1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from bzrlib.lazy_import import lazy_import
18
lazy_import(globals(), """
37
revision as _mod_revision,
42
from bzrlib.bundle import serializer
43
from bzrlib.revisiontree import RevisionTree
44
from bzrlib.store.versioned import VersionedFileStore
45
from bzrlib.testament import Testament
48
from bzrlib import registry
49
from bzrlib.decorators import needs_read_lock, needs_write_lock
50
from bzrlib.inter import InterObject
51
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
52
from bzrlib.symbol_versioning import (
58
from bzrlib.trace import (
59
log_exception_quietly, note, mutter, mutter_callsite, warning)
62
# Old formats display a warning, but only once
63
_deprecation_warning_done = False
66
class CommitBuilder(object):
67
"""Provides an interface to build up a commit.
69
This allows describing a tree to be committed without needing to
70
know the internals of the format of the repository.
73
# all clients should supply tree roots.
74
record_root_entry = True
75
# the default CommitBuilder does not manage trees whose root is versioned.
76
_versioned_root = False
78
def __init__(self, repository, parents, config, timestamp=None,
79
timezone=None, committer=None, revprops=None,
81
"""Initiate a CommitBuilder.
83
:param repository: Repository to commit to.
84
:param parents: Revision ids of the parents of the new revision.
85
:param config: Configuration to use.
86
:param timestamp: Optional timestamp recorded for commit.
87
:param timezone: Optional timezone for timestamp.
88
:param committer: Optional committer to set for commit.
89
:param revprops: Optional dictionary of revision properties.
90
:param revision_id: Optional revision id.
95
self._committer = self._config.username()
97
self._committer = committer
99
self.new_inventory = Inventory(None)
100
self._new_revision_id = revision_id
101
self.parents = parents
102
self.repository = repository
105
if revprops is not None:
106
self._validate_revprops(revprops)
107
self._revprops.update(revprops)
109
if timestamp is None:
110
timestamp = time.time()
111
# Restrict resolution to 1ms
112
self._timestamp = round(timestamp, 3)
115
self._timezone = osutils.local_time_offset()
117
self._timezone = int(timezone)
119
self._generate_revision_if_needed()
120
self.__heads = graph.HeadsCache(repository.get_graph()).heads
122
def _validate_unicode_text(self, text, context):
123
"""Verify things like commit messages don't have bogus characters."""
125
raise ValueError('Invalid value for %s: %r' % (context, text))
127
def _validate_revprops(self, revprops):
128
for key, value in revprops.iteritems():
129
# We know that the XML serializers do not round trip '\r'
130
# correctly, so refuse to accept them
131
if not isinstance(value, basestring):
132
raise ValueError('revision property (%s) is not a valid'
133
' (unicode) string: %r' % (key, value))
134
self._validate_unicode_text(value,
135
'revision property (%s)' % (key,))
137
def commit(self, message):
138
"""Make the actual commit.
140
:return: The revision id of the recorded revision.
142
self._validate_unicode_text(message, 'commit message')
143
rev = _mod_revision.Revision(
144
timestamp=self._timestamp,
145
timezone=self._timezone,
146
committer=self._committer,
148
inventory_sha1=self.inv_sha1,
149
revision_id=self._new_revision_id,
150
properties=self._revprops)
151
rev.parent_ids = self.parents
152
self.repository.add_revision(self._new_revision_id, rev,
153
self.new_inventory, self._config)
154
self.repository.commit_write_group()
155
return self._new_revision_id
158
"""Abort the commit that is being built.
160
self.repository.abort_write_group()
162
def revision_tree(self):
163
"""Return the tree that was just committed.
165
After calling commit() this can be called to get a RevisionTree
166
representing the newly committed tree. This is preferred to
167
calling Repository.revision_tree() because that may require
168
deserializing the inventory, while we already have a copy in
171
return RevisionTree(self.repository, self.new_inventory,
172
self._new_revision_id)
174
def finish_inventory(self):
175
"""Tell the builder that the inventory is finished."""
176
if self.new_inventory.root is None:
177
raise AssertionError('Root entry should be supplied to'
178
' record_entry_contents, as of bzr 0.10.')
179
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
180
self.new_inventory.revision_id = self._new_revision_id
181
self.inv_sha1 = self.repository.add_inventory(
182
self._new_revision_id,
187
def _gen_revision_id(self):
188
"""Return new revision-id."""
189
return generate_ids.gen_revision_id(self._config.username(),
192
def _generate_revision_if_needed(self):
193
"""Create a revision id if None was supplied.
195
If the repository can not support user-specified revision ids
196
they should override this function and raise CannotSetRevisionId
197
if _new_revision_id is not None.
199
:raises: CannotSetRevisionId
201
if self._new_revision_id is None:
202
self._new_revision_id = self._gen_revision_id()
203
self.random_revid = True
205
self.random_revid = False
207
def _heads(self, file_id, revision_ids):
208
"""Calculate the graph heads for revision_ids in the graph of file_id.
210
This can use either a per-file graph or a global revision graph as we
211
have an identity relationship between the two graphs.
213
return self.__heads(revision_ids)
215
def _check_root(self, ie, parent_invs, tree):
216
"""Helper for record_entry_contents.
218
:param ie: An entry being added.
219
:param parent_invs: The inventories of the parent revisions of the
221
:param tree: The tree that is being committed.
223
# In this revision format, root entries have no knit or weave When
224
# serializing out to disk and back in root.revision is always
226
ie.revision = self._new_revision_id
228
def _get_delta(self, ie, basis_inv, path):
229
"""Get a delta against the basis inventory for ie."""
230
if ie.file_id not in basis_inv:
232
return (None, path, ie.file_id, ie)
233
elif ie != basis_inv[ie.file_id]:
235
# TODO: avoid tis id2path call.
236
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
241
def record_entry_contents(self, ie, parent_invs, path, tree,
243
"""Record the content of ie from tree into the commit if needed.
245
Side effect: sets ie.revision when unchanged
247
:param ie: An inventory entry present in the commit.
248
:param parent_invs: The inventories of the parent revisions of the
250
:param path: The path the entry is at in the tree.
251
:param tree: The tree which contains this entry and should be used to
253
:param content_summary: Summary data from the tree about the paths
254
content - stat, length, exec, sha/link target. This is only
255
accessed when the entry has a revision of None - that is when it is
256
a candidate to commit.
257
:return: A tuple (change_delta, version_recorded, fs_hash).
258
change_delta is an inventory_delta change for this entry against
259
the basis tree of the commit, or None if no change occured against
261
version_recorded is True if a new version of the entry has been
262
recorded. For instance, committing a merge where a file was only
263
changed on the other side will return (delta, False).
264
fs_hash is either None, or the hash details for the path (currently
265
a tuple of the contents sha1 and the statvalue returned by
266
tree.get_file_with_stat()).
268
if self.new_inventory.root is None:
269
if ie.parent_id is not None:
270
raise errors.RootMissing()
271
self._check_root(ie, parent_invs, tree)
272
if ie.revision is None:
273
kind = content_summary[0]
275
# ie is carried over from a prior commit
277
# XXX: repository specific check for nested tree support goes here - if
278
# the repo doesn't want nested trees we skip it ?
279
if (kind == 'tree-reference' and
280
not self.repository._format.supports_tree_reference):
281
# mismatch between commit builder logic and repository:
282
# this needs the entry creation pushed down into the builder.
283
raise NotImplementedError('Missing repository subtree support.')
284
self.new_inventory.add(ie)
286
# TODO: slow, take it out of the inner loop.
288
basis_inv = parent_invs[0]
290
basis_inv = Inventory(root_id=None)
292
# ie.revision is always None if the InventoryEntry is considered
293
# for committing. We may record the previous parents revision if the
294
# content is actually unchanged against a sole head.
295
if ie.revision is not None:
296
if not self._versioned_root and path == '':
297
# repositories that do not version the root set the root's
298
# revision to the new commit even when no change occurs, and
299
# this masks when a change may have occurred against the basis,
300
# so calculate if one happened.
301
if ie.file_id in basis_inv:
302
delta = (basis_inv.id2path(ie.file_id), path,
306
delta = (None, path, ie.file_id, ie)
307
return delta, False, None
309
# we don't need to commit this, because the caller already
310
# determined that an existing revision of this file is
311
# appropriate. If its not being considered for committing then
312
# it and all its parents to the root must be unaltered so
313
# no-change against the basis.
314
if ie.revision == self._new_revision_id:
315
raise AssertionError("Impossible situation, a skipped "
316
"inventory entry (%r) claims to be modified in this "
317
"commit (%r).", (ie, self._new_revision_id))
318
return None, False, None
319
# XXX: Friction: parent_candidates should return a list not a dict
320
# so that we don't have to walk the inventories again.
321
parent_candiate_entries = ie.parent_candidates(parent_invs)
322
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
324
for inv in parent_invs:
325
if ie.file_id in inv:
326
old_rev = inv[ie.file_id].revision
327
if old_rev in head_set:
328
heads.append(inv[ie.file_id].revision)
329
head_set.remove(inv[ie.file_id].revision)
332
# now we check to see if we need to write a new record to the
334
# We write a new entry unless there is one head to the ancestors, and
335
# the kind-derived content is unchanged.
337
# Cheapest check first: no ancestors, or more the one head in the
338
# ancestors, we write a new node.
342
# There is a single head, look it up for comparison
343
parent_entry = parent_candiate_entries[heads[0]]
344
# if the non-content specific data has changed, we'll be writing a
346
if (parent_entry.parent_id != ie.parent_id or
347
parent_entry.name != ie.name):
349
# now we need to do content specific checks:
351
# if the kind changed the content obviously has
352
if kind != parent_entry.kind:
354
# Stat cache fingerprint feedback for the caller - None as we usually
355
# don't generate one.
358
if content_summary[2] is None:
359
raise ValueError("Files must not have executable = None")
361
if (# if the file length changed we have to store:
362
parent_entry.text_size != content_summary[1] or
363
# if the exec bit has changed we have to store:
364
parent_entry.executable != content_summary[2]):
366
elif parent_entry.text_sha1 == content_summary[3]:
367
# all meta and content is unchanged (using a hash cache
368
# hit to check the sha)
369
ie.revision = parent_entry.revision
370
ie.text_size = parent_entry.text_size
371
ie.text_sha1 = parent_entry.text_sha1
372
ie.executable = parent_entry.executable
373
return self._get_delta(ie, basis_inv, path), False, None
375
# Either there is only a hash change(no hash cache entry,
376
# or same size content change), or there is no change on
378
# Provide the parent's hash to the store layer, so that the
379
# content is unchanged we will not store a new node.
380
nostore_sha = parent_entry.text_sha1
382
# We want to record a new node regardless of the presence or
383
# absence of a content change in the file.
385
ie.executable = content_summary[2]
386
file_obj, stat_value = tree.get_file_with_stat(ie.file_id, path)
388
lines = file_obj.readlines()
392
ie.text_sha1, ie.text_size = self._add_text_to_weave(
393
ie.file_id, lines, heads, nostore_sha)
394
# Let the caller know we generated a stat fingerprint.
395
fingerprint = (ie.text_sha1, stat_value)
396
except errors.ExistingContent:
397
# Turns out that the file content was unchanged, and we were
398
# only going to store a new node if it was changed. Carry over
400
ie.revision = parent_entry.revision
401
ie.text_size = parent_entry.text_size
402
ie.text_sha1 = parent_entry.text_sha1
403
ie.executable = parent_entry.executable
404
return self._get_delta(ie, basis_inv, path), False, None
405
elif kind == 'directory':
407
# all data is meta here, nothing specific to directory, so
409
ie.revision = parent_entry.revision
410
return self._get_delta(ie, basis_inv, path), False, None
412
self._add_text_to_weave(ie.file_id, lines, heads, None)
413
elif kind == 'symlink':
414
current_link_target = content_summary[3]
416
# symlink target is not generic metadata, check if it has
418
if current_link_target != parent_entry.symlink_target:
421
# unchanged, carry over.
422
ie.revision = parent_entry.revision
423
ie.symlink_target = parent_entry.symlink_target
424
return self._get_delta(ie, basis_inv, path), False, None
425
ie.symlink_target = current_link_target
427
self._add_text_to_weave(ie.file_id, lines, heads, None)
428
elif kind == 'tree-reference':
430
if content_summary[3] != parent_entry.reference_revision:
433
# unchanged, carry over.
434
ie.reference_revision = parent_entry.reference_revision
435
ie.revision = parent_entry.revision
436
return self._get_delta(ie, basis_inv, path), False, None
437
ie.reference_revision = content_summary[3]
439
self._add_text_to_weave(ie.file_id, lines, heads, None)
441
raise NotImplementedError('unknown kind')
442
ie.revision = self._new_revision_id
443
return self._get_delta(ie, basis_inv, path), True, fingerprint
445
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
446
# Note: as we read the content directly from the tree, we know its not
447
# been turned into unicode or badly split - but a broken tree
448
# implementation could give us bad output from readlines() so this is
449
# not a guarantee of safety. What would be better is always checking
450
# the content during test suite execution. RBC 20070912
451
parent_keys = tuple((file_id, parent) for parent in parents)
452
return self.repository.texts.add_lines(
453
(file_id, self._new_revision_id), parent_keys, new_lines,
454
nostore_sha=nostore_sha, random_id=self.random_revid,
455
check_content=False)[0:2]
458
class RootCommitBuilder(CommitBuilder):
459
"""This commitbuilder actually records the root id"""
461
# the root entry gets versioned properly by this builder.
462
_versioned_root = True
464
def _check_root(self, ie, parent_invs, tree):
465
"""Helper for record_entry_contents.
