1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from cStringIO import StringIO
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
39
revision as _mod_revision,
45
from bzrlib.bundle import serializer
46
from bzrlib.revisiontree import RevisionTree
47
from bzrlib.store.versioned import VersionedFileStore
48
from bzrlib.store.text import TextStore
49
from bzrlib.testament import Testament
50
from bzrlib.util import bencode
53
from bzrlib.decorators import needs_read_lock, needs_write_lock
54
from bzrlib.inter import InterObject
55
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
56
from bzrlib.symbol_versioning import (
63
from bzrlib.trace import mutter, mutter_callsite, note, warning
66
# Old formats display a warning, but only once
67
_deprecation_warning_done = False
70
class CommitBuilder(object):
71
"""Provides an interface to build up a commit.
73
This allows describing a tree to be committed without needing to
74
know the internals of the format of the repository.
77
# all clients should supply tree roots.
78
record_root_entry = True
79
# the default CommitBuilder does not manage trees whose root is versioned.
80
_versioned_root = False
82
def __init__(self, repository, parents, config, timestamp=None,
83
timezone=None, committer=None, revprops=None,
85
"""Initiate a CommitBuilder.
87
:param repository: Repository to commit to.
88
:param parents: Revision ids of the parents of the new revision.
89
:param config: Configuration to use.
90
:param timestamp: Optional timestamp recorded for commit.
91
:param timezone: Optional timezone for timestamp.
92
:param committer: Optional committer to set for commit.
93
:param revprops: Optional dictionary of revision properties.
94
:param revision_id: Optional revision id.
99
self._committer = self._config.username()
101
self._committer = committer
103
self.new_inventory = Inventory(None)
104
self._new_revision_id = revision_id
105
self.parents = parents
106
self.repository = repository
109
if revprops is not None:
110
self._revprops.update(revprops)
112
if timestamp is None:
113
timestamp = time.time()
114
# Restrict resolution to 1ms
115
self._timestamp = round(timestamp, 3)
118
self._timezone = osutils.local_time_offset()
120
self._timezone = int(timezone)
122
self._generate_revision_if_needed()
123
self.__heads = graph.HeadsCache(repository.get_graph()).heads
125
def commit(self, message):
126
"""Make the actual commit.
128
:return: The revision id of the recorded revision.
130
rev = _mod_revision.Revision(
131
timestamp=self._timestamp,
132
timezone=self._timezone,
133
committer=self._committer,
135
inventory_sha1=self.inv_sha1,
136
revision_id=self._new_revision_id,
137
properties=self._revprops)
138
rev.parent_ids = self.parents
139
self.repository.add_revision(self._new_revision_id, rev,
140
self.new_inventory, self._config)
141
self.repository.commit_write_group()
142
return self._new_revision_id
145
"""Abort the commit that is being built.
147
self.repository.abort_write_group()
149
def revision_tree(self):
150
"""Return the tree that was just committed.
152
After calling commit() this can be called to get a RevisionTree
153
representing the newly committed tree. This is preferred to
154
calling Repository.revision_tree() because that may require
155
deserializing the inventory, while we already have a copy in
158
return RevisionTree(self.repository, self.new_inventory,
159
self._new_revision_id)
161
def finish_inventory(self):
162
"""Tell the builder that the inventory is finished."""
163
if self.new_inventory.root is None:
164
raise AssertionError('Root entry should be supplied to'
165
' record_entry_contents, as of bzr 0.10.')
166
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
167
self.new_inventory.revision_id = self._new_revision_id
168
self.inv_sha1 = self.repository.add_inventory(
169
self._new_revision_id,
174
def _gen_revision_id(self):
175
"""Return new revision-id."""
176
return generate_ids.gen_revision_id(self._config.username(),
179
def _generate_revision_if_needed(self):
180
"""Create a revision id if None was supplied.
182
If the repository can not support user-specified revision ids
183
they should override this function and raise CannotSetRevisionId
184
if _new_revision_id is not None.
186
:raises: CannotSetRevisionId
188
if self._new_revision_id is None:
189
self._new_revision_id = self._gen_revision_id()
190
self.random_revid = True
192
self.random_revid = False
194
def _heads(self, file_id, revision_ids):
195
"""Calculate the graph heads for revision_ids in the graph of file_id.
197
This can use either a per-file graph or a global revision graph as we
198
have an identity relationship between the two graphs.
200
return self.__heads(revision_ids)
202
def _check_root(self, ie, parent_invs, tree):
203
"""Helper for record_entry_contents.
205
:param ie: An entry being added.
206
:param parent_invs: The inventories of the parent revisions of the
208
:param tree: The tree that is being committed.
210
# In this revision format, root entries have no knit or weave When
211
# serializing out to disk and back in root.revision is always
213
ie.revision = self._new_revision_id
215
def _get_delta(self, ie, basis_inv, path):
216
"""Get a delta against the basis inventory for ie."""
217
if ie.file_id not in basis_inv:
219
return (None, path, ie.file_id, ie)
220
elif ie != basis_inv[ie.file_id]:
222
# TODO: avoid tis id2path call.
223
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
228
def record_entry_contents(self, ie, parent_invs, path, tree,
230
"""Record the content of ie from tree into the commit if needed.
232
Side effect: sets ie.revision when unchanged
234
:param ie: An inventory entry present in the commit.
235
:param parent_invs: The inventories of the parent revisions of the
237
:param path: The path the entry is at in the tree.
238
:param tree: The tree which contains this entry and should be used to
240
:param content_summary: Summary data from the tree about the paths
241
content - stat, length, exec, sha/link target. This is only
242
accessed when the entry has a revision of None - that is when it is
243
a candidate to commit.
244
:return: A tuple (change_delta, version_recorded). change_delta is
245
an inventory_delta change for this entry against the basis tree of
246
the commit, or None if no change occured against the basis tree.
247
version_recorded is True if a new version of the entry has been
248
recorded. For instance, committing a merge where a file was only
249
changed on the other side will return (delta, False).
251
if self.new_inventory.root is None:
252
if ie.parent_id is not None:
253
raise errors.RootMissing()
254
self._check_root(ie, parent_invs, tree)
255
if ie.revision is None:
256
kind = content_summary[0]
258
# ie is carried over from a prior commit
260
# XXX: repository specific check for nested tree support goes here - if
261
# the repo doesn't want nested trees we skip it ?
262
if (kind == 'tree-reference' and
263
not self.repository._format.supports_tree_reference):
264
# mismatch between commit builder logic and repository:
265
# this needs the entry creation pushed down into the builder.
266
raise NotImplementedError('Missing repository subtree support.')
267
self.new_inventory.add(ie)
269
# TODO: slow, take it out of the inner loop.
271
basis_inv = parent_invs[0]
273
basis_inv = Inventory(root_id=None)
275
# ie.revision is always None if the InventoryEntry is considered
276
# for committing. We may record the previous parents revision if the
277
# content is actually unchanged against a sole head.
278
if ie.revision is not None:
279
if not self._versioned_root and path == '':
280
# repositories that do not version the root set the root's
281
# revision to the new commit even when no change occurs, and
282
# this masks when a change may have occurred against the basis,
283
# so calculate if one happened.
284
if ie.file_id in basis_inv:
285
delta = (basis_inv.id2path(ie.file_id), path,
289
delta = (None, path, ie.file_id, ie)
292
# we don't need to commit this, because the caller already
293
# determined that an existing revision of this file is
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))
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:
338
if content_summary[2] is None:
339
raise ValueError("Files must not have executable = None")
341
if (# if the file length changed we have to store:
342
parent_entry.text_size != content_summary[1] or
343
# if the exec bit has changed we have to store:
344
parent_entry.executable != content_summary[2]):
346
elif parent_entry.text_sha1 == content_summary[3]:
347
# all meta and content is unchanged (using a hash cache
348
# hit to check the sha)
349
ie.revision = parent_entry.revision
350
ie.text_size = parent_entry.text_size
351
ie.text_sha1 = parent_entry.text_sha1
352
ie.executable = parent_entry.executable
353
return self._get_delta(ie, basis_inv, path), False
355
# Either there is only a hash change(no hash cache entry,
356
# or same size content change), or there is no change on
358
# Provide the parent's hash to the store layer, so that the
359
# content is unchanged we will not store a new node.
360
nostore_sha = parent_entry.text_sha1
362
# We want to record a new node regardless of the presence or
363
# absence of a content change in the file.
365
ie.executable = content_summary[2]
366
lines = tree.get_file(ie.file_id, path).readlines()
368
ie.text_sha1, ie.text_size = self._add_text_to_weave(
369
ie.file_id, lines, heads, nostore_sha)
370
except errors.ExistingContent:
371
# Turns out that the file content was unchanged, and we were
372
# only going to store a new node if it was changed. Carry over
374
ie.revision = parent_entry.revision
375
ie.text_size = parent_entry.text_size
376
ie.text_sha1 = parent_entry.text_sha1
377
ie.executable = parent_entry.executable
378
return self._get_delta(ie, basis_inv, path), False
379
elif kind == 'directory':
381
# all data is meta here, nothing specific to directory, so
383
ie.revision = parent_entry.revision
384
return self._get_delta(ie, basis_inv, path), False
386
self._add_text_to_weave(ie.file_id, lines, heads, None)
387
elif kind == 'symlink':
388
current_link_target = content_summary[3]
390
# symlink target is not generic metadata, check if it has
392
if current_link_target != parent_entry.symlink_target:
395
# unchanged, carry over.
396
ie.revision = parent_entry.revision
397
ie.symlink_target = parent_entry.symlink_target
398
return self._get_delta(ie, basis_inv, path), False
399
ie.symlink_target = current_link_target
401
self._add_text_to_weave(ie.file_id, lines, heads, None)
402
elif kind == 'tree-reference':
404
if content_summary[3] != parent_entry.reference_revision:
407
# unchanged, carry over.
408
ie.reference_revision = parent_entry.reference_revision
409
ie.revision = parent_entry.revision
410
return self._get_delta(ie, basis_inv, path), False
411
ie.reference_revision = content_summary[3]
413
self._add_text_to_weave(ie.file_id, lines, heads, None)
415
raise NotImplementedError('unknown kind')
416
ie.revision = self._new_revision_id
417
return self._get_delta(ie, basis_inv, path), True
419
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
420
# Note: as we read the content directly from the tree, we know its not
421
# been turned into unicode or badly split - but a broken tree
422
# implementation could give us bad output from readlines() so this is
423
# not a guarantee of safety. What would be better is always checking
424
# the content during test suite execution. RBC 20070912
425
parent_keys = tuple((file_id, parent) for parent in parents)
426
return self.repository.texts.add_lines(
427
(file_id, self._new_revision_id), parent_keys, new_lines,
428
nostore_sha=nostore_sha, random_id=self.random_revid,
429
check_content=False)[0:2]
432
class RootCommitBuilder(CommitBuilder):
433
"""This commitbuilder actually records the root id"""
435
# the root entry gets versioned properly by this builder.
436
_versioned_root = True
438
def _check_root(self, ie, parent_invs, tree):
439
"""Helper for record_entry_contents.
441
:param ie: An entry being added.
442
:param parent_invs: The inventories of the parent revisions of the
444
:param tree: The tree that is being committed.
448
######################################################################
451
class Repository(object):
452
"""Repository holding history for one or more branches.
454
The repository holds and retrieves historical information including
455
revisions and file history. It's normally accessed only by the Branch,
456
which views a particular line of development through that history.
458
The Repository builds on top of some byte storage facilies (the revisions,
459
signatures, inventories and texts attributes) and a Transport, which
460
respectively provide byte storage and a means to access the (possibly
463
The byte storage facilities are addressed via tuples, which we refer to
464
as 'keys' throughout the code base. Revision_keys, inventory_keys and
465
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
466
(file_id, revision_id). We use this interface because it allows low
467
friction with the underlying code that implements disk indices, network
468
encoding and other parts of bzrlib.
470
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
471
the serialised revisions for the repository. This can be used to obtain
472
revision graph information or to access raw serialised revisions.
473
The result of trying to insert data into the repository via this store
474
is undefined: it should be considered read-only except for implementors
476
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
477
the serialised signatures for the repository. This can be used to
478
obtain access to raw serialised signatures. The result of trying to
479
insert data into the repository via this store is undefined: it should
480
be considered read-only except for implementors of repositories.
481
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
482
the serialised inventories for the repository. This can be used to
483
obtain unserialised inventories. The result of trying to insert data
484
into the repository via this store is undefined: it should be
485
considered read-only except for implementors of repositories.
