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, fs_hash).
245
change_delta is an inventory_delta change for this entry against
246
the basis tree of the commit, or None if no change occured against
248
version_recorded is True if a new version of the entry has been
249
recorded. For instance, committing a merge where a file was only
250
changed on the other side will return (delta, False).
251
fs_hash is either None, or the hash of the path. (Currently we
252
use sha1 hashes of the entire file content, and only calculate
253
these for regular files).
255
if self.new_inventory.root is None:
256
if ie.parent_id is not None:
257
raise errors.RootMissing()
258
self._check_root(ie, parent_invs, tree)
259
if ie.revision is None:
260
kind = content_summary[0]
262
# ie is carried over from a prior commit
264
# XXX: repository specific check for nested tree support goes here - if
265
# the repo doesn't want nested trees we skip it ?
266
if (kind == 'tree-reference' and
267
not self.repository._format.supports_tree_reference):
268
# mismatch between commit builder logic and repository:
269
# this needs the entry creation pushed down into the builder.
270
raise NotImplementedError('Missing repository subtree support.')
271
self.new_inventory.add(ie)
273
# TODO: slow, take it out of the inner loop.
275
basis_inv = parent_invs[0]
277
basis_inv = Inventory(root_id=None)
279
# ie.revision is always None if the InventoryEntry is considered
280
# for committing. We may record the previous parents revision if the
281
# content is actually unchanged against a sole head.
282
if ie.revision is not None:
283
if not self._versioned_root and path == '':
284
# repositories that do not version the root set the root's
285
# revision to the new commit even when no change occurs, and
286
# this masks when a change may have occurred against the basis,
287
# so calculate if one happened.
288
if ie.file_id in basis_inv:
289
delta = (basis_inv.id2path(ie.file_id), path,
293
delta = (None, path, ie.file_id, ie)
294
return delta, False, None
296
# we don't need to commit this, because the caller already
297
# determined that an existing revision of this file is
298
# appropriate. If its not being considered for committing then
299
# it and all its parents to the root must be unaltered so
300
# no-change against the basis.
301
if ie.revision == self._new_revision_id:
302
raise AssertionError("Impossible situation, a skipped "
303
"inventory entry (%r) claims to be modified in this "
304
"commit (%r).", (ie, self._new_revision_id))
305
return None, False, None
306
# XXX: Friction: parent_candidates should return a list not a dict
307
# so that we don't have to walk the inventories again.
308
parent_candiate_entries = ie.parent_candidates(parent_invs)
309
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
311
for inv in parent_invs:
312
if ie.file_id in inv:
313
old_rev = inv[ie.file_id].revision
314
if old_rev in head_set:
315
heads.append(inv[ie.file_id].revision)
316
head_set.remove(inv[ie.file_id].revision)
319
# now we check to see if we need to write a new record to the
321
# We write a new entry unless there is one head to the ancestors, and
322
# the kind-derived content is unchanged.
324
# Cheapest check first: no ancestors, or more the one head in the
325
# ancestors, we write a new node.
329
# There is a single head, look it up for comparison
330
parent_entry = parent_candiate_entries[heads[0]]
331
# if the non-content specific data has changed, we'll be writing a
333
if (parent_entry.parent_id != ie.parent_id or
334
parent_entry.name != ie.name):
336
# now we need to do content specific checks:
338
# if the kind changed the content obviously has
339
if kind != parent_entry.kind:
342
if content_summary[2] is None:
343
raise ValueError("Files must not have executable = None")
345
if (# if the file length changed we have to store:
346
parent_entry.text_size != content_summary[1] or
347
# if the exec bit has changed we have to store:
348
parent_entry.executable != content_summary[2]):
350
elif parent_entry.text_sha1 == content_summary[3]:
351
# all meta and content is unchanged (using a hash cache
352
# hit to check the sha)
353
ie.revision = parent_entry.revision
354
ie.text_size = parent_entry.text_size
355
ie.text_sha1 = parent_entry.text_sha1
356
ie.executable = parent_entry.executable
357
return self._get_delta(ie, basis_inv, path), False, None
359
# Either there is only a hash change(no hash cache entry,
360
# or same size content change), or there is no change on
362
# Provide the parent's hash to the store layer, so that the
363
# content is unchanged we will not store a new node.
364
nostore_sha = parent_entry.text_sha1
366
# We want to record a new node regardless of the presence or
367
# absence of a content change in the file.
369
ie.executable = content_summary[2]
370
lines = tree.get_file(ie.file_id, path).readlines()
372
ie.text_sha1, ie.text_size = self._add_text_to_weave(
373
ie.file_id, lines, heads, nostore_sha)
374
except errors.ExistingContent:
375
# Turns out that the file content was unchanged, and we were
376
# only going to store a new node if it was changed. Carry over
378
ie.revision = parent_entry.revision
379
ie.text_size = parent_entry.text_size
380
ie.text_sha1 = parent_entry.text_sha1
381
ie.executable = parent_entry.executable
382
return self._get_delta(ie, basis_inv, path), False, None
383
elif kind == 'directory':
385
# all data is meta here, nothing specific to directory, so
387
ie.revision = parent_entry.revision
388
return self._get_delta(ie, basis_inv, path), False, None
390
self._add_text_to_weave(ie.file_id, lines, heads, None)
391
elif kind == 'symlink':
392
current_link_target = content_summary[3]
394
# symlink target is not generic metadata, check if it has
396
if current_link_target != parent_entry.symlink_target:
399
# unchanged, carry over.
400
ie.revision = parent_entry.revision
401
ie.symlink_target = parent_entry.symlink_target
402
return self._get_delta(ie, basis_inv, path), False, None
403
ie.symlink_target = current_link_target
405
self._add_text_to_weave(ie.file_id, lines, heads, None)
406
elif kind == 'tree-reference':
408
if content_summary[3] != parent_entry.reference_revision:
411
# unchanged, carry over.
412
ie.reference_revision = parent_entry.reference_revision
413
ie.revision = parent_entry.revision
414
return self._get_delta(ie, basis_inv, path), False, None
415
ie.reference_revision = content_summary[3]
417
self._add_text_to_weave(ie.file_id, lines, heads, None)
419
raise NotImplementedError('unknown kind')
420
ie.revision = self._new_revision_id
421
return self._get_delta(ie, basis_inv, path), True, ie.text_sha1
423
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
424
# Note: as we read the content directly from the tree, we know its not
425
# been turned into unicode or badly split - but a broken tree
426
# implementation could give us bad output from readlines() so this is
427
# not a guarantee of safety. What would be better is always checking
428
# the content during test suite execution. RBC 20070912
429
parent_keys = tuple((file_id, parent) for parent in parents)
430
return self.repository.texts.add_lines(
431
(file_id, self._new_revision_id), parent_keys, new_lines,
432
nostore_sha=nostore_sha, random_id=self.random_revid,
433
check_content=False)[0:2]
436
class RootCommitBuilder(CommitBuilder):
437
"""This commitbuilder actually records the root id"""
439
# the root entry gets versioned properly by this builder.
440
_versioned_root = True
442
def _check_root(self, ie, parent_invs, tree):
443
"""Helper for record_entry_contents.
445
:param ie: An entry being added.
446
:param parent_invs: The inventories of the parent revisions of the
448
:param tree: The tree that is being committed.
452
######################################################################
455
class Repository(object):
456
"""Repository holding history for one or more branches.
458
The repository holds and retrieves historical information including
459
revisions and file history. It's normally accessed only by the Branch,
460
which views a particular line of development through that history.
462
The Repository builds on top of some byte storage facilies (the revisions,
463
signatures, inventories and texts attributes) and a Transport, which
464
respectively provide byte storage and a means to access the (possibly
467
The byte storage facilities are addressed via tuples, which we refer to
468
as 'keys' throughout the code base. Revision_keys, inventory_keys and
469
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
470
(file_id, revision_id). We use this interface because it allows low
471
friction with the underlying code that implements disk indices, network
472
encoding and other parts of bzrlib.
474
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
475
the serialised revisions for the repository. This can be used to obtain
476
revision graph information or to access raw serialised revisions.
477
The result of trying to insert data into the repository via this store
478
is undefined: it should be considered read-only except for implementors
480
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
481
the serialised signatures for the repository. This can be used to
482
obtain access to raw serialised signatures. The result of trying to
483
insert data into the repository via this store is undefined: it should
484
be considered read-only except for implementors of repositories.
485
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
486
the serialised inventories for the repository. This can be used to
487
obtain unserialised inventories. The result of trying to insert data
488
into the repository via this store is undefined: it should be
489
considered read-only except for implementors of repositories.
490
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
491
texts of files and directories for the repository. This can be used to
492
obtain file texts or file graphs. Note that Repository.iter_file_bytes
493
is usually a better interface for accessing file texts.
494
The result of trying to insert data into the repository via this store
495
is undefined: it should be considered read-only except for implementors
497
:ivar _transport: Transport for file access to repository, typically
498
pointing to .bzr/repository.