467
:param ie: An entry being added.
468
:param parent_invs: The inventories of the parent revisions of the
470
:param tree: The tree that is being committed.
474
######################################################################
477
class Repository(object):
478
"""Repository holding history for one or more branches.
480
The repository holds and retrieves historical information including
481
revisions and file history. It's normally accessed only by the Branch,
482
which views a particular line of development through that history.
484
The Repository builds on top of some byte storage facilies (the revisions,
485
signatures, inventories and texts attributes) and a Transport, which
486
respectively provide byte storage and a means to access the (possibly
489
The byte storage facilities are addressed via tuples, which we refer to
490
as 'keys' throughout the code base. Revision_keys, inventory_keys and
491
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
492
(file_id, revision_id). We use this interface because it allows low
493
friction with the underlying code that implements disk indices, network
494
encoding and other parts of bzrlib.
496
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
497
the serialised revisions for the repository. This can be used to obtain
498
revision graph information or to access raw serialised revisions.
499
The result of trying to insert data into the repository via this store
500
is undefined: it should be considered read-only except for implementors
502
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
503
the serialised signatures for the repository. This can be used to
504
obtain access to raw serialised signatures. The result of trying to
505
insert data into the repository via this store is undefined: it should
506
be considered read-only except for implementors of repositories.
507
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
508
the serialised inventories for the repository. This can be used to
509
obtain unserialised inventories. The result of trying to insert data
510
into the repository via this store is undefined: it should be
511
considered read-only except for implementors of repositories.
512
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
513
texts of files and directories for the repository. This can be used to
514
obtain file texts or file graphs. Note that Repository.iter_file_bytes
515
is usually a better interface for accessing file texts.
516
The result of trying to insert data into the repository via this store
517
is undefined: it should be considered read-only except for implementors
519
:ivar _transport: Transport for file access to repository, typically
520
pointing to .bzr/repository.
523
# What class to use for a CommitBuilder. Often its simpler to change this
524
# in a Repository class subclass rather than to override
525
# get_commit_builder.
526
_commit_builder_class = CommitBuilder
527
# The search regex used by xml based repositories to determine what things
528
# where changed in a single commit.
529
_file_ids_altered_regex = lazy_regex.lazy_compile(
530
r'file_id="(?P<file_id>[^"]+)"'
531
r'.* revision="(?P<revision_id>[^"]+)"'
534
def abort_write_group(self, suppress_errors=False):
535
"""Commit the contents accrued within the current write group.
537
:param suppress_errors: if true, abort_write_group will catch and log
538
unexpected errors that happen during the abort, rather than
539
allowing them to propagate. Defaults to False.
541
:seealso: start_write_group.
543
if self._write_group is not self.get_transaction():
544
# has an unlock or relock occured ?
545
raise errors.BzrError('mismatched lock context and write group.')
547
self._abort_write_group()
548
except Exception, exc:
549
self._write_group = None
550
if not suppress_errors:
552
mutter('abort_write_group failed')
553
log_exception_quietly()
554
note('bzr: ERROR (ignored): %s', exc)
555
self._write_group = None
557
def _abort_write_group(self):
558
"""Template method for per-repository write group cleanup.
560
This is called during abort before the write group is considered to be
561
finished and should cleanup any internal state accrued during the write
562
group. There is no requirement that data handed to the repository be
563
*not* made available - this is not a rollback - but neither should any
564
attempt be made to ensure that data added is fully commited. Abort is
565
invoked when an error has occured so futher disk or network operations
566
may not be possible or may error and if possible should not be
570
def add_fallback_repository(self, repository):
571
"""Add a repository to use for looking up data not held locally.
573
:param repository: A repository.
575
if not self._format.supports_external_lookups:
576
raise errors.UnstackableRepositoryFormat(self._format, self.base)
577
self._check_fallback_repository(repository)
578
self._fallback_repositories.append(repository)
579
self.texts.add_fallback_versioned_files(repository.texts)
580
self.inventories.add_fallback_versioned_files(repository.inventories)
581
self.revisions.add_fallback_versioned_files(repository.revisions)
582
self.signatures.add_fallback_versioned_files(repository.signatures)
584
def _check_fallback_repository(self, repository):
585
"""Check that this repository can fallback to repository safely.
587
Raise an error if not.
589
:param repository: A repository to fallback to.
591
return InterRepository._assert_same_model(self, repository)
593
def add_inventory(self, revision_id, inv, parents):
594
"""Add the inventory inv to the repository as revision_id.
596
:param parents: The revision ids of the parents that revision_id
597
is known to have and are in the repository already.
599
:returns: The validator(which is a sha1 digest, though what is sha'd is
600
repository format specific) of the serialized inventory.
602
if not self.is_in_write_group():
603
raise AssertionError("%r not in write group" % (self,))
604
_mod_revision.check_not_reserved_id(revision_id)
605
if not (inv.revision_id is None or inv.revision_id == revision_id):
606
raise AssertionError(
607
"Mismatch between inventory revision"
608
" id and insertion revid (%r, %r)"
609
% (inv.revision_id, revision_id))
611
raise AssertionError()
612
inv_lines = self._serialise_inventory_to_lines(inv)
613
return self._inventory_add_lines(revision_id, parents,
614
inv_lines, check_content=False)
616
def _inventory_add_lines(self, revision_id, parents, lines,
618
"""Store lines in inv_vf and return the sha1 of the inventory."""
619
parents = [(parent,) for parent in parents]
620
return self.inventories.add_lines((revision_id,), parents, lines,
621
check_content=check_content)[0]
623
def add_revision(self, revision_id, rev, inv=None, config=None):
624
"""Add rev to the revision store as revision_id.
626
:param revision_id: the revision id to use.
627
:param rev: The revision object.
628
:param inv: The inventory for the revision. if None, it will be looked
629
up in the inventory storer
630
:param config: If None no digital signature will be created.
631
If supplied its signature_needed method will be used
632
to determine if a signature should be made.
634
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
636
_mod_revision.check_not_reserved_id(revision_id)
637
if config is not None and config.signature_needed():
639
inv = self.get_inventory(revision_id)
640
plaintext = Testament(rev, inv).as_short_text()
641
self.store_revision_signature(
642
gpg.GPGStrategy(config), plaintext, revision_id)
643
# check inventory present
644
if not self.inventories.get_parent_map([(revision_id,)]):
646
raise errors.WeaveRevisionNotPresent(revision_id,
649
# yes, this is not suitable for adding with ghosts.
650
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
654
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
655
self._add_revision(rev)
657
def _add_revision(self, revision):
658
text = self._serializer.write_revision_to_string(revision)
659
key = (revision.revision_id,)
660
parents = tuple((parent,) for parent in revision.parent_ids)
661
self.revisions.add_lines(key, parents, osutils.split_lines(text))
663
def all_revision_ids(self):
664
"""Returns a list of all the revision ids in the repository.
666
This is conceptually deprecated because code should generally work on
667
the graph reachable from a particular revision, and ignore any other
668
revisions that might be present. There is no direct replacement
671
if 'evil' in debug.debug_flags:
672
mutter_callsite(2, "all_revision_ids is linear with history.")
673
return self._all_revision_ids()
675
def _all_revision_ids(self):
676
"""Returns a list of all the revision ids in the repository.
678
These are in as much topological order as the underlying store can
681
raise NotImplementedError(self._all_revision_ids)
683
def break_lock(self):
684
"""Break a lock if one is present from another instance.
686
Uses the ui factory to ask for confirmation if the lock may be from
689
self.control_files.break_lock()
692
def _eliminate_revisions_not_present(self, revision_ids):
693
"""Check every revision id in revision_ids to see if we have it.
695
Returns a set of the present revisions.
698
graph = self.get_graph()
699
parent_map = graph.get_parent_map(revision_ids)
700
# The old API returned a list, should this actually be a set?
701
return parent_map.keys()
704
def create(a_bzrdir):
705
"""Construct the current default format repository in a_bzrdir."""
706
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
708
def __init__(self, _format, a_bzrdir, control_files):
709
"""instantiate a Repository.
711
:param _format: The format of the repository on disk.
712
:param a_bzrdir: The BzrDir of the repository.
714
In the future we will have a single api for all stores for
715
getting file texts, inventories and revisions, then
716
this construct will accept instances of those things.
718
super(Repository, self).__init__()
719
self._format = _format
720
# the following are part of the public API for Repository:
721
self.bzrdir = a_bzrdir
722
self.control_files = control_files
723
self._transport = control_files._transport
724
self.base = self._transport.base
726
self._reconcile_does_inventory_gc = True
727
self._reconcile_fixes_text_parents = False
728
self._reconcile_backsup_inventory = True
729
# not right yet - should be more semantically clear ?
731
# TODO: make sure to construct the right store classes, etc, depending
732
# on whether escaping is required.
733
self._warn_if_deprecated()
734
self._write_group = None
735
# Additional places to query for data.
736
self._fallback_repositories = []
737
# What order should fetch operations request streams in?
738
# The default is unordered as that is the cheapest for an origin to
740
self._fetch_order = 'unordered'
741
# Does this repository use deltas that can be fetched as-deltas ?
742
# (E.g. knits, where the knit deltas can be transplanted intact.
743
# We default to False, which will ensure that enough data to get
744
# a full text out of any fetch stream will be grabbed.
745
self._fetch_uses_deltas = False
746
# Should fetch trigger a reconcile after the fetch? Only needed for
747
# some repository formats that can suffer internal inconsistencies.
748
self._fetch_reconcile = False
751
return '%s(%r)' % (self.__class__.__name__,
754
def has_same_location(self, other):
755
"""Returns a boolean indicating if this repository is at the same
756
location as another repository.
758
This might return False even when two repository objects are accessing
759
the same physical repository via different URLs.
761
if self.__class__ is not other.__class__:
763
return (self._transport.base == other._transport.base)
765
def is_in_write_group(self):
766
"""Return True if there is an open write group.
768
:seealso: start_write_group.
770
return self._write_group is not None
773
return self.control_files.is_locked()
775
def is_write_locked(self):
776
"""Return True if this object is write locked."""
777
return self.is_locked() and self.control_files._lock_mode == 'w'
779
def lock_write(self, token=None):
780
"""Lock this repository for writing.
782
This causes caching within the repository obejct to start accumlating
783
data during reads, and allows a 'write_group' to be obtained. Write
784
groups must be used for actual data insertion.
786
:param token: if this is already locked, then lock_write will fail
787
unless the token matches the existing lock.
788
:returns: a token if this instance supports tokens, otherwise None.
789
:raises TokenLockingNotSupported: when a token is given but this
790
instance doesn't support using token locks.
791
:raises MismatchedToken: if the specified token doesn't match the token
792
of the existing lock.
793
:seealso: start_write_group.
795
A token should be passed in if you know that you have locked the object
796
some other way, and need to synchronise this object's state with that
799
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
801
result = self.control_files.lock_write(token=token)
802
for repo in self._fallback_repositories:
803
# Writes don't affect fallback repos
809
self.control_files.lock_read()
810
for repo in self._fallback_repositories:
814
def get_physical_lock_status(self):
815
return self.control_files.get_physical_lock_status()
817
def leave_lock_in_place(self):
818
"""Tell this repository not to release the physical lock when this
821
If lock_write doesn't return a token, then this method is not supported.
823
self.control_files.leave_in_place()
825
def dont_leave_lock_in_place(self):
826
"""Tell this repository to release the physical lock when this
827
object is unlocked, even if it didn't originally acquire it.
829
If lock_write doesn't return a token, then this method is not supported.
831
self.control_files.dont_leave_in_place()
834
def gather_stats(self, revid=None, committers=None):
835
"""Gather statistics from a revision id.
837
:param revid: The revision id to gather statistics from, if None, then
838
no revision specific statistics are gathered.
839
:param committers: Optional parameter controlling whether to grab
840
a count of committers from the revision specific statistics.
841
:return: A dictionary of statistics. Currently this contains:
842
committers: The number of committers if requested.
843
firstrev: A tuple with timestamp, timezone for the penultimate left
844
most ancestor of revid, if revid is not the NULL_REVISION.
845
latestrev: A tuple with timestamp, timezone for revid, if revid is
846
not the NULL_REVISION.
847
revisions: The total revision count in the repository.
848
size: An estimate disk size of the repository in bytes.
851
if revid and committers:
852
result['committers'] = 0
853
if revid and revid != _mod_revision.NULL_REVISION:
855
all_committers = set()
856
revisions = self.get_ancestry(revid)
857
# pop the leading None
859
first_revision = None
861
# ignore the revisions in the middle - just grab first and last
862
revisions = revisions[0], revisions[-1]
863
for revision in self.get_revisions(revisions):
864
if not first_revision:
865
first_revision = revision
867
all_committers.add(revision.committer)
868
last_revision = revision
870
result['committers'] = len(all_committers)
871
result['firstrev'] = (first_revision.timestamp,
872
first_revision.timezone)
873
result['latestrev'] = (last_revision.timestamp,
874
last_revision.timezone)
876
# now gather global repository information
877
# XXX: This is available for many repos regardless of listability.
878
if self.bzrdir.root_transport.listable():
879
# XXX: do we want to __define len__() ?
880
# Maybe the versionedfiles object should provide a different
881
# method to get the number of keys.
882
result['revisions'] = len(self.revisions.keys())
886
def find_branches(self, using=False):
887
"""Find branches underneath this repository.