486
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
487
texts of files and directories for the repository. This can be used to
488
obtain file texts or file graphs. Note that Repository.iter_file_bytes
489
is usually a better interface for accessing file texts.
490
The result of trying to insert data into the repository via this store
491
is undefined: it should be considered read-only except for implementors
493
:ivar _transport: Transport for file access to repository, typically
494
pointing to .bzr/repository.
497
# What class to use for a CommitBuilder. Often its simpler to change this
498
# in a Repository class subclass rather than to override
499
# get_commit_builder.
500
_commit_builder_class = CommitBuilder
501
# The search regex used by xml based repositories to determine what things
502
# where changed in a single commit.
503
_file_ids_altered_regex = lazy_regex.lazy_compile(
504
r'file_id="(?P<file_id>[^"]+)"'
505
r'.* revision="(?P<revision_id>[^"]+)"'
508
def abort_write_group(self):
509
"""Commit the contents accrued within the current write group.
511
:seealso: start_write_group.
513
if self._write_group is not self.get_transaction():
514
# has an unlock or relock occured ?
515
raise errors.BzrError('mismatched lock context and write group.')
516
self._abort_write_group()
517
self._write_group = None
519
def _abort_write_group(self):
520
"""Template method for per-repository write group cleanup.
522
This is called during abort before the write group is considered to be
523
finished and should cleanup any internal state accrued during the write
524
group. There is no requirement that data handed to the repository be
525
*not* made available - this is not a rollback - but neither should any
526
attempt be made to ensure that data added is fully commited. Abort is
527
invoked when an error has occured so futher disk or network operations
528
may not be possible or may error and if possible should not be
532
def add_fallback_repository(self, repository):
533
"""Add a repository to use for looking up data not held locally.
535
:param repository: A repository.
537
if not self._format.supports_external_lookups:
538
raise errors.UnstackableRepositoryFormat(self._format, self.base)
539
self._check_fallback_repository(repository)
540
self._fallback_repositories.append(repository)
541
self.texts.add_fallback_versioned_files(repository.texts)
542
self.inventories.add_fallback_versioned_files(repository.inventories)
543
self.revisions.add_fallback_versioned_files(repository.revisions)
544
self.signatures.add_fallback_versioned_files(repository.signatures)
546
def _check_fallback_repository(self, repository):
547
"""Check that this repository can fallback to repository safely.
549
Raise an error if not.
551
:param repository: A repository to fallback to.
553
return InterRepository._assert_same_model(self, repository)
555
def add_inventory(self, revision_id, inv, parents):
556
"""Add the inventory inv to the repository as revision_id.
558
:param parents: The revision ids of the parents that revision_id
559
is known to have and are in the repository already.
561
:returns: The validator(which is a sha1 digest, though what is sha'd is
562
repository format specific) of the serialized inventory.
564
if not self.is_in_write_group():
565
raise AssertionError("%r not in write group" % (self,))
566
_mod_revision.check_not_reserved_id(revision_id)
567
if not (inv.revision_id is None or inv.revision_id == revision_id):
568
raise AssertionError(
569
"Mismatch between inventory revision"
570
" id and insertion revid (%r, %r)"
571
% (inv.revision_id, revision_id))
573
raise AssertionError()
574
inv_lines = self._serialise_inventory_to_lines(inv)
575
return self._inventory_add_lines(revision_id, parents,
576
inv_lines, check_content=False)
578
def _inventory_add_lines(self, revision_id, parents, lines,
580
"""Store lines in inv_vf and return the sha1 of the inventory."""
581
parents = [(parent,) for parent in parents]
582
return self.inventories.add_lines((revision_id,), parents, lines,
583
check_content=check_content)[0]
585
def add_revision(self, revision_id, rev, inv=None, config=None):
586
"""Add rev to the revision store as revision_id.
588
:param revision_id: the revision id to use.
589
:param rev: The revision object.
590
:param inv: The inventory for the revision. if None, it will be looked
591
up in the inventory storer
592
:param config: If None no digital signature will be created.
593
If supplied its signature_needed method will be used
594
to determine if a signature should be made.
596
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
598
_mod_revision.check_not_reserved_id(revision_id)
599
if config is not None and config.signature_needed():
601
inv = self.get_inventory(revision_id)
602
plaintext = Testament(rev, inv).as_short_text()
603
self.store_revision_signature(
604
gpg.GPGStrategy(config), plaintext, revision_id)
605
# check inventory present
606
if not self.inventories.get_parent_map([(revision_id,)]):
608
raise errors.WeaveRevisionNotPresent(revision_id,
611
# yes, this is not suitable for adding with ghosts.
612
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
616
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
617
self._add_revision(rev)
619
def _add_revision(self, revision):
620
text = self._serializer.write_revision_to_string(revision)
621
key = (revision.revision_id,)
622
parents = tuple((parent,) for parent in revision.parent_ids)
623
self.revisions.add_lines(key, parents, osutils.split_lines(text))
625
def all_revision_ids(self):
626
"""Returns a list of all the revision ids in the repository.
628
This is conceptually deprecated because code should generally work on
629
the graph reachable from a particular revision, and ignore any other
630
revisions that might be present. There is no direct replacement
633
if 'evil' in debug.debug_flags:
634
mutter_callsite(2, "all_revision_ids is linear with history.")
635
return self._all_revision_ids()
637
def _all_revision_ids(self):
638
"""Returns a list of all the revision ids in the repository.
640
These are in as much topological order as the underlying store can
643
raise NotImplementedError(self._all_revision_ids)
645
def break_lock(self):
646
"""Break a lock if one is present from another instance.
648
Uses the ui factory to ask for confirmation if the lock may be from
651
self.control_files.break_lock()
654
def _eliminate_revisions_not_present(self, revision_ids):
655
"""Check every revision id in revision_ids to see if we have it.
657
Returns a set of the present revisions.
660
graph = self.get_graph()
661
parent_map = graph.get_parent_map(revision_ids)
662
# The old API returned a list, should this actually be a set?
663
return parent_map.keys()
666
def create(a_bzrdir):
667
"""Construct the current default format repository in a_bzrdir."""
668
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
670
def __init__(self, _format, a_bzrdir, control_files):
671
"""instantiate a Repository.
673
:param _format: The format of the repository on disk.
674
:param a_bzrdir: The BzrDir of the repository.
676
In the future we will have a single api for all stores for
677
getting file texts, inventories and revisions, then
678
this construct will accept instances of those things.
680
super(Repository, self).__init__()
681
self._format = _format
682
# the following are part of the public API for Repository:
683
self.bzrdir = a_bzrdir
684
self.control_files = control_files
685
self._transport = control_files._transport
686
self.base = self._transport.base
688
self._reconcile_does_inventory_gc = True
689
self._reconcile_fixes_text_parents = False
690
self._reconcile_backsup_inventory = True
691
# not right yet - should be more semantically clear ?
693
# TODO: make sure to construct the right store classes, etc, depending
694
# on whether escaping is required.
695
self._warn_if_deprecated()
696
self._write_group = None
697
# Additional places to query for data.
698
self._fallback_repositories = []
699
# What order should fetch operations request streams in?
700
# The default is unordered as that is the cheapest for an origin to
702
self._fetch_order = 'unordered'
703
# Does this repository use deltas that can be fetched as-deltas ?
704
# (E.g. knits, where the knit deltas can be transplanted intact.
705
# We default to False, which will ensure that enough data to get
706
# a full text out of any fetch stream will be grabbed.
707
self._fetch_uses_deltas = False
708
# Should fetch trigger a reconcile after the fetch? Only needed for
709
# some repository formats that can suffer internal inconsistencies.
710
self._fetch_reconcile = False
713
return '%s(%r)' % (self.__class__.__name__,
716
def has_same_location(self, other):
717
"""Returns a boolean indicating if this repository is at the same
718
location as another repository.
720
This might return False even when two repository objects are accessing
721
the same physical repository via different URLs.
723
if self.__class__ is not other.__class__:
725
return (self._transport.base == other._transport.base)
727
def is_in_write_group(self):
728
"""Return True if there is an open write group.
730
:seealso: start_write_group.
732
return self._write_group is not None
735
return self.control_files.is_locked()
737
def is_write_locked(self):
738
"""Return True if this object is write locked."""
739
return self.is_locked() and self.control_files._lock_mode == 'w'
741
def lock_write(self, token=None):
742
"""Lock this repository for writing.
744
This causes caching within the repository obejct to start accumlating
745
data during reads, and allows a 'write_group' to be obtained. Write
746
groups must be used for actual data insertion.
748
:param token: if this is already locked, then lock_write will fail
749
unless the token matches the existing lock.
750
:returns: a token if this instance supports tokens, otherwise None.
751
:raises TokenLockingNotSupported: when a token is given but this
752
instance doesn't support using token locks.
753
:raises MismatchedToken: if the specified token doesn't match the token
754
of the existing lock.
755
:seealso: start_write_group.
757
A token should be passed in if you know that you have locked the object
758
some other way, and need to synchronise this object's state with that
761
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
763
result = self.control_files.lock_write(token=token)
764
for repo in self._fallback_repositories:
765
# Writes don't affect fallback repos
771
self.control_files.lock_read()
772
for repo in self._fallback_repositories:
776
def get_physical_lock_status(self):
777
return self.control_files.get_physical_lock_status()
779
def leave_lock_in_place(self):
780
"""Tell this repository not to release the physical lock when this
783
If lock_write doesn't return a token, then this method is not supported.
785
self.control_files.leave_in_place()
787
def dont_leave_lock_in_place(self):
788
"""Tell this repository to release the physical lock when this
789
object is unlocked, even if it didn't originally acquire it.
791
If lock_write doesn't return a token, then this method is not supported.
793
self.control_files.dont_leave_in_place()
796
def gather_stats(self, revid=None, committers=None):
797
"""Gather statistics from a revision id.
799
:param revid: The revision id to gather statistics from, if None, then
800
no revision specific statistics are gathered.
801
:param committers: Optional parameter controlling whether to grab
802
a count of committers from the revision specific statistics.
803
:return: A dictionary of statistics. Currently this contains:
804
committers: The number of committers if requested.
805
firstrev: A tuple with timestamp, timezone for the penultimate left
806
most ancestor of revid, if revid is not the NULL_REVISION.
807
latestrev: A tuple with timestamp, timezone for revid, if revid is
808
not the NULL_REVISION.
809
revisions: The total revision count in the repository.
810
size: An estimate disk size of the repository in bytes.
813
if revid and committers:
814
result['committers'] = 0
815
if revid and revid != _mod_revision.NULL_REVISION:
817
all_committers = set()
818
revisions = self.get_ancestry(revid)
819
# pop the leading None
821
first_revision = None
823
# ignore the revisions in the middle - just grab first and last
824
revisions = revisions[0], revisions[-1]
825
for revision in self.get_revisions(revisions):
826
if not first_revision:
827
first_revision = revision
829
all_committers.add(revision.committer)
830
last_revision = revision
832
result['committers'] = len(all_committers)
833
result['firstrev'] = (first_revision.timestamp,
834
first_revision.timezone)
835
result['latestrev'] = (last_revision.timestamp,
836
last_revision.timezone)
838
# now gather global repository information
839
# XXX: This is available for many repos regardless of listability.
840
if self.bzrdir.root_transport.listable():
841
# XXX: do we want to __define len__() ?
842
# Maybe the versionedfiles object should provide a different
843
# method to get the number of keys.
844
result['revisions'] = len(self.revisions.keys())
848
def find_branches(self, using=False):
849
"""Find branches underneath this repository.
851
This will include branches inside other branches.
853
:param using: If True, list only branches using this repository.
855
if using and not self.is_shared():
857
return [self.bzrdir.open_branch()]
858
except errors.NotBranchError:
860
class Evaluator(object):
863
self.first_call = True
865
def __call__(self, bzrdir):
866
# On the first call, the parameter is always the bzrdir
867
# containing the current repo.
868
if not self.first_call:
870
repository = bzrdir.open_repository()
871
except errors.NoRepositoryPresent:
874
return False, (None, repository)
875
self.first_call = False
877
value = (bzrdir.open_branch(), None)
878
except errors.NotBranchError:
883
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
884
self.bzrdir.root_transport, evaluate=Evaluator()):
885
if branch is not None:
886
branches.append(branch)
887
if not using and repository is not None:
888
branches.extend(repository.find_branches())
892
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
893
"""Return the revision ids that other has that this does not.