501
# What class to use for a CommitBuilder. Often its simpler to change this
502
# in a Repository class subclass rather than to override
503
# get_commit_builder.
504
_commit_builder_class = CommitBuilder
505
# The search regex used by xml based repositories to determine what things
506
# where changed in a single commit.
507
_file_ids_altered_regex = lazy_regex.lazy_compile(
508
r'file_id="(?P<file_id>[^"]+)"'
509
r'.* revision="(?P<revision_id>[^"]+)"'
512
def abort_write_group(self):
513
"""Commit the contents accrued within the current write group.
515
:seealso: start_write_group.
517
if self._write_group is not self.get_transaction():
518
# has an unlock or relock occured ?
519
raise errors.BzrError('mismatched lock context and write group.')
520
self._abort_write_group()
521
self._write_group = None
523
def _abort_write_group(self):
524
"""Template method for per-repository write group cleanup.
526
This is called during abort before the write group is considered to be
527
finished and should cleanup any internal state accrued during the write
528
group. There is no requirement that data handed to the repository be
529
*not* made available - this is not a rollback - but neither should any
530
attempt be made to ensure that data added is fully commited. Abort is
531
invoked when an error has occured so futher disk or network operations
532
may not be possible or may error and if possible should not be
536
def add_fallback_repository(self, repository):
537
"""Add a repository to use for looking up data not held locally.
539
:param repository: A repository.
541
if not self._format.supports_external_lookups:
542
raise errors.UnstackableRepositoryFormat(self._format, self.base)
543
self._check_fallback_repository(repository)
544
self._fallback_repositories.append(repository)
545
self.texts.add_fallback_versioned_files(repository.texts)
546
self.inventories.add_fallback_versioned_files(repository.inventories)
547
self.revisions.add_fallback_versioned_files(repository.revisions)
548
self.signatures.add_fallback_versioned_files(repository.signatures)
550
def _check_fallback_repository(self, repository):
551
"""Check that this repository can fallback to repository safely.
553
Raise an error if not.
555
:param repository: A repository to fallback to.
557
return InterRepository._assert_same_model(self, repository)
559
def add_inventory(self, revision_id, inv, parents):
560
"""Add the inventory inv to the repository as revision_id.
562
:param parents: The revision ids of the parents that revision_id
563
is known to have and are in the repository already.
565
:returns: The validator(which is a sha1 digest, though what is sha'd is
566
repository format specific) of the serialized inventory.
568
if not self.is_in_write_group():
569
raise AssertionError("%r not in write group" % (self,))
570
_mod_revision.check_not_reserved_id(revision_id)
571
if not (inv.revision_id is None or inv.revision_id == revision_id):
572
raise AssertionError(
573
"Mismatch between inventory revision"
574
" id and insertion revid (%r, %r)"
575
% (inv.revision_id, revision_id))
577
raise AssertionError()
578
inv_lines = self._serialise_inventory_to_lines(inv)
579
return self._inventory_add_lines(revision_id, parents,
580
inv_lines, check_content=False)
582
def _inventory_add_lines(self, revision_id, parents, lines,
584
"""Store lines in inv_vf and return the sha1 of the inventory."""
585
parents = [(parent,) for parent in parents]
586
return self.inventories.add_lines((revision_id,), parents, lines,
587
check_content=check_content)[0]
589
def add_revision(self, revision_id, rev, inv=None, config=None):
590
"""Add rev to the revision store as revision_id.
592
:param revision_id: the revision id to use.
593
:param rev: The revision object.
594
:param inv: The inventory for the revision. if None, it will be looked
595
up in the inventory storer
596
:param config: If None no digital signature will be created.
597
If supplied its signature_needed method will be used
598
to determine if a signature should be made.
600
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
602
_mod_revision.check_not_reserved_id(revision_id)
603
if config is not None and config.signature_needed():
605
inv = self.get_inventory(revision_id)
606
plaintext = Testament(rev, inv).as_short_text()
607
self.store_revision_signature(
608
gpg.GPGStrategy(config), plaintext, revision_id)
609
# check inventory present
610
if not self.inventories.get_parent_map([(revision_id,)]):
612
raise errors.WeaveRevisionNotPresent(revision_id,
615
# yes, this is not suitable for adding with ghosts.
616
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
620
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
621
self._add_revision(rev)
623
def _add_revision(self, revision):
624
text = self._serializer.write_revision_to_string(revision)
625
key = (revision.revision_id,)
626
parents = tuple((parent,) for parent in revision.parent_ids)
627
self.revisions.add_lines(key, parents, osutils.split_lines(text))
629
def all_revision_ids(self):
630
"""Returns a list of all the revision ids in the repository.
632
This is conceptually deprecated because code should generally work on
633
the graph reachable from a particular revision, and ignore any other
634
revisions that might be present. There is no direct replacement
637
if 'evil' in debug.debug_flags:
638
mutter_callsite(2, "all_revision_ids is linear with history.")
639
return self._all_revision_ids()
641
def _all_revision_ids(self):
642
"""Returns a list of all the revision ids in the repository.
644
These are in as much topological order as the underlying store can
647
raise NotImplementedError(self._all_revision_ids)
649
def break_lock(self):
650
"""Break a lock if one is present from another instance.
652
Uses the ui factory to ask for confirmation if the lock may be from
655
self.control_files.break_lock()
658
def _eliminate_revisions_not_present(self, revision_ids):
659
"""Check every revision id in revision_ids to see if we have it.
661
Returns a set of the present revisions.
664
graph = self.get_graph()
665
parent_map = graph.get_parent_map(revision_ids)
666
# The old API returned a list, should this actually be a set?
667
return parent_map.keys()
670
def create(a_bzrdir):
671
"""Construct the current default format repository in a_bzrdir."""
672
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
674
def __init__(self, _format, a_bzrdir, control_files):
675
"""instantiate a Repository.
677
:param _format: The format of the repository on disk.
678
:param a_bzrdir: The BzrDir of the repository.
680
In the future we will have a single api for all stores for
681
getting file texts, inventories and revisions, then
682
this construct will accept instances of those things.
684
super(Repository, self).__init__()
685
self._format = _format
686
# the following are part of the public API for Repository:
687
self.bzrdir = a_bzrdir
688
self.control_files = control_files
689
self._transport = control_files._transport
690
self.base = self._transport.base
692
self._reconcile_does_inventory_gc = True
693
self._reconcile_fixes_text_parents = False
694
self._reconcile_backsup_inventory = True
695
# not right yet - should be more semantically clear ?
697
# TODO: make sure to construct the right store classes, etc, depending
698
# on whether escaping is required.
699
self._warn_if_deprecated()
700
self._write_group = None
701
# Additional places to query for data.
702
self._fallback_repositories = []
703
# What order should fetch operations request streams in?
704
# The default is unordered as that is the cheapest for an origin to
706
self._fetch_order = 'unordered'
707
# Does this repository use deltas that can be fetched as-deltas ?
708
# (E.g. knits, where the knit deltas can be transplanted intact.
709
# We default to False, which will ensure that enough data to get
710
# a full text out of any fetch stream will be grabbed.
711
self._fetch_uses_deltas = False
712
# Should fetch trigger a reconcile after the fetch? Only needed for
713
# some repository formats that can suffer internal inconsistencies.
714
self._fetch_reconcile = False
717
return '%s(%r)' % (self.__class__.__name__,
720
def has_same_location(self, other):
721
"""Returns a boolean indicating if this repository is at the same
722
location as another repository.
724
This might return False even when two repository objects are accessing
725
the same physical repository via different URLs.
727
if self.__class__ is not other.__class__:
729
return (self._transport.base == other._transport.base)
731
def is_in_write_group(self):
732
"""Return True if there is an open write group.
734
:seealso: start_write_group.
736
return self._write_group is not None
739
return self.control_files.is_locked()
741
def is_write_locked(self):
742
"""Return True if this object is write locked."""
743
return self.is_locked() and self.control_files._lock_mode == 'w'
745
def lock_write(self, token=None):
746
"""Lock this repository for writing.
748
This causes caching within the repository obejct to start accumlating
749
data during reads, and allows a 'write_group' to be obtained. Write
750
groups must be used for actual data insertion.
752
:param token: if this is already locked, then lock_write will fail
753
unless the token matches the existing lock.
754
:returns: a token if this instance supports tokens, otherwise None.
755
:raises TokenLockingNotSupported: when a token is given but this
756
instance doesn't support using token locks.
757
:raises MismatchedToken: if the specified token doesn't match the token
758
of the existing lock.
759
:seealso: start_write_group.
761
A token should be passed in if you know that you have locked the object
762
some other way, and need to synchronise this object's state with that
765
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
767
result = self.control_files.lock_write(token=token)
768
for repo in self._fallback_repositories:
769
# Writes don't affect fallback repos
775
self.control_files.lock_read()
776
for repo in self._fallback_repositories:
780
def get_physical_lock_status(self):
781
return self.control_files.get_physical_lock_status()
783
def leave_lock_in_place(self):
784
"""Tell this repository not to release the physical lock when this
787
If lock_write doesn't return a token, then this method is not supported.
