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._revprops.update(revprops)
108
if timestamp is None:
109
timestamp = time.time()
110
# Restrict resolution to 1ms
111
self._timestamp = round(timestamp, 3)
114
self._timezone = osutils.local_time_offset()
116
self._timezone = int(timezone)
118
self._generate_revision_if_needed()
119
self.__heads = graph.HeadsCache(repository.get_graph()).heads
121
def commit(self, message):
122
"""Make the actual commit.
124
:return: The revision id of the recorded revision.
126
rev = _mod_revision.Revision(
127
timestamp=self._timestamp,
128
timezone=self._timezone,
129
committer=self._committer,
131
inventory_sha1=self.inv_sha1,
132
revision_id=self._new_revision_id,
133
properties=self._revprops)
134
rev.parent_ids = self.parents
135
self.repository.add_revision(self._new_revision_id, rev,
136
self.new_inventory, self._config)
137
self.repository.commit_write_group()
138
return self._new_revision_id
141
"""Abort the commit that is being built.
143
self.repository.abort_write_group()
145
def revision_tree(self):
146
"""Return the tree that was just committed.
148
After calling commit() this can be called to get a RevisionTree
149
representing the newly committed tree. This is preferred to
150
calling Repository.revision_tree() because that may require
151
deserializing the inventory, while we already have a copy in
154
return RevisionTree(self.repository, self.new_inventory,
155
self._new_revision_id)
157
def finish_inventory(self):
158
"""Tell the builder that the inventory is finished."""
159
if self.new_inventory.root is None:
160
raise AssertionError('Root entry should be supplied to'
161
' record_entry_contents, as of bzr 0.10.')
162
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
163
self.new_inventory.revision_id = self._new_revision_id
164
self.inv_sha1 = self.repository.add_inventory(
165
self._new_revision_id,
170
def _gen_revision_id(self):
171
"""Return new revision-id."""
172
return generate_ids.gen_revision_id(self._config.username(),
175
def _generate_revision_if_needed(self):
176
"""Create a revision id if None was supplied.
178
If the repository can not support user-specified revision ids
179
they should override this function and raise CannotSetRevisionId
180
if _new_revision_id is not None.
182
:raises: CannotSetRevisionId
184
if self._new_revision_id is None:
185
self._new_revision_id = self._gen_revision_id()
186
self.random_revid = True
188
self.random_revid = False
190
def _heads(self, file_id, revision_ids):
191
"""Calculate the graph heads for revision_ids in the graph of file_id.
193
This can use either a per-file graph or a global revision graph as we
194
have an identity relationship between the two graphs.
196
return self.__heads(revision_ids)
198
def _check_root(self, ie, parent_invs, tree):
199
"""Helper for record_entry_contents.
201
:param ie: An entry being added.
202
:param parent_invs: The inventories of the parent revisions of the
204
:param tree: The tree that is being committed.
206
# In this revision format, root entries have no knit or weave When
207
# serializing out to disk and back in root.revision is always
209
ie.revision = self._new_revision_id
211
def _get_delta(self, ie, basis_inv, path):
212
"""Get a delta against the basis inventory for ie."""
213
if ie.file_id not in basis_inv:
215
return (None, path, ie.file_id, ie)
216
elif ie != basis_inv[ie.file_id]:
218
# TODO: avoid tis id2path call.
219
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
224
def record_entry_contents(self, ie, parent_invs, path, tree,
226
"""Record the content of ie from tree into the commit if needed.
228
Side effect: sets ie.revision when unchanged
230
:param ie: An inventory entry present in the commit.
231
:param parent_invs: The inventories of the parent revisions of the
233
:param path: The path the entry is at in the tree.
234
:param tree: The tree which contains this entry and should be used to
236
:param content_summary: Summary data from the tree about the paths
237
content - stat, length, exec, sha/link target. This is only
238
accessed when the entry has a revision of None - that is when it is
239
a candidate to commit.
240
:return: A tuple (change_delta, version_recorded, fs_hash).
241
change_delta is an inventory_delta change for this entry against
242
the basis tree of the commit, or None if no change occured against
244
version_recorded is True if a new version of the entry has been
245
recorded. For instance, committing a merge where a file was only
246
changed on the other side will return (delta, False).
247
fs_hash is either None, or the hash details for the path (currently
248
a tuple of the contents sha1 and the statvalue returned by
249
tree.get_file_with_stat()).
251
if self.new_inventory.root is None:
252
if ie.parent_id is not None:
253
raise errors.RootMissing()
254
self._check_root(ie, parent_invs, tree)
255
if ie.revision is None:
256
kind = content_summary[0]
258
# ie is carried over from a prior commit
260
# XXX: repository specific check for nested tree support goes here - if
261
# the repo doesn't want nested trees we skip it ?
262
if (kind == 'tree-reference' and
263
not self.repository._format.supports_tree_reference):
264
# mismatch between commit builder logic and repository:
265
# this needs the entry creation pushed down into the builder.
266
raise NotImplementedError('Missing repository subtree support.')
267
self.new_inventory.add(ie)
269
# TODO: slow, take it out of the inner loop.
271
basis_inv = parent_invs[0]
273
basis_inv = Inventory(root_id=None)
275
# ie.revision is always None if the InventoryEntry is considered
276
# for committing. We may record the previous parents revision if the
277
# content is actually unchanged against a sole head.
278
if ie.revision is not None:
279
if not self._versioned_root and path == '':
280
# repositories that do not version the root set the root's
281
# revision to the new commit even when no change occurs, and
282
# this masks when a change may have occurred against the basis,
283
# so calculate if one happened.
284
if ie.file_id in basis_inv:
285
delta = (basis_inv.id2path(ie.file_id), path,
289
delta = (None, path, ie.file_id, ie)
290
return delta, False, None
292
# we don't need to commit this, because the caller already
293
# determined that an existing revision of this file is
294
# appropriate. If its not being considered for committing then
295
# it and all its parents to the root must be unaltered so
296
# no-change against the basis.
297
if ie.revision == self._new_revision_id:
298
raise AssertionError("Impossible situation, a skipped "
299
"inventory entry (%r) claims to be modified in this "
300
"commit (%r).", (ie, self._new_revision_id))
301
return None, False, None
302
# XXX: Friction: parent_candidates should return a list not a dict
303
# so that we don't have to walk the inventories again.
304
parent_candiate_entries = ie.parent_candidates(parent_invs)
305
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
307
for inv in parent_invs:
308
if ie.file_id in inv:
309
old_rev = inv[ie.file_id].revision
310
if old_rev in head_set:
311
heads.append(inv[ie.file_id].revision)
312
head_set.remove(inv[ie.file_id].revision)
315
# now we check to see if we need to write a new record to the
317
# We write a new entry unless there is one head to the ancestors, and
318
# the kind-derived content is unchanged.
320
# Cheapest check first: no ancestors, or more the one head in the
321
# ancestors, we write a new node.
325
# There is a single head, look it up for comparison
326
parent_entry = parent_candiate_entries[heads[0]]
327
# if the non-content specific data has changed, we'll be writing a
329
if (parent_entry.parent_id != ie.parent_id or
330
parent_entry.name != ie.name):
332
# now we need to do content specific checks:
334
# if the kind changed the content obviously has
335
if kind != parent_entry.kind:
337
# Stat cache fingerprint feedback for the caller - None as we usually
338
# don't generate one.
341
if content_summary[2] is None:
342
raise ValueError("Files must not have executable = None")
344
if (# if the file length changed we have to store:
345
parent_entry.text_size != content_summary[1] or
346
# if the exec bit has changed we have to store:
347
parent_entry.executable != content_summary[2]):
349
elif parent_entry.text_sha1 == content_summary[3]:
350
# all meta and content is unchanged (using a hash cache
351
# hit to check the sha)
352
ie.revision = parent_entry.revision
353
ie.text_size = parent_entry.text_size
354
ie.text_sha1 = parent_entry.text_sha1
355
ie.executable = parent_entry.executable
356
return self._get_delta(ie, basis_inv, path), False, None
358
# Either there is only a hash change(no hash cache entry,
359
# or same size content change), or there is no change on
361
# Provide the parent's hash to the store layer, so that the
362
# content is unchanged we will not store a new node.
363
nostore_sha = parent_entry.text_sha1
365
# We want to record a new node regardless of the presence or
366
# absence of a content change in the file.
368
ie.executable = content_summary[2]
369
file_obj, stat_value = tree.get_file_with_stat(ie.file_id, path)
371
lines = file_obj.readlines()
375
ie.text_sha1, ie.text_size = self._add_text_to_weave(
376
ie.file_id, lines, heads, nostore_sha)
377
# Let the caller know we generated a stat fingerprint.
378
fingerprint = (ie.text_sha1, stat_value)
379
except errors.ExistingContent:
380
# Turns out that the file content was unchanged, and we were
381
# only going to store a new node if it was changed. Carry over
383
ie.revision = parent_entry.revision
384
ie.text_size = parent_entry.text_size
385
ie.text_sha1 = parent_entry.text_sha1
386
ie.executable = parent_entry.executable
387
return self._get_delta(ie, basis_inv, path), False, None
388
elif kind == 'directory':
390
# all data is meta here, nothing specific to directory, so
392
ie.revision = parent_entry.revision
393
return self._get_delta(ie, basis_inv, path), False, None
395
self._add_text_to_weave(ie.file_id, lines, heads, None)
396
elif kind == 'symlink':
397
current_link_target = content_summary[3]
399
# symlink target is not generic metadata, check if it has
401
if current_link_target != parent_entry.symlink_target:
404
# unchanged, carry over.
405
ie.revision = parent_entry.revision
406
ie.symlink_target = parent_entry.symlink_target
407
return self._get_delta(ie, basis_inv, path), False, None
408
ie.symlink_target = current_link_target
410
self._add_text_to_weave(ie.file_id, lines, heads, None)
411
elif kind == 'tree-reference':
413
if content_summary[3] != parent_entry.reference_revision:
416
# unchanged, carry over.
417
ie.reference_revision = parent_entry.reference_revision
418
ie.revision = parent_entry.revision
419
return self._get_delta(ie, basis_inv, path), False, None
420
ie.reference_revision = content_summary[3]
422
self._add_text_to_weave(ie.file_id, lines, heads, None)
424
raise NotImplementedError('unknown kind')
425
ie.revision = self._new_revision_id
426
return self._get_delta(ie, basis_inv, path), True, fingerprint
428
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
429
# Note: as we read the content directly from the tree, we know its not
430
# been turned into unicode or badly split - but a broken tree
431
# implementation could give us bad output from readlines() so this is
432
# not a guarantee of safety. What would be better is always checking
433
# the content during test suite execution. RBC 20070912
434
parent_keys = tuple((file_id, parent) for parent in parents)
435
return self.repository.texts.add_lines(
436
(file_id, self._new_revision_id), parent_keys, new_lines,
437
nostore_sha=nostore_sha, random_id=self.random_revid,
438
check_content=False)[0:2]
441
class RootCommitBuilder(CommitBuilder):
442
"""This commitbuilder actually records the root id"""
444
# the root entry gets versioned properly by this builder.
445
_versioned_root = True
447
def _check_root(self, ie, parent_invs, tree):
448
"""Helper for record_entry_contents.
450
:param ie: An entry being added.
451
:param parent_invs: The inventories of the parent revisions of the
453
:param tree: The tree that is being committed.
457
######################################################################
460
class Repository(object):
461
"""Repository holding history for one or more branches.
463
The repository holds and retrieves historical information including
464
revisions and file history. It's normally accessed only by the Branch,
465
which views a particular line of development through that history.
467
The Repository builds on top of some byte storage facilies (the revisions,
468
signatures, inventories and texts attributes) and a Transport, which
469
respectively provide byte storage and a means to access the (possibly
472
The byte storage facilities are addressed via tuples, which we refer to
473
as 'keys' throughout the code base. Revision_keys, inventory_keys and
474
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
475
(file_id, revision_id). We use this interface because it allows low
476
friction with the underlying code that implements disk indices, network
477
encoding and other parts of bzrlib.
479
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
480
the serialised revisions for the repository. This can be used to obtain
481
revision graph information or to access raw serialised revisions.
482
The result of trying to insert data into the repository via this store
483
is undefined: it should be considered read-only except for implementors
485
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
486
the serialised signatures for the repository. This can be used to
487
obtain access to raw serialised signatures. The result of trying to
488
insert data into the repository via this store is undefined: it should
489
be considered read-only except for implementors of repositories.
490
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
491
the serialised inventories for the repository. This can be used to
492
obtain unserialised inventories. The result of trying to insert data
493
into the repository via this store is undefined: it should be
494
considered read-only except for implementors of repositories.
495
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
496
texts of files and directories for the repository. This can be used to
497
obtain file texts or file graphs. Note that Repository.iter_file_bytes
498
is usually a better interface for accessing file texts.
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 _transport: Transport for file access to repository, typically
503
pointing to .bzr/repository.