895
These are returned in topological order.
897
revision_id: only return revision ids included by revision_id.
899
return InterRepository.get(other, self).search_missing_revision_ids(
900
revision_id, find_ghosts)
902
@deprecated_method(one_two)
904
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
905
"""Return the revision ids that other has that this does not.
907
These are returned in topological order.
909
revision_id: only return revision ids included by revision_id.
911
keys = self.search_missing_revision_ids(
912
other, revision_id, find_ghosts).get_keys()
915
parents = other.get_graph().get_parent_map(keys)
918
return tsort.topo_sort(parents)
922
"""Open the repository rooted at base.
924
For instance, if the repository is at URL/.bzr/repository,
925
Repository.open(URL) -> a Repository instance.
927
control = bzrdir.BzrDir.open(base)
928
return control.open_repository()
930
def copy_content_into(self, destination, revision_id=None):
931
"""Make a complete copy of the content in self into destination.
933
This is a destructive operation! Do not use it on existing
936
return InterRepository.get(self, destination).copy_content(revision_id)
938
def commit_write_group(self):
939
"""Commit the contents accrued within the current write group.
941
:seealso: start_write_group.
943
if self._write_group is not self.get_transaction():
944
# has an unlock or relock occured ?
945
raise errors.BzrError('mismatched lock context %r and '
947
(self.get_transaction(), self._write_group))
948
self._commit_write_group()
949
self._write_group = None
951
def _commit_write_group(self):
952
"""Template method for per-repository write group cleanup.
954
This is called before the write group is considered to be
955
finished and should ensure that all data handed to the repository
956
for writing during the write group is safely committed (to the
957
extent possible considering file system caching etc).
960
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
961
"""Fetch the content required to construct revision_id from source.
963
If revision_id is None all content is copied.
964
:param find_ghosts: Find and copy revisions in the source that are
965
ghosts in the target (and not reachable directly by walking out to
966
the first-present revision in target from revision_id).
968
# fast path same-url fetch operations
969
if self.has_same_location(source):
970
# check that last_revision is in 'from' and then return a
972
if (revision_id is not None and
973
not _mod_revision.is_null(revision_id)):
974
self.get_revision(revision_id)
976
# if there is no specific appropriate InterRepository, this will get
977
# the InterRepository base class, which raises an
978
# IncompatibleRepositories when asked to fetch.
979
inter = InterRepository.get(source, self)
980
return inter.fetch(revision_id=revision_id, pb=pb,
981
find_ghosts=find_ghosts)
983
def create_bundle(self, target, base, fileobj, format=None):
984
return serializer.write_bundle(self, target, base, fileobj, format)
986
def get_commit_builder(self, branch, parents, config, timestamp=None,
987
timezone=None, committer=None, revprops=None,
989
"""Obtain a CommitBuilder for this repository.
991
:param branch: Branch to commit to.
992
:param parents: Revision ids of the parents of the new revision.
993
:param config: Configuration to use.
994
:param timestamp: Optional timestamp recorded for commit.
995
:param timezone: Optional timezone for timestamp.
996
:param committer: Optional committer to set for commit.
997
:param revprops: Optional dictionary of revision properties.
998
:param revision_id: Optional revision id.
1000
result = self._commit_builder_class(self, parents, config,
1001
timestamp, timezone, committer, revprops, revision_id)
1002
self.start_write_group()
1006
if (self.control_files._lock_count == 1 and
1007
self.control_files._lock_mode == 'w'):
1008
if self._write_group is not None:
1009
self.abort_write_group()
1010
self.control_files.unlock()
1011
raise errors.BzrError(
1012
'Must end write groups before releasing write locks.')
1013
self.control_files.unlock()
1014
for repo in self._fallback_repositories:
1018
def clone(self, a_bzrdir, revision_id=None):
1019
"""Clone this repository into a_bzrdir using the current format.
1021
Currently no check is made that the format of this repository and
1022
the bzrdir format are compatible. FIXME RBC 20060201.
1024
:return: The newly created destination repository.
1026
# TODO: deprecate after 0.16; cloning this with all its settings is
1027
# probably not very useful -- mbp 20070423
1028
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1029
self.copy_content_into(dest_repo, revision_id)
1032
def start_write_group(self):
1033
"""Start a write group in the repository.
1035
Write groups are used by repositories which do not have a 1:1 mapping
1036
between file ids and backend store to manage the insertion of data from
1037
both fetch and commit operations.
1039
A write lock is required around the start_write_group/commit_write_group
1040
for the support of lock-requiring repository formats.
1042
One can only insert data into a repository inside a write group.
1046
if not self.is_write_locked():
1047
raise errors.NotWriteLocked(self)
1048
if self._write_group:
1049
raise errors.BzrError('already in a write group')
1050
self._start_write_group()
1051
# so we can detect unlock/relock - the write group is now entered.
1052
self._write_group = self.get_transaction()
1054
def _start_write_group(self):
1055
"""Template method for per-repository write group startup.
1057
This is called before the write group is considered to be
1062
def sprout(self, to_bzrdir, revision_id=None):
1063
"""Create a descendent repository for new development.
1065
Unlike clone, this does not copy the settings of the repository.
1067
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1068
dest_repo.fetch(self, revision_id=revision_id)
1071
def _create_sprouting_repo(self, a_bzrdir, shared):
1072
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1073
# use target default format.
1074
dest_repo = a_bzrdir.create_repository()
1076
# Most control formats need the repository to be specifically
1077
# created, but on some old all-in-one formats it's not needed
1079
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1080
except errors.UninitializableFormat:
1081
dest_repo = a_bzrdir.open_repository()
1085
def has_revision(self, revision_id):
1086
"""True if this repository has a copy of the revision."""
1087
return revision_id in self.has_revisions((revision_id,))
1090
def has_revisions(self, revision_ids):
1091
"""Probe to find out the presence of multiple revisions.
1093
:param revision_ids: An iterable of revision_ids.
1094
:return: A set of the revision_ids that were present.
1096
parent_map = self.revisions.get_parent_map(
1097
[(rev_id,) for rev_id in revision_ids])
1099
if _mod_revision.NULL_REVISION in revision_ids:
1100
result.add(_mod_revision.NULL_REVISION)
1101
result.update([key[0] for key in parent_map])
1105
def get_revision(self, revision_id):
1106
"""Return the Revision object for a named revision."""
1107
return self.get_revisions([revision_id])[0]
1110
def get_revision_reconcile(self, revision_id):
1111
"""'reconcile' helper routine that allows access to a revision always.
1113
This variant of get_revision does not cross check the weave graph
1114
against the revision one as get_revision does: but it should only
1115
be used by reconcile, or reconcile-alike commands that are correcting
1116
or testing the revision graph.
1118
return self._get_revisions([revision_id])[0]
1121
def get_revisions(self, revision_ids):
1122
"""Get many revisions at once."""
1123
return self._get_revisions(revision_ids)
1126
def _get_revisions(self, revision_ids):
1127
"""Core work logic to get many revisions without sanity checks."""
1128
for rev_id in revision_ids:
1129
if not rev_id or not isinstance(rev_id, basestring):
1130
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1131
keys = [(key,) for key in revision_ids]
1132
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1134
for record in stream:
1135
if record.storage_kind == 'absent':
1136
raise errors.NoSuchRevision(self, record.key[0])
1137
text = record.get_bytes_as('fulltext')
1138
rev = self._serializer.read_revision_from_string(text)
1139
revs[record.key[0]] = rev
1140
return [revs[revid] for revid in revision_ids]
1143
def get_revision_xml(self, revision_id):
1144
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1145
# would have already do it.
1146
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1147
rev = self.get_revision(revision_id)
1148
rev_tmp = StringIO()
1149
# the current serializer..
1150
self._serializer.write_revision(rev, rev_tmp)
1152
return rev_tmp.getvalue()
1154
def get_deltas_for_revisions(self, revisions):
1155
"""Produce a generator of revision deltas.
1157
Note that the input is a sequence of REVISIONS, not revision_ids.
1158
Trees will be held in memory until the generator exits.
1159
Each delta is relative to the revision's lefthand predecessor.
1161
required_trees = set()
1162
for revision in revisions:
1163
required_trees.add(revision.revision_id)
1164
required_trees.update(revision.parent_ids[:1])
1165
trees = dict((t.get_revision_id(), t) for
1166
t in self.revision_trees(required_trees))
1167
for revision in revisions:
1168
if not revision.parent_ids:
1169
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
1171
old_tree = trees[revision.parent_ids[0]]
1172
yield trees[revision.revision_id].changes_from(old_tree)
1175
def get_revision_delta(self, revision_id):
1176
"""Return the delta for one revision.
1178
The delta is relative to the left-hand predecessor of the
1181
r = self.get_revision(revision_id)
1182
return list(self.get_deltas_for_revisions([r]))[0]
1185
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1186
signature = gpg_strategy.sign(plaintext)
1187
self.add_signature_text(revision_id, signature)
1190
def add_signature_text(self, revision_id, signature):
1191
self.signatures.add_lines((revision_id,), (),
1192
osutils.split_lines(signature))
1194
def find_text_key_references(self):
1195
"""Find the text key references within the repository.
1197
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1198
revision_ids. Each altered file-ids has the exact revision_ids that
1199
altered it listed explicitly.
1200
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1201
to whether they were referred to by the inventory of the
1202
revision_id that they contain. The inventory texts from all present
1203
revision ids are assessed to generate this report.
1205
revision_keys = self.revisions.keys()
1206
w = self.inventories
1207
pb = ui.ui_factory.nested_progress_bar()
1209
return self._find_text_key_references_from_xml_inventory_lines(
1210
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1214
def _find_text_key_references_from_xml_inventory_lines(self,
1216
"""Core routine for extracting references to texts from inventories.
1218
This performs the translation of xml lines to revision ids.
1220
:param line_iterator: An iterator of lines, origin_version_id
1221
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1222
to whether they were referred to by the inventory of the
1223
revision_id that they contain. Note that if that revision_id was
1224
not part of the line_iterator's output then False will be given -
1225
even though it may actually refer to that key.
1227
if not self._serializer.support_altered_by_hack:
1228
raise AssertionError(
1229
"_find_text_key_references_from_xml_inventory_lines only "
1230
"supported for branches which store inventory as unnested xml"
1231
", not on %r" % self)
1234
# this code needs to read every new line in every inventory for the
1235
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1236
# not present in one of those inventories is unnecessary but not
1237
# harmful because we are filtering by the revision id marker in the
1238
# inventory lines : we only select file ids altered in one of those
1239
# revisions. We don't need to see all lines in the inventory because
1240
# only those added in an inventory in rev X can contain a revision=X
1242
unescape_revid_cache = {}
1243
unescape_fileid_cache = {}
1245
# jam 20061218 In a big fetch, this handles hundreds of thousands
1246
# of lines, so it has had a lot of inlining and optimizing done.
1247
# Sorry that it is a little bit messy.
1248
# Move several functions to be local variables, since this is a long
1250
search = self._file_ids_altered_regex.search
1251
unescape = _unescape_xml
1252
setdefault = result.setdefault
1253
for line, line_key in line_iterator:
1254
match = search(line)
1257
# One call to match.group() returning multiple items is quite a
1258
# bit faster than 2 calls to match.group() each returning 1
1259
file_id, revision_id = match.group('file_id', 'revision_id')
1261
# Inlining the cache lookups helps a lot when you make 170,000
1262
# lines and 350k ids, versus 8.4 unique ids.
1263
# Using a cache helps in 2 ways:
1264
# 1) Avoids unnecessary decoding calls
1265
# 2) Re-uses cached strings, which helps in future set and
1267
# (2) is enough that removing encoding entirely along with
1268
# the cache (so we are using plain strings) results in no
1269
# performance improvement.
1271
revision_id = unescape_revid_cache[revision_id]
1273
unescaped = unescape(revision_id)
1274
unescape_revid_cache[revision_id] = unescaped
1275
revision_id = unescaped
1277
# Note that unconditionally unescaping means that we deserialise
1278
# every fileid, which for general 'pull' is not great, but we don't
1279
# really want to have some many fulltexts that this matters anyway.
1282
file_id = unescape_fileid_cache[file_id]
1284
unescaped = unescape(file_id)
1285
unescape_fileid_cache[file_id] = unescaped
1288
key = (file_id, revision_id)
1289
setdefault(key, False)
1290
if revision_id == line_key[-1]:
1294
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1296
"""Helper routine for fileids_altered_by_revision_ids.