789
self.control_files.leave_in_place()
791
def dont_leave_lock_in_place(self):
792
"""Tell this repository to release the physical lock when this
793
object is unlocked, even if it didn't originally acquire it.
795
If lock_write doesn't return a token, then this method is not supported.
797
self.control_files.dont_leave_in_place()
800
def gather_stats(self, revid=None, committers=None):
801
"""Gather statistics from a revision id.
803
:param revid: The revision id to gather statistics from, if None, then
804
no revision specific statistics are gathered.
805
:param committers: Optional parameter controlling whether to grab
806
a count of committers from the revision specific statistics.
807
:return: A dictionary of statistics. Currently this contains:
808
committers: The number of committers if requested.
809
firstrev: A tuple with timestamp, timezone for the penultimate left
810
most ancestor of revid, if revid is not the NULL_REVISION.
811
latestrev: A tuple with timestamp, timezone for revid, if revid is
812
not the NULL_REVISION.
813
revisions: The total revision count in the repository.
814
size: An estimate disk size of the repository in bytes.
817
if revid and committers:
818
result['committers'] = 0
819
if revid and revid != _mod_revision.NULL_REVISION:
821
all_committers = set()
822
revisions = self.get_ancestry(revid)
823
# pop the leading None
825
first_revision = None
827
# ignore the revisions in the middle - just grab first and last
828
revisions = revisions[0], revisions[-1]
829
for revision in self.get_revisions(revisions):
830
if not first_revision:
831
first_revision = revision
833
all_committers.add(revision.committer)
834
last_revision = revision
836
result['committers'] = len(all_committers)
837
result['firstrev'] = (first_revision.timestamp,
838
first_revision.timezone)
839
result['latestrev'] = (last_revision.timestamp,
840
last_revision.timezone)
842
# now gather global repository information
843
# XXX: This is available for many repos regardless of listability.
844
if self.bzrdir.root_transport.listable():
845
# XXX: do we want to __define len__() ?
846
# Maybe the versionedfiles object should provide a different
847
# method to get the number of keys.
848
result['revisions'] = len(self.revisions.keys())
852
def find_branches(self, using=False):
853
"""Find branches underneath this repository.
855
This will include branches inside other branches.
857
:param using: If True, list only branches using this repository.
859
if using and not self.is_shared():
861
return [self.bzrdir.open_branch()]
862
except errors.NotBranchError:
864
class Evaluator(object):
867
self.first_call = True
869
def __call__(self, bzrdir):
870
# On the first call, the parameter is always the bzrdir
871
# containing the current repo.
872
if not self.first_call:
874
repository = bzrdir.open_repository()
875
except errors.NoRepositoryPresent:
878
return False, (None, repository)
879
self.first_call = False
881
value = (bzrdir.open_branch(), None)
882
except errors.NotBranchError:
887
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
888
self.bzrdir.root_transport, evaluate=Evaluator()):
889
if branch is not None:
890
branches.append(branch)
891
if not using and repository is not None:
892
branches.extend(repository.find_branches())
896
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
897
"""Return the revision ids that other has that this does not.
899
These are returned in topological order.
901
revision_id: only return revision ids included by revision_id.
903
return InterRepository.get(other, self).search_missing_revision_ids(
904
revision_id, find_ghosts)
906
@deprecated_method(one_two)
908
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
909
"""Return the revision ids that other has that this does not.
911
These are returned in topological order.
913
revision_id: only return revision ids included by revision_id.
915
keys = self.search_missing_revision_ids(
916
other, revision_id, find_ghosts).get_keys()
919
parents = other.get_graph().get_parent_map(keys)
922
return tsort.topo_sort(parents)
926
"""Open the repository rooted at base.
928
For instance, if the repository is at URL/.bzr/repository,
929
Repository.open(URL) -> a Repository instance.
931
control = bzrdir.BzrDir.open(base)
932
return control.open_repository()
934
def copy_content_into(self, destination, revision_id=None):
935
"""Make a complete copy of the content in self into destination.
937
This is a destructive operation! Do not use it on existing
940
return InterRepository.get(self, destination).copy_content(revision_id)
942
def commit_write_group(self):
943
"""Commit the contents accrued within the current write group.
945
:seealso: start_write_group.
947
if self._write_group is not self.get_transaction():
948
# has an unlock or relock occured ?
949
raise errors.BzrError('mismatched lock context %r and '
951
(self.get_transaction(), self._write_group))
952
self._commit_write_group()
953
self._write_group = None
955
def _commit_write_group(self):
956
"""Template method for per-repository write group cleanup.
958
This is called before the write group is considered to be
959
finished and should ensure that all data handed to the repository
960
for writing during the write group is safely committed (to the
961
extent possible considering file system caching etc).
964
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
965
"""Fetch the content required to construct revision_id from source.
967
If revision_id is None all content is copied.
968
:param find_ghosts: Find and copy revisions in the source that are
969
ghosts in the target (and not reachable directly by walking out to
970
the first-present revision in target from revision_id).
972
# fast path same-url fetch operations
973
if self.has_same_location(source):
974
# check that last_revision is in 'from' and then return a
976
if (revision_id is not None and
977
not _mod_revision.is_null(revision_id)):
978
self.get_revision(revision_id)
980
# if there is no specific appropriate InterRepository, this will get
981
# the InterRepository base class, which raises an
982
# IncompatibleRepositories when asked to fetch.
983
inter = InterRepository.get(source, self)
984
return inter.fetch(revision_id=revision_id, pb=pb,
985
find_ghosts=find_ghosts)
987
def create_bundle(self, target, base, fileobj, format=None):
988
return serializer.write_bundle(self, target, base, fileobj, format)
990
def get_commit_builder(self, branch, parents, config, timestamp=None,
991
timezone=None, committer=None, revprops=None,
993
"""Obtain a CommitBuilder for this repository.
995
:param branch: Branch to commit to.
996
:param parents: Revision ids of the parents of the new revision.
997
:param config: Configuration to use.
998
:param timestamp: Optional timestamp recorded for commit.
999
:param timezone: Optional timezone for timestamp.
1000
:param committer: Optional committer to set for commit.
1001
:param revprops: Optional dictionary of revision properties.
1002
:param revision_id: Optional revision id.
1004
result = self._commit_builder_class(self, parents, config,
1005
timestamp, timezone, committer, revprops, revision_id)
1006
self.start_write_group()
1010
if (self.control_files._lock_count == 1 and
1011
self.control_files._lock_mode == 'w'):
1012
if self._write_group is not None:
1013
self.abort_write_group()
1014
self.control_files.unlock()
1015
raise errors.BzrError(
1016
'Must end write groups before releasing write locks.')
1017
self.control_files.unlock()
1018
for repo in self._fallback_repositories:
1022
def clone(self, a_bzrdir, revision_id=None):
1023
"""Clone this repository into a_bzrdir using the current format.
1025
Currently no check is made that the format of this repository and
1026
the bzrdir format are compatible. FIXME RBC 20060201.
1028
:return: The newly created destination repository.
1030
# TODO: deprecate after 0.16; cloning this with all its settings is
1031
# probably not very useful -- mbp 20070423
1032
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1033
self.copy_content_into(dest_repo, revision_id)
1036
def start_write_group(self):
1037
"""Start a write group in the repository.
1039
Write groups are used by repositories which do not have a 1:1 mapping
1040
between file ids and backend store to manage the insertion of data from
1041
both fetch and commit operations.
1043
A write lock is required around the start_write_group/commit_write_group
1044
for the support of lock-requiring repository formats.
1046
One can only insert data into a repository inside a write group.
1050
if not self.is_write_locked():
1051
raise errors.NotWriteLocked(self)
1052
if self._write_group:
1053
raise errors.BzrError('already in a write group')
1054
self._start_write_group()
1055
# so we can detect unlock/relock - the write group is now entered.
1056
self._write_group = self.get_transaction()
1058
def _start_write_group(self):
1059
"""Template method for per-repository write group startup.
1061
This is called before the write group is considered to be
1066
def sprout(self, to_bzrdir, revision_id=None):
1067
"""Create a descendent repository for new development.
1069
Unlike clone, this does not copy the settings of the repository.
1071
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1072
dest_repo.fetch(self, revision_id=revision_id)
1075
def _create_sprouting_repo(self, a_bzrdir, shared):
1076
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1077
# use target default format.
1078
dest_repo = a_bzrdir.create_repository()
1080
# Most control formats need the repository to be specifically
1081
# created, but on some old all-in-one formats it's not needed
1083
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1084
except errors.UninitializableFormat:
1085
dest_repo = a_bzrdir.open_repository()
1089
def has_revision(self, revision_id):
1090
"""True if this repository has a copy of the revision."""
1091
return revision_id in self.has_revisions((revision_id,))
1094
def has_revisions(self, revision_ids):
1095
"""Probe to find out the presence of multiple revisions.