889
This will include branches inside other branches.
891
:param using: If True, list only branches using this repository.
893
if using and not self.is_shared():
895
return [self.bzrdir.open_branch()]
896
except errors.NotBranchError:
898
class Evaluator(object):
901
self.first_call = True
903
def __call__(self, bzrdir):
904
# On the first call, the parameter is always the bzrdir
905
# containing the current repo.
906
if not self.first_call:
908
repository = bzrdir.open_repository()
909
except errors.NoRepositoryPresent:
912
return False, (None, repository)
913
self.first_call = False
915
value = (bzrdir.open_branch(), None)
916
except errors.NotBranchError:
921
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
922
self.bzrdir.root_transport, evaluate=Evaluator()):
923
if branch is not None:
924
branches.append(branch)
925
if not using and repository is not None:
926
branches.extend(repository.find_branches())
930
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
931
"""Return the revision ids that other has that this does not.
933
These are returned in topological order.
935
revision_id: only return revision ids included by revision_id.
937
return InterRepository.get(other, self).search_missing_revision_ids(
938
revision_id, find_ghosts)
940
@deprecated_method(one_two)
942
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
943
"""Return the revision ids that other has that this does not.
945
These are returned in topological order.
947
revision_id: only return revision ids included by revision_id.
949
keys = self.search_missing_revision_ids(
950
other, revision_id, find_ghosts).get_keys()
953
parents = other.get_graph().get_parent_map(keys)
956
return tsort.topo_sort(parents)
960
"""Open the repository rooted at base.
962
For instance, if the repository is at URL/.bzr/repository,
963
Repository.open(URL) -> a Repository instance.
965
control = bzrdir.BzrDir.open(base)
966
return control.open_repository()
968
def copy_content_into(self, destination, revision_id=None):
969
"""Make a complete copy of the content in self into destination.
971
This is a destructive operation! Do not use it on existing
974
return InterRepository.get(self, destination).copy_content(revision_id)
976
def commit_write_group(self):
977
"""Commit the contents accrued within the current write group.
979
:seealso: start_write_group.
981
if self._write_group is not self.get_transaction():
982
# has an unlock or relock occured ?
983
raise errors.BzrError('mismatched lock context %r and '
985
(self.get_transaction(), self._write_group))
986
self._commit_write_group()
987
self._write_group = None
989
def _commit_write_group(self):
990
"""Template method for per-repository write group cleanup.
992
This is called before the write group is considered to be
993
finished and should ensure that all data handed to the repository
994
for writing during the write group is safely committed (to the
995
extent possible considering file system caching etc).
998
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
999
"""Fetch the content required to construct revision_id from source.
1001
If revision_id is None all content is copied.
1002
:param find_ghosts: Find and copy revisions in the source that are
1003
ghosts in the target (and not reachable directly by walking out to
1004
the first-present revision in target from revision_id).
1006
# fast path same-url fetch operations
1007
if self.has_same_location(source):
1008
# check that last_revision is in 'from' and then return a
1010
if (revision_id is not None and
1011
not _mod_revision.is_null(revision_id)):
1012
self.get_revision(revision_id)
1014
# if there is no specific appropriate InterRepository, this will get
1015
# the InterRepository base class, which raises an
1016
# IncompatibleRepositories when asked to fetch.
1017
inter = InterRepository.get(source, self)
1018
return inter.fetch(revision_id=revision_id, pb=pb,
1019
find_ghosts=find_ghosts)
1021
def create_bundle(self, target, base, fileobj, format=None):
1022
return serializer.write_bundle(self, target, base, fileobj, format)
1024
def get_commit_builder(self, branch, parents, config, timestamp=None,
1025
timezone=None, committer=None, revprops=None,
1027
"""Obtain a CommitBuilder for this repository.
1029
:param branch: Branch to commit to.
1030
:param parents: Revision ids of the parents of the new revision.
1031
:param config: Configuration to use.
1032
:param timestamp: Optional timestamp recorded for commit.
1033
:param timezone: Optional timezone for timestamp.
1034
:param committer: Optional committer to set for commit.
1035
:param revprops: Optional dictionary of revision properties.
1036
:param revision_id: Optional revision id.
1038
result = self._commit_builder_class(self, parents, config,
1039
timestamp, timezone, committer, revprops, revision_id)
1040
self.start_write_group()
1044
if (self.control_files._lock_count == 1 and
1045
self.control_files._lock_mode == 'w'):
1046
if self._write_group is not None:
1047
self.abort_write_group()
1048
self.control_files.unlock()
1049
raise errors.BzrError(
1050
'Must end write groups before releasing write locks.')
1051
self.control_files.unlock()
1052
for repo in self._fallback_repositories:
1056
def clone(self, a_bzrdir, revision_id=None):
1057
"""Clone this repository into a_bzrdir using the current format.
1059
Currently no check is made that the format of this repository and
1060
the bzrdir format are compatible. FIXME RBC 20060201.
1062
:return: The newly created destination repository.
1064
# TODO: deprecate after 0.16; cloning this with all its settings is
1065
# probably not very useful -- mbp 20070423
1066
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1067
self.copy_content_into(dest_repo, revision_id)
1070
def start_write_group(self):
1071
"""Start a write group in the repository.
1073
Write groups are used by repositories which do not have a 1:1 mapping
1074
between file ids and backend store to manage the insertion of data from
1075
both fetch and commit operations.
1077
A write lock is required around the start_write_group/commit_write_group
1078
for the support of lock-requiring repository formats.
1080
One can only insert data into a repository inside a write group.
1084
if not self.is_write_locked():
1085
raise errors.NotWriteLocked(self)
1086
if self._write_group:
1087
raise errors.BzrError('already in a write group')
1088
self._start_write_group()
1089
# so we can detect unlock/relock - the write group is now entered.
1090
self._write_group = self.get_transaction()
1092
def _start_write_group(self):
1093
"""Template method for per-repository write group startup.
1095
This is called before the write group is considered to be
1100
def sprout(self, to_bzrdir, revision_id=None):
1101
"""Create a descendent repository for new development.
1103
Unlike clone, this does not copy the settings of the repository.
1105
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1106
dest_repo.fetch(self, revision_id=revision_id)
1109
def _create_sprouting_repo(self, a_bzrdir, shared):
1110
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1111
# use target default format.
1112
dest_repo = a_bzrdir.create_repository()
1114
# Most control formats need the repository to be specifically
1115
# created, but on some old all-in-one formats it's not needed
1117
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1118
except errors.UninitializableFormat:
1119
dest_repo = a_bzrdir.open_repository()
1123
def has_revision(self, revision_id):
1124
"""True if this repository has a copy of the revision."""
1125
return revision_id in self.has_revisions((revision_id,))
1128
def has_revisions(self, revision_ids):
1129
"""Probe to find out the presence of multiple revisions.
1131
:param revision_ids: An iterable of revision_ids.
1132
:return: A set of the revision_ids that were present.
1134
parent_map = self.revisions.get_parent_map(
1135
[(rev_id,) for rev_id in revision_ids])
1137
if _mod_revision.NULL_REVISION in revision_ids:
1138
result.add(_mod_revision.NULL_REVISION)
1139
result.update([key[0] for key in parent_map])
1143
def get_revision(self, revision_id):
1144
"""Return the Revision object for a named revision."""
1145
return self.get_revisions([revision_id])[0]
1148
def get_revision_reconcile(self, revision_id):
1149
"""'reconcile' helper routine that allows access to a revision always.
1151
This variant of get_revision does not cross check the weave graph
1152
against the revision one as get_revision does: but it should only
1153
be used by reconcile, or reconcile-alike commands that are correcting
1154
or testing the revision graph.
1156
return self._get_revisions([revision_id])[0]
1159
def get_revisions(self, revision_ids):
1160
"""Get many revisions at once."""
1161
return self._get_revisions(revision_ids)
1164
def _get_revisions(self, revision_ids):
1165
"""Core work logic to get many revisions without sanity checks."""
1166
for rev_id in revision_ids:
1167
if not rev_id or not isinstance(rev_id, basestring):
1168
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1169
keys = [(key,) for key in revision_ids]
1170
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1172
for record in stream:
1173
if record.storage_kind == 'absent':
1174
raise errors.NoSuchRevision(self, record.key[0])
1175
text = record.get_bytes_as('fulltext')
1176
rev = self._serializer.read_revision_from_string(text)
1177
revs[record.key[0]] = rev
1178
return [revs[revid] for revid in revision_ids]
1181
def get_revision_xml(self, revision_id):
1182
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1183
# would have already do it.
1184
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1185
rev = self.get_revision(revision_id)
1186
rev_tmp = cStringIO.StringIO()
1187
# the current serializer..
1188
self._serializer.write_revision(rev, rev_tmp)
1190
return rev_tmp.getvalue()
1192
def get_deltas_for_revisions(self, revisions):
1193
"""Produce a generator of revision deltas.
1195
Note that the input is a sequence of REVISIONS, not revision_ids.
1196
Trees will be held in memory until the generator exits.
1197
Each delta is relative to the revision's lefthand predecessor.
1199
required_trees = set()
1200
for revision in revisions:
1201
required_trees.add(revision.revision_id)
1202
required_trees.update(revision.parent_ids[:1])
1203
trees = dict((t.get_revision_id(), t) for
1204
t in self.revision_trees(required_trees))
1205
for revision in revisions:
1206
if not revision.parent_ids:
1207
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
1209
old_tree = trees[revision.parent_ids[0]]
1210
yield trees[revision.revision_id].changes_from(old_tree)
1213
def get_revision_delta(self, revision_id):
1214
"""Return the delta for one revision.
1216
The delta is relative to the left-hand predecessor of the
1219
r = self.get_revision(revision_id)
1220
return list(self.get_deltas_for_revisions([r]))[0]
1223
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1224
signature = gpg_strategy.sign(plaintext)
1225
self.add_signature_text(revision_id, signature)
1228
def add_signature_text(self, revision_id, signature):
1229
self.signatures.add_lines((revision_id,), (),
1230
osutils.split_lines(signature))
1232
def find_text_key_references(self):
1233
"""Find the text key references within the repository.
1235
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1236
revision_ids. Each altered file-ids has the exact revision_ids that
1237
altered it listed explicitly.
1238
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1239
to whether they were referred to by the inventory of the
1240
revision_id that they contain. The inventory texts from all present
1241
revision ids are assessed to generate this report.
1243
revision_keys = self.revisions.keys()
1244
w = self.inventories
1245
pb = ui.ui_factory.nested_progress_bar()
1247
return self._find_text_key_references_from_xml_inventory_lines(
1248
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1252
def _find_text_key_references_from_xml_inventory_lines(self,
1254
"""Core routine for extracting references to texts from inventories.
1256
This performs the translation of xml lines to revision ids.
1258
:param line_iterator: An iterator of lines, origin_version_id
1259
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1260
to whether they were referred to by the inventory of the
1261
revision_id that they contain. Note that if that revision_id was
1262
not part of the line_iterator's output then False will be given -
1263
even though it may actually refer to that key.
1265
if not self._serializer.support_altered_by_hack:
1266
raise AssertionError(
1267
"_find_text_key_references_from_xml_inventory_lines only "
1268
"supported for branches which store inventory as unnested xml"
1269
", not on %r" % self)
1272
# this code needs to read every new line in every inventory for the
1273
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1274
# not present in one of those inventories is unnecessary but not
1275
# harmful because we are filtering by the revision id marker in the
1276
# inventory lines : we only select file ids altered in one of those
1277
# revisions. We don't need to see all lines in the inventory because
1278
# only those added in an inventory in rev X can contain a revision=X
1280
unescape_revid_cache = {}
1281
unescape_fileid_cache = {}
1283
# jam 20061218 In a big fetch, this handles hundreds of thousands
1284
# of lines, so it has had a lot of inlining and optimizing done.
1285
# Sorry that it is a little bit messy.
1286
# Move several functions to be local variables, since this is a long
1288
search = self._file_ids_altered_regex.search
1289
unescape = _unescape_xml
1290
setdefault = result.setdefault
1291
for line, line_key in line_iterator:
1292
match = search(line)
1295
# One call to match.group() returning multiple items is quite a
1296
# bit faster than 2 calls to match.group() each returning 1
1297
file_id, revision_id = match.group('file_id', 'revision_id')
1299
# Inlining the cache lookups helps a lot when you make 170,000
1300
# lines and 350k ids, versus 8.4 unique ids.
1301
# Using a cache helps in 2 ways:
1302
# 1) Avoids unnecessary decoding calls
1303
# 2) Re-uses cached strings, which helps in future set and
1305
# (2) is enough that removing encoding entirely along with
1306
# the cache (so we are using plain strings) results in no
1307
# performance improvement.
1309
revision_id = unescape_revid_cache[revision_id]
1311
unescaped = unescape(revision_id)
1312
unescape_revid_cache[revision_id] = unescaped
1313
revision_id = unescaped
1315
# Note that unconditionally unescaping means that we deserialise
1316
# every fileid, which for general 'pull' is not great, but we don't
1317
# really want to have some many fulltexts that this matters anyway.
1320
file_id = unescape_fileid_cache[file_id]
1322
unescaped = unescape(file_id)
1323
unescape_fileid_cache[file_id] = unescaped
1326
key = (file_id, revision_id)
1327
setdefault(key, False)
1328
if revision_id == line_key[-1]:
1332
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1334
"""Helper routine for fileids_altered_by_revision_ids.
