFormatKnit3)
2905
from bzrlib.repofmt.pack_repo import (
2906
RepositoryFormatKnitPack1,
2907
RepositoryFormatKnitPack3,
2908
RepositoryFormatPackDevelopment0,
2909
RepositoryFormatPackDevelopment0Subtree,
2912
RepositoryFormatKnit1,
2913
RepositoryFormatKnitPack1,
2914
RepositoryFormatPackDevelopment0,
2917
RepositoryFormatKnit3,
2918
RepositoryFormatKnitPack3,
2919
RepositoryFormatPackDevelopment0Subtree,
2921
return (isinstance(source._format, nosubtrees) and
2922
isinstance(target._format, subtrees))
2923
except AttributeError:
2927
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2928
"""See InterRepository.fetch()."""
2929
from bzrlib.fetch import Knit1to2Fetcher
2930
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2931
self.source, self.source._format, self.target,
2932
self.target._format)
2933
f = Knit1to2Fetcher(to_repository=self.target,
2934
from_repository=self.source,
2935
last_revision=revision_id,
2936
pb=pb, find_ghosts=find_ghosts)
2937
return f.count_copied, f.failed_revisions
2940
class InterDifferingSerializer(InterKnitRepo):
2943
def _get_repo_format_to_test(self):
2947
def is_compatible(source, target):
2948
"""Be compatible with Knit2 source and Knit3 target"""
2949
if source.supports_rich_root() != target.supports_rich_root():
2951
# Ideally, we'd support fetching if the source had no tree references
2952
# even if it supported them...
2953
if (getattr(source, '_format.supports_tree_reference', False) and
2954
not getattr(target, '_format.supports_tree_reference', False)):
2959
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2960
"""See InterRepository.fetch()."""
2961
revision_ids = self.target.search_missing_revision_ids(self.source,
2962
revision_id, find_ghosts=find_ghosts).get_keys()
2963
revision_ids = tsort.topo_sort(
2964
self.source.get_graph().get_parent_map(revision_ids))
2965
def revisions_iterator():
2966
for current_revision_id in revision_ids:
2967
revision = self.source.get_revision(current_revision_id)
2968
tree = self.source.revision_tree(current_revision_id)
2970
signature = self.source.get_signature_text(
2971
current_revision_id)
2972
except errors.NoSuchRevision:
2974
yield revision, tree, signature
2976
my_pb = ui.ui_factory.nested_progress_bar()
2981
install_revisions(self.target, revisions_iterator(),
2982
len(revision_ids), pb)
2984
if my_pb is not None:
2986
return len(revision_ids), 0
2989
class InterOtherToRemote(InterRepository):
2991
def __init__(self, source, target):
2992
InterRepository.__init__(self, source, target)
2993
self._real_inter = None
2996
def is_compatible(source, target):
2997
if isinstance(target, remote.RemoteRepository):
3001
def _ensure_real_inter(self):
3002
if self._real_inter is None:
3003
self.target._ensure_real()
3004
real_target = self.target._real_repository
3005
self._real_inter = InterRepository.get(self.source, real_target)
3007
def copy_content(self, revision_id=None):
3008
self._ensure_real_inter()
3009
self._real_inter.copy_content(revision_id=revision_id)
3011
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3012
self._ensure_real_inter()
3013
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3014
find_ghosts=find_ghosts)
3017
def _get_repo_format_to_test(self):
3021
class InterRemoteToOther(InterRepository):
3023
def __init__(self, source, target):
3024
InterRepository.__init__(self, source, target)
3025
self._real_inter = None
3028
def is_compatible(source, target):
3029
if not isinstance(source, remote.RemoteRepository):
3031
# Is source's model compatible with target's model?
3032
source._ensure_real()
3033
real_source = source._real_repository
3034
if isinstance(real_source, remote.RemoteRepository):
3035
raise NotImplementedError(
3036
"We don't support remote repos backed by remote repos yet.")
3037
return InterRepository._same_model(real_source, target)
3039
def _ensure_real_inter(self):
3040
if self._real_inter is None:
3041
self.source._ensure_real()
3042
real_source = self.source._real_repository
3043
self._real_inter = InterRepository.get(real_source, self.target)
3045
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3046
self._ensure_real_inter()
3047
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3048
find_ghosts=find_ghosts)
3050
def copy_content(self, revision_id=None):
3051
self._ensure_real_inter()
3052
self._real_inter.copy_content(revision_id=revision_id)
3055
def _get_repo_format_to_test(self):
3060
InterRepository.register_optimiser(InterDifferingSerializer)
3061
InterRepository.register_optimiser(InterSameDataRepository)
3062
InterRepository.register_optimiser(InterWeaveRepo)
3063
InterRepository.register_optimiser(InterKnitRepo)
3064
InterRepository.register_optimiser(InterModel1and2)
3065
InterRepository.register_optimiser(InterKnit1and2)
3066
InterRepository.register_optimiser(InterPackRepo)
3067
InterRepository.register_optimiser(InterOtherToRemote)
3068
InterRepository.register_optimiser(InterRemoteToOther)
3071
class CopyConverter(object):
3072
"""A repository conversion tool which just performs a copy of the content.