1097
:param revision_ids: An iterable of revision_ids.
1098
:return: A set of the revision_ids that were present.
1100
parent_map = self.revisions.get_parent_map(
1101
[(rev_id,) for rev_id in revision_ids])
1103
if _mod_revision.NULL_REVISION in revision_ids:
1104
result.add(_mod_revision.NULL_REVISION)
1105
result.update([key[0] for key in parent_map])
1109
def get_revision(self, revision_id):
1110
"""Return the Revision object for a named revision."""
1111
return self.get_revisions([revision_id])[0]
1114
def get_revision_reconcile(self, revision_id):
1115
"""'reconcile' helper routine that allows access to a revision always.
1117
This variant of get_revision does not cross check the weave graph
1118
against the revision one as get_revision does: but it should only
1119
be used by reconcile, or reconcile-alike commands that are correcting
1120
or testing the revision graph.
1122
return self._get_revisions([revision_id])[0]
1125
def get_revisions(self, revision_ids):
1126
"""Get many revisions at once."""
1127
return self._get_revisions(revision_ids)
1130
def _get_revisions(self, revision_ids):
1131
"""Core work logic to get many revisions without sanity checks."""
1132
for rev_id in revision_ids:
1133
if not rev_id or not isinstance(rev_id, basestring):
1134
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1135
keys = [(key,) for key in revision_ids]
1136
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1138
for record in stream:
1139
if record.storage_kind == 'absent':
1140
raise errors.NoSuchRevision(self, record.key[0])
1141
text = record.get_bytes_as('fulltext')
1142
rev = self._serializer.read_revision_from_string(text)
1143
revs[record.key[0]] = rev
1144
return [revs[revid] for revid in revision_ids]
1147
def get_revision_xml(self, revision_id):
1148
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1149
# would have already do it.
1150
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1151
rev = self.get_revision(revision_id)
1152
rev_tmp = StringIO()
1153
# the current serializer..
1154
self._serializer.write_revision(rev, rev_tmp)
1156
return rev_tmp.getvalue()
1158
def get_deltas_for_revisions(self, revisions):
1159
"""Produce a generator of revision deltas.
1161
Note that the input is a sequence of REVISIONS, not revision_ids.
1162
Trees will be held in memory until the generator exits.
1163
Each delta is relative to the revision's lefthand predecessor.
1165
required_trees = set()
1166
for revision in revisions:
1167
required_trees.add(revision.revision_id)
1168
required_trees.update(revision.parent_ids[:1])
1169
trees = dict((t.get_revision_id(), t) for
1170
t in self.revision_trees(required_trees))
1171
for revision in revisions:
1172
if not revision.parent_ids:
1173
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
1175
old_tree = trees[revision.parent_ids[0]]
1176
yield trees[revision.revision_id].changes_from(old_tree)
1179
def get_revision_delta(self, revision_id):
1180
"""Return the delta for one revision.
1182
The delta is relative to the left-hand predecessor of the
1185
r = self.get_revision(revision_id)
1186
return list(self.get_deltas_for_revisions([r]))[0]
1189
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1190
signature = gpg_strategy.sign(plaintext)
1191
self.add_signature_text(revision_id, signature)
1194
def add_signature_text(self, revision_id, signature):
1195
self.signatures.add_lines((revision_id,), (),
1196
osutils.split_lines(signature))
1198
def find_text_key_references(self):
1199
"""Find the text key references within the repository.
1201
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1202
revision_ids. Each altered file-ids has the exact revision_ids that
1203
altered it listed explicitly.
1204
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1205
to whether they were referred to by the inventory of the
1206
revision_id that they contain. The inventory texts from all present
1207
revision ids are assessed to generate this report.
1209
revision_keys = self.revisions.keys()
1210
w = self.inventories
1211
pb = ui.ui_factory.nested_progress_bar()
1213
return self._find_text_key_references_from_xml_inventory_lines(
1214
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1218
def _find_text_key_references_from_xml_inventory_lines(self,
1220
"""Core routine for extracting references to texts from inventories.
1222
This performs the translation of xml lines to revision ids.
1224
:param line_iterator: An iterator of lines, origin_version_id
1225
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1226
to whether they were referred to by the inventory of the
1227
revision_id that they contain. Note that if that revision_id was
1228
not part of the line_iterator's output then False will be given -
1229
even though it may actually refer to that key.
1231
if not self._serializer.support_altered_by_hack:
1232
raise AssertionError(
1233
"_find_text_key_references_from_xml_inventory_lines only "
1234
"supported for branches which store inventory as unnested xml"
1235
", not on %r" % self)
1238
# this code needs to read every new line in every inventory for the
1239
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1240
# not present in one of those inventories is unnecessary but not
1241
# harmful because we are filtering by the revision id marker in the
1242
# inventory lines : we only select file ids altered in one of those
1243
# revisions. We don't need to see all lines in the inventory because
1244
# only those added in an inventory in rev X can contain a revision=X
1246
unescape_revid_cache = {}
1247
unescape_fileid_cache = {}
1249
# jam 20061218 In a big fetch, this handles hundreds of thousands
1250
# of lines, so it has had a lot of inlining and optimizing done.
1251
# Sorry that it is a little bit messy.
1252
# Move several functions to be local variables, since this is a long
1254
search = self._file_ids_altered_regex.search
1255
unescape = _unescape_xml
1256
setdefault = result.setdefault
1257
for line, line_key in line_iterator:
1258
match = search(line)
1261
# One call to match.group() returning multiple items is quite a
1262
# bit faster than 2 calls to match.group() each returning 1
1263
file_id, revision_id = match.group('file_id', 'revision_id')
1265
# Inlining the cache lookups helps a lot when you make 170,000
1266
# lines and 350k ids, versus 8.4 unique ids.
1267
# Using a cache helps in 2 ways:
1268
# 1) Avoids unnecessary decoding calls
1269
# 2) Re-uses cached strings, which helps in future set and
1271
# (2) is enough that removing encoding entirely along with
1272
# the cache (so we are using plain strings) results in no
1273
# performance improvement.
1275
revision_id = unescape_revid_cache[revision_id]
1277
unescaped = unescape(revision_id)
1278
unescape_revid_cache[revision_id] = unescaped
1279
revision_id = unescaped
1281
# Note that unconditionally unescaping means that we deserialise
1282
# every fileid, which for general 'pull' is not great, but we don't
1283
# really want to have some many fulltexts that this matters anyway.
1286
file_id = unescape_fileid_cache[file_id]
1288
unescaped = unescape(file_id)
1289
unescape_fileid_cache[file_id] = unescaped
1292
key = (file_id, revision_id)
1293
setdefault(key, False)
1294
if revision_id == line_key[-1]:
1298
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1300
"""Helper routine for fileids_altered_by_revision_ids.
1302
This performs the translation of xml lines to revision ids.
1304
:param line_iterator: An iterator of lines, origin_version_id
1305
:param revision_ids: The revision ids to filter for. This should be a
1306
set or other type which supports efficient __contains__ lookups, as
1307
the revision id from each parsed line will be looked up in the
1308
revision_ids filter.
1309
:return: a dictionary mapping altered file-ids to an iterable of
1310
revision_ids. Each altered file-ids has the exact revision_ids that
1311
altered it listed explicitly.
1314
setdefault = result.setdefault
1316
self._find_text_key_references_from_xml_inventory_lines(
1317
line_iterator).iterkeys():
1318
# once data is all ensured-consistent; then this is
1319
# if revision_id == version_id
1320
if key[-1:] in revision_ids:
1321
setdefault(key[0], set()).add(key[-1])
1324
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1325
"""Find the file ids and versions affected by revisions.
1327
:param revisions: an iterable containing revision ids.
1328
:param _inv_weave: The inventory weave from this repository or None.
1329
If None, the inventory weave will be opened automatically.
1330
:return: a dictionary mapping altered file-ids to an iterable of
1331
revision_ids. Each altered file-ids has the exact revision_ids that
1332
altered it listed explicitly.
1334
selected_keys = set((revid,) for revid in revision_ids)
1335
w = _inv_weave or self.inventories
1336
pb = ui.ui_factory.nested_progress_bar()
1338
return self._find_file_ids_from_xml_inventory_lines(
1339
w.iter_lines_added_or_present_in_keys(
1340
selected_keys, pb=pb),
1345
def iter_files_bytes(self, desired_files):
1346
"""Iterate through file versions.
1348
Files will not necessarily be returned in the order they occur in
1349
desired_files. No specific order is guaranteed.
1351
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1352
value supplied by the caller as part of desired_files. It should
1353
uniquely identify the file version in the caller's context. (Examples:
1354
an index number or a TreeTransform trans_id.)
1356
bytes_iterator is an iterable of bytestrings for the file. The
1357
kind of iterable and length of the bytestrings are unspecified, but for
1358
this implementation, it is a list of bytes produced by
1359
VersionedFile.get_record_stream().