1336
This performs the translation of xml lines to revision ids.
1338
:param line_iterator: An iterator of lines, origin_version_id
1339
:param revision_ids: The revision ids to filter for. This should be a
1340
set or other type which supports efficient __contains__ lookups, as
1341
the revision id from each parsed line will be looked up in the
1342
revision_ids filter.
1343
:return: a dictionary mapping altered file-ids to an iterable of
1344
revision_ids. Each altered file-ids has the exact revision_ids that
1345
altered it listed explicitly.
1348
setdefault = result.setdefault
1350
self._find_text_key_references_from_xml_inventory_lines(
1351
line_iterator).iterkeys():
1352
# once data is all ensured-consistent; then this is
1353
# if revision_id == version_id
1354
if key[-1:] in revision_ids:
1355
setdefault(key[0], set()).add(key[-1])
1358
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1359
"""Find the file ids and versions affected by revisions.
1361
:param revisions: an iterable containing revision ids.
1362
:param _inv_weave: The inventory weave from this repository or None.
1363
If None, the inventory weave will be opened automatically.
1364
:return: a dictionary mapping altered file-ids to an iterable of
1365
revision_ids. Each altered file-ids has the exact revision_ids that
1366
altered it listed explicitly.
1368
selected_keys = set((revid,) for revid in revision_ids)
1369
w = _inv_weave or self.inventories
1370
pb = ui.ui_factory.nested_progress_bar()
1372
return self._find_file_ids_from_xml_inventory_lines(
1373
w.iter_lines_added_or_present_in_keys(
1374
selected_keys, pb=pb),
1379
def iter_files_bytes(self, desired_files):
1380
"""Iterate through file versions.
1382
Files will not necessarily be returned in the order they occur in
1383
desired_files. No specific order is guaranteed.
1385
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1386
value supplied by the caller as part of desired_files. It should
1387
uniquely identify the file version in the caller's context. (Examples:
1388
an index number or a TreeTransform trans_id.)
1390
bytes_iterator is an iterable of bytestrings for the file. The
1391
kind of iterable and length of the bytestrings are unspecified, but for
1392
this implementation, it is a list of bytes produced by
1393
VersionedFile.get_record_stream().
1395
:param desired_files: a list of (file_id, revision_id, identifier)
1398
transaction = self.get_transaction()
1400
for file_id, revision_id, callable_data in desired_files:
1401
text_keys[(file_id, revision_id)] = callable_data
1402
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1403
if record.storage_kind == 'absent':
1404
raise errors.RevisionNotPresent(record.key, self)
1405
yield text_keys[record.key], record.get_bytes_as('fulltext')
1407
def _generate_text_key_index(self, text_key_references=None,
1409
"""Generate a new text key index for the repository.
1411
This is an expensive function that will take considerable time to run.
1413
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1414
list of parents, also text keys. When a given key has no parents,
1415
the parents list will be [NULL_REVISION].
1417
# All revisions, to find inventory parents.
1418
if ancestors is None:
1419
graph = self.get_graph()
1420
ancestors = graph.get_parent_map(self.all_revision_ids())
1421
if text_key_references is None:
1422
text_key_references = self.find_text_key_references()
1423
pb = ui.ui_factory.nested_progress_bar()
1425
return self._do_generate_text_key_index(ancestors,
1426
text_key_references, pb)
1430
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1431
"""Helper for _generate_text_key_index to avoid deep nesting."""
1432
revision_order = tsort.topo_sort(ancestors)
1433
invalid_keys = set()
1435
for revision_id in revision_order:
1436
revision_keys[revision_id] = set()
1437
text_count = len(text_key_references)
1438
# a cache of the text keys to allow reuse; costs a dict of all the
1439
# keys, but saves a 2-tuple for every child of a given key.
1441
for text_key, valid in text_key_references.iteritems():
1443
invalid_keys.add(text_key)
1445
revision_keys[text_key[1]].add(text_key)
1446
text_key_cache[text_key] = text_key
1447
del text_key_references
1449
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1450
NULL_REVISION = _mod_revision.NULL_REVISION
1451
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1452
# too small for large or very branchy trees. However, for 55K path
1453
# trees, it would be easy to use too much memory trivially. Ideally we
1454
# could gauge this by looking at available real memory etc, but this is
1455
# always a tricky proposition.
1456
inventory_cache = lru_cache.LRUCache(10)
1457
batch_size = 10 # should be ~150MB on a 55K path tree
1458
batch_count = len(revision_order) / batch_size + 1
1460
pb.update("Calculating text parents.", processed_texts, text_count)
1461
for offset in xrange(batch_count):
1462
to_query = revision_order[offset * batch_size:(offset + 1) *
1466
for rev_tree in self.revision_trees(to_query):
1467
revision_id = rev_tree.get_revision_id()
1468
parent_ids = ancestors[revision_id]
1469
for text_key in revision_keys[revision_id]:
1470
pb.update("Calculating text parents.", processed_texts)
1471
processed_texts += 1
1472
candidate_parents = []
1473
for parent_id in parent_ids:
1474
parent_text_key = (text_key[0], parent_id)
1476
check_parent = parent_text_key not in \
1477
revision_keys[parent_id]
1479
# the parent parent_id is a ghost:
1480
check_parent = False
1481
# truncate the derived graph against this ghost.
1482
parent_text_key = None
1484
# look at the parent commit details inventories to
1485
# determine possible candidates in the per file graph.
1488
inv = inventory_cache[parent_id]
1490
inv = self.revision_tree(parent_id).inventory
1491
inventory_cache[parent_id] = inv
1492
parent_entry = inv._byid.get(text_key[0], None)
1493
if parent_entry is not None:
1495
text_key[0], parent_entry.revision)
1497
parent_text_key = None
1498
if parent_text_key is not None:
1499
candidate_parents.append(
1500
text_key_cache[parent_text_key])
1501
parent_heads = text_graph.heads(candidate_parents)
1502
new_parents = list(parent_heads)
1503
new_parents.sort(key=lambda x:candidate_parents.index(x))
1504
if new_parents == []:
1505
new_parents = [NULL_REVISION]
1506
text_index[text_key] = new_parents
1508
for text_key in invalid_keys:
1509
text_index[text_key] = [NULL_REVISION]
1512
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1513
"""Get an iterable listing the keys of all the data introduced by a set
1516
The keys will be ordered so that the corresponding items can be safely
1517
fetched and inserted in that order.
1519
:returns: An iterable producing tuples of (knit-kind, file-id,
1520
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1521
'revisions'. file-id is None unless knit-kind is 'file'.
1523
# XXX: it's a bit weird to control the inventory weave caching in this
1524
# generator. Ideally the caching would be done in fetch.py I think. Or
1525
# maybe this generator should explicitly have the contract that it
1526
# should not be iterated until the previously yielded item has been
1528
inv_w = self.inventories
1530
# file ids that changed
1531
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1533
num_file_ids = len(file_ids)
1534
for file_id, altered_versions in file_ids.iteritems():
1535
if _files_pb is not None:
1536
_files_pb.update("fetch texts", count, num_file_ids)
1538
yield ("file", file_id, altered_versions)
1539
# We're done with the files_pb. Note that it finished by the caller,
1540
# just as it was created by the caller.
1544
yield ("inventory", None, revision_ids)
1547
revisions_with_signatures = set()
1548
for rev_id in revision_ids:
1550
self.get_signature_text(rev_id)
1551
except errors.NoSuchRevision:
1555
revisions_with_signatures.add(rev_id)
1556
yield ("signatures", None, revisions_with_signatures)
1559
yield ("revisions", None, revision_ids)
1562
def get_inventory(self, revision_id):
1563
"""Get Inventory object by revision id."""
1564
return self.iter_inventories([revision_id]).next()
1566
def iter_inventories(self, revision_ids):
1567
"""Get many inventories by revision_ids.
1569
This will buffer some or all of the texts used in constructing the
1570
inventories in memory, but will only parse a single inventory at a
1573
:return: An iterator of inventories.
1575
if ((None in revision_ids)
1576
or (_mod_revision.NULL_REVISION in revision_ids)):
1577
raise ValueError('cannot get null revision inventory')
1578
return self._iter_inventories(revision_ids)
1580
def _iter_inventories(self, revision_ids):
1581
"""single-document based inventory iteration."""
1582
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1583
yield self.deserialise_inventory(revision_id, text)
1585
def _iter_inventory_xmls(self, revision_ids):
1586
keys = [(revision_id,) for revision_id in revision_ids]
1587
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1589
for record in stream:
1590
if record.storage_kind != 'absent':
1591
texts[record.key] = record.get_bytes_as('fulltext')
1593
raise errors.NoSuchRevision(self, record.key)
1595
yield texts[key], key[-1]
1597
def deserialise_inventory(self, revision_id, xml):
1598
"""Transform the xml into an inventory object.
1600
:param revision_id: The expected revision id of the inventory.
1601
:param xml: A serialised inventory.
1603
result = self._serializer.read_inventory_from_string(xml, revision_id)
1604
if result.revision_id != revision_id:
1605
raise AssertionError('revision id mismatch %s != %s' % (
1606
result.revision_id, revision_id))
1609
def serialise_inventory(self, inv):
1610
return self._serializer.write_inventory_to_string(inv)
1612
def _serialise_inventory_to_lines(self, inv):
1613
return self._serializer.write_inventory_to_lines(inv)
1615
def get_serializer_format(self):
1616
return self._serializer.format_num
1619
def get_inventory_xml(self, revision_id):
1620
"""Get inventory XML as a file object."""
1621
texts = self._iter_inventory_xmls([revision_id])
1623
text, revision_id = texts.next()
1624
except StopIteration:
1625
raise errors.HistoryMissing(self, 'inventory', revision_id)
1629
def get_inventory_sha1(self, revision_id):
1630
"""Return the sha1 hash of the inventory entry
1632
return self.get_revision(revision_id).inventory_sha1
1634
def iter_reverse_revision_history(self, revision_id):
1635
"""Iterate backwards through revision ids in the lefthand history
1637
:param revision_id: The revision id to start with. All its lefthand
1638
ancestors will be traversed.
1640
graph = self.get_graph()
1641
next_id = revision_id
1643
if next_id in (None, _mod_revision.NULL_REVISION):
1646
# Note: The following line may raise KeyError in the event of
1647
# truncated history. We decided not to have a try:except:raise
1648
# RevisionNotPresent here until we see a use for it, because of the
1649
# cost in an inner loop that is by its very nature O(history).
1650
# Robert Collins 20080326
1651
parents = graph.get_parent_map([next_id])[next_id]
1652
if len(parents) == 0:
1655
next_id = parents[0]
1658
def get_revision_inventory(self, revision_id):
1659
"""Return inventory of a past revision."""
1660
# TODO: Unify this with get_inventory()
1661
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1662
# must be the same as its revision, so this is trivial.
1663
if revision_id is None:
1664
# This does not make sense: if there is no revision,
1665
# then it is the current tree inventory surely ?!
1666
# and thus get_root_id() is something that looks at the last
1667
# commit on the branch, and the get_root_id is an inventory check.
1668
raise NotImplementedError
1669
# return Inventory(self.get_root_id())
1671
return self.get_inventory(revision_id)
1673
def is_shared(self):
1674
"""Return True if this repository is flagged as a shared repository."""
1675
raise NotImplementedError(self.is_shared)
1678
def reconcile(self, other=None, thorough=False):
1679
"""Reconcile this repository."""
1680
from bzrlib.reconcile import RepoReconciler
1681
reconciler = RepoReconciler(self, thorough=thorough)
1682
reconciler.reconcile()
1685
def _refresh_data(self):
1686
"""Helper called from lock_* to ensure coherency with disk.
1688
The default implementation does nothing; it is however possible
1689
for repositories to maintain loaded indices across multiple locks
1690
by checking inside their implementation of this method to see
1691
whether their indices are still valid. This depends of course on
1692
the disk format being validatable in this manner.
1696
def revision_tree(self, revision_id):
1697
"""Return Tree for a revision on this branch.
1699
`revision_id` may be NULL_REVISION for the empty tree revision.
1701
revision_id = _mod_revision.ensure_null(revision_id)
1702
# TODO: refactor this to use an existing revision object
1703
# so we don't need to read it in twice.
1704
if revision_id == _mod_revision.NULL_REVISION:
1705
return RevisionTree(self, Inventory(root_id=None),
1706
_mod_revision.NULL_REVISION)
1708
inv = self.get_revision_inventory(revision_id)
1709
return RevisionTree(self, inv, revision_id)
1711
def revision_trees(self, revision_ids):
1712
"""Return Tree for a revision on this branch.