1298
This performs the translation of xml lines to revision ids.
1300
:param line_iterator: An iterator of lines, origin_version_id
1301
:param revision_ids: The revision ids to filter for. This should be a
1302
set or other type which supports efficient __contains__ lookups, as
1303
the revision id from each parsed line will be looked up in the
1304
revision_ids filter.
1305
:return: a dictionary mapping altered file-ids to an iterable of
1306
revision_ids. Each altered file-ids has the exact revision_ids that
1307
altered it listed explicitly.
1310
setdefault = result.setdefault
1312
self._find_text_key_references_from_xml_inventory_lines(
1313
line_iterator).iterkeys():
1314
# once data is all ensured-consistent; then this is
1315
# if revision_id == version_id
1316
if key[-1:] in revision_ids:
1317
setdefault(key[0], set()).add(key[-1])
1320
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1321
"""Find the file ids and versions affected by revisions.
1323
:param revisions: an iterable containing revision ids.
1324
:param _inv_weave: The inventory weave from this repository or None.
1325
If None, the inventory weave will be opened automatically.
1326
:return: a dictionary mapping altered file-ids to an iterable of
1327
revision_ids. Each altered file-ids has the exact revision_ids that
1328
altered it listed explicitly.
1330
selected_keys = set((revid,) for revid in revision_ids)
1331
w = _inv_weave or self.inventories
1332
pb = ui.ui_factory.nested_progress_bar()
1334
return self._find_file_ids_from_xml_inventory_lines(
1335
w.iter_lines_added_or_present_in_keys(
1336
selected_keys, pb=pb),
1341
def iter_files_bytes(self, desired_files):
1342
"""Iterate through file versions.
1344
Files will not necessarily be returned in the order they occur in
1345
desired_files. No specific order is guaranteed.
1347
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1348
value supplied by the caller as part of desired_files. It should
1349
uniquely identify the file version in the caller's context. (Examples:
1350
an index number or a TreeTransform trans_id.)
1352
bytes_iterator is an iterable of bytestrings for the file. The
1353
kind of iterable and length of the bytestrings are unspecified, but for
1354
this implementation, it is a list of bytes produced by
1355
VersionedFile.get_record_stream().
1357
:param desired_files: a list of (file_id, revision_id, identifier)
1360
transaction = self.get_transaction()
1362
for file_id, revision_id, callable_data in desired_files:
1363
text_keys[(file_id, revision_id)] = callable_data
1364
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1365
if record.storage_kind == 'absent':
1366
raise errors.RevisionNotPresent(record.key, self)
1367
yield text_keys[record.key], record.get_bytes_as('fulltext')
1369
def _generate_text_key_index(self, text_key_references=None,
1371
"""Generate a new text key index for the repository.
1373
This is an expensive function that will take considerable time to run.
1375
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1376
list of parents, also text keys. When a given key has no parents,
1377
the parents list will be [NULL_REVISION].
1379
# All revisions, to find inventory parents.
1380
if ancestors is None:
1381
graph = self.get_graph()
1382
ancestors = graph.get_parent_map(self.all_revision_ids())
1383
if text_key_references is None:
1384
text_key_references = self.find_text_key_references()
1385
pb = ui.ui_factory.nested_progress_bar()
1387
return self._do_generate_text_key_index(ancestors,
1388
text_key_references, pb)
1392
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1393
"""Helper for _generate_text_key_index to avoid deep nesting."""
1394
revision_order = tsort.topo_sort(ancestors)
1395
invalid_keys = set()
1397
for revision_id in revision_order:
1398
revision_keys[revision_id] = set()
1399
text_count = len(text_key_references)
1400
# a cache of the text keys to allow reuse; costs a dict of all the
1401
# keys, but saves a 2-tuple for every child of a given key.
1403
for text_key, valid in text_key_references.iteritems():
1405
invalid_keys.add(text_key)
1407
revision_keys[text_key[1]].add(text_key)
1408
text_key_cache[text_key] = text_key
1409
del text_key_references
1411
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1412
NULL_REVISION = _mod_revision.NULL_REVISION
1413
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1414
# too small for large or very branchy trees. However, for 55K path
1415
# trees, it would be easy to use too much memory trivially. Ideally we
1416
# could gauge this by looking at available real memory etc, but this is
1417
# always a tricky proposition.
1418
inventory_cache = lru_cache.LRUCache(10)
1419
batch_size = 10 # should be ~150MB on a 55K path tree
1420
batch_count = len(revision_order) / batch_size + 1
1422
pb.update("Calculating text parents.", processed_texts, text_count)
1423
for offset in xrange(batch_count):
1424
to_query = revision_order[offset * batch_size:(offset + 1) *
1428
for rev_tree in self.revision_trees(to_query):
1429
revision_id = rev_tree.get_revision_id()
1430
parent_ids = ancestors[revision_id]
1431
for text_key in revision_keys[revision_id]:
1432
pb.update("Calculating text parents.", processed_texts)
1433
processed_texts += 1
1434
candidate_parents = []
1435
for parent_id in parent_ids:
1436
parent_text_key = (text_key[0], parent_id)
1438
check_parent = parent_text_key not in \
1439
revision_keys[parent_id]
1441
# the parent parent_id is a ghost:
1442
check_parent = False
1443
# truncate the derived graph against this ghost.
1444
parent_text_key = None
1446
# look at the parent commit details inventories to
1447
# determine possible candidates in the per file graph.
1450
inv = inventory_cache[parent_id]
1452
inv = self.revision_tree(parent_id).inventory
1453
inventory_cache[parent_id] = inv
1454
parent_entry = inv._byid.get(text_key[0], None)
1455
if parent_entry is not None:
1457
text_key[0], parent_entry.revision)
1459
parent_text_key = None
1460
if parent_text_key is not None:
1461
candidate_parents.append(
1462
text_key_cache[parent_text_key])
1463
parent_heads = text_graph.heads(candidate_parents)
1464
new_parents = list(parent_heads)
1465
new_parents.sort(key=lambda x:candidate_parents.index(x))
1466
if new_parents == []:
1467
new_parents = [NULL_REVISION]
1468
text_index[text_key] = new_parents
1470
for text_key in invalid_keys:
1471
text_index[text_key] = [NULL_REVISION]
1474
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1475
"""Get an iterable listing the keys of all the data introduced by a set
1478
The keys will be ordered so that the corresponding items can be safely
1479
fetched and inserted in that order.
1481
:returns: An iterable producing tuples of (knit-kind, file-id,
1482
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1483
'revisions'. file-id is None unless knit-kind is 'file'.
1485
# XXX: it's a bit weird to control the inventory weave caching in this
1486
# generator. Ideally the caching would be done in fetch.py I think. Or
1487
# maybe this generator should explicitly have the contract that it
1488
# should not be iterated until the previously yielded item has been
1490
inv_w = self.inventories
1492
# file ids that changed
1493
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1495
num_file_ids = len(file_ids)
1496
for file_id, altered_versions in file_ids.iteritems():
1497
if _files_pb is not None:
1498
_files_pb.update("fetch texts", count, num_file_ids)
1500
yield ("file", file_id, altered_versions)
1501
# We're done with the files_pb. Note that it finished by the caller,
1502
# just as it was created by the caller.
1506
yield ("inventory", None, revision_ids)
1509
revisions_with_signatures = set()
1510
for rev_id in revision_ids:
1512
self.get_signature_text(rev_id)
1513
except errors.NoSuchRevision:
1517
revisions_with_signatures.add(rev_id)
1518
yield ("signatures", None, revisions_with_signatures)
1521
yield ("revisions", None, revision_ids)
1524
def get_inventory(self, revision_id):
1525
"""Get Inventory object by revision id."""
1526
return self.iter_inventories([revision_id]).next()
1528
def iter_inventories(self, revision_ids):
1529
"""Get many inventories by revision_ids.
1531
This will buffer some or all of the texts used in constructing the
1532
inventories in memory, but will only parse a single inventory at a
1535
:return: An iterator of inventories.
1537
if ((None in revision_ids)
1538
or (_mod_revision.NULL_REVISION in revision_ids)):
1539
raise ValueError('cannot get null revision inventory')
1540
return self._iter_inventories(revision_ids)
1542
def _iter_inventories(self, revision_ids):
1543
"""single-document based inventory iteration."""
1544
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1545
yield self.deserialise_inventory(revision_id, text)
1547
def _iter_inventory_xmls(self, revision_ids):
1548
keys = [(revision_id,) for revision_id in revision_ids]
1549
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1551
for record in stream:
1552
if record.storage_kind != 'absent':
1553
texts[record.key] = record.get_bytes_as('fulltext')
1555
raise errors.NoSuchRevision(self, record.key)
1557
yield texts[key], key[-1]
1559
def deserialise_inventory(self, revision_id, xml):
1560
"""Transform the xml into an inventory object.
1562
:param revision_id: The expected revision id of the inventory.
1563
:param xml: A serialised inventory.
1565
result = self._serializer.read_inventory_from_string(xml, revision_id)
1566
if result.revision_id != revision_id:
1567
raise AssertionError('revision id mismatch %s != %s' % (
1568
result.revision_id, revision_id))
1571
def serialise_inventory(self, inv):
1572
return self._serializer.write_inventory_to_string(inv)
1574
def _serialise_inventory_to_lines(self, inv):
1575
return self._serializer.write_inventory_to_lines(inv)
1577
def get_serializer_format(self):
1578
return self._serializer.format_num
1581
def get_inventory_xml(self, revision_id):
1582
"""Get inventory XML as a file object."""
1583
texts = self._iter_inventory_xmls([revision_id])
1585
text, revision_id = texts.next()
1586
except StopIteration:
1587
raise errors.HistoryMissing(self, 'inventory', revision_id)
1591
def get_inventory_sha1(self, revision_id):
1592
"""Return the sha1 hash of the inventory entry
1594
return self.get_revision(revision_id).inventory_sha1
1596
def iter_reverse_revision_history(self, revision_id):
1597
"""Iterate backwards through revision ids in the lefthand history
1599
:param revision_id: The revision id to start with. All its lefthand
1600
ancestors will be traversed.
1602
graph = self.get_graph()
1603
next_id = revision_id
1605
if next_id in (None, _mod_revision.NULL_REVISION):
1608
# Note: The following line may raise KeyError in the event of
1609
# truncated history. We decided not to have a try:except:raise
1610
# RevisionNotPresent here until we see a use for it, because of the
1611
# cost in an inner loop that is by its very nature O(history).
1612
# Robert Collins 20080326
1613
parents = graph.get_parent_map([next_id])[next_id]
1614
if len(parents) == 0:
1617
next_id = parents[0]
1620
def get_revision_inventory(self, revision_id):
1621
"""Return inventory of a past revision."""
1622
# TODO: Unify this with get_inventory()
1623
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1624
# must be the same as its revision, so this is trivial.
1625
if revision_id is None:
1626
# This does not make sense: if there is no revision,
1627
# then it is the current tree inventory surely ?!
1628
# and thus get_root_id() is something that looks at the last
1629
# commit on the branch, and the get_root_id is an inventory check.
1630
raise NotImplementedError
1631
# return Inventory(self.get_root_id())
1633
return self.get_inventory(revision_id)
1635
def is_shared(self):
1636
"""Return True if this repository is flagged as a shared repository."""
1637
raise NotImplementedError(self.is_shared)
1640
def reconcile(self, other=None, thorough=False):
1641
"""Reconcile this repository."""
1642
from bzrlib.reconcile import RepoReconciler
1643
reconciler = RepoReconciler(self, thorough=thorough)
1644
reconciler.reconcile()
1647
def _refresh_data(self):
1648
"""Helper called from lock_* to ensure coherency with disk.
1650
The default implementation does nothing; it is however possible
1651
for repositories to maintain loaded indices across multiple locks
1652
by checking inside their implementation of this method to see
1653
whether their indices are still valid. This depends of course on
1654
the disk format being validatable in this manner.
1658
def revision_tree(self, revision_id):
1659
"""Return Tree for a revision on this branch.
1661
`revision_id` may be NULL_REVISION for the empty tree revision.
1663
revision_id = _mod_revision.ensure_null(revision_id)
1664
# TODO: refactor this to use an existing revision object
1665
# so we don't need to read it in twice.
1666
if revision_id == _mod_revision.NULL_REVISION:
1667
return RevisionTree(self, Inventory(root_id=None),
1668
_mod_revision.NULL_REVISION)
1670
inv = self.get_revision_inventory(revision_id)
1671
return RevisionTree(self, inv, revision_id)
1673
def revision_trees(self, revision_ids):
1674
"""Return Tree for a revision on this branch.