506
# What class to use for a CommitBuilder. Often its simpler to change this
507
# in a Repository class subclass rather than to override
508
# get_commit_builder.
509
_commit_builder_class = CommitBuilder
510
# The search regex used by xml based repositories to determine what things
511
# where changed in a single commit.
512
_file_ids_altered_regex = lazy_regex.lazy_compile(
513
r'file_id="(?P<file_id>[^"]+)"'
514
r'.* revision="(?P<revision_id>[^"]+)"'
517
def abort_write_group(self, suppress_errors=False):
518
"""Commit the contents accrued within the current write group.
520
:seealso: start_write_group.
522
if self._write_group is not self.get_transaction():
523
# has an unlock or relock occured ?
524
raise errors.BzrError('mismatched lock context and write group.')
526
self._abort_write_group()
527
except Exception, exc:
528
self._write_group = None
529
if not suppress_errors:
531
mutter('abort_write_group failed')
532
log_exception_quietly()
533
note('bzr: ERROR (ignored): %s', exc)
534
self._write_group = None
536
def _abort_write_group(self):
537
"""Template method for per-repository write group cleanup.
539
This is called during abort before the write group is considered to be
540
finished and should cleanup any internal state accrued during the write
541
group. There is no requirement that data handed to the repository be
542
*not* made available - this is not a rollback - but neither should any
543
attempt be made to ensure that data added is fully commited. Abort is
544
invoked when an error has occured so futher disk or network operations
545
may not be possible or may error and if possible should not be
549
def add_fallback_repository(self, repository):
550
"""Add a repository to use for looking up data not held locally.
552
:param repository: A repository.
554
if not self._format.supports_external_lookups:
555
raise errors.UnstackableRepositoryFormat(self._format, self.base)
556
self._check_fallback_repository(repository)
557
self._fallback_repositories.append(repository)
558
self.texts.add_fallback_versioned_files(repository.texts)
559
self.inventories.add_fallback_versioned_files(repository.inventories)
560
self.revisions.add_fallback_versioned_files(repository.revisions)
561
self.signatures.add_fallback_versioned_files(repository.signatures)
563
def _check_fallback_repository(self, repository):
564
"""Check that this repository can fallback to repository safely.
566
Raise an error if not.
568
:param repository: A repository to fallback to.
570
return InterRepository._assert_same_model(self, repository)
572
def add_inventory(self, revision_id, inv, parents):
573
"""Add the inventory inv to the repository as revision_id.
575
:param parents: The revision ids of the parents that revision_id
576
is known to have and are in the repository already.
578
:returns: The validator(which is a sha1 digest, though what is sha'd is
579
repository format specific) of the serialized inventory.
581
if not self.is_in_write_group():
582
raise AssertionError("%r not in write group" % (self,))
583
_mod_revision.check_not_reserved_id(revision_id)
584
if not (inv.revision_id is None or inv.revision_id == revision_id):
585
raise AssertionError(
586
"Mismatch between inventory revision"
587
" id and insertion revid (%r, %r)"
588
% (inv.revision_id, revision_id))
590
raise AssertionError()
591
inv_lines = self._serialise_inventory_to_lines(inv)
592
return self._inventory_add_lines(revision_id, parents,
593
inv_lines, check_content=False)
595
def _inventory_add_lines(self, revision_id, parents, lines,
597
"""Store lines in inv_vf and return the sha1 of the inventory."""
598
parents = [(parent,) for parent in parents]
599
return self.inventories.add_lines((revision_id,), parents, lines,
600
check_content=check_content)[0]
602
def add_revision(self, revision_id, rev, inv=None, config=None):
603
"""Add rev to the revision store as revision_id.
605
:param revision_id: the revision id to use.
606
:param rev: The revision object.
607
:param inv: The inventory for the revision. if None, it will be looked
608
up in the inventory storer
609
:param config: If None no digital signature will be created.
610
If supplied its signature_needed method will be used
611
to determine if a signature should be made.
613
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
615
_mod_revision.check_not_reserved_id(revision_id)
616
if config is not None and config.signature_needed():
618
inv = self.get_inventory(revision_id)
619
plaintext = Testament(rev, inv).as_short_text()
620
self.store_revision_signature(
621
gpg.GPGStrategy(config), plaintext, revision_id)
622
# check inventory present
623
if not self.inventories.get_parent_map([(revision_id,)]):
625
raise errors.WeaveRevisionNotPresent(revision_id,
628
# yes, this is not suitable for adding with ghosts.
629
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
633
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
634
self._add_revision(rev)
636
def _add_revision(self, revision):
637
text = self._serializer.write_revision_to_string(revision)
638
key = (revision.revision_id,)
639
parents = tuple((parent,) for parent in revision.parent_ids)
640
self.revisions.add_lines(key, parents, osutils.split_lines(text))
642
def all_revision_ids(self):
643
"""Returns a list of all the revision ids in the repository.
645
This is conceptually deprecated because code should generally work on
646
the graph reachable from a particular revision, and ignore any other
647
revisions that might be present. There is no direct replacement
650
if 'evil' in debug.debug_flags:
651
mutter_callsite(2, "all_revision_ids is linear with history.")
652
return self._all_revision_ids()
654
def _all_revision_ids(self):
655
"""Returns a list of all the revision ids in the repository.
657
These are in as much topological order as the underlying store can
660
raise NotImplementedError(self._all_revision_ids)
662
def break_lock(self):
663
"""Break a lock if one is present from another instance.
665
Uses the ui factory to ask for confirmation if the lock may be from
668
self.control_files.break_lock()
671
def _eliminate_revisions_not_present(self, revision_ids):
672
"""Check every revision id in revision_ids to see if we have it.
674
Returns a set of the present revisions.
677
graph = self.get_graph()
678
parent_map = graph.get_parent_map(revision_ids)
679
# The old API returned a list, should this actually be a set?
680
return parent_map.keys()
683
def create(a_bzrdir):
684
"""Construct the current default format repository in a_bzrdir."""
685
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
687
def __init__(self, _format, a_bzrdir, control_files):
688
"""instantiate a Repository.
690
:param _format: The format of the repository on disk.
691
:param a_bzrdir: The BzrDir of the repository.
693
In the future we will have a single api for all stores for
694
getting file texts, inventories and revisions, then
695
this construct will accept instances of those things.
697
super(Repository, self).__init__()
698
self._format = _format
699
# the following are part of the public API for Repository:
700
self.bzrdir = a_bzrdir
701
self.control_files = control_files
702
self._transport = control_files._transport
703
self.base = self._transport.base
705
self._reconcile_does_inventory_gc = True
706
self._reconcile_fixes_text_parents = False
707
self._reconcile_backsup_inventory = True
708
# not right yet - should be more semantically clear ?
710
# TODO: make sure to construct the right store classes, etc, depending
711
# on whether escaping is required.
712
self._warn_if_deprecated()
713
self._write_group = None
714
# Additional places to query for data.
715
self._fallback_repositories = []
716
# What order should fetch operations request streams in?
717
# The default is unordered as that is the cheapest for an origin to
719
self._fetch_order = 'unordered'
720
# Does this repository use deltas that can be fetched as-deltas ?
721
# (E.g. knits, where the knit deltas can be transplanted intact.
722
# We default to False, which will ensure that enough data to get
723
# a full text out of any fetch stream will be grabbed.
724
self._fetch_uses_deltas = False
725
# Should fetch trigger a reconcile after the fetch? Only needed for
726
# some repository formats that can suffer internal inconsistencies.
727
self._fetch_reconcile = False
730
return '%s(%r)' % (self.__class__.__name__,
733
def has_same_location(self, other):
734
"""Returns a boolean indicating if this repository is at the same
735
location as another repository.
737
This might return False even when two repository objects are accessing
738
the same physical repository via different URLs.
740
if self.__class__ is not other.__class__:
742
return (self._transport.base == other._transport.base)
744
def is_in_write_group(self):
745
"""Return True if there is an open write group.
747
:seealso: start_write_group.
749
return self._write_group is not None
752
return self.control_files.is_locked()
754
def is_write_locked(self):
755
"""Return True if this object is write locked."""
756
return self.is_locked() and self.control_files._lock_mode == 'w'
758
def lock_write(self, token=None):
759
"""Lock this repository for writing.
761
This causes caching within the repository obejct to start accumlating
762
data during reads, and allows a 'write_group' to be obtained. Write
763
groups must be used for actual data insertion.
765
:param token: if this is already locked, then lock_write will fail
766
unless the token matches the existing lock.
767
:returns: a token if this instance supports tokens, otherwise None.
768
:raises TokenLockingNotSupported: when a token is given but this
769
instance doesn't support using token locks.
770
:raises MismatchedToken: if the specified token doesn't match the token
771
of the existing lock.
772
:seealso: start_write_group.
774
A token should be passed in if you know that you have locked the object
775
some other way, and need to synchronise this object's state with that
778
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
780
result = self.control_files.lock_write(token=token)
781
for repo in self._fallback_repositories:
782
# Writes don't affect fallback repos
788
self.control_files.lock_read()
789
for repo in self._fallback_repositories:
793
def get_physical_lock_status(self):
794
return self.control_files.get_physical_lock_status()
796
def leave_lock_in_place(self):
797
"""Tell this repository not to release the physical lock when this
800
If lock_write doesn't return a token, then this method is not supported.
802
self.control_files.leave_in_place()
804
def dont_leave_lock_in_place(self):
805
"""Tell this repository to release the physical lock when this
806
object is unlocked, even if it didn't originally acquire it.
808
If lock_write doesn't return a token, then this method is not supported.
810
self.control_files.dont_leave_in_place()
813
def gather_stats(self, revid=None, committers=None):
814
"""Gather statistics from a revision id.
816
:param revid: The revision id to gather statistics from, if None, then
817
no revision specific statistics are gathered.
818
:param committers: Optional parameter controlling whether to grab
819
a count of committers from the revision specific statistics.
820
:return: A dictionary of statistics. Currently this contains:
821
committers: The number of committers if requested.
822
firstrev: A tuple with timestamp, timezone for the penultimate left
823
most ancestor of revid, if revid is not the NULL_REVISION.
824
latestrev: A tuple with timestamp, timezone for revid, if revid is
825
not the NULL_REVISION.
826
revisions: The total revision count in the repository.
827
size: An estimate disk size of the repository in bytes.
830
if revid and committers:
831
result['committers'] = 0
832
if revid and revid != _mod_revision.NULL_REVISION:
834
all_committers = set()
835
revisions = self.get_ancestry(revid)
836
# pop the leading None
838
first_revision = None
840
# ignore the revisions in the middle - just grab first and last
841
revisions = revisions[0], revisions[-1]
842
for revision in self.get_revisions(revisions):
843
if not first_revision:
844
first_revision = revision
846
all_committers.add(revision.committer)
847
last_revision = revision
849
result['committers'] = len(all_committers)
850
result['firstrev'] = (first_revision.timestamp,
851
first_revision.timezone)
852
result['latestrev'] = (last_revision.timestamp,
853
last_revision.timezone)
855
# now gather global repository information
856
# XXX: This is available for many repos regardless of listability.
857
if self.bzrdir.root_transport.listable():
858
# XXX: do we want to __define len__() ?
859
# Maybe the versionedfiles object should provide a different
860
# method to get the number of keys.
861
result['revisions'] = len(self.revisions.keys())
865
def find_branches(self, using=False):
866
"""Find branches underneath this repository.
868
This will include branches inside other branches.
870
:param using: If True, list only branches using this repository.
872
if using and not self.is_shared():
874
return [self.bzrdir.open_branch()]
875
except errors.NotBranchError:
877
class Evaluator(object):
880
self.first_call = True
882
def __call__(self, bzrdir):
883
# On the first call, the parameter is always the bzrdir
884
# containing the current repo.
885
if not self.first_call:
887
repository = bzrdir.open_repository()
888
except errors.NoRepositoryPresent:
891
return False, (None, repository)
892
self.first_call = False
894
value = (bzrdir.open_branch(), None)
895
except errors.NotBranchError:
900
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
901
self.bzrdir.root_transport, evaluate=Evaluator()):
902
if branch is not None:
903
branches.append(branch)
904
if not using and repository is not None:
905
branches.extend(repository.find_branches())
909
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
910
"""Return the revision ids that other has that this does not.
912
These are returned in topological order.
914
revision_id: only return revision ids included by revision_id.
916
return InterRepository.get(other, self).search_missing_revision_ids(
917
revision_id, find_ghosts)
919
@deprecated_method(one_two)
921
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
922
"""Return the revision ids that other has that this does not.
924
These are returned in topological order.
926
revision_id: only return revision ids included by revision_id.
928
keys = self.search_missing_revision_ids(
929
other, revision_id, find_ghosts).get_keys()
932
parents = other.get_graph().get_parent_map(keys)
935
return tsort.topo_sort(parents)
939
"""Open the repository rooted at base.