1714
`revision_id` may not be None or 'null:'"""
1715
inventories = self.iter_inventories(revision_ids)
1716
for inv in inventories:
1717
yield RevisionTree(self, inv, inv.revision_id)
1720
def get_ancestry(self, revision_id, topo_sorted=True):
1721
"""Return a list of revision-ids integrated by a revision.
1723
The first element of the list is always None, indicating the origin
1724
revision. This might change when we have history horizons, or
1725
perhaps we should have a new API.
1727
This is topologically sorted.
1729
if _mod_revision.is_null(revision_id):
1731
if not self.has_revision(revision_id):
1732
raise errors.NoSuchRevision(self, revision_id)
1733
graph = self.get_graph()
1735
search = graph._make_breadth_first_searcher([revision_id])
1738
found, ghosts = search.next_with_ghosts()
1739
except StopIteration:
1742
if _mod_revision.NULL_REVISION in keys:
1743
keys.remove(_mod_revision.NULL_REVISION)
1745
parent_map = graph.get_parent_map(keys)
1746
keys = tsort.topo_sort(parent_map)
1747
return [None] + list(keys)
1750
"""Compress the data within the repository.
1752
This operation only makes sense for some repository types. For other
1753
types it should be a no-op that just returns.
1755
This stub method does not require a lock, but subclasses should use
1756
@needs_write_lock as this is a long running call its reasonable to
1757
implicitly lock for the user.
1761
@deprecated_method(one_six)
1762
def print_file(self, file, revision_id):
1763
"""Print `file` to stdout.
1765
FIXME RBC 20060125 as John Meinel points out this is a bad api
1766
- it writes to stdout, it assumes that that is valid etc. Fix
1767
by creating a new more flexible convenience function.
1769
tree = self.revision_tree(revision_id)
1770
# use inventory as it was in that revision
1771
file_id = tree.inventory.path2id(file)
1773
# TODO: jam 20060427 Write a test for this code path
1774
# it had a bug in it, and was raising the wrong
1776
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1777
tree.print_file(file_id)
1779
def get_transaction(self):
1780
return self.control_files.get_transaction()
1782
@deprecated_method(one_one)
1783
def get_parents(self, revision_ids):
1784
"""See StackedParentsProvider.get_parents"""
1785
parent_map = self.get_parent_map(revision_ids)
1786
return [parent_map.get(r, None) for r in revision_ids]
1788
def get_parent_map(self, revision_ids):
1789
"""See graph._StackedParentsProvider.get_parent_map"""
1790
# revisions index works in keys; this just works in revisions
1791
# therefore wrap and unwrap
1794
for revision_id in revision_ids:
1795
if revision_id == _mod_revision.NULL_REVISION:
1796
result[revision_id] = ()
1797
elif revision_id is None:
1798
raise ValueError('get_parent_map(None) is not valid')
1800
query_keys.append((revision_id ,))
1801
for ((revision_id,), parent_keys) in \
1802
self.revisions.get_parent_map(query_keys).iteritems():
1804
result[revision_id] = tuple(parent_revid
1805
for (parent_revid,) in parent_keys)
1807
result[revision_id] = (_mod_revision.NULL_REVISION,)
1810
def _make_parents_provider(self):
1813
def get_graph(self, other_repository=None):
1814
"""Return the graph walker for this repository format"""
1815
parents_provider = self._make_parents_provider()
1816
if (other_repository is not None and
1817
not self.has_same_location(other_repository)):
1818
parents_provider = graph._StackedParentsProvider(
1819
[parents_provider, other_repository._make_parents_provider()])
1820
return graph.Graph(parents_provider)
1822
def _get_versioned_file_checker(self):
1823
"""Return an object suitable for checking versioned files."""
1824
return _VersionedFileChecker(self)
1826
def revision_ids_to_search_result(self, result_set):
1827
"""Convert a set of revision ids to a graph SearchResult."""
1828
result_parents = set()
1829
for parents in self.get_graph().get_parent_map(
1830
result_set).itervalues():
1831
result_parents.update(parents)
1832
included_keys = result_set.intersection(result_parents)
1833
start_keys = result_set.difference(included_keys)
1834
exclude_keys = result_parents.difference(result_set)
1835
result = graph.SearchResult(start_keys, exclude_keys,
1836
len(result_set), result_set)
1840
def set_make_working_trees(self, new_value):
1841
"""Set the policy flag for making working trees when creating branches.
1843
This only applies to branches that use this repository.
1845
The default is 'True'.
1846
:param new_value: True to restore the default, False to disable making
1849
raise NotImplementedError(self.set_make_working_trees)
1851
def make_working_trees(self):
1852
"""Returns the policy for making working trees on new branches."""
1853
raise NotImplementedError(self.make_working_trees)
1856
def sign_revision(self, revision_id, gpg_strategy):
1857
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1858
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1861
def has_signature_for_revision_id(self, revision_id):
1862
"""Query for a revision signature for revision_id in the repository."""
1863
if not self.has_revision(revision_id):
1864
raise errors.NoSuchRevision(self, revision_id)
1865
sig_present = (1 == len(
1866
self.signatures.get_parent_map([(revision_id,)])))
1870
def get_signature_text(self, revision_id):
1871
"""Return the text for a signature."""
1872
stream = self.signatures.get_record_stream([(revision_id,)],
1874
record = stream.next()
1875
if record.storage_kind == 'absent':
1876
raise errors.NoSuchRevision(self, revision_id)
1877
return record.get_bytes_as('fulltext')
1880
def check(self, revision_ids=None):
1881
"""Check consistency of all history of given revision_ids.
1883
Different repository implementations should override _check().
1885
:param revision_ids: A non-empty list of revision_ids whose ancestry
1886
will be checked. Typically the last revision_id of a branch.
1888
return self._check(revision_ids)
1890
def _check(self, revision_ids):
1891
result = check.Check(self)
1895
def _warn_if_deprecated(self):
1896
global _deprecation_warning_done
1897
if _deprecation_warning_done:
1899
_deprecation_warning_done = True
1900
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1901
% (self._format, self.bzrdir.transport.base))
1903
def supports_rich_root(self):
1904
return self._format.rich_root_data
1906
def _check_ascii_revisionid(self, revision_id, method):
1907
"""Private helper for ascii-only repositories."""
1908
# weave repositories refuse to store revisionids that are non-ascii.
1909
if revision_id is not None:
1910
# weaves require ascii revision ids.
1911
if isinstance(revision_id, unicode):
1913
revision_id.encode('ascii')
1914
except UnicodeEncodeError:
1915
raise errors.NonAsciiRevisionId(method, self)
1918
revision_id.decode('ascii')
1919
except UnicodeDecodeError:
1920
raise errors.NonAsciiRevisionId(method, self)
1922
def revision_graph_can_have_wrong_parents(self):
1923
"""Is it possible for this repository to have a revision graph with
1926
If True, then this repository must also implement
1927
_find_inconsistent_revision_parents so that check and reconcile can
1928
check for inconsistencies before proceeding with other checks that may
1929
depend on the revision index being consistent.
1931
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1934
# remove these delegates a while after bzr 0.15
1935
def __make_delegated(name, from_module):
1936
def _deprecated_repository_forwarder():
1937
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1938
% (name, from_module),
1941
m = __import__(from_module, globals(), locals(), [name])
1943
return getattr(m, name)
1944
except AttributeError:
1945
raise AttributeError('module %s has no name %s'
1947
globals()[name] = _deprecated_repository_forwarder
1950
'AllInOneRepository',
1951
'WeaveMetaDirRepository',
1952
'PreSplitOutRepositoryFormat',
1953
'RepositoryFormat4',
1954
'RepositoryFormat5',
1955
'RepositoryFormat6',
1956
'RepositoryFormat7',
1958
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1962
'RepositoryFormatKnit',
1963
'RepositoryFormatKnit1',
1965
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1968
def install_revision(repository, rev, revision_tree):
1969
"""Install all revision data into a repository."""
1970
install_revisions(repository, [(rev, revision_tree, None)])
1973
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1974
"""Install all revision data into a repository.
1976
Accepts an iterable of revision, tree, signature tuples. The signature
1979
repository.start_write_group()
1981
for n, (revision, revision_tree, signature) in enumerate(iterable):
1982
_install_revision(repository, revision, revision_tree, signature)
1984
pb.update('Transferring revisions', n + 1, num_revisions)
1986
repository.abort_write_group()
1989
repository.commit_write_group()
1992
def _install_revision(repository, rev, revision_tree, signature):
1993
"""Install all revision data into a repository."""
1994
present_parents = []
1996
for p_id in rev.parent_ids:
1997
if repository.has_revision(p_id):
1998
present_parents.append(p_id)
1999
parent_trees[p_id] = repository.revision_tree(p_id)
2001
parent_trees[p_id] = repository.revision_tree(
2002
_mod_revision.NULL_REVISION)
2004
inv = revision_tree.inventory
2005
entries = inv.iter_entries()
2006
# backwards compatibility hack: skip the root id.
2007
if not repository.supports_rich_root():
2008
path, root = entries.next()
2009
if root.revision != rev.revision_id:
2010
raise errors.IncompatibleRevision(repr(repository))
2012
for path, ie in entries:
2013
text_keys[(ie.file_id, ie.revision)] = ie
2014
text_parent_map = repository.texts.get_parent_map(text_keys)
2015
missing_texts = set(text_keys) - set(text_parent_map)
2016
# Add the texts that are not already present
2017
for text_key in missing_texts:
2018
ie = text_keys[text_key]
2020
# FIXME: TODO: The following loop overlaps/duplicates that done by
2021
# commit to determine parents. There is a latent/real bug here where
2022
# the parents inserted are not those commit would do - in particular
2023
# they are not filtered by heads(). RBC, AB
2024
for revision, tree in parent_trees.iteritems():
2025
if ie.file_id not in tree:
2027
parent_id = tree.inventory[ie.file_id].revision
2028
if parent_id in text_parents:
2030
text_parents.append((ie.file_id, parent_id))
2031
lines = revision_tree.get_file(ie.file_id).readlines()
2032
repository.texts.add_lines(text_key, text_parents, lines)
2034
# install the inventory
2035
repository.add_inventory(rev.revision_id, inv, present_parents)
2036
except errors.RevisionAlreadyPresent:
2038
if signature is not None:
2039
repository.add_signature_text(rev.revision_id, signature)
2040
repository.add_revision(rev.revision_id, rev, inv)
2043
class MetaDirRepository(Repository):
2044
"""Repositories in the new meta-dir layout.
2046
:ivar _transport: Transport for access to repository control files,
2047
typically pointing to .bzr/repository.
2050
def __init__(self, _format, a_bzrdir, control_files):
2051
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2052
self._transport = control_files._transport
2054
def is_shared(self):
2055
"""Return True if this repository is flagged as a shared repository."""
2056
return self._transport.has('shared-storage')
2059
def set_make_working_trees(self, new_value):
2060
"""Set the policy flag for making working trees when creating branches.
2062
This only applies to branches that use this repository.
2064
The default is 'True'.
2065
:param new_value: True to restore the default, False to disable making
2070
self._transport.delete('no-working-trees')
2071
except errors.NoSuchFile:
2074
self._transport.put_bytes('no-working-trees', '',
2075
mode=self.bzrdir._get_file_mode())
2077
def make_working_trees(self):
2078
"""Returns the policy for making working trees on new branches."""
2079
return not self._transport.has('no-working-trees')
2082
class MetaDirVersionedFileRepository(MetaDirRepository):
2083
"""Repositories in a meta-dir, that work via versioned file objects."""
2085
def __init__(self, _format, a_bzrdir, control_files):
2086
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2090
class RepositoryFormatRegistry(registry.Registry):
2091
"""Registry of RepositoryFormats."""
2093
def get(self, format_string):
2094
r = registry.Registry.get(self, format_string)
2100
format_registry = RepositoryFormatRegistry()
2101
"""Registry of formats, indexed by their identifying format string.
2103
This can contain either format instances themselves, or classes/factories that
2104
can be called to obtain one.
2108
#####################################################################
2109
# Repository Formats
2111
class RepositoryFormat(object):
2112
"""A repository format.
2114
Formats provide three things:
2115
* An initialization routine to construct repository data on disk.
2116
* a format string which is used when the BzrDir supports versioned
2118
* an open routine which returns a Repository instance.
2120
There is one and only one Format subclass for each on-disk format. But
2121
there can be one Repository subclass that is used for several different
2122
formats. The _format attribute on a Repository instance can be used to
2123
determine the disk format.
2125
Formats are placed in an dict by their format string for reference
2126
during opening. These should be subclasses of RepositoryFormat
2129
Once a format is deprecated, just deprecate the initialize and open
2130
methods on the format class. Do not deprecate the object, as the
2131
object will be created every system load.
2133
Common instance attributes:
2134
_matchingbzrdir - the bzrdir format that the repository format was
2135
originally written to work with. This can be used if manually
2136
constructing a bzrdir and repository, or more commonly for test suite
2140
# Set to True or False in derived classes. True indicates that the format
2141
# supports ghosts gracefully.
2142
supports_ghosts = None
2143
# Can this repository be given external locations to lookup additional
2144
# data. Set to True or False in derived classes.
2145
supports_external_lookups = None
2148
return "<%s>" % self.__class__.__name__
2150
def __eq__(self, other):
2151
# format objects are generally stateless
2152
return isinstance(other, self.__class__)
2154
def __ne__(self, other):
2155
return not self == other
2158
def find_format(klass, a_bzrdir):
2159
"""Return the format for the repository object in a_bzrdir.