1676
`revision_id` may not be None or 'null:'"""
1677
inventories = self.iter_inventories(revision_ids)
1678
for inv in inventories:
1679
yield RevisionTree(self, inv, inv.revision_id)
1682
def get_ancestry(self, revision_id, topo_sorted=True):
1683
"""Return a list of revision-ids integrated by a revision.
1685
The first element of the list is always None, indicating the origin
1686
revision. This might change when we have history horizons, or
1687
perhaps we should have a new API.
1689
This is topologically sorted.
1691
if _mod_revision.is_null(revision_id):
1693
if not self.has_revision(revision_id):
1694
raise errors.NoSuchRevision(self, revision_id)
1695
graph = self.get_graph()
1697
search = graph._make_breadth_first_searcher([revision_id])
1700
found, ghosts = search.next_with_ghosts()
1701
except StopIteration:
1704
if _mod_revision.NULL_REVISION in keys:
1705
keys.remove(_mod_revision.NULL_REVISION)
1707
parent_map = graph.get_parent_map(keys)
1708
keys = tsort.topo_sort(parent_map)
1709
return [None] + list(keys)
1712
"""Compress the data within the repository.
1714
This operation only makes sense for some repository types. For other
1715
types it should be a no-op that just returns.
1717
This stub method does not require a lock, but subclasses should use
1718
@needs_write_lock as this is a long running call its reasonable to
1719
implicitly lock for the user.
1723
@deprecated_method(one_six)
1724
def print_file(self, file, revision_id):
1725
"""Print `file` to stdout.
1727
FIXME RBC 20060125 as John Meinel points out this is a bad api
1728
- it writes to stdout, it assumes that that is valid etc. Fix
1729
by creating a new more flexible convenience function.
1731
tree = self.revision_tree(revision_id)
1732
# use inventory as it was in that revision
1733
file_id = tree.inventory.path2id(file)
1735
# TODO: jam 20060427 Write a test for this code path
1736
# it had a bug in it, and was raising the wrong
1738
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1739
tree.print_file(file_id)
1741
def get_transaction(self):
1742
return self.control_files.get_transaction()
1744
@deprecated_method(one_one)
1745
def get_parents(self, revision_ids):
1746
"""See StackedParentsProvider.get_parents"""
1747
parent_map = self.get_parent_map(revision_ids)
1748
return [parent_map.get(r, None) for r in revision_ids]
1750
def get_parent_map(self, revision_ids):
1751
"""See graph._StackedParentsProvider.get_parent_map"""
1752
# revisions index works in keys; this just works in revisions
1753
# therefore wrap and unwrap
1756
for revision_id in revision_ids:
1757
if revision_id == _mod_revision.NULL_REVISION:
1758
result[revision_id] = ()
1759
elif revision_id is None:
1760
raise ValueError('get_parent_map(None) is not valid')
1762
query_keys.append((revision_id ,))
1763
for ((revision_id,), parent_keys) in \
1764
self.revisions.get_parent_map(query_keys).iteritems():
1766
result[revision_id] = tuple(parent_revid
1767
for (parent_revid,) in parent_keys)
1769
result[revision_id] = (_mod_revision.NULL_REVISION,)
1772
def _make_parents_provider(self):
1775
def get_graph(self, other_repository=None):
1776
"""Return the graph walker for this repository format"""
1777
parents_provider = self._make_parents_provider()
1778
if (other_repository is not None and
1779
not self.has_same_location(other_repository)):
1780
parents_provider = graph._StackedParentsProvider(
1781
[parents_provider, other_repository._make_parents_provider()])
1782
return graph.Graph(parents_provider)
1784
def _get_versioned_file_checker(self):
1785
"""Return an object suitable for checking versioned files."""
1786
return _VersionedFileChecker(self)
1788
def revision_ids_to_search_result(self, result_set):
1789
"""Convert a set of revision ids to a graph SearchResult."""
1790
result_parents = set()
1791
for parents in self.get_graph().get_parent_map(
1792
result_set).itervalues():
1793
result_parents.update(parents)
1794
included_keys = result_set.intersection(result_parents)
1795
start_keys = result_set.difference(included_keys)
1796
exclude_keys = result_parents.difference(result_set)
1797
result = graph.SearchResult(start_keys, exclude_keys,
1798
len(result_set), result_set)
1802
def set_make_working_trees(self, new_value):
1803
"""Set the policy flag for making working trees when creating branches.
1805
This only applies to branches that use this repository.
1807
The default is 'True'.
1808
:param new_value: True to restore the default, False to disable making
1811
raise NotImplementedError(self.set_make_working_trees)
1813
def make_working_trees(self):
1814
"""Returns the policy for making working trees on new branches."""
1815
raise NotImplementedError(self.make_working_trees)
1818
def sign_revision(self, revision_id, gpg_strategy):
1819
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1820
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1823
def has_signature_for_revision_id(self, revision_id):
1824
"""Query for a revision signature for revision_id in the repository."""
1825
if not self.has_revision(revision_id):
1826
raise errors.NoSuchRevision(self, revision_id)
1827
sig_present = (1 == len(
1828
self.signatures.get_parent_map([(revision_id,)])))
1832
def get_signature_text(self, revision_id):
1833
"""Return the text for a signature."""
1834
stream = self.signatures.get_record_stream([(revision_id,)],
1836
record = stream.next()
1837
if record.storage_kind == 'absent':
1838
raise errors.NoSuchRevision(self, revision_id)
1839
return record.get_bytes_as('fulltext')
1842
def check(self, revision_ids=None):
1843
"""Check consistency of all history of given revision_ids.
1845
Different repository implementations should override _check().
1847
:param revision_ids: A non-empty list of revision_ids whose ancestry
1848
will be checked. Typically the last revision_id of a branch.
1850
return self._check(revision_ids)
1852
def _check(self, revision_ids):
1853
result = check.Check(self)
1857
def _warn_if_deprecated(self):
1858
global _deprecation_warning_done
1859
if _deprecation_warning_done:
1861
_deprecation_warning_done = True
1862
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1863
% (self._format, self.bzrdir.transport.base))
1865
def supports_rich_root(self):
1866
return self._format.rich_root_data
1868
def _check_ascii_revisionid(self, revision_id, method):
1869
"""Private helper for ascii-only repositories."""
1870
# weave repositories refuse to store revisionids that are non-ascii.
1871
if revision_id is not None:
1872
# weaves require ascii revision ids.
1873
if isinstance(revision_id, unicode):
1875
revision_id.encode('ascii')
1876
except UnicodeEncodeError:
1877
raise errors.NonAsciiRevisionId(method, self)
1880
revision_id.decode('ascii')
1881
except UnicodeDecodeError:
1882
raise errors.NonAsciiRevisionId(method, self)
1884
def revision_graph_can_have_wrong_parents(self):
1885
"""Is it possible for this repository to have a revision graph with
1888
If True, then this repository must also implement
1889
_find_inconsistent_revision_parents so that check and reconcile can
1890
check for inconsistencies before proceeding with other checks that may
1891
depend on the revision index being consistent.
1893
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1896
# remove these delegates a while after bzr 0.15
1897
def __make_delegated(name, from_module):
1898
def _deprecated_repository_forwarder():
1899
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1900
% (name, from_module),
1903
m = __import__(from_module, globals(), locals(), [name])
1905
return getattr(m, name)
1906
except AttributeError:
1907
raise AttributeError('module %s has no name %s'
1909
globals()[name] = _deprecated_repository_forwarder
1912
'AllInOneRepository',
1913
'WeaveMetaDirRepository',
1914
'PreSplitOutRepositoryFormat',
1915
'RepositoryFormat4',
1916
'RepositoryFormat5',
1917
'RepositoryFormat6',
1918
'RepositoryFormat7',
1920
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1924
'RepositoryFormatKnit',
1925
'RepositoryFormatKnit1',
1927
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1930
def install_revision(repository, rev, revision_tree):
1931
"""Install all revision data into a repository."""
1932
install_revisions(repository, [(rev, revision_tree, None)])
1935
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1936
"""Install all revision data into a repository.
1938
Accepts an iterable of revision, tree, signature tuples. The signature
1941
repository.start_write_group()
1943
for n, (revision, revision_tree, signature) in enumerate(iterable):
1944
_install_revision(repository, revision, revision_tree, signature)
1946
pb.update('Transferring revisions', n + 1, num_revisions)
1948
repository.abort_write_group()
1951
repository.commit_write_group()
1954
def _install_revision(repository, rev, revision_tree, signature):
1955
"""Install all revision data into a repository."""
1956
present_parents = []
1958
for p_id in rev.parent_ids:
1959
if repository.has_revision(p_id):
1960
present_parents.append(p_id)
1961
parent_trees[p_id] = repository.revision_tree(p_id)
1963
parent_trees[p_id] = repository.revision_tree(
1964
_mod_revision.NULL_REVISION)
1966
inv = revision_tree.inventory
1967
entries = inv.iter_entries()
1968
# backwards compatibility hack: skip the root id.
1969
if not repository.supports_rich_root():
1970
path, root = entries.next()
1971
if root.revision != rev.revision_id:
1972
raise errors.IncompatibleRevision(repr(repository))
1974
for path, ie in entries:
1975
text_keys[(ie.file_id, ie.revision)] = ie
1976
text_parent_map = repository.texts.get_parent_map(text_keys)
1977
missing_texts = set(text_keys) - set(text_parent_map)
1978
# Add the texts that are not already present
1979
for text_key in missing_texts:
1980
ie = text_keys[text_key]
1982
# FIXME: TODO: The following loop overlaps/duplicates that done by
1983
# commit to determine parents. There is a latent/real bug here where
1984
# the parents inserted are not those commit would do - in particular
1985
# they are not filtered by heads(). RBC, AB
1986
for revision, tree in parent_trees.iteritems():
1987
if ie.file_id not in tree:
1989
parent_id = tree.inventory[ie.file_id].revision
1990
if parent_id in text_parents:
1992
text_parents.append((ie.file_id, parent_id))
1993
lines = revision_tree.get_file(ie.file_id).readlines()
1994
repository.texts.add_lines(text_key, text_parents, lines)
1996
# install the inventory
1997
repository.add_inventory(rev.revision_id, inv, present_parents)
1998
except errors.RevisionAlreadyPresent:
2000
if signature is not None:
2001
repository.add_signature_text(rev.revision_id, signature)
2002
repository.add_revision(rev.revision_id, rev, inv)
2005
class MetaDirRepository(Repository):
2006
"""Repositories in the new meta-dir layout.
2008
:ivar _transport: Transport for access to repository control files,
2009
typically pointing to .bzr/repository.
2012
def __init__(self, _format, a_bzrdir, control_files):
2013
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2014
self._transport = control_files._transport
2016
def is_shared(self):
2017
"""Return True if this repository is flagged as a shared repository."""
2018
return self._transport.has('shared-storage')
2021
def set_make_working_trees(self, new_value):
2022
"""Set the policy flag for making working trees when creating branches.
2024
This only applies to branches that use this repository.
2026
The default is 'True'.
2027
:param new_value: True to restore the default, False to disable making
2032
self._transport.delete('no-working-trees')
2033
except errors.NoSuchFile:
2036
self._transport.put_bytes('no-working-trees', '',
2037
mode=self.bzrdir._get_file_mode())
2039
def make_working_trees(self):
2040
"""Returns the policy for making working trees on new branches."""
2041
return not self._transport.has('no-working-trees')
2044
class MetaDirVersionedFileRepository(MetaDirRepository):
2045
"""Repositories in a meta-dir, that work via versioned file objects."""
2047
def __init__(self, _format, a_bzrdir, control_files):
2048
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2052
class RepositoryFormatRegistry(registry.Registry):
2053
"""Registry of RepositoryFormats."""
2055
def get(self, format_string):
2056
r = registry.Registry.get(self, format_string)
2062
format_registry = RepositoryFormatRegistry()
2063
"""Registry of formats, indexed by their identifying format string.
2065
This can contain either format instances themselves, or classes/factories that
2066
can be called to obtain one.
2070
#####################################################################
2071
# Repository Formats
2073
class RepositoryFormat(object):
2074
"""A repository format.