1361
:param desired_files: a list of (file_id, revision_id, identifier)
1364
transaction = self.get_transaction()
1366
for file_id, revision_id, callable_data in desired_files:
1367
text_keys[(file_id, revision_id)] = callable_data
1368
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1369
if record.storage_kind == 'absent':
1370
raise errors.RevisionNotPresent(record.key, self)
1371
yield text_keys[record.key], record.get_bytes_as('fulltext')
1373
def _generate_text_key_index(self, text_key_references=None,
1375
"""Generate a new text key index for the repository.
1377
This is an expensive function that will take considerable time to run.
1379
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1380
list of parents, also text keys. When a given key has no parents,
1381
the parents list will be [NULL_REVISION].
1383
# All revisions, to find inventory parents.
1384
if ancestors is None:
1385
graph = self.get_graph()
1386
ancestors = graph.get_parent_map(self.all_revision_ids())
1387
if text_key_references is None:
1388
text_key_references = self.find_text_key_references()
1389
pb = ui.ui_factory.nested_progress_bar()
1391
return self._do_generate_text_key_index(ancestors,
1392
text_key_references, pb)
1396
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1397
"""Helper for _generate_text_key_index to avoid deep nesting."""
1398
revision_order = tsort.topo_sort(ancestors)
1399
invalid_keys = set()
1401
for revision_id in revision_order:
1402
revision_keys[revision_id] = set()
1403
text_count = len(text_key_references)
1404
# a cache of the text keys to allow reuse; costs a dict of all the
1405
# keys, but saves a 2-tuple for every child of a given key.
1407
for text_key, valid in text_key_references.iteritems():
1409
invalid_keys.add(text_key)
1411
revision_keys[text_key[1]].add(text_key)
1412
text_key_cache[text_key] = text_key
1413
del text_key_references
1415
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1416
NULL_REVISION = _mod_revision.NULL_REVISION
1417
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1418
# too small for large or very branchy trees. However, for 55K path
1419
# trees, it would be easy to use too much memory trivially. Ideally we
1420
# could gauge this by looking at available real memory etc, but this is
1421
# always a tricky proposition.
1422
inventory_cache = lru_cache.LRUCache(10)
1423
batch_size = 10 # should be ~150MB on a 55K path tree
1424
batch_count = len(revision_order) / batch_size + 1
1426
pb.update("Calculating text parents.", processed_texts, text_count)
1427
for offset in xrange(batch_count):
1428
to_query = revision_order[offset * batch_size:(offset + 1) *
1432
for rev_tree in self.revision_trees(to_query):
1433
revision_id = rev_tree.get_revision_id()
1434
parent_ids = ancestors[revision_id]
1435
for text_key in revision_keys[revision_id]:
1436
pb.update("Calculating text parents.", processed_texts)
1437
processed_texts += 1
1438
candidate_parents = []
1439
for parent_id in parent_ids:
1440
parent_text_key = (text_key[0], parent_id)
1442
check_parent = parent_text_key not in \
1443
revision_keys[parent_id]
1445
# the parent parent_id is a ghost:
1446
check_parent = False
1447
# truncate the derived graph against this ghost.
1448
parent_text_key = None
1450
# look at the parent commit details inventories to
1451
# determine possible candidates in the per file graph.
1454
inv = inventory_cache[parent_id]
1456
inv = self.revision_tree(parent_id).inventory
1457
inventory_cache[parent_id] = inv
1458
parent_entry = inv._byid.get(text_key[0], None)
1459
if parent_entry is not None:
1461
text_key[0], parent_entry.revision)
1463
parent_text_key = None
1464
if parent_text_key is not None:
1465
candidate_parents.append(
1466
text_key_cache[parent_text_key])
1467
parent_heads = text_graph.heads(candidate_parents)
1468
new_parents = list(parent_heads)
1469
new_parents.sort(key=lambda x:candidate_parents.index(x))
1470
if new_parents == []:
1471
new_parents = [NULL_REVISION]
1472
text_index[text_key] = new_parents
1474
for text_key in invalid_keys:
1475
text_index[text_key] = [NULL_REVISION]
1478
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1479
"""Get an iterable listing the keys of all the data introduced by a set
1482
The keys will be ordered so that the corresponding items can be safely
1483
fetched and inserted in that order.
1485
:returns: An iterable producing tuples of (knit-kind, file-id,
1486
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1487
'revisions'. file-id is None unless knit-kind is 'file'.
1489
# XXX: it's a bit weird to control the inventory weave caching in this
1490
# generator. Ideally the caching would be done in fetch.py I think. Or
1491
# maybe this generator should explicitly have the contract that it
1492
# should not be iterated until the previously yielded item has been
1494
inv_w = self.inventories
1496
# file ids that changed
1497
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1499
num_file_ids = len(file_ids)
1500
for file_id, altered_versions in file_ids.iteritems():
1501
if _files_pb is not None:
1502
_files_pb.update("fetch texts", count, num_file_ids)
1504
yield ("file", file_id, altered_versions)
1505
# We're done with the files_pb. Note that it finished by the caller,
1506
# just as it was created by the caller.
1510
yield ("inventory", None, revision_ids)
1513
revisions_with_signatures = set()
1514
for rev_id in revision_ids:
1516
self.get_signature_text(rev_id)
1517
except errors.NoSuchRevision:
1521
revisions_with_signatures.add(rev_id)
1522
yield ("signatures", None, revisions_with_signatures)
1525
yield ("revisions", None, revision_ids)
1528
def get_inventory(self, revision_id):
1529
"""Get Inventory object by revision id."""
1530
return self.iter_inventories([revision_id]).next()
1532
def iter_inventories(self, revision_ids):
1533
"""Get many inventories by revision_ids.
1535
This will buffer some or all of the texts used in constructing the
1536
inventories in memory, but will only parse a single inventory at a
1539
:return: An iterator of inventories.
1541
if ((None in revision_ids)
1542
or (_mod_revision.NULL_REVISION in revision_ids)):
1543
raise ValueError('cannot get null revision inventory')
1544
return self._iter_inventories(revision_ids)
1546
def _iter_inventories(self, revision_ids):
1547
"""single-document based inventory iteration."""
1548
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1549
yield self.deserialise_inventory(revision_id, text)
1551
def _iter_inventory_xmls(self, revision_ids):
1552
keys = [(revision_id,) for revision_id in revision_ids]
1553
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1555
for record in stream:
1556
if record.storage_kind != 'absent':
1557
texts[record.key] = record.get_bytes_as('fulltext')
1559
raise errors.NoSuchRevision(self, record.key)
1561
yield texts[key], key[-1]
1563
def deserialise_inventory(self, revision_id, xml):
1564
"""Transform the xml into an inventory object.
1566
:param revision_id: The expected revision id of the inventory.
1567
:param xml: A serialised inventory.
1569
result = self._serializer.read_inventory_from_string(xml, revision_id)
1570
if result.revision_id != revision_id:
1571
raise AssertionError('revision id mismatch %s != %s' % (
1572
result.revision_id, revision_id))
1575
def serialise_inventory(self, inv):
1576
return self._serializer.write_inventory_to_string(inv)
1578
def _serialise_inventory_to_lines(self, inv):
1579
return self._serializer.write_inventory_to_lines(inv)
1581
def get_serializer_format(self):
1582
return self._serializer.format_num
1585
def get_inventory_xml(self, revision_id):
1586
"""Get inventory XML as a file object."""
1587
texts = self._iter_inventory_xmls([revision_id])
1589
text, revision_id = texts.next()
1590
except StopIteration:
1591
raise errors.HistoryMissing(self, 'inventory', revision_id)
1595
def get_inventory_sha1(self, revision_id):
1596
"""Return the sha1 hash of the inventory entry
1598
return self.get_revision(revision_id).inventory_sha1
1600
def iter_reverse_revision_history(self, revision_id):
1601
"""Iterate backwards through revision ids in the lefthand history
1603
:param revision_id: The revision id to start with. All its lefthand
1604
ancestors will be traversed.
1606
graph = self.get_graph()
1607
next_id = revision_id
1609
if next_id in (None, _mod_revision.NULL_REVISION):
1612
# Note: The following line may raise KeyError in the event of
1613
# truncated history. We decided not to have a try:except:raise
1614
# RevisionNotPresent here until we see a use for it, because of the
1615
# cost in an inner loop that is by its very nature O(history).
1616
# Robert Collins 20080326
1617
parents = graph.get_parent_map([next_id])[next_id]
1618
if len(parents) == 0:
1621
next_id = parents[0]
1624
def get_revision_inventory(self, revision_id):
1625
"""Return inventory of a past revision."""
1626
# TODO: Unify this with get_inventory()
1627
# bzr 0.0.6 and later imposes the constraint that the inventory_id
1628
# must be the same as its revision, so this is trivial.
1629
if revision_id is None:
1630
# This does not make sense: if there is no revision,
1631
# then it is the current tree inventory surely ?!
1632
# and thus get_root_id() is something that looks at the last
1633
# commit on the branch, and the get_root_id is an inventory check.
1634
raise NotImplementedError
1635
# return Inventory(self.get_root_id())
1637
return self.get_inventory(revision_id)
1639
def is_shared(self):
1640
"""Return True if this repository is flagged as a shared repository."""