941
For instance, if the repository is at URL/.bzr/repository,
942
Repository.open(URL) -> a Repository instance.
944
control = bzrdir.BzrDir.open(base)
945
return control.open_repository()
947
def copy_content_into(self, destination, revision_id=None):
948
"""Make a complete copy of the content in self into destination.
950
This is a destructive operation! Do not use it on existing
953
return InterRepository.get(self, destination).copy_content(revision_id)
955
def commit_write_group(self):
956
"""Commit the contents accrued within the current write group.
958
:seealso: start_write_group.
960
if self._write_group is not self.get_transaction():
961
# has an unlock or relock occured ?
962
raise errors.BzrError('mismatched lock context %r and '
964
(self.get_transaction(), self._write_group))
965
self._commit_write_group()
966
self._write_group = None
968
def _commit_write_group(self):
969
"""Template method for per-repository write group cleanup.
971
This is called before the write group is considered to be
972
finished and should ensure that all data handed to the repository
973
for writing during the write group is safely committed (to the
974
extent possible considering file system caching etc).
977
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
978
"""Fetch the content required to construct revision_id from source.
980
If revision_id is None all content is copied.
981
:param find_ghosts: Find and copy revisions in the source that are
982
ghosts in the target (and not reachable directly by walking out to
983
the first-present revision in target from revision_id).
985
# fast path same-url fetch operations
986
if self.has_same_location(source):
987
# check that last_revision is in 'from' and then return a
989
if (revision_id is not None and
990
not _mod_revision.is_null(revision_id)):
991
self.get_revision(revision_id)
993
# if there is no specific appropriate InterRepository, this will get
994
# the InterRepository base class, which raises an
995
# IncompatibleRepositories when asked to fetch.
996
inter = InterRepository.get(source, self)
997
return inter.fetch(revision_id=revision_id, pb=pb,
998
find_ghosts=find_ghosts)
1000
def create_bundle(self, target, base, fileobj, format=None):
1001
return serializer.write_bundle(self, target, base, fileobj, format)
1003
def get_commit_builder(self, branch, parents, config, timestamp=None,
1004
timezone=None, committer=None, revprops=None,
1006
"""Obtain a CommitBuilder for this repository.
1008
:param branch: Branch to commit to.
1009
:param parents: Revision ids of the parents of the new revision.
1010
:param config: Configuration to use.
1011
:param timestamp: Optional timestamp recorded for commit.
1012
:param timezone: Optional timezone for timestamp.
1013
:param committer: Optional committer to set for commit.
1014
:param revprops: Optional dictionary of revision properties.
1015
:param revision_id: Optional revision id.
1017
result = self._commit_builder_class(self, parents, config,
1018
timestamp, timezone, committer, revprops, revision_id)
1019
self.start_write_group()
1023
if (self.control_files._lock_count == 1 and
1024
self.control_files._lock_mode == 'w'):
1025
if self._write_group is not None:
1026
self.abort_write_group()
1027
self.control_files.unlock()
1028
raise errors.BzrError(
1029
'Must end write groups before releasing write locks.')
1030
self.control_files.unlock()
1031
for repo in self._fallback_repositories:
1035
def clone(self, a_bzrdir, revision_id=None):
1036
"""Clone this repository into a_bzrdir using the current format.
1038
Currently no check is made that the format of this repository and
1039
the bzrdir format are compatible. FIXME RBC 20060201.
1041
:return: The newly created destination repository.
1043
# TODO: deprecate after 0.16; cloning this with all its settings is
1044
# probably not very useful -- mbp 20070423
1045
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1046
self.copy_content_into(dest_repo, revision_id)
1049
def start_write_group(self):
1050
"""Start a write group in the repository.
1052
Write groups are used by repositories which do not have a 1:1 mapping
1053
between file ids and backend store to manage the insertion of data from
1054
both fetch and commit operations.
1056
A write lock is required around the start_write_group/commit_write_group
1057
for the support of lock-requiring repository formats.
1059
One can only insert data into a repository inside a write group.
1063
if not self.is_write_locked():
1064
raise errors.NotWriteLocked(self)
1065
if self._write_group:
1066
raise errors.BzrError('already in a write group')
1067
self._start_write_group()
1068
# so we can detect unlock/relock - the write group is now entered.
1069
self._write_group = self.get_transaction()
1071
def _start_write_group(self):
1072
"""Template method for per-repository write group startup.
1074
This is called before the write group is considered to be
1079
def sprout(self, to_bzrdir, revision_id=None):
1080
"""Create a descendent repository for new development.
1082
Unlike clone, this does not copy the settings of the repository.
1084
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1085
dest_repo.fetch(self, revision_id=revision_id)
1088
def _create_sprouting_repo(self, a_bzrdir, shared):
1089
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1090
# use target default format.
1091
dest_repo = a_bzrdir.create_repository()
1093
# Most control formats need the repository to be specifically
1094
# created, but on some old all-in-one formats it's not needed
1096
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1097
except errors.UninitializableFormat:
1098
dest_repo = a_bzrdir.open_repository()
1102
def has_revision(self, revision_id):
1103
"""True if this repository has a copy of the revision."""
1104
return revision_id in self.has_revisions((revision_id,))
1107
def has_revisions(self, revision_ids):
1108
"""Probe to find out the presence of multiple revisions.
1110
:param revision_ids: An iterable of revision_ids.
1111
:return: A set of the revision_ids that were present.
1113
parent_map = self.revisions.get_parent_map(
1114
[(rev_id,) for rev_id in revision_ids])
1116
if _mod_revision.NULL_REVISION in revision_ids:
1117
result.add(_mod_revision.NULL_REVISION)
1118
result.update([key[0] for key in parent_map])
1122
def get_revision(self, revision_id):
1123
"""Return the Revision object for a named revision."""
1124
return self.get_revisions([revision_id])[0]
1127
def get_revision_reconcile(self, revision_id):
1128
"""'reconcile' helper routine that allows access to a revision always.
1130
This variant of get_revision does not cross check the weave graph
1131
against the revision one as get_revision does: but it should only
1132
be used by reconcile, or reconcile-alike commands that are correcting
1133
or testing the revision graph.
1135
return self._get_revisions([revision_id])[0]
1138
def get_revisions(self, revision_ids):
1139
"""Get many revisions at once."""
1140
return self._get_revisions(revision_ids)
1143
def _get_revisions(self, revision_ids):
1144
"""Core work logic to get many revisions without sanity checks."""
1145
for rev_id in revision_ids:
1146
if not rev_id or not isinstance(rev_id, basestring):
1147
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1148
keys = [(key,) for key in revision_ids]
1149
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1151
for record in stream:
1152
if record.storage_kind == 'absent':
1153
raise errors.NoSuchRevision(self, record.key[0])
1154
text = record.get_bytes_as('fulltext')
1155
rev = self._serializer.read_revision_from_string(text)
1156
revs[record.key[0]] = rev
1157
return [revs[revid] for revid in revision_ids]
1160
def get_revision_xml(self, revision_id):
1161
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1162
# would have already do it.
1163
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1164
rev = self.get_revision(revision_id)
1165
rev_tmp = cStringIO.StringIO()
1166
# the current serializer..
1167
self._serializer.write_revision(rev, rev_tmp)
1169
return rev_tmp.getvalue()
1171
def get_deltas_for_revisions(self, revisions):
1172
"""Produce a generator of revision deltas.
1174
Note that the input is a sequence of REVISIONS, not revision_ids.
1175
Trees will be held in memory until the generator exits.
1176
Each delta is relative to the revision's lefthand predecessor.
1178
required_trees = set()
1179
for revision in revisions:
1180
required_trees.add(revision.revision_id)
1181
required_trees.update(revision.parent_ids[:1])
1182
trees = dict((t.get_revision_id(), t) for
1183
t in self.revision_trees(required_trees))
1184
for revision in revisions:
1185
if not revision.parent_ids:
1186
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
1188
old_tree = trees[revision.parent_ids[0]]
1189
yield trees[revision.revision_id].changes_from(old_tree)
1192
def get_revision_delta(self, revision_id):
1193
"""Return the delta for one revision.
1195
The delta is relative to the left-hand predecessor of the
1198
r = self.get_revision(revision_id)
1199
return list(self.get_deltas_for_revisions([r]))[0]
1202
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1203
signature = gpg_strategy.sign(plaintext)
1204
self.add_signature_text(revision_id, signature)
1207
def add_signature_text(self, revision_id, signature):
1208
self.signatures.add_lines((revision_id,), (),
1209
osutils.split_lines(signature))
1211
def find_text_key_references(self):
1212
"""Find the text key references within the repository.
1214
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1215
revision_ids. Each altered file-ids has the exact revision_ids that
1216
altered it listed explicitly.
1217
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1218
to whether they were referred to by the inventory of the
1219
revision_id that they contain. The inventory texts from all present
1220
revision ids are assessed to generate this report.
1222
revision_keys = self.revisions.keys()
1223
w = self.inventories
1224
pb = ui.ui_factory.nested_progress_bar()
1226
return self._find_text_key_references_from_xml_inventory_lines(
1227
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1231
def _find_text_key_references_from_xml_inventory_lines(self,
1233
"""Core routine for extracting references to texts from inventories.
1235
This performs the translation of xml lines to revision ids.
1237
:param line_iterator: An iterator of lines, origin_version_id
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. Note that if that revision_id was
1241
not part of the line_iterator's output then False will be given -
1242
even though it may actually refer to that key.
1244
if not self._serializer.support_altered_by_hack:
1245
raise AssertionError(
1246
"_find_text_key_references_from_xml_inventory_lines only "
1247
"supported for branches which store inventory as unnested xml"
1248
", not on %r" % self)
1251
# this code needs to read every new line in every inventory for the
1252
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1253
# not present in one of those inventories is unnecessary but not
1254
# harmful because we are filtering by the revision id marker in the
1255
# inventory lines : we only select file ids altered in one of those
1256
# revisions. We don't need to see all lines in the inventory because
1257
# only those added in an inventory in rev X can contain a revision=X
1259
unescape_revid_cache = {}
1260
unescape_fileid_cache = {}
1262
# jam 20061218 In a big fetch, this handles hundreds of thousands
1263
# of lines, so it has had a lot of inlining and optimizing done.
1264
# Sorry that it is a little bit messy.
1265
# Move several functions to be local variables, since this is a long
1267
search = self._file_ids_altered_regex.search
1268
unescape = _unescape_xml
1269
setdefault = result.setdefault
1270
for line, line_key in line_iterator:
1271
match = search(line)
1274
# One call to match.group() returning multiple items is quite a
1275
# bit faster than 2 calls to match.group() each returning 1
1276
file_id, revision_id = match.group('file_id', 'revision_id')
1278
# Inlining the cache lookups helps a lot when you make 170,000
1279
# lines and 350k ids, versus 8.4 unique ids.
1280
# Using a cache helps in 2 ways:
1281
# 1) Avoids unnecessary decoding calls
1282
# 2) Re-uses cached strings, which helps in future set and
1284
# (2) is enough that removing encoding entirely along with
1285
# the cache (so we are using plain strings) results in no
1286
# performance improvement.
1288
revision_id = unescape_revid_cache[revision_id]
1290
unescaped = unescape(revision_id)
1291
unescape_revid_cache[revision_id] = unescaped
1292
revision_id = unescaped
1294
# Note that unconditionally unescaping means that we deserialise
1295
# every fileid, which for general 'pull' is not great, but we don't
1296
# really want to have some many fulltexts that this matters anyway.
1299
file_id = unescape_fileid_cache[file_id]
1301
unescaped = unescape(file_id)
1302
unescape_fileid_cache[file_id] = unescaped
1305
key = (file_id, revision_id)
1306
setdefault(key, False)
1307
if revision_id == line_key[-1]:
1311
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1313
"""Helper routine for fileids_altered_by_revision_ids.
1315
This performs the translation of xml lines to revision ids.
1317
:param line_iterator: An iterator of lines, origin_version_id
1318
:param revision_ids: The revision ids to filter for. This should be a
1319
set or other type which supports efficient __contains__ lookups, as
1320
the revision id from each parsed line will be looked up in the
1321
revision_ids filter.
1322
:return: a dictionary mapping altered file-ids to an iterable of
1323
revision_ids. Each altered file-ids has the exact revision_ids that
1324
altered it listed explicitly.
1327
setdefault = result.setdefault
1329
self._find_text_key_references_from_xml_inventory_lines(
1330
line_iterator).iterkeys():
1331
# once data is all ensured-consistent; then this is
1332
# if revision_id == version_id
1333
if key[-1:] in revision_ids:
1334
setdefault(key[0], set()).add(key[-1])
1337
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1338
"""Find the file ids and versions affected by revisions.
1340
:param revisions: an iterable containing revision ids.
1341
:param _inv_weave: The inventory weave from this repository or None.
1342
If None, the inventory weave will be opened automatically.
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.