2076
Formats provide three things:
2077
* An initialization routine to construct repository data on disk.
2078
* a format string which is used when the BzrDir supports versioned
2080
* an open routine which returns a Repository instance.
2082
There is one and only one Format subclass for each on-disk format. But
2083
there can be one Repository subclass that is used for several different
2084
formats. The _format attribute on a Repository instance can be used to
2085
determine the disk format.
2087
Formats are placed in an dict by their format string for reference
2088
during opening. These should be subclasses of RepositoryFormat
2091
Once a format is deprecated, just deprecate the initialize and open
2092
methods on the format class. Do not deprecate the object, as the
2093
object will be created every system load.
2095
Common instance attributes:
2096
_matchingbzrdir - the bzrdir format that the repository format was
2097
originally written to work with. This can be used if manually
2098
constructing a bzrdir and repository, or more commonly for test suite
2102
# Set to True or False in derived classes. True indicates that the format
2103
# supports ghosts gracefully.
2104
supports_ghosts = None
2105
# Can this repository be given external locations to lookup additional
2106
# data. Set to True or False in derived classes.
2107
supports_external_lookups = None
2110
return "<%s>" % self.__class__.__name__
2112
def __eq__(self, other):
2113
# format objects are generally stateless
2114
return isinstance(other, self.__class__)
2116
def __ne__(self, other):
2117
return not self == other
2120
def find_format(klass, a_bzrdir):
2121
"""Return the format for the repository object in a_bzrdir.
2123
This is used by bzr native formats that have a "format" file in
2124
the repository. Other methods may be used by different types of
2128
transport = a_bzrdir.get_repository_transport(None)
2129
format_string = transport.get("format").read()
2130
return format_registry.get(format_string)
2131
except errors.NoSuchFile:
2132
raise errors.NoRepositoryPresent(a_bzrdir)
2134
raise errors.UnknownFormatError(format=format_string,
2138
def register_format(klass, format):
2139
format_registry.register(format.get_format_string(), format)
2142
def unregister_format(klass, format):
2143
format_registry.remove(format.get_format_string())
2146
def get_default_format(klass):
2147
"""Return the current default format."""
2148
from bzrlib import bzrdir
2149
return bzrdir.format_registry.make_bzrdir('default').repository_format
2151
def get_format_string(self):
2152
"""Return the ASCII format string that identifies this format.
2154
Note that in pre format ?? repositories the format string is
2155
not permitted nor written to disk.
2157
raise NotImplementedError(self.get_format_string)
2159
def get_format_description(self):
2160
"""Return the short description for this format."""
2161
raise NotImplementedError(self.get_format_description)
2163
# TODO: this shouldn't be in the base class, it's specific to things that
2164
# use weaves or knits -- mbp 20070207
2165
def _get_versioned_file_store(self,
2170
versionedfile_class=None,
2171
versionedfile_kwargs={},
2173
if versionedfile_class is None:
2174
versionedfile_class = self._versionedfile_class
2175
weave_transport = control_files._transport.clone(name)
2176
dir_mode = control_files._dir_mode
2177
file_mode = control_files._file_mode
2178
return VersionedFileStore(weave_transport, prefixed=prefixed,
2180
file_mode=file_mode,
2181
versionedfile_class=versionedfile_class,
2182
versionedfile_kwargs=versionedfile_kwargs,
2185
def initialize(self, a_bzrdir, shared=False):
2186
"""Initialize a repository of this format in a_bzrdir.
2188
:param a_bzrdir: The bzrdir to put the new repository in it.
2189
:param shared: The repository should be initialized as a sharable one.
2190
:returns: The new repository object.
2192
This may raise UninitializableFormat if shared repository are not
2193
compatible the a_bzrdir.
2195
raise NotImplementedError(self.initialize)
2197
def is_supported(self):
2198
"""Is this format supported?
2200
Supported formats must be initializable and openable.
2201
Unsupported formats may not support initialization or committing or
2202
some other features depending on the reason for not being supported.
2206
def check_conversion_target(self, target_format):
2207
raise NotImplementedError(self.check_conversion_target)
2209
def open(self, a_bzrdir, _found=False):
2210
"""Return an instance of this format for the bzrdir a_bzrdir.
2212
_found is a private parameter, do not use it.
2214
raise NotImplementedError(self.open)
2217
class MetaDirRepositoryFormat(RepositoryFormat):
2218
"""Common base class for the new repositories using the metadir layout."""
2220
rich_root_data = False
2221
supports_tree_reference = False
2222
supports_external_lookups = False
2223
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2226
super(MetaDirRepositoryFormat, self).__init__()
2228
def _create_control_files(self, a_bzrdir):
2229
"""Create the required files and the initial control_files object."""
2230
# FIXME: RBC 20060125 don't peek under the covers
2231
# NB: no need to escape relative paths that are url safe.
2232
repository_transport = a_bzrdir.get_repository_transport(self)
2233
control_files = lockable_files.LockableFiles(repository_transport,
2234
'lock', lockdir.LockDir)
2235
control_files.create_lock()
2236
return control_files
2238
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
2239
"""Upload the initial blank content."""
2240
control_files = self._create_control_files(a_bzrdir)
2241
control_files.lock_write()
2242
transport = control_files._transport
2244
utf8_files += [('shared-storage', '')]
2246
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2247
for (filename, content_stream) in files:
2248
transport.put_file(filename, content_stream,
2249
mode=a_bzrdir._get_file_mode())
2250
for (filename, content_bytes) in utf8_files:
2251
transport.put_bytes_non_atomic(filename, content_bytes,
2252
mode=a_bzrdir._get_file_mode())
2254
control_files.unlock()
2257
# formats which have no format string are not discoverable
2258
# and not independently creatable, so are not registered. They're
2259
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2260
# needed, it's constructed directly by the BzrDir. Non-native formats where
2261
# the repository is not separately opened are similar.
2263
format_registry.register_lazy(
2264
'Bazaar-NG Repository format 7',
2265
'bzrlib.repofmt.weaverepo',
2269
format_registry.register_lazy(
2270
'Bazaar-NG Knit Repository Format 1',
2271
'bzrlib.repofmt.knitrepo',
2272
'RepositoryFormatKnit1',
2275
format_registry.register_lazy(
2276
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2277
'bzrlib.repofmt.knitrepo',
2278
'RepositoryFormatKnit3',
2281
format_registry.register_lazy(
2282
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2283
'bzrlib.repofmt.knitrepo',
2284
'RepositoryFormatKnit4',
2287
# Pack-based formats. There is one format for pre-subtrees, and one for
2288
# post-subtrees to allow ease of testing.
2289
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2290
format_registry.register_lazy(
2291
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2292
'bzrlib.repofmt.pack_repo',
2293
'RepositoryFormatKnitPack1',
2295
format_registry.register_lazy(
2296
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2297
'bzrlib.repofmt.pack_repo',
2298
'RepositoryFormatKnitPack3',
2300
format_registry.register_lazy(
2301
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2302
'bzrlib.repofmt.pack_repo',
2303
'RepositoryFormatKnitPack4',
2305
format_registry.register_lazy(
2306
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2307
'bzrlib.repofmt.pack_repo',
2308
'RepositoryFormatKnitPack5',
2310
format_registry.register_lazy(
2311
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2312
'bzrlib.repofmt.pack_repo',
2313
'RepositoryFormatKnitPack5RichRoot',
2315
format_registry.register_lazy(
2316
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2317
'bzrlib.repofmt.pack_repo',
2318
'RepositoryFormatKnitPack5RichRootBroken',
2321
# Development formats.
2323
format_registry.register_lazy(
2324
"Bazaar development format 1 (needs bzr.dev from before 1.6)\n",
2325
'bzrlib.repofmt.pack_repo',
2326
'RepositoryFormatPackDevelopment1',
2328
format_registry.register_lazy(
2329
("Bazaar development format 1 with subtree support "
2330
"(needs bzr.dev from before 1.6)\n"),
2331
'bzrlib.repofmt.pack_repo',
2332
'RepositoryFormatPackDevelopment1Subtree',
2334
# 1.6->1.7 go below here
2337
class InterRepository(InterObject):
2338
"""This class represents operations taking place between two repositories.
2340
Its instances have methods like copy_content and fetch, and contain
2341
references to the source and target repositories these operations can be
2344
Often we will provide convenience methods on 'repository' which carry out
2345
operations with another repository - they will always forward to
2346
InterRepository.get(other).method_name(parameters).
2350
"""The available optimised InterRepository types."""
2352
def copy_content(self, revision_id=None):
2353
raise NotImplementedError(self.copy_content)
2355
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2356
"""Fetch the content required to construct revision_id.
2358
The content is copied from self.source to self.target.
2360
:param revision_id: if None all content is copied, if NULL_REVISION no
2362
:param pb: optional progress bar to use for progress reports. If not
2363
provided a default one will be created.
2365
:returns: (copied_revision_count, failures).
2367
# Normally we should find a specific InterRepository subclass to do
2368
# the fetch; if nothing else then at least InterSameDataRepository.
2369
# If none of them is suitable it looks like fetching is not possible;
2370
# we try to give a good message why. _assert_same_model will probably
2371
# give a helpful message; otherwise a generic one.
2372
self._assert_same_model(self.source, self.target)
2373
raise errors.IncompatibleRepositories(self.source, self.target,
2374
"no suitableInterRepository found")
2376
def _walk_to_common_revisions(self, revision_ids):
2377
"""Walk out from revision_ids in source to revisions target has.
2379
:param revision_ids: The start point for the search.
2380
:return: A set of revision ids.
2382
target_graph = self.target.get_graph()
2383
revision_ids = frozenset(revision_ids)
2384
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2385
return graph.SearchResult(revision_ids, set(), 0, set())
2386
missing_revs = set()
2387
source_graph = self.source.get_graph()
2388
# ensure we don't pay silly lookup costs.
2389
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2390
null_set = frozenset([_mod_revision.NULL_REVISION])
2393
next_revs, ghosts = searcher.next_with_ghosts()
2394
except StopIteration:
2396
if revision_ids.intersection(ghosts):
2397
absent_ids = set(revision_ids.intersection(ghosts))
2398
# If all absent_ids are present in target, no error is needed.
2399
absent_ids.difference_update(
2400
set(target_graph.get_parent_map(absent_ids)))
2402
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2403
# we don't care about other ghosts as we can't fetch them and
2404
# haven't been asked to.
2405
next_revs = set(next_revs)
2406
# we always have NULL_REVISION present.
2407
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2408
missing_revs.update(next_revs - have_revs)
2409
searcher.stop_searching_any(have_revs)
2410
return searcher.get_result()
2412
@deprecated_method(one_two)
2414
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2415
"""Return the revision ids that source has that target does not.
2417
These are returned in topological order.
2419
:param revision_id: only return revision ids included by this
2421
:param find_ghosts: If True find missing revisions in deep history
2422
rather than just finding the surface difference.
2424
return list(self.search_missing_revision_ids(
2425
revision_id, find_ghosts).get_keys())
2428
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2429
"""Return the revision ids that source has that target does not.
2431
:param revision_id: only return revision ids included by this
2433
:param find_ghosts: If True find missing revisions in deep history
2434
rather than just finding the surface difference.
2435
:return: A bzrlib.graph.SearchResult.
2437
# stop searching at found target revisions.
2438
if not find_ghosts and revision_id is not None:
2439
return self._walk_to_common_revisions([revision_id])
2440
# generic, possibly worst case, slow code path.
2441
target_ids = set(self.target.all_revision_ids())
2442
if revision_id is not None:
2443
source_ids = self.source.get_ancestry(revision_id)
2444
if source_ids[0] is not None:
2445
raise AssertionError()
2448
source_ids = self.source.all_revision_ids()
2449
result_set = set(source_ids).difference(target_ids)
2450
return self.source.revision_ids_to_search_result(result_set)
2453
def _same_model(source, target):
2454
"""True if source and target have the same data representation.
2456
Note: this is always called on the base class; overriding it in a
2457
subclass will have no effect.
2460
InterRepository._assert_same_model(source, target)
2462
except errors.IncompatibleRepositories, e:
2466
def _assert_same_model(source, target):
2467
"""Raise an exception if two repositories do not use the same model.
2469
if source.supports_rich_root() != target.supports_rich_root():
2470
raise errors.IncompatibleRepositories(source, target,
2471
"different rich-root support")
2472
if source._serializer != target._serializer:
2473
raise errors.IncompatibleRepositories(source, target,
2474
"different serializers")
2477
class InterSameDataRepository(InterRepository):
2478
"""Code for converting between repositories that represent the same data.