2161
This is used by bzr native formats that have a "format" file in
2162
the repository. Other methods may be used by different types of
2166
transport = a_bzrdir.get_repository_transport(None)
2167
format_string = transport.get("format").read()
2168
return format_registry.get(format_string)
2169
except errors.NoSuchFile:
2170
raise errors.NoRepositoryPresent(a_bzrdir)
2172
raise errors.UnknownFormatError(format=format_string,
2176
def register_format(klass, format):
2177
format_registry.register(format.get_format_string(), format)
2180
def unregister_format(klass, format):
2181
format_registry.remove(format.get_format_string())
2184
def get_default_format(klass):
2185
"""Return the current default format."""
2186
from bzrlib import bzrdir
2187
return bzrdir.format_registry.make_bzrdir('default').repository_format
2189
def get_format_string(self):
2190
"""Return the ASCII format string that identifies this format.
2192
Note that in pre format ?? repositories the format string is
2193
not permitted nor written to disk.
2195
raise NotImplementedError(self.get_format_string)
2197
def get_format_description(self):
2198
"""Return the short description for this format."""
2199
raise NotImplementedError(self.get_format_description)
2201
# TODO: this shouldn't be in the base class, it's specific to things that
2202
# use weaves or knits -- mbp 20070207
2203
def _get_versioned_file_store(self,
2208
versionedfile_class=None,
2209
versionedfile_kwargs={},
2211
if versionedfile_class is None:
2212
versionedfile_class = self._versionedfile_class
2213
weave_transport = control_files._transport.clone(name)
2214
dir_mode = control_files._dir_mode
2215
file_mode = control_files._file_mode
2216
return VersionedFileStore(weave_transport, prefixed=prefixed,
2218
file_mode=file_mode,
2219
versionedfile_class=versionedfile_class,
2220
versionedfile_kwargs=versionedfile_kwargs,
2223
def initialize(self, a_bzrdir, shared=False):
2224
"""Initialize a repository of this format in a_bzrdir.
2226
:param a_bzrdir: The bzrdir to put the new repository in it.
2227
:param shared: The repository should be initialized as a sharable one.
2228
:returns: The new repository object.
2230
This may raise UninitializableFormat if shared repository are not
2231
compatible the a_bzrdir.
2233
raise NotImplementedError(self.initialize)
2235
def is_supported(self):
2236
"""Is this format supported?
2238
Supported formats must be initializable and openable.
2239
Unsupported formats may not support initialization or committing or
2240
some other features depending on the reason for not being supported.
2244
def check_conversion_target(self, target_format):
2245
raise NotImplementedError(self.check_conversion_target)
2247
def open(self, a_bzrdir, _found=False):
2248
"""Return an instance of this format for the bzrdir a_bzrdir.
2250
_found is a private parameter, do not use it.
2252
raise NotImplementedError(self.open)
2255
class MetaDirRepositoryFormat(RepositoryFormat):
2256
"""Common base class for the new repositories using the metadir layout."""
2258
rich_root_data = False
2259
supports_tree_reference = False
2260
supports_external_lookups = False
2261
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2264
super(MetaDirRepositoryFormat, self).__init__()
2266
def _create_control_files(self, a_bzrdir):
2267
"""Create the required files and the initial control_files object."""
2268
# FIXME: RBC 20060125 don't peek under the covers
2269
# NB: no need to escape relative paths that are url safe.
2270
repository_transport = a_bzrdir.get_repository_transport(self)
2271
control_files = lockable_files.LockableFiles(repository_transport,
2272
'lock', lockdir.LockDir)
2273
control_files.create_lock()
2274
return control_files
2276
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
2277
"""Upload the initial blank content."""
2278
control_files = self._create_control_files(a_bzrdir)
2279
control_files.lock_write()
2280
transport = control_files._transport
2282
utf8_files += [('shared-storage', '')]
2284
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2285
for (filename, content_stream) in files:
2286
transport.put_file(filename, content_stream,
2287
mode=a_bzrdir._get_file_mode())
2288
for (filename, content_bytes) in utf8_files:
2289
transport.put_bytes_non_atomic(filename, content_bytes,
2290
mode=a_bzrdir._get_file_mode())
2292
control_files.unlock()
2295
# formats which have no format string are not discoverable
2296
# and not independently creatable, so are not registered. They're
2297
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2298
# needed, it's constructed directly by the BzrDir. Non-native formats where
2299
# the repository is not separately opened are similar.
2301
format_registry.register_lazy(
2302
'Bazaar-NG Repository format 7',
2303
'bzrlib.repofmt.weaverepo',
2307
format_registry.register_lazy(
2308
'Bazaar-NG Knit Repository Format 1',
2309
'bzrlib.repofmt.knitrepo',
2310
'RepositoryFormatKnit1',
2313
format_registry.register_lazy(
2314
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2315
'bzrlib.repofmt.knitrepo',
2316
'RepositoryFormatKnit3',
2319
format_registry.register_lazy(
2320
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2321
'bzrlib.repofmt.knitrepo',
2322
'RepositoryFormatKnit4',
2325
# Pack-based formats. There is one format for pre-subtrees, and one for
2326
# post-subtrees to allow ease of testing.
2327
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2328
format_registry.register_lazy(
2329
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2330
'bzrlib.repofmt.pack_repo',
2331
'RepositoryFormatKnitPack1',
2333
format_registry.register_lazy(
2334
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2335
'bzrlib.repofmt.pack_repo',
2336
'RepositoryFormatKnitPack3',
2338
format_registry.register_lazy(
2339
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2340
'bzrlib.repofmt.pack_repo',
2341
'RepositoryFormatKnitPack4',
2343
format_registry.register_lazy(
2344
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2345
'bzrlib.repofmt.pack_repo',
2346
'RepositoryFormatKnitPack5',
2348
format_registry.register_lazy(
2349
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2350
'bzrlib.repofmt.pack_repo',
2351
'RepositoryFormatKnitPack5RichRoot',
2353
format_registry.register_lazy(
2354
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2355
'bzrlib.repofmt.pack_repo',
2356
'RepositoryFormatKnitPack5RichRootBroken',
2358
format_registry.register_lazy(
2359
'Bazaar RepositoryFormatKnitPack6 (bzr 1.9)\n',
2360
'bzrlib.repofmt.pack_repo',
2361
'RepositoryFormatKnitPack6',
2363
format_registry.register_lazy(
2364
'Bazaar RepositoryFormatKnitPack6RichRoot (bzr 1.9)\n',
2365
'bzrlib.repofmt.pack_repo',
2366
'RepositoryFormatKnitPack6RichRoot',
2369
# Development formats.
2370
# 1.7->1.8 go below here
2371
format_registry.register_lazy(
2372
"Bazaar development format 2 (needs bzr.dev from before 1.8)\n",
2373
'bzrlib.repofmt.pack_repo',
2374
'RepositoryFormatPackDevelopment2',
2376
format_registry.register_lazy(
2377
("Bazaar development format 2 with subtree support "
2378
"(needs bzr.dev from before 1.8)\n"),
2379
'bzrlib.repofmt.pack_repo',
2380
'RepositoryFormatPackDevelopment2Subtree',
2384
class InterRepository(InterObject):
2385
"""This class represents operations taking place between two repositories.
2387
Its instances have methods like copy_content and fetch, and contain
2388
references to the source and target repositories these operations can be
2391
Often we will provide convenience methods on 'repository' which carry out
2392
operations with another repository - they will always forward to
2393
InterRepository.get(other).method_name(parameters).
2396
_walk_to_common_revisions_batch_size = 1
2398
"""The available optimised InterRepository types."""
2400
def __init__(self, source, target):
2401
InterObject.__init__(self, source, target)
2402
# These two attributes may be overridden by e.g. InterOtherToRemote to
2403
# provide a faster implementation.
2404
self.target_get_graph = self.target.get_graph
2405
self.target_get_parent_map = self.target.get_parent_map
2407
def copy_content(self, revision_id=None):
2408
raise NotImplementedError(self.copy_content)
2410
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2411
"""Fetch the content required to construct revision_id.
2413
The content is copied from self.source to self.target.
2415
:param revision_id: if None all content is copied, if NULL_REVISION no
2417
:param pb: optional progress bar to use for progress reports. If not
2418
provided a default one will be created.
2420
:returns: (copied_revision_count, failures).
2422
# Normally we should find a specific InterRepository subclass to do
2423
# the fetch; if nothing else then at least InterSameDataRepository.
2424
# If none of them is suitable it looks like fetching is not possible;
2425
# we try to give a good message why. _assert_same_model will probably
2426
# give a helpful message; otherwise a generic one.
2427
self._assert_same_model(self.source, self.target)
2428
raise errors.IncompatibleRepositories(self.source, self.target,
2429
"no suitableInterRepository found")
2431
def _walk_to_common_revisions(self, revision_ids):
2432
"""Walk out from revision_ids in source to revisions target has.
2434
:param revision_ids: The start point for the search.
2435
:return: A set of revision ids.
2437
target_graph = self.target_get_graph()
2438
revision_ids = frozenset(revision_ids)
2439
# Fast path for the case where all the revisions are already in the
2441
# (Although this does incur an extra round trip for the
2442
# fairly common case where the target doesn't already have the revision
2444
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2445
return graph.SearchResult(revision_ids, set(), 0, set())
2446
missing_revs = set()
2447
source_graph = self.source.get_graph()
2448
# ensure we don't pay silly lookup costs.
2449
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2450
null_set = frozenset([_mod_revision.NULL_REVISION])
2451
searcher_exhausted = False
2455
# Iterate the searcher until we have enough next_revs
2456
while len(next_revs) < self._walk_to_common_revisions_batch_size:
2458
next_revs_part, ghosts_part = searcher.next_with_ghosts()
2459
next_revs.update(next_revs_part)
2460
ghosts.update(ghosts_part)
2461
except StopIteration:
2462
searcher_exhausted = True
2464
# If there are ghosts in the source graph, and the caller asked for
2465
# them, make sure that they are present in the target.
2466
# We don't care about other ghosts as we can't fetch them and
2467
# haven't been asked to.
2468
ghosts_to_check = set(revision_ids.intersection(ghosts))
2469
revs_to_get = set(next_revs).union(ghosts_to_check)
2471
have_revs = set(target_graph.get_parent_map(revs_to_get))
2472
# we always have NULL_REVISION present.
2473
have_revs = have_revs.union(null_set)
2474
# Check if the target is missing any ghosts we need.
2475
ghosts_to_check.difference_update(have_revs)
2477
# One of the caller's revision_ids is a ghost in both the
2478
# source and the target.
2479
raise errors.NoSuchRevision(
2480
self.source, ghosts_to_check.pop())
2481
missing_revs.update(next_revs - have_revs)
2482
# Because we may have walked past the original stop point, make
2483
# sure everything is stopped
2484
stop_revs = searcher.find_seen_ancestors(have_revs)
2485
searcher.stop_searching_any(stop_revs)
2486
if searcher_exhausted:
2488
return searcher.get_result()
2490
@deprecated_method(one_two)
2492
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2493
"""Return the revision ids that source has that target does not.
2495
These are returned in topological order.
2497
:param revision_id: only return revision ids included by this
2499
:param find_ghosts: If True find missing revisions in deep history
2500
rather than just finding the surface difference.
2502
return list(self.search_missing_revision_ids(
2503
revision_id, find_ghosts).get_keys())
2506
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2507
"""Return the revision ids that source has that target does not.
2509
:param revision_id: only return revision ids included by this
2511
:param find_ghosts: If True find missing revisions in deep history
2512
rather than just finding the surface difference.
2513
:return: A bzrlib.graph.SearchResult.
2515
# stop searching at found target revisions.
2516
if not find_ghosts and revision_id is not None:
2517
return self._walk_to_common_revisions([revision_id])
2518
# generic, possibly worst case, slow code path.
2519
target_ids = set(self.target.all_revision_ids())
2520
if revision_id is not None:
2521
source_ids = self.source.get_ancestry(revision_id)
2522
if source_ids[0] is not None:
2523
raise AssertionError()
2526
source_ids = self.source.all_revision_ids()
2527
result_set = set(source_ids).difference(target_ids)
2528
return self.source.revision_ids_to_search_result(result_set)
2531
def _same_model(source, target):
2532
"""True if source and target have the same data representation.
2534
Note: this is always called on the base class; overriding it in a
2535
subclass will have no effect.
2538
InterRepository._assert_same_model(source, target)
2540
except errors.IncompatibleRepositories, e:
2544
def _assert_same_model(source, target):
2545
"""Raise an exception if two repositories do not use the same model.
2547
if source.supports_rich_root() != target.supports_rich_root():
2548
raise errors.IncompatibleRepositories(source, target,
2549
"different rich-root support")
2550
if source._serializer != target._serializer:
2551
raise errors.IncompatibleRepositories(source, target,
2552
"different serializers")
2555
class InterSameDataRepository(InterRepository):
2556
"""Code for converting between repositories that represent the same data.
2558
Data format and model must match for this to work.
2562
def _get_repo_format_to_test(self):
2563
"""Repository format for testing with.