1347
selected_keys = set((revid,) for revid in revision_ids)
1348
w = _inv_weave or self.inventories
1349
pb = ui.ui_factory.nested_progress_bar()
1351
return self._find_file_ids_from_xml_inventory_lines(
1352
w.iter_lines_added_or_present_in_keys(
1353
selected_keys, pb=pb),
1358
def iter_files_bytes(self, desired_files):
1359
"""Iterate through file versions.
1361
Files will not necessarily be returned in the order they occur in
1362
desired_files. No specific order is guaranteed.
1364
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1365
value supplied by the caller as part of desired_files. It should
1366
uniquely identify the file version in the caller's context. (Examples:
1367
an index number or a TreeTransform trans_id.)
1369
bytes_iterator is an iterable of bytestrings for the file. The
1370
kind of iterable and length of the bytestrings are unspecified, but for
1371
this implementation, it is a list of bytes produced by
1372
VersionedFile.get_record_stream().
1374
:param desired_files: a list of (file_id, revision_id, identifier)
1377
transaction = self.get_transaction()
1379
for file_id, revision_id, callable_data in desired_files:
1380
text_keys[(file_id, revision_id)] = callable_data
1381
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1382
if record.storage_kind == 'absent':
1383
raise errors.RevisionNotPresent(record.key, self)
1384
yield text_keys[record.key], record.get_bytes_as('fulltext')
1386
def _generate_text_key_index(self, text_key_references=None,
1388
"""Generate a new text key index for the repository.
1390
This is an expensive function that will take considerable time to run.
1392
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1393
list of parents, also text keys. When a given key has no parents,
1394
the parents list will be [NULL_REVISION].
1396
# All revisions, to find inventory parents.
1397
if ancestors is None:
1398
graph = self.get_graph()
1399
ancestors = graph.get_parent_map(self.all_revision_ids())
1400
if text_key_references is None:
1401
text_key_references = self.find_text_key_references()
1402
pb = ui.ui_factory.nested_progress_bar()
1404
return self._do_generate_text_key_index(ancestors,
1405
text_key_references, pb)
1409
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1410
"""Helper for _generate_text_key_index to avoid deep nesting."""
1411
revision_order = tsort.topo_sort(ancestors)
1412
invalid_keys = set()
1414
for revision_id in revision_order:
1415
revision_keys[revision_id] = set()
1416
text_count = len(text_key_references)
1417
# a cache of the text keys to allow reuse; costs a dict of all the
1418
# keys, but saves a 2-tuple for every child of a given key.
1420
for text_key, valid in text_key_references.iteritems():
1422
invalid_keys.add(text_key)
1424
revision_keys[text_key[1]].add(text_key)
1425
text_key_cache[text_key] = text_key
1426
del text_key_references
1428
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1429
NULL_REVISION = _mod_revision.NULL_REVISION
1430
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1431
# too small for large or very branchy trees. However, for 55K path
1432
# trees, it would be easy to use too much memory trivially. Ideally we
1433
# could gauge this by looking at available real memory etc, but this is
1434
# always a tricky proposition.
1435
inventory_cache = lru_cache.LRUCache(10)
1436
batch_size = 10 # should be ~150MB on a 55K path tree
1437
batch_count = len(revision_order) / batch_size + 1
1439
pb.update("Calculating text parents.", processed_texts, text_count)
1440
for offset in xrange(batch_count):
1441
to_query = revision_order[offset * batch_size:(offset + 1) *
1445
for rev_tree in self.revision_trees(to_query):
1446
revision_id = rev_tree.get_revision_id()
1447
parent_ids = ancestors[revision_id]
1448
for text_key in revision_keys[revision_id]:
1449
pb.update("Calculating text parents.", processed_texts)
1450
processed_texts += 1
1451
candidate_parents = []
1452
for parent_id in parent_ids:
1453
parent_text_key = (text_key[0], parent_id)
1455
check_parent = parent_text_key not in \
1456
revision_keys[parent_id]
1458
# the parent parent_id is a ghost:
1459
check_parent = False
1460
# truncate the derived graph against this ghost.
1461
parent_text_key = None
1463
# look at the parent commit details inventories to
1464
# determine possible candidates in the per file graph.
1467
inv = inventory_cache[parent_id]
1469
inv = self.revision_tree(parent_id).inventory
1470
inventory_cache[parent_id] = inv
1471
parent_entry = inv._byid.get(text_key[0], None)
1472
if parent_entry is not None:
1474
text_key[0], parent_entry.revision)
1476
parent_text_key = None
1477
if parent_text_key is not None:
1478
candidate_parents.append(
1479
text_key_cache[parent_text_key])
1480
parent_heads = text_graph.heads(candidate_parents)
1481
new_parents = list(parent_heads)
1482
new_parents.sort(key=lambda x:candidate_parents.index(x))
1483
if new_parents == []:
1484
new_parents = [NULL_REVISION]
1485
text_index[text_key] = new_parents
1487
for text_key in invalid_keys:
1488
text_index[text_key] = [NULL_REVISION]
1491
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1492
"""Get an iterable listing the keys of all the data introduced by a set
1495
The keys will be ordered so that the corresponding items can be safely
1496
fetched and inserted in that order.
1498
:returns: An iterable producing tuples of (knit-kind, file-id,
1499
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1500
'revisions'. file-id is None unless knit-kind is 'file'.
1502
# XXX: it's a bit weird to control the inventory weave caching in this
1503
# generator. Ideally the caching would be done in fetch.py I think. Or
1504
# maybe this generator should explicitly have the contract that it
1505
# should not be iterated until the previously yielded item has been
1507
inv_w = self.inventories
1509
# file ids that changed
1510
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1512
num_file_ids = len(file_ids)
1513
for file_id, altered_versions in file_ids.iteritems():
1514
if _files_pb is not None:
1515
_files_pb.update("fetch texts", count, num_file_ids)
1517
yield ("file", file_id, altered_versions)
1518
# We're done with the files_pb. Note that it finished by the caller,
1519
# just as it was created by the caller.
1523
yield ("inventory", None, revision_ids)
1526
revisions_with_signatures = set()
1527
for rev_id in revision_ids:
1529
self.get_signature_text(rev_id)
1530
except errors.NoSuchRevision:
1534
revisions_with_signatures.add(rev_id)
1535
yield ("signatures", None, revisions_with_signatures)
1538
yield ("revisions", None, revision_ids)
1541
def get_inventory(self, revision_id):
1542
"""Get Inventory object by revision id."""
1543
return self.iter_inventories([revision_id]).next()
1545
def iter_inventories(self, revision_ids):
1546
"""Get many inventories by revision_ids.
1548
This will buffer some or all of the texts used in constructing the
1549
inventories in memory, but will only parse a single inventory at a
1552
:return: An iterator of inventories.
1554
if ((None in revision_ids)
1555
or (_mod_revision.NULL_REVISION in revision_ids)):
1556
raise ValueError('cannot get null revision inventory')
1557
return self._iter_inventories(revision_ids)
1559
def _iter_inventories(self, revision_ids):
1560
"""single-document based inventory iteration."""
1561
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1562
yield self.deserialise_inventory(revision_id, text)
1564
def _iter_inventory_xmls(self, revision_ids):
1565
keys = [(revision_id,) for revision_id in revision_ids]
1566
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1568
for record in stream:
1569
if record.storage_kind != 'absent':
1570
texts[record.key] = record.get_bytes_as('fulltext')
1572
raise errors.NoSuchRevision(self, record.key)
1574
yield texts[key], key[-1]
1576
def deserialise_inventory(self, revision_id, xml):
1577
"""Transform the xml into an inventory object.
1579
:param revision_id: The expected revision id of the inventory.
1580
:param xml: A serialised inventory.
1582
result = self._serializer.read_inventory_from_string(xml, revision_id)
1583
if result.revision_id != revision_id:
1584
raise AssertionError('revision id mismatch %s != %s' % (
1585
result.revision_id, revision_id))
1588
def serialise_inventory(self, inv):
1589
return self._serializer.write_inventory_to_string(inv)
1591
def _serialise_inventory_to_lines(self, inv):
1592
return self._serializer.write_inventory_to_lines(inv)
1594
def get_serializer_format(self):
1595
return self._serializer.format_num
1598
def get_inventory_xml(self, revision_id):
1599
"""Get inventory XML as a file object."""
1600
texts = self._iter_inventory_xmls([revision_id])
1602
text, revision_id = texts.next()
1603
except StopIteration:
1604
raise errors.HistoryMissing(self, 'inventory', revision_id)
1608
def get_inventory_sha1(self, revision_id):
1609
"""Return the sha1 hash of the inventory entry
1611
return self.get_revision(revision_id).inventory_sha1
1613
def iter_reverse_revision_history(self, revision_id):
1614
"""Iterate backwards through revision ids in the lefthand history
1616
:param revision_id: The revision id to start with. All its lefthand
1617
ancestors will be traversed.
1619
graph = self.get_graph()
1620
next_id = revision_id
1622
if next_id in (None, _mod_revision.NULL_REVISION):
1625
# Note: The following line may raise KeyError in the event of
1626
# truncated history. We decided not to have a try:except:raise
1627
# RevisionNotPresent here until we see a use for it, because of the
1628
# cost in an inner loop that is by its very nature O(history).
1629
# Robert Collins 20080326
1630
parents = graph.get_parent_map([next_id])[next_id]
1631
if len(parents) == 0:
1634
next_id = parents[0]
1637
def get_revision_inventory(self, revision_id):
1638
"""Return inventory of a past revision."""
1639
# TODO: Unify this with get_inventory()
1640
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1641
# must be the same as its revision, so this is trivial.
1642
if revision_id is None:
1643
# This does not make sense: if there is no revision,
1644
# then it is the current tree inventory surely ?!
1645
# and thus get_root_id() is something that looks at the last
1646
# commit on the branch, and the get_root_id is an inventory check.
1647
raise NotImplementedError
1648
# return Inventory(self.get_root_id())
1650
return self.get_inventory(revision_id)
1652
def is_shared(self):
1653
"""Return True if this repository is flagged as a shared repository."""
1654
raise NotImplementedError(self.is_shared)
1657
def reconcile(self, other=None, thorough=False):
1658
"""Reconcile this repository."""
1659
from bzrlib.reconcile import RepoReconciler
1660
reconciler = RepoReconciler(self, thorough=thorough)
1661
reconciler.reconcile()
1664
def _refresh_data(self):
1665
"""Helper called from lock_* to ensure coherency with disk.
1667
The default implementation does nothing; it is however possible
1668
for repositories to maintain loaded indices across multiple locks
1669
by checking inside their implementation of this method to see
1670
whether their indices are still valid. This depends of course on
1671
the disk format being validatable in this manner.
1675
def revision_tree(self, revision_id):
1676
"""Return Tree for a revision on this branch.
1678
`revision_id` may be NULL_REVISION for the empty tree revision.
1680
revision_id = _mod_revision.ensure_null(revision_id)
1681
# TODO: refactor this to use an existing revision object
1682
# so we don't need to read it in twice.
1683
if revision_id == _mod_revision.NULL_REVISION:
1684
return RevisionTree(self, Inventory(root_id=None),
1685
_mod_revision.NULL_REVISION)
1687
inv = self.get_revision_inventory(revision_id)
1688
return RevisionTree(self, inv, revision_id)
1690
def revision_trees(self, revision_ids):
1691
"""Return Tree for a revision on this branch.
1693
`revision_id` may not be None or 'null:'"""
1694
inventories = self.iter_inventories(revision_ids)
1695
for inv in inventories:
1696
yield RevisionTree(self, inv, inv.revision_id)
1699
def get_ancestry(self, revision_id, topo_sorted=True):
1700
"""Return a list of revision-ids integrated by a revision.
1702
The first element of the list is always None, indicating the origin
1703
revision. This might change when we have history horizons, or
1704
perhaps we should have a new API.
1706
This is topologically sorted.
1708
if _mod_revision.is_null(revision_id):
1710
if not self.has_revision(revision_id):
1711
raise errors.NoSuchRevision(self, revision_id)
1712
graph = self.get_graph()
1714
search = graph._make_breadth_first_searcher([revision_id])
1717
found, ghosts = search.next_with_ghosts()
1718
except StopIteration:
1721
if _mod_revision.NULL_REVISION in keys:
1722
keys.remove(_mod_revision.NULL_REVISION)
1724
parent_map = graph.get_parent_map(keys)
1725
keys = tsort.topo_sort(parent_map)
1726
return [None] + list(keys)
1729
"""Compress the data within the repository.
1731
This operation only makes sense for some repository types. For other
1732
types it should be a no-op that just returns.
1734
This stub method does not require a lock, but subclasses should use
1735
@needs_write_lock as this is a long running call its reasonable to
1736
implicitly lock for the user.
1740
@deprecated_method(one_six)
1741
def print_file(self, file, revision_id):
1742
"""Print `file` to stdout.