2480
Data format and model must match for this to work.
2484
def _get_repo_format_to_test(self):
2485
"""Repository format for testing with.
2487
InterSameData can pull from subtree to subtree and from non-subtree to
2488
non-subtree, so we test this with the richest repository format.
2490
from bzrlib.repofmt import knitrepo
2491
return knitrepo.RepositoryFormatKnit3()
2494
def is_compatible(source, target):
2495
return InterRepository._same_model(source, target)
2498
def copy_content(self, revision_id=None):
2499
"""Make a complete copy of the content in self into destination.
2501
This copies both the repository's revision data, and configuration information
2502
such as the make_working_trees setting.
2504
This is a destructive operation! Do not use it on existing
2507
:param revision_id: Only copy the content needed to construct
2508
revision_id and its parents.
2511
self.target.set_make_working_trees(self.source.make_working_trees())
2512
except NotImplementedError:
2514
# but don't bother fetching if we have the needed data now.
2515
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2516
self.target.has_revision(revision_id)):
2518
self.target.fetch(self.source, revision_id=revision_id)
2521
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2522
"""See InterRepository.fetch()."""
2523
from bzrlib.fetch import RepoFetcher
2524
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2525
self.source, self.source._format, self.target,
2526
self.target._format)
2527
f = RepoFetcher(to_repository=self.target,
2528
from_repository=self.source,
2529
last_revision=revision_id,
2530
pb=pb, find_ghosts=find_ghosts)
2531
return f.count_copied, f.failed_revisions
2534
class InterWeaveRepo(InterSameDataRepository):
2535
"""Optimised code paths between Weave based repositories.
2537
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2538
implemented lazy inter-object optimisation.
2542
def _get_repo_format_to_test(self):
2543
from bzrlib.repofmt import weaverepo
2544
return weaverepo.RepositoryFormat7()
2547
def is_compatible(source, target):
2548
"""Be compatible with known Weave formats.
2550
We don't test for the stores being of specific types because that
2551
could lead to confusing results, and there is no need to be
2554
from bzrlib.repofmt.weaverepo import (
2560
return (isinstance(source._format, (RepositoryFormat5,
2562
RepositoryFormat7)) and
2563
isinstance(target._format, (RepositoryFormat5,
2565
RepositoryFormat7)))
2566
except AttributeError:
2570
def copy_content(self, revision_id=None):
2571
"""See InterRepository.copy_content()."""
2572
# weave specific optimised path:
2574
self.target.set_make_working_trees(self.source.make_working_trees())
2575
except (errors.RepositoryUpgradeRequired, NotImplemented):
2577
# FIXME do not peek!
2578
if self.source._transport.listable():
2579
pb = ui.ui_factory.nested_progress_bar()
2581
self.target.texts.insert_record_stream(
2582
self.source.texts.get_record_stream(
2583
self.source.texts.keys(), 'topological', False))
2584
pb.update('copying inventory', 0, 1)
2585
self.target.inventories.insert_record_stream(
2586
self.source.inventories.get_record_stream(
2587
self.source.inventories.keys(), 'topological', False))
2588
self.target.signatures.insert_record_stream(
2589
self.source.signatures.get_record_stream(
2590
self.source.signatures.keys(),
2592
self.target.revisions.insert_record_stream(
2593
self.source.revisions.get_record_stream(
2594
self.source.revisions.keys(),
2595
'topological', True))
2599
self.target.fetch(self.source, revision_id=revision_id)
2602
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2603
"""See InterRepository.fetch()."""
2604
from bzrlib.fetch import RepoFetcher
2605
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2606
self.source, self.source._format, self.target, self.target._format)
2607
f = RepoFetcher(to_repository=self.target,
2608
from_repository=self.source,
2609
last_revision=revision_id,
2610
pb=pb, find_ghosts=find_ghosts)
2611
return f.count_copied, f.failed_revisions
2614
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2615
"""See InterRepository.missing_revision_ids()."""
2616
# we want all revisions to satisfy revision_id in source.
2617
# but we don't want to stat every file here and there.
2618
# we want then, all revisions other needs to satisfy revision_id
2619
# checked, but not those that we have locally.
2620
# so the first thing is to get a subset of the revisions to
2621
# satisfy revision_id in source, and then eliminate those that
2622
# we do already have.
2623
# this is slow on high latency connection to self, but as as this
2624
# disk format scales terribly for push anyway due to rewriting
2625
# inventory.weave, this is considered acceptable.
2627
if revision_id is not None:
2628
source_ids = self.source.get_ancestry(revision_id)
2629
if source_ids[0] is not None:
2630
raise AssertionError()
2633
source_ids = self.source._all_possible_ids()
2634
source_ids_set = set(source_ids)
2635
# source_ids is the worst possible case we may need to pull.
2636
# now we want to filter source_ids against what we actually
2637
# have in target, but don't try to check for existence where we know
2638
# we do not have a revision as that would be pointless.
2639
target_ids = set(self.target._all_possible_ids())
2640
possibly_present_revisions = target_ids.intersection(source_ids_set)
2641
actually_present_revisions = set(
2642
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2643
required_revisions = source_ids_set.difference(actually_present_revisions)
2644
if revision_id is not None:
2645
# we used get_ancestry to determine source_ids then we are assured all
2646
# revisions referenced are present as they are installed in topological order.
2647
# and the tip revision was validated by get_ancestry.
2648
result_set = required_revisions
2650
# if we just grabbed the possibly available ids, then
2651
# we only have an estimate of whats available and need to validate
2652
# that against the revision records.
2654
self.source._eliminate_revisions_not_present(required_revisions))
2655
return self.source.revision_ids_to_search_result(result_set)
2658
class InterKnitRepo(InterSameDataRepository):
2659
"""Optimised code paths between Knit based repositories."""
2662
def _get_repo_format_to_test(self):
2663
from bzrlib.repofmt import knitrepo
2664
return knitrepo.RepositoryFormatKnit1()
2667
def is_compatible(source, target):
2668
"""Be compatible with known Knit formats.
2670
We don't test for the stores being of specific types because that
2671
could lead to confusing results, and there is no need to be
2674
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2676
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2677
isinstance(target._format, RepositoryFormatKnit))
2678
except AttributeError:
2680
return are_knits and InterRepository._same_model(source, target)
2683
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2684
"""See InterRepository.fetch()."""
2685
from bzrlib.fetch import RepoFetcher
2686
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2687
self.source, self.source._format, self.target, self.target._format)
2688
f = RepoFetcher(to_repository=self.target,
2689
from_repository=self.source,
2690
last_revision=revision_id,
2691
pb=pb, find_ghosts=find_ghosts)
2692
return f.count_copied, f.failed_revisions
2695
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2696
"""See InterRepository.missing_revision_ids()."""
2697
if revision_id is not None:
2698
source_ids = self.source.get_ancestry(revision_id)
2699
if source_ids[0] is not None:
2700
raise AssertionError()
2703
source_ids = self.source.all_revision_ids()
2704
source_ids_set = set(source_ids)
2705
# source_ids is the worst possible case we may need to pull.
2706
# now we want to filter source_ids against what we actually
2707
# have in target, but don't try to check for existence where we know
2708
# we do not have a revision as that would be pointless.
2709
target_ids = set(self.target.all_revision_ids())
2710
possibly_present_revisions = target_ids.intersection(source_ids_set)
2711
actually_present_revisions = set(
2712
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2713
required_revisions = source_ids_set.difference(actually_present_revisions)
2714
if revision_id is not None:
2715
# we used get_ancestry to determine source_ids then we are assured all
2716
# revisions referenced are present as they are installed in topological order.
2717
# and the tip revision was validated by get_ancestry.
2718
result_set = required_revisions
2720
# if we just grabbed the possibly available ids, then
2721
# we only have an estimate of whats available and need to validate
2722
# that against the revision records.
2724
self.source._eliminate_revisions_not_present(required_revisions))
2725
return self.source.revision_ids_to_search_result(result_set)
2728
class InterPackRepo(InterSameDataRepository):
2729
"""Optimised code paths between Pack based repositories."""
2732
def _get_repo_format_to_test(self):
2733
from bzrlib.repofmt import pack_repo
2734
return pack_repo.RepositoryFormatKnitPack1()
2737
def is_compatible(source, target):
2738
"""Be compatible with known Pack formats.
2740
We don't test for the stores being of specific types because that
2741
could lead to confusing results, and there is no need to be
2744
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2746
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2747
isinstance(target._format, RepositoryFormatPack))
2748
except AttributeError:
2750
return are_packs and InterRepository._same_model(source, target)
2753
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2754
"""See InterRepository.fetch()."""
2755
if (len(self.source._fallback_repositories) > 0 or
2756
len(self.target._fallback_repositories) > 0):
2757
# The pack layer is not aware of fallback repositories, so when
2758
# fetching from a stacked repository or into a stacked repository
2759
# we use the generic fetch logic which uses the VersionedFiles
2760
# attributes on repository.
2761
from bzrlib.fetch import RepoFetcher
2762
fetcher = RepoFetcher(self.target, self.source, revision_id,
2764
return fetcher.count_copied, fetcher.failed_revisions
2765
from bzrlib.repofmt.pack_repo import Packer
2766
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2767
self.source, self.source._format, self.target, self.target._format)
2768
self.count_copied = 0
2769
if revision_id is None:
2771
# everything to do - use pack logic
2772
# to fetch from all packs to one without
2773
# inventory parsing etc, IFF nothing to be copied is in the target.
2775
source_revision_ids = frozenset(self.source.all_revision_ids())
2776
revision_ids = source_revision_ids - \
2777
frozenset(self.target.get_parent_map(source_revision_ids))
2778
revision_keys = [(revid,) for revid in revision_ids]
2779
index = self.target._pack_collection.revision_index.combined_index
2780
present_revision_ids = set(item[1][0] for item in
2781
index.iter_entries(revision_keys))
2782
revision_ids = set(revision_ids) - present_revision_ids
2783
# implementing the TODO will involve:
2784
# - detecting when all of a pack is selected
2785
# - avoiding as much as possible pre-selection, so the
2786
# more-core routines such as create_pack_from_packs can filter in
2787
# a just-in-time fashion. (though having a HEADS list on a
2788
# repository might make this a lot easier, because we could
2789
# sensibly detect 'new revisions' without doing a full index scan.
2790
elif _mod_revision.is_null(revision_id):
2795
revision_ids = self.search_missing_revision_ids(revision_id,
2796
find_ghosts=find_ghosts).get_keys()
2797
except errors.NoSuchRevision:
2798
raise errors.InstallFailed([revision_id])
2799
if len(revision_ids) == 0:
2801
packs = self.source._pack_collection.all_packs()
2802
pack = Packer(self.target._pack_collection, packs, '.fetch',
2803
revision_ids).pack()
2804
if pack is not None:
2805
self.target._pack_collection._save_pack_names()
2806
# Trigger an autopack. This may duplicate effort as we've just done
2807
# a pack creation, but for now it is simpler to think about as
2808
# 'upload data, then repack if needed'.
2809
self.target._pack_collection.autopack()
2810
return (pack.get_revision_count(), [])
2815
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2816
"""See InterRepository.missing_revision_ids().
2818
:param find_ghosts: Find ghosts throughout the ancestry of
2821
if not find_ghosts and revision_id is not None:
2822
return self._walk_to_common_revisions([revision_id])
2823
elif revision_id is not None:
2824
# Find ghosts: search for revisions pointing from one repository to
2825
# the other, and vice versa, anywhere in the history of revision_id.
2826
graph = self.target.get_graph(other_repository=self.source)
2827
searcher = graph._make_breadth_first_searcher([revision_id])
2831
next_revs, ghosts = searcher.next_with_ghosts()
2832
except StopIteration:
2834
if revision_id in ghosts:
2835
raise errors.NoSuchRevision(self.source, revision_id)
2836
found_ids.update(next_revs)
2837
found_ids.update(ghosts)
2838
found_ids = frozenset(found_ids)
2839
# Double query here: should be able to avoid this by changing the
2840
# graph api further.
2841
result_set = found_ids - frozenset(
2842
self.target.get_parent_map(found_ids))
2844
source_ids = self.source.all_revision_ids()
2845
# source_ids is the worst possible case we may need to pull.
2846
# now we want to filter source_ids against what we actually
2847
# have in target, but don't try to check for existence where we know
2848
# we do not have a revision as that would be pointless.
2849
target_ids = set(self.target.all_revision_ids())
2850
result_set = set(source_ids).difference(target_ids)
2851
return self.source.revision_ids_to_search_result(result_set)
2854
class InterModel1and2(InterRepository):
2857
def _get_repo_format_to_test(self):
2861
def is_compatible(source, target):
2862
if not source.supports_rich_root() and target.supports_rich_root():
2868
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2869
"""See InterRepository.fetch()."""