1641
raise NotImplementedError(self.is_shared)
1644
def reconcile(self, other=None, thorough=False):
1645
"""Reconcile this repository."""
1646
from bzrlib.reconcile import RepoReconciler
1647
reconciler = RepoReconciler(self, thorough=thorough)
1648
reconciler.reconcile()
1651
def _refresh_data(self):
1652
"""Helper called from lock_* to ensure coherency with disk.
1654
The default implementation does nothing; it is however possible
1655
for repositories to maintain loaded indices across multiple locks
1656
by checking inside their implementation of this method to see
1657
whether their indices are still valid. This depends of course on
1658
the disk format being validatable in this manner.
1662
def revision_tree(self, revision_id):
1663
"""Return Tree for a revision on this branch.
1665
`revision_id` may be NULL_REVISION for the empty tree revision.
1667
revision_id = _mod_revision.ensure_null(revision_id)
1668
# TODO: refactor this to use an existing revision object
1669
# so we don't need to read it in twice.
1670
if revision_id == _mod_revision.NULL_REVISION:
1671
return RevisionTree(self, Inventory(root_id=None),
1672
_mod_revision.NULL_REVISION)
1674
inv = self.get_revision_inventory(revision_id)
1675
return RevisionTree(self, inv, revision_id)
1677
def revision_trees(self, revision_ids):
1678
"""Return Tree for a revision on this branch.
1680
`revision_id` may not be None or 'null:'"""
1681
inventories = self.iter_inventories(revision_ids)
1682
for inv in inventories:
1683
yield RevisionTree(self, inv, inv.revision_id)
1686
def get_ancestry(self, revision_id, topo_sorted=True):
1687
"""Return a list of revision-ids integrated by a revision.
1689
The first element of the list is always None, indicating the origin
1690
revision. This might change when we have history horizons, or
1691
perhaps we should have a new API.
1693
This is topologically sorted.
1695
if _mod_revision.is_null(revision_id):
1697
if not self.has_revision(revision_id):
1698
raise errors.NoSuchRevision(self, revision_id)
1699
graph = self.get_graph()
1701
search = graph._make_breadth_first_searcher([revision_id])
1704
found, ghosts = search.next_with_ghosts()
1705
except StopIteration:
1708
if _mod_revision.NULL_REVISION in keys:
1709
keys.remove(_mod_revision.NULL_REVISION)
1711
parent_map = graph.get_parent_map(keys)
1712
keys = tsort.topo_sort(parent_map)
1713
return [None] + list(keys)
1716
"""Compress the data within the repository.
1718
This operation only makes sense for some repository types. For other
1719
types it should be a no-op that just returns.
1721
This stub method does not require a lock, but subclasses should use
1722
@needs_write_lock as this is a long running call its reasonable to
1723
implicitly lock for the user.
1727
@deprecated_method(one_six)
1728
def print_file(self, file, revision_id):
1729
"""Print `file` to stdout.
1731
FIXME RBC 20060125 as John Meinel points out this is a bad api
1732
- it writes to stdout, it assumes that that is valid etc. Fix
1733
by creating a new more flexible convenience function.
1735
tree = self.revision_tree(revision_id)
1736
# use inventory as it was in that revision
1737
file_id = tree.inventory.path2id(file)
1739
# TODO: jam 20060427 Write a test for this code path
1740
# it had a bug in it, and was raising the wrong
1742
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1743
tree.print_file(file_id)
1745
def get_transaction(self):
1746
return self.control_files.get_transaction()
1748
@deprecated_method(one_one)
1749
def get_parents(self, revision_ids):
1750
"""See StackedParentsProvider.get_parents"""
1751
parent_map = self.get_parent_map(revision_ids)
1752
return [parent_map.get(r, None) for r in revision_ids]
1754
def get_parent_map(self, revision_ids):
1755
"""See graph._StackedParentsProvider.get_parent_map"""
1756
# revisions index works in keys; this just works in revisions
1757
# therefore wrap and unwrap
1760
for revision_id in revision_ids:
1761
if revision_id == _mod_revision.NULL_REVISION:
1762
result[revision_id] = ()
1763
elif revision_id is None:
1764
raise ValueError('get_parent_map(None) is not valid')
1766
query_keys.append((revision_id ,))
1767
for ((revision_id,), parent_keys) in \
1768
self.revisions.get_parent_map(query_keys).iteritems():
1770
result[revision_id] = tuple(parent_revid
1771
for (parent_revid,) in parent_keys)
1773
result[revision_id] = (_mod_revision.NULL_REVISION,)
1776
def _make_parents_provider(self):
1779
def get_graph(self, other_repository=None):
1780
"""Return the graph walker for this repository format"""
1781
parents_provider = self._make_parents_provider()
1782
if (other_repository is not None and
1783
not self.has_same_location(other_repository)):
1784
parents_provider = graph._StackedParentsProvider(
1785
[parents_provider, other_repository._make_parents_provider()])
1786
return graph.Graph(parents_provider)
1788
def _get_versioned_file_checker(self):
1789
"""Return an object suitable for checking versioned files."""
1790
return _VersionedFileChecker(self)
1792
def revision_ids_to_search_result(self, result_set):
1793
"""Convert a set of revision ids to a graph SearchResult."""
1794
result_parents = set()
1795
for parents in self.get_graph().get_parent_map(
1796
result_set).itervalues():
1797
result_parents.update(parents)
1798
included_keys = result_set.intersection(result_parents)
1799
start_keys = result_set.difference(included_keys)
1800
exclude_keys = result_parents.difference(result_set)
1801
result = graph.SearchResult(start_keys, exclude_keys,
1802
len(result_set), result_set)
1806
def set_make_working_trees(self, new_value):
1807
"""Set the policy flag for making working trees when creating branches.
1809
This only applies to branches that use this repository.
1811
The default is 'True'.
1812
:param new_value: True to restore the default, False to disable making
1815
raise NotImplementedError(self.set_make_working_trees)
1817
def make_working_trees(self):
1818
"""Returns the policy for making working trees on new branches."""
1819
raise NotImplementedError(self.make_working_trees)
1822
def sign_revision(self, revision_id, gpg_strategy):
1823
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1824
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1827
def has_signature_for_revision_id(self, revision_id):
1828
"""Query for a revision signature for revision_id in the repository."""
1829
if not self.has_revision(revision_id):
1830
raise errors.NoSuchRevision(self, revision_id)
1831
sig_present = (1 == len(
1832
self.signatures.get_parent_map([(revision_id,)])))
1836
def get_signature_text(self, revision_id):
1837
"""Return the text for a signature."""
1838
stream = self.signatures.get_record_stream([(revision_id,)],
1840
record = stream.next()
1841
if record.storage_kind == 'absent':
1842
raise errors.NoSuchRevision(self, revision_id)
1843
return record.get_bytes_as('fulltext')
1846
def check(self, revision_ids=None):
1847
"""Check consistency of all history of given revision_ids.
1849
Different repository implementations should override _check().
1851
:param revision_ids: A non-empty list of revision_ids whose ancestry
1852
will be checked. Typically the last revision_id of a branch.
1854
return self._check(revision_ids)
1856
def _check(self, revision_ids):
1857
result = check.Check(self)
1861
def _warn_if_deprecated(self):
1862
global _deprecation_warning_done
1863
if _deprecation_warning_done:
1865
_deprecation_warning_done = True
1866
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1867
% (self._format, self.bzrdir.transport.base))
1869
def supports_rich_root(self):
1870
return self._format.rich_root_data
1872
def _check_ascii_revisionid(self, revision_id, method):
1873
"""Private helper for ascii-only repositories."""
1874
# weave repositories refuse to store revisionids that are non-ascii.
1875
if revision_id is not None:
1876
# weaves require ascii revision ids.
1877
if isinstance(revision_id, unicode):
1879
revision_id.encode('ascii')
1880
except UnicodeEncodeError:
1881
raise errors.NonAsciiRevisionId(method, self)
1884
revision_id.decode('ascii')
1885
except UnicodeDecodeError:
1886
raise errors.NonAsciiRevisionId(method, self)
1888
def revision_graph_can_have_wrong_parents(self):
1889
"""Is it possible for this repository to have a revision graph with
1892
If True, then this repository must also implement
1893
_find_inconsistent_revision_parents so that check and reconcile can
1894
check for inconsistencies before proceeding with other checks that may
1895
depend on the revision index being consistent.