1744
FIXME RBC 20060125 as John Meinel points out this is a bad api
1745
- it writes to stdout, it assumes that that is valid etc. Fix
1746
by creating a new more flexible convenience function.
1748
tree = self.revision_tree(revision_id)
1749
# use inventory as it was in that revision
1750
file_id = tree.inventory.path2id(file)
1752
# TODO: jam 20060427 Write a test for this code path
1753
# it had a bug in it, and was raising the wrong
1755
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1756
tree.print_file(file_id)
1758
def get_transaction(self):
1759
return self.control_files.get_transaction()
1761
@deprecated_method(one_one)
1762
def get_parents(self, revision_ids):
1763
"""See StackedParentsProvider.get_parents"""
1764
parent_map = self.get_parent_map(revision_ids)
1765
return [parent_map.get(r, None) for r in revision_ids]
1767
def get_parent_map(self, revision_ids):
1768
"""See graph._StackedParentsProvider.get_parent_map"""
1769
# revisions index works in keys; this just works in revisions
1770
# therefore wrap and unwrap
1773
for revision_id in revision_ids:
1774
if revision_id == _mod_revision.NULL_REVISION:
1775
result[revision_id] = ()
1776
elif revision_id is None:
1777
raise ValueError('get_parent_map(None) is not valid')
1779
query_keys.append((revision_id ,))
1780
for ((revision_id,), parent_keys) in \
1781
self.revisions.get_parent_map(query_keys).iteritems():
1783
result[revision_id] = tuple(parent_revid
1784
for (parent_revid,) in parent_keys)
1786
result[revision_id] = (_mod_revision.NULL_REVISION,)
1789
def _make_parents_provider(self):
1792
def get_graph(self, other_repository=None):
1793
"""Return the graph walker for this repository format"""
1794
parents_provider = self._make_parents_provider()
1795
if (other_repository is not None and
1796
not self.has_same_location(other_repository)):
1797
parents_provider = graph._StackedParentsProvider(
1798
[parents_provider, other_repository._make_parents_provider()])
1799
return graph.Graph(parents_provider)
1801
def _get_versioned_file_checker(self):
1802
"""Return an object suitable for checking versioned files."""
1803
return _VersionedFileChecker(self)
1805
def revision_ids_to_search_result(self, result_set):
1806
"""Convert a set of revision ids to a graph SearchResult."""
1807
result_parents = set()
1808
for parents in self.get_graph().get_parent_map(
1809
result_set).itervalues():
1810
result_parents.update(parents)
1811
included_keys = result_set.intersection(result_parents)
1812
start_keys = result_set.difference(included_keys)
1813
exclude_keys = result_parents.difference(result_set)
1814
result = graph.SearchResult(start_keys, exclude_keys,
1815
len(result_set), result_set)
1819
def set_make_working_trees(self, new_value):
1820
"""Set the policy flag for making working trees when creating branches.
1822
This only applies to branches that use this repository.
1824
The default is 'True'.
1825
:param new_value: True to restore the default, False to disable making
1828
raise NotImplementedError(self.set_make_working_trees)
1830
def make_working_trees(self):
1831
"""Returns the policy for making working trees on new branches."""
1832
raise NotImplementedError(self.make_working_trees)
1835
def sign_revision(self, revision_id, gpg_strategy):
1836
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1837
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1840
def has_signature_for_revision_id(self, revision_id):
1841
"""Query for a revision signature for revision_id in the repository."""
1842
if not self.has_revision(revision_id):
1843
raise errors.NoSuchRevision(self, revision_id)
1844
sig_present = (1 == len(
1845
self.signatures.get_parent_map([(revision_id,)])))
1849
def get_signature_text(self, revision_id):
1850
"""Return the text for a signature."""
1851
stream = self.signatures.get_record_stream([(revision_id,)],
1853
record = stream.next()
1854
if record.storage_kind == 'absent':
1855
raise errors.NoSuchRevision(self, revision_id)
1856
return record.get_bytes_as('fulltext')
1859
def check(self, revision_ids=None):
1860
"""Check consistency of all history of given revision_ids.
1862
Different repository implementations should override _check().
1864
:param revision_ids: A non-empty list of revision_ids whose ancestry
1865
will be checked. Typically the last revision_id of a branch.
1867
return self._check(revision_ids)
1869
def _check(self, revision_ids):
1870
result = check.Check(self)
1874
def _warn_if_deprecated(self):
1875
global _deprecation_warning_done
1876
if _deprecation_warning_done:
1878
_deprecation_warning_done = True
1879
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1880
% (self._format, self.bzrdir.transport.base))
1882
def supports_rich_root(self):
1883
return self._format.rich_root_data
1885
def _check_ascii_revisionid(self, revision_id, method):
1886
"""Private helper for ascii-only repositories."""
1887
# weave repositories refuse to store revisionids that are non-ascii.
1888
if revision_id is not None:
1889
# weaves require ascii revision ids.
1890
if isinstance(revision_id, unicode):
1892
revision_id.encode('ascii')
1893
except UnicodeEncodeError:
1894
raise errors.NonAsciiRevisionId(method, self)
1897
revision_id.decode('ascii')
1898
except UnicodeDecodeError:
1899
raise errors.NonAsciiRevisionId(method, self)
1901
def revision_graph_can_have_wrong_parents(self):
1902
"""Is it possible for this repository to have a revision graph with
1905
If True, then this repository must also implement
1906
_find_inconsistent_revision_parents so that check and reconcile can
1907
check for inconsistencies before proceeding with other checks that may
1908
depend on the revision index being consistent.
1910
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1913
# remove these delegates a while after bzr 0.15
1914
def __make_delegated(name, from_module):
1915
def _deprecated_repository_forwarder():
1916
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1917
% (name, from_module),
1920
m = __import__(from_module, globals(), locals(), [name])
1922
return getattr(m, name)
1923
except AttributeError:
1924
raise AttributeError('module %s has no name %s'
1926
globals()[name] = _deprecated_repository_forwarder
1929
'AllInOneRepository',
1930
'WeaveMetaDirRepository',
1931
'PreSplitOutRepositoryFormat',
1932
'RepositoryFormat4',
1933
'RepositoryFormat5',
1934
'RepositoryFormat6',
1935
'RepositoryFormat7',
1937
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1941
'RepositoryFormatKnit',
1942
'RepositoryFormatKnit1',
1944
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1947
def install_revision(repository, rev, revision_tree):
1948
"""Install all revision data into a repository."""
1949
install_revisions(repository, [(rev, revision_tree, None)])
1952
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1953
"""Install all revision data into a repository.
1955
Accepts an iterable of revision, tree, signature tuples. The signature
1958
repository.start_write_group()
1960
for n, (revision, revision_tree, signature) in enumerate(iterable):
1961
_install_revision(repository, revision, revision_tree, signature)
1963
pb.update('Transferring revisions', n + 1, num_revisions)
1965
repository.abort_write_group()
1968
repository.commit_write_group()
1971
def _install_revision(repository, rev, revision_tree, signature):
1972
"""Install all revision data into a repository."""
1973
present_parents = []
1975
for p_id in rev.parent_ids:
1976
if repository.has_revision(p_id):
1977
present_parents.append(p_id)
1978
parent_trees[p_id] = repository.revision_tree(p_id)
1980
parent_trees[p_id] = repository.revision_tree(
1981
_mod_revision.NULL_REVISION)
1983
inv = revision_tree.inventory
1984
entries = inv.iter_entries()
1985
# backwards compatibility hack: skip the root id.
1986
if not repository.supports_rich_root():
1987
path, root = entries.next()
1988
if root.revision != rev.revision_id:
1989
raise errors.IncompatibleRevision(repr(repository))
1991
for path, ie in entries:
1992
text_keys[(ie.file_id, ie.revision)] = ie
1993
text_parent_map = repository.texts.get_parent_map(text_keys)
1994
missing_texts = set(text_keys) - set(text_parent_map)
1995
# Add the texts that are not already present
1996
for text_key in missing_texts:
1997
ie = text_keys[text_key]
1999
# FIXME: TODO: The following loop overlaps/duplicates that done by
2000
# commit to determine parents. There is a latent/real bug here where
2001
# the parents inserted are not those commit would do - in particular
2002
# they are not filtered by heads(). RBC, AB
2003
for revision, tree in parent_trees.iteritems():
2004
if ie.file_id not in tree:
2006
parent_id = tree.inventory[ie.file_id].revision
2007
if parent_id in text_parents:
2009
text_parents.append((ie.file_id, parent_id))
2010
lines = revision_tree.get_file(ie.file_id).readlines()
2011
repository.texts.add_lines(text_key, text_parents, lines)
2013
# install the inventory
2014
repository.add_inventory(rev.revision_id, inv, present_parents)
2015
except errors.RevisionAlreadyPresent:
2017
if signature is not None:
2018
repository.add_signature_text(rev.revision_id, signature)
2019
repository.add_revision(rev.revision_id, rev, inv)
2022
class MetaDirRepository(Repository):
2023
"""Repositories in the new meta-dir layout.
2025
:ivar _transport: Transport for access to repository control files,
2026
typically pointing to .bzr/repository.
2029
def __init__(self, _format, a_bzrdir, control_files):
2030
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2031
self._transport = control_files._transport
2033
def is_shared(self):
2034
"""Return True if this repository is flagged as a shared repository."""
2035
return self._transport.has('shared-storage')
2038
def set_make_working_trees(self, new_value):
2039
"""Set the policy flag for making working trees when creating branches.
2041
This only applies to branches that use this repository.
2043
The default is 'True'.
2044
:param new_value: True to restore the default, False to disable making
2049
self._transport.delete('no-working-trees')
2050
except errors.NoSuchFile:
2053
self._transport.put_bytes('no-working-trees', '',
2054
mode=self.bzrdir._get_file_mode())
2056
def make_working_trees(self):
2057
"""Returns the policy for making working trees on new branches."""
2058
return not self._transport.has('no-working-trees')
2061
class MetaDirVersionedFileRepository(MetaDirRepository):
2062
"""Repositories in a meta-dir, that work via versioned file objects."""
2064
def __init__(self, _format, a_bzrdir, control_files):
2065
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2069
class RepositoryFormatRegistry(registry.Registry):
2070
"""Registry of RepositoryFormats."""
2072
def get(self, format_string):
2073
r = registry.Registry.get(self, format_string)
2079
format_registry = RepositoryFormatRegistry()
2080
"""Registry of formats, indexed by their identifying format string.
2082
This can contain either format instances themselves, or classes/factories that
2083
can be called to obtain one.
2087
#####################################################################
2088
# Repository Formats
2090
class RepositoryFormat(object):
2091
"""A repository format.
2093
Formats provide three things:
2094
* An initialization routine to construct repository data on disk.
2095
* a format string which is used when the BzrDir supports versioned
2097
* an open routine which returns a Repository instance.
2099
There is one and only one Format subclass for each on-disk format. But
2100
there can be one Repository subclass that is used for several different
2101
formats. The _format attribute on a Repository instance can be used to
2102
determine the disk format.
2104
Formats are placed in an dict by their format string for reference
2105
during opening. These should be subclasses of RepositoryFormat
2108
Once a format is deprecated, just deprecate the initialize and open
2109
methods on the format class. Do not deprecate the object, as the
2110
object will be created every system load.
2112
Common instance attributes:
2113
_matchingbzrdir - the bzrdir format that the repository format was
2114
originally written to work with. This can be used if manually
2115
constructing a bzrdir and repository, or more commonly for test suite
2119
# Set to True or False in derived classes. True indicates that the format
2120
# supports ghosts gracefully.
2121
supports_ghosts = None
2122
# Can this repository be given external locations to lookup additional
2123
# data. Set to True or False in derived classes.
2124
supports_external_lookups = None
2127
return "<%s>" % self.__class__.__name__
2129
def __eq__(self, other):
2130
# format objects are generally stateless
2131
return isinstance(other, self.__class__)
2133
def __ne__(self, other):
2134
return not self == other
2137
def find_format(klass, a_bzrdir):
2138
"""Return the format for the repository object in a_bzrdir.
2140
This is used by bzr native formats that have a "format" file in
2141
the repository. Other methods may be used by different types of
2145
transport = a_bzrdir.get_repository_transport(None)
2146
format_string = transport.get("format").read()
2147
return format_registry.get(format_string)
2148
except errors.NoSuchFile:
2149
raise errors.NoRepositoryPresent(a_bzrdir)
2151
raise errors.UnknownFormatError(format=format_string,
2155
def register_format(klass, format):
2156
format_registry.register(format.get_format_string(), format)
2159
def unregister_format(klass, format):
2160
format_registry.remove(format.get_format_string())
2163
def get_default_format(klass):
2164
"""Return the current default format."""
2165
from bzrlib import bzrdir
2166
return bzrdir.format_registry.make_bzrdir('default').repository_format
2168
def get_format_string(self):
2169
"""Return the ASCII format string that identifies this format.