2870
from bzrlib.fetch import Model1toKnit2Fetcher
2871
f = Model1toKnit2Fetcher(to_repository=self.target,
2872
from_repository=self.source,
2873
last_revision=revision_id,
2874
pb=pb, find_ghosts=find_ghosts)
2875
return f.count_copied, f.failed_revisions
2878
def copy_content(self, revision_id=None):
2879
"""Make a complete copy of the content in self into destination.
2881
This is a destructive operation! Do not use it on existing
2884
:param revision_id: Only copy the content needed to construct
2885
revision_id and its parents.
2888
self.target.set_make_working_trees(self.source.make_working_trees())
2889
except NotImplementedError:
2891
# but don't bother fetching if we have the needed data now.
2892
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2893
self.target.has_revision(revision_id)):
2895
self.target.fetch(self.source, revision_id=revision_id)
2898
class InterKnit1and2(InterKnitRepo):
2901
def _get_repo_format_to_test(self):
2905
def is_compatible(source, target):
2906
"""Be compatible with Knit1 source and Knit3 target"""
2908
from bzrlib.repofmt.knitrepo import (
2909
RepositoryFormatKnit1,
2910
RepositoryFormatKnit3,
2912
from bzrlib.repofmt.pack_repo import (
2913
RepositoryFormatKnitPack1,
2914
RepositoryFormatKnitPack3,
2915
RepositoryFormatKnitPack4,
2916
RepositoryFormatKnitPack5,
2917
RepositoryFormatKnitPack5RichRoot,
2918
RepositoryFormatPackDevelopment1,
2919
RepositoryFormatPackDevelopment1Subtree,
2922
RepositoryFormatKnit1, # no rr, no subtree
2923
RepositoryFormatKnitPack1, # no rr, no subtree
2924
RepositoryFormatPackDevelopment1, # no rr, no subtree
2925
RepositoryFormatKnitPack5, # no rr, no subtree
2928
RepositoryFormatKnit3, # rr, subtree
2929
RepositoryFormatKnitPack3, # rr, subtree
2930
RepositoryFormatKnitPack4, # rr, no subtree
2931
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
2932
RepositoryFormatPackDevelopment1Subtree, # rr, subtree
2934
for format in norichroot:
2935
if format.rich_root_data:
2936
raise AssertionError('Format %s is a rich-root format'
2937
' but is included in the non-rich-root list'
2939
for format in richroot:
2940
if not format.rich_root_data:
2941
raise AssertionError('Format %s is not a rich-root format'
2942
' but is included in the rich-root list'
2944
# TODO: One alternative is to just check format.rich_root_data,
2945
# instead of keeping membership lists. However, the formats
2946
# *also* have to use the same 'Knit' style of storage
2947
# (line-deltas, fulltexts, etc.)
2948
return (isinstance(source._format, norichroot) and
2949
isinstance(target._format, richroot))
2950
except AttributeError:
2954
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2955
"""See InterRepository.fetch()."""
2956
from bzrlib.fetch import Knit1to2Fetcher
2957
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2958
self.source, self.source._format, self.target,
2959
self.target._format)
2960
f = Knit1to2Fetcher(to_repository=self.target,
2961
from_repository=self.source,
2962
last_revision=revision_id,
2963
pb=pb, find_ghosts=find_ghosts)
2964
return f.count_copied, f.failed_revisions
2967
class InterDifferingSerializer(InterKnitRepo):
2970
def _get_repo_format_to_test(self):
2974
def is_compatible(source, target):
2975
"""Be compatible with Knit2 source and Knit3 target"""
2976
if source.supports_rich_root() != target.supports_rich_root():
2978
# Ideally, we'd support fetching if the source had no tree references
2979
# even if it supported them...
2980
if (getattr(source, '_format.supports_tree_reference', False) and
2981
not getattr(target, '_format.supports_tree_reference', False)):
2986
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2987
"""See InterRepository.fetch()."""
2988
revision_ids = self.target.search_missing_revision_ids(self.source,
2989
revision_id, find_ghosts=find_ghosts).get_keys()
2990
revision_ids = tsort.topo_sort(
2991
self.source.get_graph().get_parent_map(revision_ids))
2992
def revisions_iterator():
2993
for current_revision_id in revision_ids:
2994
revision = self.source.get_revision(current_revision_id)
2995
tree = self.source.revision_tree(current_revision_id)
2997
signature = self.source.get_signature_text(
2998
current_revision_id)
2999
except errors.NoSuchRevision:
3001
yield revision, tree, signature
3003
my_pb = ui.ui_factory.nested_progress_bar()
3008
install_revisions(self.target, revisions_iterator(),
3009
len(revision_ids), pb)
3011
if my_pb is not None:
3013
return len(revision_ids), 0
3016
class InterOtherToRemote(InterRepository):
3018
def __init__(self, source, target):
3019
InterRepository.__init__(self, source, target)
3020
self._real_inter = None
3023
def is_compatible(source, target):
3024
if isinstance(target, remote.RemoteRepository):
3028
def _ensure_real_inter(self):
3029
if self._real_inter is None:
3030
self.target._ensure_real()
3031
real_target = self.target._real_repository
3032
self._real_inter = InterRepository.get(self.source, real_target)
3034
def copy_content(self, revision_id=None):
3035
self._ensure_real_inter()
3036
self._real_inter.copy_content(revision_id=revision_id)
3038
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3039
self._ensure_real_inter()
3040
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3041
find_ghosts=find_ghosts)
3044
def _get_repo_format_to_test(self):
3048
class InterRemoteToOther(InterRepository):
3050
def __init__(self, source, target):
3051
InterRepository.__init__(self, source, target)
3052
self._real_inter = None
3055
def is_compatible(source, target):
3056
if not isinstance(source, remote.RemoteRepository):
3058
# Is source's model compatible with target's model?
3059
source._ensure_real()
3060
real_source = source._real_repository
3061
if isinstance(real_source, remote.RemoteRepository):
3062
raise NotImplementedError(
3063
"We don't support remote repos backed by remote repos yet.")
3064
return InterRepository._same_model(real_source, target)
3066
def _ensure_real_inter(self):
3067
if self._real_inter is None:
3068
self.source._ensure_real()
3069
real_source = self.source._real_repository
3070
self._real_inter = InterRepository.get(real_source, self.target)
3072
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3073
self._ensure_real_inter()
3074
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3075
find_ghosts=find_ghosts)
3077
def copy_content(self, revision_id=None):
3078
self._ensure_real_inter()
3079
self._real_inter.copy_content(revision_id=revision_id)
3082
def _get_repo_format_to_test(self):
3087
InterRepository.register_optimiser(InterDifferingSerializer)
3088
InterRepository.register_optimiser(InterSameDataRepository)
3089
InterRepository.register_optimiser(InterWeaveRepo)
3090
InterRepository.register_optimiser(InterKnitRepo)
3091
InterRepository.register_optimiser(InterModel1and2)
3092
InterRepository.register_optimiser(InterKnit1and2)
3093
InterRepository.register_optimiser(InterPackRepo)
3094
InterRepository.register_optimiser(InterOtherToRemote)
3095
InterRepository.register_optimiser(InterRemoteToOther)
3098
class CopyConverter(object):
3099
"""A repository conversion tool which just performs a copy of the content.
3101
This is slow but quite reliable.
3104
def __init__(self, target_format):
3105
"""Create a CopyConverter.
3107
:param target_format: The format the resulting repository should be.
3109
self.target_format = target_format
3111
def convert(self, repo, pb):
3112
"""Perform the conversion of to_convert, giving feedback via pb.
3114
:param to_convert: The disk object to convert.
3115
:param pb: a progress bar to use for progress information.
3120
# this is only useful with metadir layouts - separated repo content.
3121
# trigger an assertion if not such
3122
repo._format.get_format_string()
3123
self.repo_dir = repo.bzrdir
3124
self.step('Moving repository to repository.backup')
3125
self.repo_dir.transport.move('repository', 'repository.backup')
3126
backup_transport = self.repo_dir.transport.clone('repository.backup')
3127
repo._format.check_conversion_target(self.target_format)
3128
self.source_repo = repo._format.open(self.repo_dir,
3130
_override_transport=backup_transport)
3131
self.step('Creating new repository')
3132
converted = self.target_format.initialize(self.repo_dir,
3133
self.source_repo.is_shared())
3134
converted.lock_write()
3136
self.step('Copying content into repository.')
3137
self.source_repo.copy_content_into(converted)
3140
self.step('Deleting old repository content.')
3141
self.repo_dir.transport.delete_tree('repository.backup')
3142
self.pb.note('repository converted')
3144
def step(self, message):
3145
"""Update the pb by a step."""
3147
self.pb.update(message, self.count, self.total)
3159
def _unescaper(match, _map=_unescape_map):
3160
code = match.group(1)
3164
if not code.startswith('#'):
3166
return unichr(int(code[1:])).encode('utf8')
3172
def _unescape_xml(data):
3173
"""Unescape predefined XML entities in a string of data."""
3175
if _unescape_re is None:
3176
_unescape_re = re.compile('\&([^;]*);')
3177
return _unescape_re.sub(_unescaper, data)
3180
class _VersionedFileChecker(object):
3182
def __init__(self, repository):
3183
self.repository = repository
3184
self.text_index = self.repository._generate_text_key_index()
3186
def calculate_file_version_parents(self, text_key):
3187
"""Calculate the correct parents for a file version according to
3190
parent_keys = self.text_index[text_key]
3191
if parent_keys == [_mod_revision.NULL_REVISION]:
3193
return tuple(parent_keys)
3195
def check_file_version_parents(self, texts, progress_bar=None):
3196
"""Check the parents stored in a versioned file are correct.
3198
It also detects file versions that are not referenced by their
3199
corresponding revision's inventory.
3201
:returns: A tuple of (wrong_parents, dangling_file_versions).
3202
wrong_parents is a dict mapping {revision_id: (stored_parents,
3203
correct_parents)} for each revision_id where the stored parents
3204
are not correct. dangling_file_versions is a set of (file_id,
3205
revision_id) tuples for versions that are present in this versioned
3206
file, but not used by the corresponding inventory.
3209
self.file_ids = set([file_id for file_id, _ in
3210
self.text_index.iterkeys()])
3211
# text keys is now grouped by file_id
3212
n_weaves = len(self.file_ids)
3213
files_in_revisions = {}
3214
revisions_of_files = {}
3215
n_versions = len(self.text_index)
3216
progress_bar.update('loading text store', 0, n_versions)
3217
parent_map = self.repository.texts.get_parent_map(self.text_index)
3218
# On unlistable transports this could well be empty/error...
3219
text_keys = self.repository.texts.keys()
3220
unused_keys = frozenset(text_keys) - set(self.text_index)
3221
for num, key in enumerate(self.text_index.iterkeys()):
3222
if progress_bar is not None:
3223
progress_bar.update('checking text graph', num, n_versions)
3224
correct_parents = self.calculate_file_version_parents(key)
3226
knit_parents = parent_map[key]
3227
except errors.RevisionNotPresent:
3230
if correct_parents != knit_parents:
3231
wrong_parents[key] = (knit_parents, correct_parents)
3232
return wrong_parents, unused_keys
3235
def _old_get_graph(repository, revision_id):
3236
"""DO NOT USE. That is all. I'm serious."""
3237
graph = repository.get_graph()
3238
revision_graph = dict(((key, value) for key, value in
3239
graph.iter_ancestry([revision_id]) if value is not None))
3240
return _strip_NULL_ghosts(revision_graph)
3243
def _strip_NULL_ghosts(revision_graph):
3244
"""Also don't use this. more compatibility code for unmigrated clients."""
3245
# Filter ghosts, and null:
3246
if _mod_revision.NULL_REVISION in revision_graph:
3247
del revision_graph[_mod_revision.NULL_REVISION]
3248
for key, parents in revision_graph.items():
3249
revision_graph[key] = tuple(parent for parent in parents if parent
3251
return revision_graph
added
removed
bzrlib/repository.py