1897
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1900
# remove these delegates a while after bzr 0.15
1901
def __make_delegated(name, from_module):
1902
def _deprecated_repository_forwarder():
1903
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1904
% (name, from_module),
1907
m = __import__(from_module, globals(), locals(), [name])
1909
return getattr(m, name)
1910
except AttributeError:
1911
raise AttributeError('module %s has no name %s'
1913
globals()[name] = _deprecated_repository_forwarder
1916
'AllInOneRepository',
1917
'WeaveMetaDirRepository',
1918
'PreSplitOutRepositoryFormat',
1919
'RepositoryFormat4',
1920
'RepositoryFormat5',
1921
'RepositoryFormat6',
1922
'RepositoryFormat7',
1924
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1928
'RepositoryFormatKnit',
1929
'RepositoryFormatKnit1',
1931
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1934
def install_revision(repository, rev, revision_tree):
1935
"""Install all revision data into a repository."""
1936
install_revisions(repository, [(rev, revision_tree, None)])
1939
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1940
"""Install all revision data into a repository.
1942
Accepts an iterable of revision, tree, signature tuples. The signature
1945
repository.start_write_group()
1947
for n, (revision, revision_tree, signature) in enumerate(iterable):
1948
_install_revision(repository, revision, revision_tree, signature)
1950
pb.update('Transferring revisions', n + 1, num_revisions)
1952
repository.abort_write_group()
1955
repository.commit_write_group()
1958
def _install_revision(repository, rev, revision_tree, signature):
1959
"""Install all revision data into a repository."""
1960
present_parents = []
1962
for p_id in rev.parent_ids:
1963
if repository.has_revision(p_id):
1964
present_parents.append(p_id)
1965
parent_trees[p_id] = repository.revision_tree(p_id)
1967
parent_trees[p_id] = repository.revision_tree(
1968
_mod_revision.NULL_REVISION)
1970
inv = revision_tree.inventory
1971
entries = inv.iter_entries()
1972
# backwards compatibility hack: skip the root id.
1973
if not repository.supports_rich_root():
1974
path, root = entries.next()
1975
if root.revision != rev.revision_id:
1976
raise errors.IncompatibleRevision(repr(repository))
1978
for path, ie in entries:
1979
text_keys[(ie.file_id, ie.revision)] = ie
1980
text_parent_map = repository.texts.get_parent_map(text_keys)
1981
missing_texts = set(text_keys) - set(text_parent_map)
1982
# Add the texts that are not already present
1983
for text_key in missing_texts:
1984
ie = text_keys[text_key]
1986
# FIXME: TODO: The following loop overlaps/duplicates that done by
1987
# commit to determine parents. There is a latent/real bug here where
1988
# the parents inserted are not those commit would do - in particular
1989
# they are not filtered by heads(). RBC, AB
1990
for revision, tree in parent_trees.iteritems():
1991
if ie.file_id not in tree:
1993
parent_id = tree.inventory[ie.file_id].revision
1994
if parent_id in text_parents:
1996
text_parents.append((ie.file_id, parent_id))
1997
lines = revision_tree.get_file(ie.file_id).readlines()
1998
repository.texts.add_lines(text_key, text_parents, lines)
2000
# install the inventory
2001
repository.add_inventory(rev.revision_id, inv, present_parents)
2002
except errors.RevisionAlreadyPresent:
2004
if signature is not None:
2005
repository.add_signature_text(rev.revision_id, signature)
2006
repository.add_revision(rev.revision_id, rev, inv)
2009
class MetaDirRepository(Repository):
2010
"""Repositories in the new meta-dir layout.
2012
:ivar _transport: Transport for access to repository control files,
2013
typically pointing to .bzr/repository.
2016
def __init__(self, _format, a_bzrdir, control_files):
2017
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2018
self._transport = control_files._transport
2020
def is_shared(self):
2021
"""Return True if this repository is flagged as a shared repository."""
2022
return self._transport.has('shared-storage')
2025
def set_make_working_trees(self, new_value):
2026
"""Set the policy flag for making working trees when creating branches.
2028
This only applies to branches that use this repository.
2030
The default is 'True'.
2031
:param new_value: True to restore the default, False to disable making
2036
self._transport.delete('no-working-trees')
2037
except errors.NoSuchFile:
2040
self._transport.put_bytes('no-working-trees', '',
2041
mode=self.bzrdir._get_file_mode())
2043
def make_working_trees(self):
2044
"""Returns the policy for making working trees on new branches."""
2045
return not self._transport.has('no-working-trees')
2048
class MetaDirVersionedFileRepository(MetaDirRepository):
2049
"""Repositories in a meta-dir, that work via versioned file objects."""
2051
def __init__(self, _format, a_bzrdir, control_files):
2052
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2056
class RepositoryFormatRegistry(registry.Registry):
2057
"""Registry of RepositoryFormats."""
2059
def get(self, format_string):
2060
r = registry.Registry.get(self, format_string)
2066
format_registry = RepositoryFormatRegistry()
2067
"""Registry of formats, indexed by their identifying format string.
2069
This can contain either format instances themselves, or classes/factories that
2070
can be called to obtain one.
2074
#####################################################################
2075
# Repository Formats
2077
class RepositoryFormat(object):
2078
"""A repository format.
2080
Formats provide three things:
2081
* An initialization routine to construct repository data on disk.
2082
* a format string which is used when the BzrDir supports versioned
2084
* an open routine which returns a Repository instance.
2086
There is one and only one Format subclass for each on-disk format. But
2087
there can be one Repository subclass that is used for several different
2088
formats. The _format attribute on a Repository instance can be used to
2089
determine the disk format.
2091
Formats are placed in an dict by their format string for reference
2092
during opening. These should be subclasses of RepositoryFormat
2095
Once a format is deprecated, just deprecate the initialize and open
2096
methods on the format class. Do not deprecate the object, as the
2097
object will be created every system load.
2099
Common instance attributes:
2100
_matchingbzrdir - the bzrdir format that the repository format was
2101
originally written to work with. This can be used if manually
2102
constructing a bzrdir and repository, or more commonly for test suite
2106
# Set to True or False in derived classes. True indicates that the format
2107
# supports ghosts gracefully.
2108
supports_ghosts = None
2109
# Can this repository be given external locations to lookup additional
2110
# data. Set to True or False in derived classes.
2111
supports_external_lookups = None
2114
return "<%s>" % self.__class__.__name__
2116
def __eq__(self, other):
2117
# format objects are generally stateless
2118
return isinstance(other, self.__class__)
2120
def __ne__(self, other):
2121
return not self == other
2124
def find_format(klass, a_bzrdir):
2125
"""Return the format for the repository object in a_bzrdir.
2127
This is used by bzr native formats that have a "format" file in
2128
the repository. Other methods may be used by different types of
2132
transport = a_bzrdir.get_repository_transport(None)
2133
format_string = transport.get("format").read()
2134
return format_registry.get(format_string)
2135
except errors.NoSuchFile:
2136
raise errors.NoRepositoryPresent(a_bzrdir)
2138
raise errors.UnknownFormatError(format=format_string,
2142
def register_format(klass, format):
2143
format_registry.register(format.get_format_string(), format)
2146
def unregister_format(klass, format):
2147
format_registry.remove(format.get_format_string())
2150
def get_default_format(klass):
2151
"""Return the current default format."""
2152
from bzrlib import bzrdir
2153
return bzrdir.format_registry.make_bzrdir('default').repository_format
2155
def get_format_string(self):
2156
"""Return the ASCII format string that identifies this format.
2158
Note that in pre format ?? repositories the format string is
2159
not permitted nor written to disk.
2161
raise NotImplementedError(self.get_format_string)
2163
def get_format_description(self):
2164
"""Return the short description for this format."""
2165
raise NotImplementedError(self.get_format_description)
2167
# TODO: this shouldn't be in the base class, it's specific to things that
2168
# use weaves or knits -- mbp 20070207
2169
def _get_versioned_file_store(self,
2174
versionedfile_class=None,
2175
versionedfile_kwargs={},
2177
if versionedfile_class is None:
2178
versionedfile_class = self._versionedfile_class
2179
weave_transport = control_files._transport.clone(name)
2180
dir_mode = control_files._dir_mode
2181
file_mode = control_files._file_mode
2182
return VersionedFileStore(weave_transport, prefixed=prefixed,
2184
file_mode=file_mode,
2185
versionedfile_class=versionedfile_class,
2186
versionedfile_kwargs=versionedfile_kwargs,
2189
def initialize(self, a_bzrdir, shared=False):
2190
"""Initialize a repository of this format in a_bzrdir.
2192
:param a_bzrdir: The bzrdir to put the new repository in it.
2193
:param shared: The repository should be initialized as a sharable one.
2194
:returns: The new repository object.
2196
This may raise UninitializableFormat if shared repository are not
2197
compatible the a_bzrdir.
2199
raise NotImplementedError(self.initialize)
2201
def is_supported(self):
2202
"""Is this format supported?
2204
Supported formats must be initializable and openable.
2205
Unsupported formats may not support initialization or committing or
2206
some other features depending on the reason for not being supported.
2210
def check_conversion_target(self, target_format):
2211
raise NotImplementedError(self.check_conversion_target)
2213
def open(self, a_bzrdir, _found=False):
2214
"""Return an instance of this format for the bzrdir a_bzrdir.