2565
InterSameData can pull from subtree to subtree and from non-subtree to
2566
non-subtree, so we test this with the richest repository format.
2568
from bzrlib.repofmt import knitrepo
2569
return knitrepo.RepositoryFormatKnit3()
2572
def is_compatible(source, target):
2573
return InterRepository._same_model(source, target)
2576
def copy_content(self, revision_id=None):
2577
"""Make a complete copy of the content in self into destination.
2579
This copies both the repository's revision data, and configuration information
2580
such as the make_working_trees setting.
2582
This is a destructive operation! Do not use it on existing
2585
:param revision_id: Only copy the content needed to construct
2586
revision_id and its parents.
2589
self.target.set_make_working_trees(self.source.make_working_trees())
2590
except NotImplementedError:
2592
# but don't bother fetching if we have the needed data now.
2593
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2594
self.target.has_revision(revision_id)):
2596
self.target.fetch(self.source, revision_id=revision_id)
2599
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2600
"""See InterRepository.fetch()."""
2601
from bzrlib.fetch import RepoFetcher
2602
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2603
self.source, self.source._format, self.target,
2604
self.target._format)
2605
f = RepoFetcher(to_repository=self.target,
2606
from_repository=self.source,
2607
last_revision=revision_id,
2608
pb=pb, find_ghosts=find_ghosts)
2609
return f.count_copied, f.failed_revisions
2612
class InterWeaveRepo(InterSameDataRepository):
2613
"""Optimised code paths between Weave based repositories.
2615
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2616
implemented lazy inter-object optimisation.
2620
def _get_repo_format_to_test(self):
2621
from bzrlib.repofmt import weaverepo
2622
return weaverepo.RepositoryFormat7()
2625
def is_compatible(source, target):
2626
"""Be compatible with known Weave formats.
2628
We don't test for the stores being of specific types because that
2629
could lead to confusing results, and there is no need to be
2632
from bzrlib.repofmt.weaverepo import (
2638
return (isinstance(source._format, (RepositoryFormat5,
2640
RepositoryFormat7)) and
2641
isinstance(target._format, (RepositoryFormat5,
2643
RepositoryFormat7)))
2644
except AttributeError:
2648
def copy_content(self, revision_id=None):
2649
"""See InterRepository.copy_content()."""
2650
# weave specific optimised path:
2652
self.target.set_make_working_trees(self.source.make_working_trees())
2653
except (errors.RepositoryUpgradeRequired, NotImplemented):
2655
# FIXME do not peek!
2656
if self.source._transport.listable():
2657
pb = ui.ui_factory.nested_progress_bar()
2659
self.target.texts.insert_record_stream(
2660
self.source.texts.get_record_stream(
2661
self.source.texts.keys(), 'topological', False))
2662
pb.update('copying inventory', 0, 1)
2663
self.target.inventories.insert_record_stream(
2664
self.source.inventories.get_record_stream(
2665
self.source.inventories.keys(), 'topological', False))
2666
self.target.signatures.insert_record_stream(
2667
self.source.signatures.get_record_stream(
2668
self.source.signatures.keys(),
2670
self.target.revisions.insert_record_stream(
2671
self.source.revisions.get_record_stream(
2672
self.source.revisions.keys(),
2673
'topological', True))
2677
self.target.fetch(self.source, revision_id=revision_id)
2680
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2681
"""See InterRepository.fetch()."""
2682
from bzrlib.fetch import RepoFetcher
2683
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2684
self.source, self.source._format, self.target, self.target._format)
2685
f = RepoFetcher(to_repository=self.target,
2686
from_repository=self.source,
2687
last_revision=revision_id,
2688
pb=pb, find_ghosts=find_ghosts)
2689
return f.count_copied, f.failed_revisions
2692
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2693
"""See InterRepository.missing_revision_ids()."""
2694
# we want all revisions to satisfy revision_id in source.
2695
# but we don't want to stat every file here and there.
2696
# we want then, all revisions other needs to satisfy revision_id
2697
# checked, but not those that we have locally.
2698
# so the first thing is to get a subset of the revisions to
2699
# satisfy revision_id in source, and then eliminate those that
2700
# we do already have.
2701
# this is slow on high latency connection to self, but as as this
2702
# disk format scales terribly for push anyway due to rewriting
2703
# inventory.weave, this is considered acceptable.
2705
if revision_id is not None:
2706
source_ids = self.source.get_ancestry(revision_id)
2707
if source_ids[0] is not None:
2708
raise AssertionError()
2711
source_ids = self.source._all_possible_ids()
2712
source_ids_set = set(source_ids)
2713
# source_ids is the worst possible case we may need to pull.
2714
# now we want to filter source_ids against what we actually
2715
# have in target, but don't try to check for existence where we know
2716
# we do not have a revision as that would be pointless.
2717
target_ids = set(self.target._all_possible_ids())
2718
possibly_present_revisions = target_ids.intersection(source_ids_set)
2719
actually_present_revisions = set(
2720
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2721
required_revisions = source_ids_set.difference(actually_present_revisions)
2722
if revision_id is not None:
2723
# we used get_ancestry to determine source_ids then we are assured all
2724
# revisions referenced are present as they are installed in topological order.
2725
# and the tip revision was validated by get_ancestry.
2726
result_set = required_revisions
2728
# if we just grabbed the possibly available ids, then
2729
# we only have an estimate of whats available and need to validate
2730
# that against the revision records.
2732
self.source._eliminate_revisions_not_present(required_revisions))
2733
return self.source.revision_ids_to_search_result(result_set)
2736
class InterKnitRepo(InterSameDataRepository):
2737
"""Optimised code paths between Knit based repositories."""
2740
def _get_repo_format_to_test(self):
2741
from bzrlib.repofmt import knitrepo
2742
return knitrepo.RepositoryFormatKnit1()
2745
def is_compatible(source, target):
2746
"""Be compatible with known Knit formats.
2748
We don't test for the stores being of specific types because that
2749
could lead to confusing results, and there is no need to be
2752
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2754
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2755
isinstance(target._format, RepositoryFormatKnit))
2756
except AttributeError:
2758
return are_knits and InterRepository._same_model(source, target)
2761
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2762
"""See InterRepository.fetch()."""
2763
from bzrlib.fetch import RepoFetcher
2764
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2765
self.source, self.source._format, self.target, self.target._format)
2766
f = RepoFetcher(to_repository=self.target,
2767
from_repository=self.source,
2768
last_revision=revision_id,
2769
pb=pb, find_ghosts=find_ghosts)
2770
return f.count_copied, f.failed_revisions
2773
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2774
"""See InterRepository.missing_revision_ids()."""
2775
if revision_id is not None:
2776
source_ids = self.source.get_ancestry(revision_id)
2777
if source_ids[0] is not None:
2778
raise AssertionError()
2781
source_ids = self.source.all_revision_ids()
2782
source_ids_set = set(source_ids)
2783
# source_ids is the worst possible case we may need to pull.
2784
# now we want to filter source_ids against what we actually
2785
# have in target, but don't try to check for existence where we know
2786
# we do not have a revision as that would be pointless.
2787
target_ids = set(self.target.all_revision_ids())
2788
possibly_present_revisions = target_ids.intersection(source_ids_set)
2789
actually_present_revisions = set(
2790
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2791
required_revisions = source_ids_set.difference(actually_present_revisions)
2792
if revision_id is not None:
2793
# we used get_ancestry to determine source_ids then we are assured all
2794
# revisions referenced are present as they are installed in topological order.
2795
# and the tip revision was validated by get_ancestry.
2796
result_set = required_revisions
2798
# if we just grabbed the possibly available ids, then
2799
# we only have an estimate of whats available and need to validate
2800
# that against the revision records.
2802
self.source._eliminate_revisions_not_present(required_revisions))
2803
return self.source.revision_ids_to_search_result(result_set)
2806
class InterPackRepo(InterSameDataRepository):
2807
"""Optimised code paths between Pack based repositories."""
2810
def _get_repo_format_to_test(self):
2811
from bzrlib.repofmt import pack_repo
2812
return pack_repo.RepositoryFormatKnitPack1()
2815
def is_compatible(source, target):
2816
"""Be compatible with known Pack formats.
2818
We don't test for the stores being of specific types because that
2819
could lead to confusing results, and there is no need to be
2822
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2824
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2825
isinstance(target._format, RepositoryFormatPack))
2826
except AttributeError:
2828
return are_packs and InterRepository._same_model(source, target)
2831
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2832
"""See InterRepository.fetch()."""
2833
if (len(self.source._fallback_repositories) > 0 or
2834
len(self.target._fallback_repositories) > 0):
2835
# The pack layer is not aware of fallback repositories, so when
2836
# fetching from a stacked repository or into a stacked repository
2837
# we use the generic fetch logic which uses the VersionedFiles
2838
# attributes on repository.
2840
# XXX: Andrew suggests removing the check on the target
2842
from bzrlib.fetch import RepoFetcher
2843
fetcher = RepoFetcher(self.target, self.source, revision_id,
2845
return fetcher.count_copied, fetcher.failed_revisions
2846
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2847
self.source, self.source._format, self.target, self.target._format)
2848
self.count_copied = 0
2849
if revision_id is None:
2851
# everything to do - use pack logic
2852
# to fetch from all packs to one without
2853
# inventory parsing etc, IFF nothing to be copied is in the target.
2855
source_revision_ids = frozenset(self.source.all_revision_ids())
2856
revision_ids = source_revision_ids - \
2857
frozenset(self.target_get_parent_map(source_revision_ids))
2858
revision_keys = [(revid,) for revid in revision_ids]
2859
target_pack_collection = self._get_target_pack_collection()
2860
index = target_pack_collection.revision_index.combined_index
2861
present_revision_ids = set(item[1][0] for item in
2862
index.iter_entries(revision_keys))
2863
revision_ids = set(revision_ids) - present_revision_ids
2864
# implementing the TODO will involve:
2865
# - detecting when all of a pack is selected
2866
# - avoiding as much as possible pre-selection, so the
2867
# more-core routines such as create_pack_from_packs can filter in
2868
# a just-in-time fashion. (though having a HEADS list on a
2869
# repository might make this a lot easier, because we could
2870
# sensibly detect 'new revisions' without doing a full index scan.
2871
elif _mod_revision.is_null(revision_id):
2876
revision_ids = self.search_missing_revision_ids(revision_id,
2877
find_ghosts=find_ghosts).get_keys()
2878
except errors.NoSuchRevision:
2879
raise errors.InstallFailed([revision_id])
2880
if len(revision_ids) == 0:
2882
return self._pack(self.source, self.target, revision_ids)
2884
def _pack(self, source, target, revision_ids):
2885
from bzrlib.repofmt.pack_repo import Packer
2886
target_pack_collection = self._get_target_pack_collection()
2887
packs = source._pack_collection.all_packs()
2888
pack = Packer(target_pack_collection, packs, '.fetch',
2889
revision_ids).pack()
2890
if pack is not None:
2891
target_pack_collection._save_pack_names()
2892
copied_revs = pack.get_revision_count()
2893
# Trigger an autopack. This may duplicate effort as we've just done
2894
# a pack creation, but for now it is simpler to think about as
2895
# 'upload data, then repack if needed'.
2897
return (copied_revs, [])
2901
def _autopack(self):
2902
self.target._pack_collection.autopack()
2904
def _get_target_pack_collection(self):
2905
return self.target._pack_collection
2908
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2909
"""See InterRepository.missing_revision_ids().
2911
:param find_ghosts: Find ghosts throughout the ancestry of
2914
if not find_ghosts and revision_id is not None:
2915
return self._walk_to_common_revisions([revision_id])
2916
elif revision_id is not None:
2917
# Find ghosts: search for revisions pointing from one repository to
2918
# the other, and vice versa, anywhere in the history of revision_id.
2919
graph = self.target_get_graph(other_repository=self.source)
2920
searcher = graph._make_breadth_first_searcher([revision_id])
2924
next_revs, ghosts = searcher.next_with_ghosts()
2925
except StopIteration:
2927
if revision_id in ghosts:
2928
raise errors.NoSuchRevision(self.source, revision_id)
2929
found_ids.update(next_revs)
2930
found_ids.update(ghosts)
2931
found_ids = frozenset(found_ids)
2932
# Double query here: should be able to avoid this by changing the
2933
# graph api further.
2934
result_set = found_ids - frozenset(
2935
self.target_get_parent_map(found_ids))
2937
source_ids = self.source.all_revision_ids()
2938
# source_ids is the worst possible case we may need to pull.
2939
# now we want to filter source_ids against what we actually
2940
# have in target, but don't try to check for existence where we know
2941
# we do not have a revision as that would be pointless.
2942
target_ids = set(self.target.all_revision_ids())
2943
result_set = set(source_ids).difference(target_ids)
2944
return self.source.revision_ids_to_search_result(result_set)
2947
class InterModel1and2(InterRepository):
2950
def _get_repo_format_to_test(self):
2954
def is_compatible(source, target):
2955
if not source.supports_rich_root() and target.supports_rich_root():
2961
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2962
"""See InterRepository.fetch()."""