2171
Note that in pre format ?? repositories the format string is
2172
not permitted nor written to disk.
2174
raise NotImplementedError(self.get_format_string)
2176
def get_format_description(self):
2177
"""Return the short description for this format."""
2178
raise NotImplementedError(self.get_format_description)
2180
# TODO: this shouldn't be in the base class, it's specific to things that
2181
# use weaves or knits -- mbp 20070207
2182
def _get_versioned_file_store(self,
2187
versionedfile_class=None,
2188
versionedfile_kwargs={},
2190
if versionedfile_class is None:
2191
versionedfile_class = self._versionedfile_class
2192
weave_transport = control_files._transport.clone(name)
2193
dir_mode = control_files._dir_mode
2194
file_mode = control_files._file_mode
2195
return VersionedFileStore(weave_transport, prefixed=prefixed,
2197
file_mode=file_mode,
2198
versionedfile_class=versionedfile_class,
2199
versionedfile_kwargs=versionedfile_kwargs,
2202
def initialize(self, a_bzrdir, shared=False):
2203
"""Initialize a repository of this format in a_bzrdir.
2205
:param a_bzrdir: The bzrdir to put the new repository in it.
2206
:param shared: The repository should be initialized as a sharable one.
2207
:returns: The new repository object.
2209
This may raise UninitializableFormat if shared repository are not
2210
compatible the a_bzrdir.
2212
raise NotImplementedError(self.initialize)
2214
def is_supported(self):
2215
"""Is this format supported?
2217
Supported formats must be initializable and openable.
2218
Unsupported formats may not support initialization or committing or
2219
some other features depending on the reason for not being supported.
2223
def check_conversion_target(self, target_format):
2224
raise NotImplementedError(self.check_conversion_target)
2226
def open(self, a_bzrdir, _found=False):
2227
"""Return an instance of this format for the bzrdir a_bzrdir.
2229
_found is a private parameter, do not use it.
2231
raise NotImplementedError(self.open)
2234
class MetaDirRepositoryFormat(RepositoryFormat):
2235
"""Common base class for the new repositories using the metadir layout."""
2237
rich_root_data = False
2238
supports_tree_reference = False
2239
supports_external_lookups = False
2240
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2243
super(MetaDirRepositoryFormat, self).__init__()
2245
def _create_control_files(self, a_bzrdir):
2246
"""Create the required files and the initial control_files object."""
2247
# FIXME: RBC 20060125 don't peek under the covers
2248
# NB: no need to escape relative paths that are url safe.
2249
repository_transport = a_bzrdir.get_repository_transport(self)
2250
control_files = lockable_files.LockableFiles(repository_transport,
2251
'lock', lockdir.LockDir)
2252
control_files.create_lock()
2253
return control_files
2255
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
2256
"""Upload the initial blank content."""
2257
control_files = self._create_control_files(a_bzrdir)
2258
control_files.lock_write()
2259
transport = control_files._transport
2261
utf8_files += [('shared-storage', '')]
2263
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2264
for (filename, content_stream) in files:
2265
transport.put_file(filename, content_stream,
2266
mode=a_bzrdir._get_file_mode())
2267
for (filename, content_bytes) in utf8_files:
2268
transport.put_bytes_non_atomic(filename, content_bytes,
2269
mode=a_bzrdir._get_file_mode())
2271
control_files.unlock()
2274
# formats which have no format string are not discoverable
2275
# and not independently creatable, so are not registered. They're
2276
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2277
# needed, it's constructed directly by the BzrDir. Non-native formats where
2278
# the repository is not separately opened are similar.
2280
format_registry.register_lazy(
2281
'Bazaar-NG Repository format 7',
2282
'bzrlib.repofmt.weaverepo',
2286
format_registry.register_lazy(
2287
'Bazaar-NG Knit Repository Format 1',
2288
'bzrlib.repofmt.knitrepo',
2289
'RepositoryFormatKnit1',
2292
format_registry.register_lazy(
2293
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2294
'bzrlib.repofmt.knitrepo',
2295
'RepositoryFormatKnit3',
2298
format_registry.register_lazy(
2299
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2300
'bzrlib.repofmt.knitrepo',
2301
'RepositoryFormatKnit4',
2304
# Pack-based formats. There is one format for pre-subtrees, and one for
2305
# post-subtrees to allow ease of testing.
2306
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2307
format_registry.register_lazy(
2308
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2309
'bzrlib.repofmt.pack_repo',
2310
'RepositoryFormatKnitPack1',
2312
format_registry.register_lazy(
2313
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2314
'bzrlib.repofmt.pack_repo',
2315
'RepositoryFormatKnitPack3',
2317
format_registry.register_lazy(
2318
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2319
'bzrlib.repofmt.pack_repo',
2320
'RepositoryFormatKnitPack4',
2322
format_registry.register_lazy(
2323
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2324
'bzrlib.repofmt.pack_repo',
2325
'RepositoryFormatKnitPack5',
2327
format_registry.register_lazy(
2328
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2329
'bzrlib.repofmt.pack_repo',
2330
'RepositoryFormatKnitPack5RichRoot',
2332
format_registry.register_lazy(
2333
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2334
'bzrlib.repofmt.pack_repo',
2335
'RepositoryFormatKnitPack5RichRootBroken',
2337
format_registry.register_lazy(
2338
'Bazaar RepositoryFormatKnitPack6 (bzr 1.9)\n',
2339
'bzrlib.repofmt.pack_repo',
2340
'RepositoryFormatKnitPack6',
2342
format_registry.register_lazy(
2343
'Bazaar RepositoryFormatKnitPack6RichRoot (bzr 1.9)\n',
2344
'bzrlib.repofmt.pack_repo',
2345
'RepositoryFormatKnitPack6RichRoot',
2348
# Development formats.
2349
# 1.7->1.8 go below here
2350
format_registry.register_lazy(
2351
"Bazaar development format 2 (needs bzr.dev from before 1.8)\n",
2352
'bzrlib.repofmt.pack_repo',
2353
'RepositoryFormatPackDevelopment2',
2355
format_registry.register_lazy(
2356
("Bazaar development format 2 with subtree support "
2357
"(needs bzr.dev from before 1.8)\n"),
2358
'bzrlib.repofmt.pack_repo',
2359
'RepositoryFormatPackDevelopment2Subtree',
2363
class InterRepository(InterObject):
2364
"""This class represents operations taking place between two repositories.
2366
Its instances have methods like copy_content and fetch, and contain
2367
references to the source and target repositories these operations can be
2370
Often we will provide convenience methods on 'repository' which carry out
2371
operations with another repository - they will always forward to
2372
InterRepository.get(other).method_name(parameters).
2375
_walk_to_common_revisions_batch_size = 1
2377
"""The available optimised InterRepository types."""
2379
def __init__(self, source, target):
2380
InterObject.__init__(self, source, target)
2381
# These two attributes may be overridden by e.g. InterOtherToRemote to
2382
# provide a faster implementation.
2383
self.target_get_graph = self.target.get_graph
2384
self.target_get_parent_map = self.target.get_parent_map
2386
def copy_content(self, revision_id=None):
2387
raise NotImplementedError(self.copy_content)
2389
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2390
"""Fetch the content required to construct revision_id.
2392
The content is copied from self.source to self.target.
2394
:param revision_id: if None all content is copied, if NULL_REVISION no
2396
:param pb: optional progress bar to use for progress reports. If not
2397
provided a default one will be created.
2399
:returns: (copied_revision_count, failures).
2401
# Normally we should find a specific InterRepository subclass to do
2402
# the fetch; if nothing else then at least InterSameDataRepository.
2403
# If none of them is suitable it looks like fetching is not possible;
2404
# we try to give a good message why. _assert_same_model will probably
2405
# give a helpful message; otherwise a generic one.
2406
self._assert_same_model(self.source, self.target)
2407
raise errors.IncompatibleRepositories(self.source, self.target,
2408
"no suitableInterRepository found")
2410
def _walk_to_common_revisions(self, revision_ids):
2411
"""Walk out from revision_ids in source to revisions target has.
2413
:param revision_ids: The start point for the search.
2414
:return: A set of revision ids.
2416
target_graph = self.target_get_graph()
2417
revision_ids = frozenset(revision_ids)
2418
# Fast path for the case where all the revisions are already in the
2420
# (Although this does incur an extra round trip for the
2421
# fairly common case where the target doesn't already have the revision
2423
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2424
return graph.SearchResult(revision_ids, set(), 0, set())
2425
missing_revs = set()
2426
source_graph = self.source.get_graph()
2427
# ensure we don't pay silly lookup costs.
2428
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2429
null_set = frozenset([_mod_revision.NULL_REVISION])
2430
searcher_exhausted = False
2434
# Iterate the searcher until we have enough next_revs
2435
while len(next_revs) < self._walk_to_common_revisions_batch_size:
2437
next_revs_part, ghosts_part = searcher.next_with_ghosts()
2438
next_revs.update(next_revs_part)
2439
ghosts.update(ghosts_part)
2440
except StopIteration:
2441
searcher_exhausted = True
2443
# If there are ghosts in the source graph, and the caller asked for
2444
# them, make sure that they are present in the target.
2445
# We don't care about other ghosts as we can't fetch them and
2446
# haven't been asked to.
2447
ghosts_to_check = set(revision_ids.intersection(ghosts))
2448
revs_to_get = set(next_revs).union(ghosts_to_check)
2450
have_revs = set(target_graph.get_parent_map(revs_to_get))
2451
# we always have NULL_REVISION present.
2452
have_revs = have_revs.union(null_set)
2453
# Check if the target is missing any ghosts we need.
2454
ghosts_to_check.difference_update(have_revs)
2456
# One of the caller's revision_ids is a ghost in both the
2457
# source and the target.
2458
raise errors.NoSuchRevision(
2459
self.source, ghosts_to_check.pop())
2460
missing_revs.update(next_revs - have_revs)
2461
# Because we may have walked past the original stop point, make
2462
# sure everything is stopped
2463
stop_revs = searcher.find_seen_ancestors(have_revs)
2464
searcher.stop_searching_any(stop_revs)
2465
if searcher_exhausted:
2467
return searcher.get_result()
2469
@deprecated_method(one_two)
2471
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2472
"""Return the revision ids that source has that target does not.
2474
These are returned in topological order.
2476
:param revision_id: only return revision ids included by this
2478
:param find_ghosts: If True find missing revisions in deep history
2479
rather than just finding the surface difference.
2481
return list(self.search_missing_revision_ids(
2482
revision_id, find_ghosts).get_keys())
2485
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2486
"""Return the revision ids that source has that target does not.
2488
:param revision_id: only return revision ids included by this
2490
:param find_ghosts: If True find missing revisions in deep history
2491
rather than just finding the surface difference.
2492
:return: A bzrlib.graph.SearchResult.
2494
# stop searching at found target revisions.
2495
if not find_ghosts and revision_id is not None:
2496
return self._walk_to_common_revisions([revision_id])
2497
# generic, possibly worst case, slow code path.
2498
target_ids = set(self.target.all_revision_ids())
2499
if revision_id is not None:
2500
source_ids = self.source.get_ancestry(revision_id)
2501
if source_ids[0] is not None:
2502
raise AssertionError()
2505
source_ids = self.source.all_revision_ids()
2506
result_set = set(source_ids).difference(target_ids)
2507
return self.source.revision_ids_to_search_result(result_set)
2510
def _same_model(source, target):
2511
"""True if source and target have the same data representation.
2513
Note: this is always called on the base class; overriding it in a
2514
subclass will have no effect.
2517
InterRepository._assert_same_model(source, target)
2519
except errors.IncompatibleRepositories, e:
2523
def _assert_same_model(source, target):
2524
"""Raise an exception if two repositories do not use the same model.
2526
if source.supports_rich_root() != target.supports_rich_root():
2527
raise errors.IncompatibleRepositories(source, target,
2528
"different rich-root support")
2529
if source._serializer != target._serializer:
2530
raise errors.IncompatibleRepositories(source, target,
2531
"different serializers")
2534
class InterSameDataRepository(InterRepository):
2535
"""Code for converting between repositories that represent the same data.
2537
Data format and model must match for this to work.
2541
def _get_repo_format_to_test(self):
2542
"""Repository format for testing with.
2544
InterSameData can pull from subtree to subtree and from non-subtree to
2545
non-subtree, so we test this with the richest repository format.
2547
from bzrlib.repofmt import knitrepo
2548
return knitrepo.RepositoryFormatKnit3()
2551
def is_compatible(source, target):
2552
return InterRepository._same_model(source, target)
2555
def copy_content(self, revision_id=None):
2556
"""Make a complete copy of the content in self into destination.
2558
This copies both the repository's revision data, and configuration information
2559
such as the make_working_trees setting.
2561
This is a destructive operation! Do not use it on existing
2564
:param revision_id: Only copy the content needed to construct
2565
revision_id and its parents.