3074
This is slow but quite reliable.
3077
def __init__(self, target_format):
3078
"""Create a CopyConverter.
3080
:param target_format: The format the resulting repository should be.
3082
self.target_format = target_format
3084
def convert(self, repo, pb):
3085
"""Perform the conversion of to_convert, giving feedback via pb.
3087
:param to_convert: The disk object to convert.
3088
:param pb: a progress bar to use for progress information.
3093
# this is only useful with metadir layouts - separated repo content.
3094
# trigger an assertion if not such
3095
repo._format.get_format_string()
3096
self.repo_dir = repo.bzrdir
3097
self.step('Moving repository to repository.backup')
3098
self.repo_dir.transport.move('repository', 'repository.backup')
3099
backup_transport = self.repo_dir.transport.clone('repository.backup')
3100
repo._format.check_conversion_target(self.target_format)
3101
self.source_repo = repo._format.open(self.repo_dir,
3103
_override_transport=backup_transport)
3104
self.step('Creating new repository')
3105
converted = self.target_format.initialize(self.repo_dir,
3106
self.source_repo.is_shared())
3107
converted.lock_write()
3109
self.step('Copying content into repository.')
3110
self.source_repo.copy_content_into(converted)
3113
self.step('Deleting old repository content.')
3114
self.repo_dir.transport.delete_tree('repository.backup')
3115
self.pb.note('repository converted')
3117
def step(self, message):
3118
"""Update the pb by a step."""
3120
self.pb.update(message, self.count, self.total)
3132
def _unescaper(match, _map=_unescape_map):
3133
code = match.group(1)
3137
if not code.startswith('#'):
3139
return unichr(int(code[1:])).encode('utf8')
3145
def _unescape_xml(data):
3146
"""Unescape predefined XML entities in a string of data."""
3148
if _unescape_re is None:
3149
_unescape_re = re.compile('\&([^;]*);')
3150
return _unescape_re.sub(_unescaper, data)
3153
class _VersionedFileChecker(object):
3155
def __init__(self, repository):
3156
self.repository = repository
3157
self.text_index = self.repository._generate_text_key_index()
3159
def calculate_file_version_parents(self, text_key):
3160
"""Calculate the correct parents for a file version according to
3163
parent_keys = self.text_index[text_key]
3164
if parent_keys == [_mod_revision.NULL_REVISION]:
3166
return tuple(parent_keys)
3168
def check_file_version_parents(self, texts, progress_bar=None):
3169
"""Check the parents stored in a versioned file are correct.
3171
It also detects file versions that are not referenced by their
3172
corresponding revision's inventory.
3174
:returns: A tuple of (wrong_parents, dangling_file_versions).
3175
wrong_parents is a dict mapping {revision_id: (stored_parents,
3176
correct_parents)} for each revision_id where the stored parents
3177
are not correct. dangling_file_versions is a set of (file_id,
3178
revision_id) tuples for versions that are present in this versioned
3179
file, but not used by the corresponding inventory.
3182
self.file_ids = set([file_id for file_id, _ in
3183
self.text_index.iterkeys()])
3184
# text keys is now grouped by file_id
3185
n_weaves = len(self.file_ids)
3186
files_in_revisions = {}
3187
revisions_of_files = {}
3188
n_versions = len(self.text_index)
3189
progress_bar.update('loading text store', 0, n_versions)
3190
parent_map = self.repository.texts.get_parent_map(self.text_index)
3191
# On unlistable transports this could well be empty/error...
3192
text_keys = self.repository.texts.keys()
3193
unused_keys = frozenset(text_keys) - set(self.text_index)
3194
for num, key in enumerate(self.text_index.iterkeys()):
3195
if progress_bar is not None:
3196
progress_bar.update('checking text graph', num, n_versions)
3197
correct_parents = self.calculate_file_version_parents(key)
3199
knit_parents = parent_map[key]
3200
except errors.RevisionNotPresent:
3203
if correct_parents != knit_parents:
3204
wrong_parents[key] = (knit_parents, correct_parents)
3205
return wrong_parents, unused_keys
3208
def _old_get_graph(repository, revision_id):
3209
"""DO NOT USE. That is all. I'm serious."""
3210
graph = repository.get_graph()
3211
revision_graph = dict(((key, value) for key, value in
3212
graph.iter_ancestry([revision_id]) if value is not None))
3213
return _strip_NULL_ghosts(revision_graph)
3216
def _strip_NULL_ghosts(revision_graph):
3217
"""Also don't use this. more compatibility code for unmigrated clients."""
3218
# Filter ghosts, and null:
3219
if _mod_revision.NULL_REVISION in revision_graph:
3220
del revision_graph[_mod_revision.NULL_REVISION]
3221
for key, parents in revision_graph.items():
3222
revision_graph[key] = tuple(parent for parent in parents if parent
3224
return revision_graph
added
removed
bzrlib/repository.py