2963
from bzrlib.fetch import Model1toKnit2Fetcher
2964
f = Model1toKnit2Fetcher(to_repository=self.target,
2965
from_repository=self.source,
2966
last_revision=revision_id,
2967
pb=pb, find_ghosts=find_ghosts)
2968
return f.count_copied, f.failed_revisions
2971
def copy_content(self, revision_id=None):
2972
"""Make a complete copy of the content in self into destination.
2974
This is a destructive operation! Do not use it on existing
2977
:param revision_id: Only copy the content needed to construct
2978
revision_id and its parents.
2981
self.target.set_make_working_trees(self.source.make_working_trees())
2982
except NotImplementedError:
2984
# but don't bother fetching if we have the needed data now.
2985
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2986
self.target.has_revision(revision_id)):
2988
self.target.fetch(self.source, revision_id=revision_id)
2991
class InterKnit1and2(InterKnitRepo):
2994
def _get_repo_format_to_test(self):
2998
def is_compatible(source, target):
2999
"""Be compatible with Knit1 source and Knit3 target"""
3001
from bzrlib.repofmt.knitrepo import (
3002
RepositoryFormatKnit1,
3003
RepositoryFormatKnit3,
3005
from bzrlib.repofmt.pack_repo import (
3006
RepositoryFormatKnitPack1,
3007
RepositoryFormatKnitPack3,
3008
RepositoryFormatKnitPack4,
3009
RepositoryFormatKnitPack5,
3010
RepositoryFormatKnitPack5RichRoot,
3011
RepositoryFormatKnitPack6,
3012
RepositoryFormatKnitPack6RichRoot,
3013
RepositoryFormatPackDevelopment2,
3014
RepositoryFormatPackDevelopment2Subtree,
3017
RepositoryFormatKnit1, # no rr, no subtree
3018
RepositoryFormatKnitPack1, # no rr, no subtree
3019
RepositoryFormatPackDevelopment2, # no rr, no subtree
3020
RepositoryFormatKnitPack5, # no rr, no subtree
3021
RepositoryFormatKnitPack6, # no rr, no subtree
3024
RepositoryFormatKnit3, # rr, subtree
3025
RepositoryFormatKnitPack3, # rr, subtree
3026
RepositoryFormatKnitPack4, # rr, no subtree
3027
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
3028
RepositoryFormatKnitPack6RichRoot,# rr, no subtree
3029
RepositoryFormatPackDevelopment2Subtree, # rr, subtree
3031
for format in norichroot:
3032
if format.rich_root_data:
3033
raise AssertionError('Format %s is a rich-root format'
3034
' but is included in the non-rich-root list'
3036
for format in richroot:
3037
if not format.rich_root_data:
3038
raise AssertionError('Format %s is not a rich-root format'
3039
' but is included in the rich-root list'
3041
# TODO: One alternative is to just check format.rich_root_data,
3042
# instead of keeping membership lists. However, the formats
3043
# *also* have to use the same 'Knit' style of storage
3044
# (line-deltas, fulltexts, etc.)
3045
return (isinstance(source._format, norichroot) and
3046
isinstance(target._format, richroot))
3047
except AttributeError:
3051
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3052
"""See InterRepository.fetch()."""
3053
from bzrlib.fetch import Knit1to2Fetcher
3054
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
3055
self.source, self.source._format, self.target,
3056
self.target._format)
3057
f = Knit1to2Fetcher(to_repository=self.target,
3058
from_repository=self.source,
3059
last_revision=revision_id,
3060
pb=pb, find_ghosts=find_ghosts)
3061
return f.count_copied, f.failed_revisions
3064
class InterDifferingSerializer(InterKnitRepo):
3067
def _get_repo_format_to_test(self):
3071
def is_compatible(source, target):
3072
"""Be compatible with Knit2 source and Knit3 target"""
3073
if source.supports_rich_root() != target.supports_rich_root():
3075
# Ideally, we'd support fetching if the source had no tree references
3076
# even if it supported them...
3077
if (getattr(source, '_format.supports_tree_reference', False) and
3078
not getattr(target, '_format.supports_tree_reference', False)):
3083
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3084
"""See InterRepository.fetch()."""
3085
revision_ids = self.target.search_missing_revision_ids(self.source,
3086
revision_id, find_ghosts=find_ghosts).get_keys()
3087
revision_ids = tsort.topo_sort(
3088
self.source.get_graph().get_parent_map(revision_ids))
3089
def revisions_iterator():
3090
for current_revision_id in revision_ids:
3091
revision = self.source.get_revision(current_revision_id)
3092
tree = self.source.revision_tree(current_revision_id)
3094
signature = self.source.get_signature_text(
3095
current_revision_id)
3096
except errors.NoSuchRevision:
3098
yield revision, tree, signature
3100
my_pb = ui.ui_factory.nested_progress_bar()
3105
install_revisions(self.target, revisions_iterator(),
3106
len(revision_ids), pb)
3108
if my_pb is not None:
3110
return len(revision_ids), 0
3113
class InterOtherToRemote(InterRepository):
3114
"""An InterRepository that simply delegates to the 'real' InterRepository
3115
calculated for (source, target._real_repository).
3118
_walk_to_common_revisions_batch_size = 50
3120
def __init__(self, source, target):
3121
InterRepository.__init__(self, source, target)
3122
self._real_inter = None
3125
def is_compatible(source, target):
3126
if isinstance(target, remote.RemoteRepository):
3130
def _ensure_real_inter(self):
3131
if self._real_inter is None:
3132
self.target._ensure_real()
3133
real_target = self.target._real_repository
3134
self._real_inter = InterRepository.get(self.source, real_target)
3135
# Make _real_inter use the RemoteRepository for get_parent_map
3136
self._real_inter.target_get_graph = self.target.get_graph
3137
self._real_inter.target_get_parent_map = self.target.get_parent_map
3139
def copy_content(self, revision_id=None):
3140
self._ensure_real_inter()
3141
self._real_inter.copy_content(revision_id=revision_id)
3143
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3144
self._ensure_real_inter()
3145
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3146
find_ghosts=find_ghosts)
3149
def _get_repo_format_to_test(self):
3153
class InterRemoteToOther(InterRepository):
3155
def __init__(self, source, target):
3156
InterRepository.__init__(self, source, target)
3157
self._real_inter = None
3160
def is_compatible(source, target):
3161
if not isinstance(source, remote.RemoteRepository):
3163
# Is source's model compatible with target's model?
3164
source._ensure_real()
3165
real_source = source._real_repository
3166
if isinstance(real_source, remote.RemoteRepository):
3167
raise NotImplementedError(
3168
"We don't support remote repos backed by remote repos yet.")
3169
return InterRepository._same_model(real_source, target)
3171
def _ensure_real_inter(self):
3172
if self._real_inter is None:
3173
self.source._ensure_real()
3174
real_source = self.source._real_repository
3175
self._real_inter = InterRepository.get(real_source, self.target)
3177
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3178
self._ensure_real_inter()
3179
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3180
find_ghosts=find_ghosts)
3182
def copy_content(self, revision_id=None):
3183
self._ensure_real_inter()
3184
self._real_inter.copy_content(revision_id=revision_id)
3187
def _get_repo_format_to_test(self):
3192
class InterPackToRemotePack(InterPackRepo):
3193
"""A specialisation of InterPackRepo for a target that is a
3196
This will use the get_parent_map RPC rather than plain readvs, and also
3197
uses an RPC for autopacking.
3200
_walk_to_common_revisions_batch_size = 50
3203
def is_compatible(source, target):
3204
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
3205
if isinstance(source._format, RepositoryFormatPack):
3206
if isinstance(target, remote.RemoteRepository):
3207
target._ensure_real()
3208
if isinstance(target._real_repository._format,
3209
RepositoryFormatPack):
3210
if InterRepository._same_model(source, target):
3214
def _autopack(self):
3215
self.target.autopack()
3217
def _get_target_pack_collection(self):
3218
return self.target._real_repository._pack_collection
3221
def _get_repo_format_to_test(self):
3225
InterRepository.register_optimiser(InterDifferingSerializer)
3226
InterRepository.register_optimiser(InterSameDataRepository)
3227
InterRepository.register_optimiser(InterWeaveRepo)
3228
InterRepository.register_optimiser(InterKnitRepo)
3229
InterRepository.register_optimiser(InterModel1and2)
3230
InterRepository.register_optimiser(InterKnit1and2)
3231
InterRepository.register_optimiser(InterPackRepo)
3232
InterRepository.register_optimiser(InterOtherToRemote)
3233
InterRepository.register_optimiser(InterRemoteToOther)
3234
InterRepository.register_optimiser(InterPackToRemotePack)
3237
class CopyConverter(object):
3238
"""A repository conversion tool which just performs a copy of the content.
3240
This is slow but quite reliable.
3243
def __init__(self, target_format):
3244
"""Create a CopyConverter.
3246
:param target_format: The format the resulting repository should be.
3248
self.target_format = target_format
3250
def convert(self, repo, pb):
3251
"""Perform the conversion of to_convert, giving feedback via pb.
3253
:param to_convert: The disk object to convert.
3254
:param pb: a progress bar to use for progress information.
3259
# this is only useful with metadir layouts - separated repo content.
3260
# trigger an assertion if not such
3261
repo._format.get_format_string()
3262
self.repo_dir = repo.bzrdir
3263
self.step('Moving repository to repository.backup')
3264
self.repo_dir.transport.move('repository', 'repository.backup')
3265
backup_transport = self.repo_dir.transport.clone('repository.backup')
3266
repo._format.check_conversion_target(self.target_format)
3267
self.source_repo = repo._format.open(self.repo_dir,
3269
_override_transport=backup_transport)
3270
self.step('Creating new repository')
3271
converted = self.target_format.initialize(self.repo_dir,
3272
self.source_repo.is_shared())
3273
converted.lock_write()
3275
self.step('Copying content into repository.')
3276
self.source_repo.copy_content_into(converted)
3279
self.step('Deleting old repository content.')
3280
self.repo_dir.transport.delete_tree('repository.backup')
3281
self.pb.note('repository converted')
3283
def step(self, message):
3284
"""Update the pb by a step."""
3286
self.pb.update(message, self.count, self.total)
3298
def _unescaper(match, _map=_unescape_map):
3299
code = match.group(1)
3303
if not code.startswith('#'):
3305
return unichr(int(code[1:])).encode('utf8')
3311
def _unescape_xml(data):
3312
"""Unescape predefined XML entities in a string of data."""
3314
if _unescape_re is None:
3315
_unescape_re = re.compile('\&([^;]*);')
3316
return _unescape_re.sub(_unescaper, data)
3319
class _VersionedFileChecker(object):
3321
def __init__(self, repository):
3322
self.repository = repository
3323
self.text_index = self.repository._generate_text_key_index()
3325
def calculate_file_version_parents(self, text_key):
3326
"""Calculate the correct parents for a file version according to
3329
parent_keys = self.text_index[text_key]
3330
if parent_keys == [_mod_revision.NULL_REVISION]:
3332
return tuple(parent_keys)
3334
def check_file_version_parents(self, texts, progress_bar=None):
3335
"""Check the parents stored in a versioned file are correct.
3337
It also detects file versions that are not referenced by their
3338
corresponding revision's inventory.
3340
:returns: A tuple of (wrong_parents, dangling_file_versions).
3341
wrong_parents is a dict mapping {revision_id: (stored_parents,
3342
correct_parents)} for each revision_id where the stored parents
3343
are not correct. dangling_file_versions is a set of (file_id,
3344
revision_id) tuples for versions that are present in this versioned
3345
file, but not used by the corresponding inventory.
3348
self.file_ids = set([file_id for file_id, _ in
3349
self.text_index.iterkeys()])
3350
# text keys is now grouped by file_id
3351
n_weaves = len(self.file_ids)
3352
files_in_revisions = {}
3353
revisions_of_files = {}
3354
n_versions = len(self.text_index)
3355
progress_bar.update('loading text store', 0, n_versions)
3356
parent_map = self.repository.texts.get_parent_map(self.text_index)
3357
# On unlistable transports this could well be empty/error...
3358
text_keys = self.repository.texts.keys()
3359
unused_keys = frozenset(text_keys) - set(self.text_index)
3360
for num, key in enumerate(self.text_index.iterkeys()):
3361
if progress_bar is not None:
3362
progress_bar.update('checking text graph', num, n_versions)
3363
correct_parents = self.calculate_file_version_parents(key)
3365
knit_parents = parent_map[key]
3366
except errors.RevisionNotPresent:
3369
if correct_parents != knit_parents:
3370
wrong_parents[key] = (knit_parents, correct_parents)
3371
return wrong_parents, unused_keys
3374
def _old_get_graph(repository, revision_id):
3375
"""DO NOT USE. That is all. I'm serious."""
3376
graph = repository.get_graph()
3377
revision_graph = dict(((key, value) for key, value in
3378
graph.iter_ancestry([revision_id]) if value is not None))
3379
return _strip_NULL_ghosts(revision_graph)
3382
def _strip_NULL_ghosts(revision_graph):
3383
"""Also don't use this. more compatibility code for unmigrated clients."""
3384
# Filter ghosts, and null:
3385
if _mod_revision.NULL_REVISION in revision_graph:
3386
del revision_graph[_mod_revision.NULL_REVISION]
3387
for key, parents in revision_graph.items():
3388
revision_graph[key] = tuple(parent for parent in parents if parent
3390
return revision_graph
added
removed
bzrlib/repository.py