2568
self.target.set_make_working_trees(self.source.make_working_trees())
2569
except NotImplementedError:
2571
# but don't bother fetching if we have the needed data now.
2572
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2573
self.target.has_revision(revision_id)):
2575
self.target.fetch(self.source, revision_id=revision_id)
2578
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2579
"""See InterRepository.fetch()."""
2580
from bzrlib.fetch import RepoFetcher
2581
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2582
self.source, self.source._format, self.target,
2583
self.target._format)
2584
f = RepoFetcher(to_repository=self.target,
2585
from_repository=self.source,
2586
last_revision=revision_id,
2587
pb=pb, find_ghosts=find_ghosts)
2588
return f.count_copied, f.failed_revisions
2591
class InterWeaveRepo(InterSameDataRepository):
2592
"""Optimised code paths between Weave based repositories.
2594
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2595
implemented lazy inter-object optimisation.
2599
def _get_repo_format_to_test(self):
2600
from bzrlib.repofmt import weaverepo
2601
return weaverepo.RepositoryFormat7()
2604
def is_compatible(source, target):
2605
"""Be compatible with known Weave formats.
2607
We don't test for the stores being of specific types because that
2608
could lead to confusing results, and there is no need to be
2611
from bzrlib.repofmt.weaverepo import (
2617
return (isinstance(source._format, (RepositoryFormat5,
2619
RepositoryFormat7)) and
2620
isinstance(target._format, (RepositoryFormat5,
2622
RepositoryFormat7)))
2623
except AttributeError:
2627
def copy_content(self, revision_id=None):
2628
"""See InterRepository.copy_content()."""
2629
# weave specific optimised path:
2631
self.target.set_make_working_trees(self.source.make_working_trees())
2632
except (errors.RepositoryUpgradeRequired, NotImplemented):
2634
# FIXME do not peek!
2635
if self.source._transport.listable():
2636
pb = ui.ui_factory.nested_progress_bar()
2638
self.target.texts.insert_record_stream(
2639
self.source.texts.get_record_stream(
2640
self.source.texts.keys(), 'topological', False))
2641
pb.update('copying inventory', 0, 1)
2642
self.target.inventories.insert_record_stream(
2643
self.source.inventories.get_record_stream(
2644
self.source.inventories.keys(), 'topological', False))
2645
self.target.signatures.insert_record_stream(
2646
self.source.signatures.get_record_stream(
2647
self.source.signatures.keys(),
2649
self.target.revisions.insert_record_stream(
2650
self.source.revisions.get_record_stream(
2651
self.source.revisions.keys(),
2652
'topological', True))
2656
self.target.fetch(self.source, revision_id=revision_id)
2659
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2660
"""See InterRepository.fetch()."""
2661
from bzrlib.fetch import RepoFetcher
2662
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2663
self.source, self.source._format, self.target, self.target._format)
2664
f = RepoFetcher(to_repository=self.target,
2665
from_repository=self.source,
2666
last_revision=revision_id,
2667
pb=pb, find_ghosts=find_ghosts)
2668
return f.count_copied, f.failed_revisions
2671
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2672
"""See InterRepository.missing_revision_ids()."""
2673
# we want all revisions to satisfy revision_id in source.
2674
# but we don't want to stat every file here and there.
2675
# we want then, all revisions other needs to satisfy revision_id
2676
# checked, but not those that we have locally.
2677
# so the first thing is to get a subset of the revisions to
2678
# satisfy revision_id in source, and then eliminate those that
2679
# we do already have.
2680
# this is slow on high latency connection to self, but as as this
2681
# disk format scales terribly for push anyway due to rewriting
2682
# inventory.weave, this is considered acceptable.
2684
if revision_id is not None:
2685
source_ids = self.source.get_ancestry(revision_id)
2686
if source_ids[0] is not None:
2687
raise AssertionError()
2690
source_ids = self.source._all_possible_ids()
2691
source_ids_set = set(source_ids)
2692
# source_ids is the worst possible case we may need to pull.
2693
# now we want to filter source_ids against what we actually
2694
# have in target, but don't try to check for existence where we know
2695
# we do not have a revision as that would be pointless.
2696
target_ids = set(self.target._all_possible_ids())
2697
possibly_present_revisions = target_ids.intersection(source_ids_set)
2698
actually_present_revisions = set(
2699
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2700
required_revisions = source_ids_set.difference(actually_present_revisions)
2701
if revision_id is not None:
2702
# we used get_ancestry to determine source_ids then we are assured all
2703
# revisions referenced are present as they are installed in topological order.
2704
# and the tip revision was validated by get_ancestry.
2705
result_set = required_revisions
2707
# if we just grabbed the possibly available ids, then
2708
# we only have an estimate of whats available and need to validate
2709
# that against the revision records.
2711
self.source._eliminate_revisions_not_present(required_revisions))
2712
return self.source.revision_ids_to_search_result(result_set)
2715
class InterKnitRepo(InterSameDataRepository):
2716
"""Optimised code paths between Knit based repositories."""
2719
def _get_repo_format_to_test(self):
2720
from bzrlib.repofmt import knitrepo
2721
return knitrepo.RepositoryFormatKnit1()
2724
def is_compatible(source, target):
2725
"""Be compatible with known Knit formats.
2727
We don't test for the stores being of specific types because that
2728
could lead to confusing results, and there is no need to be
2731
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2733
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2734
isinstance(target._format, RepositoryFormatKnit))
2735
except AttributeError:
2737
return are_knits and InterRepository._same_model(source, target)
2740
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2741
"""See InterRepository.fetch()."""
2742
from bzrlib.fetch import RepoFetcher
2743
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2744
self.source, self.source._format, self.target, self.target._format)
2745
f = RepoFetcher(to_repository=self.target,
2746
from_repository=self.source,
2747
last_revision=revision_id,
2748
pb=pb, find_ghosts=find_ghosts)
2749
return f.count_copied, f.failed_revisions
2752
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2753
"""See InterRepository.missing_revision_ids()."""
2754
if revision_id is not None:
2755
source_ids = self.source.get_ancestry(revision_id)
2756
if source_ids[0] is not None:
2757
raise AssertionError()
2760
source_ids = self.source.all_revision_ids()
2761
source_ids_set = set(source_ids)
2762
# source_ids is the worst possible case we may need to pull.
2763
# now we want to filter source_ids against what we actually
2764
# have in target, but don't try to check for existence where we know
2765
# we do not have a revision as that would be pointless.
2766
target_ids = set(self.target.all_revision_ids())
2767
possibly_present_revisions = target_ids.intersection(source_ids_set)
2768
actually_present_revisions = set(
2769
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2770
required_revisions = source_ids_set.difference(actually_present_revisions)
2771
if revision_id is not None:
2772
# we used get_ancestry to determine source_ids then we are assured all
2773
# revisions referenced are present as they are installed in topological order.
2774
# and the tip revision was validated by get_ancestry.
2775
result_set = required_revisions
2777
# if we just grabbed the possibly available ids, then
2778
# we only have an estimate of whats available and need to validate
2779
# that against the revision records.
2781
self.source._eliminate_revisions_not_present(required_revisions))
2782
return self.source.revision_ids_to_search_result(result_set)
2785
class InterPackRepo(InterSameDataRepository):
2786
"""Optimised code paths between Pack based repositories."""
2789
def _get_repo_format_to_test(self):
2790
from bzrlib.repofmt import pack_repo
2791
return pack_repo.RepositoryFormatKnitPack1()
2794
def is_compatible(source, target):
2795
"""Be compatible with known Pack formats.
2797
We don't test for the stores being of specific types because that
2798
could lead to confusing results, and there is no need to be
2801
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2803
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2804
isinstance(target._format, RepositoryFormatPack))
2805
except AttributeError:
2807
return are_packs and InterRepository._same_model(source, target)
2810
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2811
"""See InterRepository.fetch()."""
2812
if (len(self.source._fallback_repositories) > 0 or
2813
len(self.target._fallback_repositories) > 0):
2814
# The pack layer is not aware of fallback repositories, so when
2815
# fetching from a stacked repository or into a stacked repository
2816
# we use the generic fetch logic which uses the VersionedFiles
2817
# attributes on repository.
2818
from bzrlib.fetch import RepoFetcher
2819
fetcher = RepoFetcher(self.target, self.source, revision_id,
2821
return fetcher.count_copied, fetcher.failed_revisions
2822
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2823
self.source, self.source._format, self.target, self.target._format)
2824
self.count_copied = 0
2825
if revision_id is None:
2827
# everything to do - use pack logic
2828
# to fetch from all packs to one without
2829
# inventory parsing etc, IFF nothing to be copied is in the target.
2831
source_revision_ids = frozenset(self.source.all_revision_ids())
2832
revision_ids = source_revision_ids - \
2833
frozenset(self.target_get_parent_map(source_revision_ids))
2834
revision_keys = [(revid,) for revid in revision_ids]
2835
target_pack_collection = self._get_target_pack_collection()
2836
index = target_pack_collection.revision_index.combined_index
2837
present_revision_ids = set(item[1][0] for item in
2838
index.iter_entries(revision_keys))
2839
revision_ids = set(revision_ids) - present_revision_ids
2840
# implementing the TODO will involve:
2841
# - detecting when all of a pack is selected
2842
# - avoiding as much as possible pre-selection, so the
2843
# more-core routines such as create_pack_from_packs can filter in
2844
# a just-in-time fashion. (though having a HEADS list on a
2845
# repository might make this a lot easier, because we could
2846
# sensibly detect 'new revisions' without doing a full index scan.
2847
elif _mod_revision.is_null(revision_id):
2852
revision_ids = self.search_missing_revision_ids(revision_id,
2853
find_ghosts=find_ghosts).get_keys()
2854
except errors.NoSuchRevision:
2855
raise errors.InstallFailed([revision_id])
2856
if len(revision_ids) == 0:
2858
return self._pack(self.source, self.target, revision_ids)
2860
def _pack(self, source, target, revision_ids):
2861
from bzrlib.repofmt.pack_repo import Packer
2862
target_pack_collection = self._get_target_pack_collection()
2863
packs = source._pack_collection.all_packs()
2864
pack = Packer(target_pack_collection, packs, '.fetch',
2865
revision_ids).pack()
2866
if pack is not None:
2867
target_pack_collection._save_pack_names()
2868
copied_revs = pack.get_revision_count()
2869
# Trigger an autopack. This may duplicate effort as we've just done
2870
# a pack creation, but for now it is simpler to think about as
2871
# 'upload data, then repack if needed'.
2873
return (copied_revs, [])
2877
def _autopack(self):
2878
self.target._pack_collection.autopack()
2880
def _get_target_pack_collection(self):
2881
return self.target._pack_collection
2884
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2885
"""See InterRepository.missing_revision_ids().
2887
:param find_ghosts: Find ghosts throughout the ancestry of
2890
if not find_ghosts and revision_id is not None:
2891
return self._walk_to_common_revisions([revision_id])
2892
elif revision_id is not None:
2893
# Find ghosts: search for revisions pointing from one repository to
2894
# the other, and vice versa, anywhere in the history of revision_id.
2895
graph = self.target_get_graph(other_repository=self.source)
2896
searcher = graph._make_breadth_first_searcher([revision_id])
2900
next_revs, ghosts = searcher.next_with_ghosts()
2901
except StopIteration:
2903
if revision_id in ghosts:
2904
raise errors.NoSuchRevision(self.source, revision_id)
2905
found_ids.update(next_revs)
2906
found_ids.update(ghosts)
2907
found_ids = frozenset(found_ids)
2908
# Double query here: should be able to avoid this by changing the
2909
# graph api further.
2910
result_set = found_ids - frozenset(
2911
self.target_get_parent_map(found_ids))
2913
source_ids = self.source.all_revision_ids()
2914
# source_ids is the worst possible case we may need to pull.
2915
# now we want to filter source_ids against what we actually
2916
# have in target, but don't try to check for existence where we know
2917
# we do not have a revision as that would be pointless.
2918
target_ids = set(self.target.all_revision_ids())
2919
result_set = set(source_ids).difference(target_ids)
2920
return self.source.revision_ids_to_search_result(result_set)
2923
class InterModel1and2(InterRepository):
2926
def _get_repo_format_to_test(self):
2930
def is_compatible(source, target):
2931
if not source.supports_rich_root() and target.supports_rich_root():
2937
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2938
"""See InterRepository.fetch()."""
2939
from bzrlib.fetch import Model1toKnit2Fetcher
2940
f = Model1toKnit2Fetcher(to_repository=self.target,
2941
from_repository=self.source,
2942
last_revision=revision_id,
2943
pb=pb, find_ghosts=find_ghosts)
2944
return f.count_copied, f.failed_revisions
2947
def copy_content(self, revision_id=None):
2948
"""Make a complete copy of the content in self into destination.
2950
This is a destructive operation! Do not use it on existing
2953
:param revision_id: Only copy the content needed to construct
2954
revision_id and its parents.
2957
self.target.set_make_working_trees(self.source.make_working_trees())
2958
except NotImplementedError:
2960
# but don't bother fetching if we have the needed data now.
2961
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2962
self.target.has_revision(revision_id)):
2964
self.target.fetch(self.source, revision_id=revision_id)
2967
class InterKnit1and2(InterKnitRepo):
2970
def _get_repo_format_to_test(self):
2974
def is_compatible(source, target):
2975
"""Be compatible with Knit1 source and Knit3 target"""
2977
from bzrlib.repofmt.knitrepo import (
2978
RepositoryFormatKnit1,
2979
RepositoryFormatKnit3,
2981
from bzrlib.repofmt.pack_repo import (
2982
RepositoryFormatKnitPack1,
2983
RepositoryFormatKnitPack3,
2984
RepositoryFormatKnitPack4,
2985
RepositoryFormatKnitPack5,
2986
RepositoryFormatKnitPack5RichRoot,
2987
RepositoryFormatKnitPack6,
2988
RepositoryFormatKnitPack6RichRoot,
2989
RepositoryFormatPackDevelopment2,
2990
RepositoryFormatPackDevelopment2Subtree,
2993
RepositoryFormatKnit1, # no rr, no subtree
2994
RepositoryFormatKnitPack1, # no rr, no subtree
2995
RepositoryFormatPackDevelopment2, # no rr, no subtree
2996
RepositoryFormatKnitPack5, # no rr, no subtree
2997
RepositoryFormatKnitPack6, # no rr, no subtree
3000
RepositoryFormatKnit3, # rr, subtree
3001
RepositoryFormatKnitPack3, # rr, subtree
3002
RepositoryFormatKnitPack4, # rr, no subtree
3003
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
3004
RepositoryFormatKnitPack6RichRoot,# rr, no subtree
3005
RepositoryFormatPackDevelopment2Subtree, # rr, subtree
3007
for format in norichroot:
3008
if format.rich_root_data:
3009
raise AssertionError('Format %s is a rich-root format'
3010
' but is included in the non-rich-root list'
3012
for format in richroot:
3013
if not format.rich_root_data:
3014
raise AssertionError('Format %s is not a rich-root format'
3015
' but is included in the rich-root list'
3017
# TODO: One alternative is to just check format.rich_root_data,
3018
# instead of keeping membership lists. However, the formats
3019
# *also* have to use the same 'Knit' style of storage
3020
# (line-deltas, fulltexts, etc.)
3021
return (isinstance(source._format, norichroot) and
3022
isinstance(target._format, richroot))
3023
except AttributeError:
3027
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3028
"""See InterRepository.fetch()."""
3029
from bzrlib.fetch import Knit1to2Fetcher
3030
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
3031
self.source, self.source._format, self.target,
3032
self.target._format)
3033
f = Knit1to2Fetcher(to_repository=self.target,
3034
from_repository=self.source,
3035
last_revision=revision_id,
3036
pb=pb, find_ghosts=find_ghosts)
3037
return f.count_copied, f.failed_revisions
3040
class InterDifferingSerializer(InterKnitRepo):
3043
def _get_repo_format_to_test(self):
3047
def is_compatible(source, target):
3048
"""Be compatible with Knit2 source and Knit3 target"""
3049
if source.supports_rich_root() != target.supports_rich_root():
3051
# Ideally, we'd support fetching if the source had no tree references
3052
# even if it supported them...
3053
if (getattr(source, '_format.supports_tree_reference', False) and
3054
not getattr(target, '_format.supports_tree_reference', False)):
3059
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3060
"""See InterRepository.fetch()."""
3061
revision_ids = self.target.search_missing_revision_ids(self.source,
3062
revision_id, find_ghosts=find_ghosts).get_keys()
3063
revision_ids = tsort.topo_sort(
3064
self.source.get_graph().get_parent_map(revision_ids))
3065
def revisions_iterator():
3066
for current_revision_id in revision_ids:
3067
revision = self.source.get_revision(current_revision_id)
3068
tree = self.source.revision_tree(current_revision_id)
3070
signature = self.source.get_signature_text(
3071
current_revision_id)
3072
except errors.NoSuchRevision:
3074
yield revision, tree, signature
3076
my_pb = ui.ui_factory.nested_progress_bar()
3081
install_revisions(self.target, revisions_iterator(),
3082
len(revision_ids), pb)
3084
if my_pb is not None:
3086
return len(revision_ids), 0
3089
class InterOtherToRemote(InterRepository):
3090
"""An InterRepository that simply delegates to the 'real' InterRepository
3091
calculated for (source, target._real_repository).
3094
_walk_to_common_revisions_batch_size = 50
3096
def __init__(self, source, target):
3097
InterRepository.__init__(self, source, target)
3098
self._real_inter = None
3101
def is_compatible(source, target):
3102
if isinstance(target, remote.RemoteRepository):
3106
def _ensure_real_inter(self):
3107
if self._real_inter is None:
3108
self.target._ensure_real()
3109
real_target = self.target._real_repository
3110
self._real_inter = InterRepository.get(self.source, real_target)
3111
# Make _real_inter use the RemoteRepository for get_parent_map
3112
self._real_inter.target_get_graph = self.target.get_graph
3113
self._real_inter.target_get_parent_map = self.target.get_parent_map
3115
def copy_content(self, revision_id=None):
3116
self._ensure_real_inter()
3117
self._real_inter.copy_content(revision_id=revision_id)
3119
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3120
self._ensure_real_inter()
3121
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3122
find_ghosts=find_ghosts)
3125
def _get_repo_format_to_test(self):
3129
class InterRemoteToOther(InterRepository):
3131
def __init__(self, source, target):
3132
InterRepository.__init__(self, source, target)
3133
self._real_inter = None
3136
def is_compatible(source, target):
3137
if not isinstance(source, remote.RemoteRepository):
3139
# Is source's model compatible with target's model?
3140
source._ensure_real()
3141
real_source = source._real_repository
3142
if isinstance(real_source, remote.RemoteRepository):
3143
raise NotImplementedError(
3144
"We don't support remote repos backed by remote repos yet.")
3145
return InterRepository._same_model(real_source, target)
3147
def _ensure_real_inter(self):
3148
if self._real_inter is None:
3149
self.source._ensure_real()
3150
real_source = self.source._real_repository
3151
self._real_inter = InterRepository.get(real_source, self.target)
3153
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3154
self._ensure_real_inter()
3155
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3156
find_ghosts=find_ghosts)
3158
def copy_content(self, revision_id=None):
3159
self._ensure_real_inter()
3160
self._real_inter.copy_content(revision_id=revision_id)
3163
def _get_repo_format_to_test(self):
3168
class InterPackToRemotePack(InterPackRepo):
3169
"""A specialisation of InterPackRepo for a target that is a
3172
This will use the get_parent_map RPC rather than plain readvs, and also
3173
uses an RPC for autopacking.
3176
_walk_to_common_revisions_batch_size = 50
3179
def is_compatible(source, target):
3180
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
3181
if isinstance(source._format, RepositoryFormatPack):
3182
if isinstance(target, remote.RemoteRepository):
3183
target._ensure_real()
3184
if isinstance(target._real_repository._format,
3185
RepositoryFormatPack):
3186
if InterRepository._same_model(source, target):
3190
def _autopack(self):
3191
self.target.autopack()
3193
def _get_target_pack_collection(self):
3194
return self.target._real_repository._pack_collection
3197
def _get_repo_format_to_test(self):
3201
InterRepository.register_optimiser(InterDifferingSerializer)
3202
InterRepository.register_optimiser(InterSameDataRepository)
3203
InterRepository.register_optimiser(InterWeaveRepo)
3204
InterRepository.register_optimiser(InterKnitRepo)
3205
InterRepository.register_optimiser(InterModel1and2)
3206
InterRepository.register_optimiser(InterKnit1and2)
3207
InterRepository.register_optimiser(InterPackRepo)
3208
InterRepository.register_optimiser(InterOtherToRemote)
3209
InterRepository.register_optimiser(InterRemoteToOther)
3210
InterRepository.register_optimiser(InterPackToRemotePack)
3213
class CopyConverter(object):
3214
"""A repository conversion tool which just performs a copy of the content.
3216
This is slow but quite reliable.
3219
def __init__(self, target_format):
3220
"""Create a CopyConverter.
3222
:param target_format: The format the resulting repository should be.
3224
self.target_format = target_format
3226
def convert(self, repo, pb):
3227
"""Perform the conversion of to_convert, giving feedback via pb.
3229
:param to_convert: The disk object to convert.
3230
:param pb: a progress bar to use for progress information.
3235
# this is only useful with metadir layouts - separated repo content.
3236
# trigger an assertion if not such
3237
repo._format.get_format_string()
3238
self.repo_dir = repo.bzrdir
3239
self.step('Moving repository to repository.backup')
3240
self.repo_dir.transport.move('repository', 'repository.backup')
3241
backup_transport = self.repo_dir.transport.clone('repository.backup')
3242
repo._format.check_conversion_target(self.target_format)
3243
self.source_repo = repo._format.open(self.repo_dir,
3245
_override_transport=backup_transport)
3246
self.step('Creating new repository')
3247
converted = self.target_format.initialize(self.repo_dir,
3248
self.source_repo.is_shared())
3249
converted.lock_write()
3251
self.step('Copying content into repository.')
3252
self.source_repo.copy_content_into(converted)
3255
self.step('Deleting old repository content.')
3256
self.repo_dir.transport.delete_tree('repository.backup')
3257
self.pb.note('repository converted')
3259
def step(self, message):
3260
"""Update the pb by a step."""
3262
self.pb.update(message, self.count, self.total)
3274
def _unescaper(match, _map=_unescape_map):
3275
code = match.group(1)
3279
if not code.startswith('#'):
3281
return unichr(int(code[1:])).encode('utf8')
3287
def _unescape_xml(data):
3288
"""Unescape predefined XML entities in a string of data."""
3290
if _unescape_re is None:
3291
_unescape_re = re.compile('\&([^;]*);')
3292
return _unescape_re.sub(_unescaper, data)
3295
class _VersionedFileChecker(object):
3297
def __init__(self, repository):
3298
self.repository = repository
3299
self.text_index = self.repository._generate_text_key_index()
3301
def calculate_file_version_parents(self, text_key):
3302
"""Calculate the correct parents for a file version according to
3305
parent_keys = self.text_index[text_key]
3306
if parent_keys == [_mod_revision.NULL_REVISION]:
3308
return tuple(parent_keys)
3310
def check_file_version_parents(self, texts, progress_bar=None):
3311
"""Check the parents stored in a versioned file are correct.
3313
It also detects file versions that are not referenced by their
3314
corresponding revision's inventory.
3316
:returns: A tuple of (wrong_parents, dangling_file_versions).
3317
wrong_parents is a dict mapping {revision_id: (stored_parents,
3318
correct_parents)} for each revision_id where the stored parents
3319
are not correct. dangling_file_versions is a set of (file_id,
3320
revision_id) tuples for versions that are present in this versioned
3321
file, but not used by the corresponding inventory.
3324
self.file_ids = set([file_id for file_id, _ in
3325
self.text_index.iterkeys()])
3326
# text keys is now grouped by file_id
3327
n_weaves = len(self.file_ids)
3328
files_in_revisions = {}
3329
revisions_of_files = {}
3330
n_versions = len(self.text_index)
3331
progress_bar.update('loading text store', 0, n_versions)
3332
parent_map = self.repository.texts.get_parent_map(self.text_index)
3333
# On unlistable transports this could well be empty/error...
3334
text_keys = self.repository.texts.keys()
3335
unused_keys = frozenset(text_keys) - set(self.text_index)
3336
for num, key in enumerate(self.text_index.iterkeys()):
3337
if progress_bar is not None:
3338
progress_bar.update('checking text graph', num, n_versions)
3339
correct_parents = self.calculate_file_version_parents(key)
3341
knit_parents = parent_map[key]
3342
except errors.RevisionNotPresent:
3345
if correct_parents != knit_parents:
3346
wrong_parents[key] = (knit_parents, correct_parents)
3347
return wrong_parents, unused_keys
3350
def _old_get_graph(repository, revision_id):
3351
"""DO NOT USE. That is all. I'm serious."""
3352
graph = repository.get_graph()
3353
revision_graph = dict(((key, value) for key, value in
3354
graph.iter_ancestry([revision_id]) if value is not None))
3355
return _strip_NULL_ghosts(revision_graph)
3358
def _strip_NULL_ghosts(revision_graph):
3359
"""Also don't use this. more compatibility code for unmigrated clients."""
3360
# Filter ghosts, and null:
3361
if _mod_revision.NULL_REVISION in revision_graph:
3362
del revision_graph[_mod_revision.NULL_REVISION]
3363
for key, parents in revision_graph.items():
3364
revision_graph[key] = tuple(parent for parent in parents if parent
3366
return revision_graph
added
removed
bzrlib/repository.py