2216
_found is a private parameter, do not use it.
2218
raise NotImplementedError(self.open)
2221
class MetaDirRepositoryFormat(RepositoryFormat):
2222
"""Common base class for the new repositories using the metadir layout."""
2224
rich_root_data = False
2225
supports_tree_reference = False
2226
supports_external_lookups = False
2227
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2230
super(MetaDirRepositoryFormat, self).__init__()
2232
def _create_control_files(self, a_bzrdir):
2233
"""Create the required files and the initial control_files object."""
2234
# FIXME: RBC 20060125 don't peek under the covers
2235
# NB: no need to escape relative paths that are url safe.
2236
repository_transport = a_bzrdir.get_repository_transport(self)
2237
control_files = lockable_files.LockableFiles(repository_transport,
2238
'lock', lockdir.LockDir)
2239
control_files.create_lock()
2240
return control_files
2242
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
2243
"""Upload the initial blank content."""
2244
control_files = self._create_control_files(a_bzrdir)
2245
control_files.lock_write()
2246
transport = control_files._transport
2248
utf8_files += [('shared-storage', '')]
2250
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2251
for (filename, content_stream) in files:
2252
transport.put_file(filename, content_stream,
2253
mode=a_bzrdir._get_file_mode())
2254
for (filename, content_bytes) in utf8_files:
2255
transport.put_bytes_non_atomic(filename, content_bytes,
2256
mode=a_bzrdir._get_file_mode())
2258
control_files.unlock()
2261
# formats which have no format string are not discoverable
2262
# and not independently creatable, so are not registered. They're
2263
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2264
# needed, it's constructed directly by the BzrDir. Non-native formats where
2265
# the repository is not separately opened are similar.
2267
format_registry.register_lazy(
2268
'Bazaar-NG Repository format 7',
2269
'bzrlib.repofmt.weaverepo',
2273
format_registry.register_lazy(
2274
'Bazaar-NG Knit Repository Format 1',
2275
'bzrlib.repofmt.knitrepo',
2276
'RepositoryFormatKnit1',
2279
format_registry.register_lazy(
2280
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2281
'bzrlib.repofmt.knitrepo',
2282
'RepositoryFormatKnit3',
2285
format_registry.register_lazy(
2286
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2287
'bzrlib.repofmt.knitrepo',
2288
'RepositoryFormatKnit4',
2291
# Pack-based formats. There is one format for pre-subtrees, and one for
2292
# post-subtrees to allow ease of testing.
2293
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2294
format_registry.register_lazy(
2295
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2296
'bzrlib.repofmt.pack_repo',
2297
'RepositoryFormatKnitPack1',
2299
format_registry.register_lazy(
2300
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2301
'bzrlib.repofmt.pack_repo',
2302
'RepositoryFormatKnitPack3',
2304
format_registry.register_lazy(
2305
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2306
'bzrlib.repofmt.pack_repo',
2307
'RepositoryFormatKnitPack4',
2309
format_registry.register_lazy(
2310
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2311
'bzrlib.repofmt.pack_repo',
2312
'RepositoryFormatKnitPack5',
2314
format_registry.register_lazy(
2315
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2316
'bzrlib.repofmt.pack_repo',
2317
'RepositoryFormatKnitPack5RichRoot',
2319
format_registry.register_lazy(
2320
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2321
'bzrlib.repofmt.pack_repo',
2322
'RepositoryFormatKnitPack5RichRootBroken',
2325
# Development formats.
2327
format_registry.register_lazy(
2328
"Bazaar development format 1 (needs bzr.dev from before 1.6)\n",
2329
'bzrlib.repofmt.pack_repo',
2330
'RepositoryFormatPackDevelopment1',
2332
format_registry.register_lazy(
2333
("Bazaar development format 1 with subtree support "
2334
"(needs bzr.dev from before 1.6)\n"),
2335
'bzrlib.repofmt.pack_repo',
2336
'RepositoryFormatPackDevelopment1Subtree',
2338
# 1.6->1.7 go below here
2341
class InterRepository(InterObject):
2342
"""This class represents operations taking place between two repositories.
2344
Its instances have methods like copy_content and fetch, and contain
2345
references to the source and target repositories these operations can be
2348
Often we will provide convenience methods on 'repository' which carry out
2349
operations with another repository - they will always forward to
2350
InterRepository.get(other).method_name(parameters).
2354
"""The available optimised InterRepository types."""
2356
def copy_content(self, revision_id=None):
2357
raise NotImplementedError(self.copy_content)
2359
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2360
"""Fetch the content required to construct revision_id.
2362
The content is copied from self.source to self.target.
2364
:param revision_id: if None all content is copied, if NULL_REVISION no
2366
:param pb: optional progress bar to use for progress reports. If not
2367
provided a default one will be created.
2369
:returns: (copied_revision_count, failures).
2371
# Normally we should find a specific InterRepository subclass to do
2372
# the fetch; if nothing else then at least InterSameDataRepository.
2373
# If none of them is suitable it looks like fetching is not possible;
2374
# we try to give a good message why. _assert_same_model will probably
2375
# give a helpful message; otherwise a generic one.
2376
self._assert_same_model(self.source, self.target)
2377
raise errors.IncompatibleRepositories(self.source, self.target,
2378
"no suitableInterRepository found")
2380
def _walk_to_common_revisions(self, revision_ids):
2381
"""Walk out from revision_ids in source to revisions target has.
2383
:param revision_ids: The start point for the search.
2384
:return: A set of revision ids.
2386
target_graph = self.target.get_graph()
2387
revision_ids = frozenset(revision_ids)
2388
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2389
return graph.SearchResult(revision_ids, set(), 0, set())
2390
missing_revs = set()
2391
source_graph = self.source.get_graph()
2392
# ensure we don't pay silly lookup costs.
2393
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2394
null_set = frozenset([_mod_revision.NULL_REVISION])
2397
next_revs, ghosts = searcher.next_with_ghosts()
2398
except StopIteration:
2400
if revision_ids.intersection(ghosts):
2401
absent_ids = set(revision_ids.intersection(ghosts))
2402
# If all absent_ids are present in target, no error is needed.
2403
absent_ids.difference_update(
2404
set(target_graph.get_parent_map(absent_ids)))
2406
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2407
# we don't care about other ghosts as we can't fetch them and
2408
# haven't been asked to.
2409
next_revs = set(next_revs)
2410
# we always have NULL_REVISION present.
2411
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2412
missing_revs.update(next_revs - have_revs)
2413
searcher.stop_searching_any(have_revs)
2414
return searcher.get_result()
2416
@deprecated_method(one_two)
2418
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
2419
"""Return the revision ids that source has that target does not.
2421
These are returned in topological order.
2423
:param revision_id: only return revision ids included by this
2425
:param find_ghosts: If True find missing revisions in deep history
2426
rather than just finding the surface difference.
2428
return list(self.search_missing_revision_ids(
2429
revision_id, find_ghosts).get_keys())
2432
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2433
"""Return the revision ids that source has that target does not.
2435
:param revision_id: only return revision ids included by this
2437
:param find_ghosts: If True find missing revisions in deep history
2438
rather than just finding the surface difference.
2439
:return: A bzrlib.graph.SearchResult.
2441
# stop searching at found target revisions.
2442
if not find_ghosts and revision_id is not None:
2443
return self._walk_to_common_revisions([revision_id])
2444
# generic, possibly worst case, slow code path.
2445
target_ids = set(self.target.all_revision_ids())
2446
if revision_id is not None:
2447
source_ids = self.source.get_ancestry(revision_id)
2448
if source_ids[0] is not None:
2449
raise AssertionError()
2452
source_ids = self.source.all_revision_ids()
2453
result_set = set(source_ids).difference(target_ids)
2454
return self.source.revision_ids_to_search_result(result_set)
2457
def _same_model(source, target):
2458
"""True if source and target have the same data representation.
2460
Note: this is always called on the base class; overriding it in a
2461
subclass will have no effect.
2464
InterRepository._assert_same_model(source, target)
2466
except errors.IncompatibleRepositories, e:
2470
def _assert_same_model(source, target):
2471
"""Raise an exception if two repositories do not use the same model.
2473
if source.supports_rich_root() != target.supports_rich_root():
2474
raise errors.IncompatibleRepositories(source, target,
2475
"different rich-root support")
2476
if source._serializer != target._serializer:
2477
raise errors.IncompatibleRepositories(source, target,
2478
"different serializers")
2481
class InterSameDataRepository(InterRepository):
2482
"""Code for converting between repositories that represent the same data.
2484
Data format and model must match for this to work.
2488
def _get_repo_format_to_test(self):
2489
"""Repository format for testing with.
2491
InterSameData can pull from subtree to subtree and from non-subtree to
2492
non-subtree, so we test this with the richest repository format.
2494
from bzrlib.repofmt import knitrepo
2495
return knitrepo.RepositoryFormatKnit3()
2498
def is_compatible(source, target):
2499
return InterRepository._same_model(source, target)
2502
def copy_content(self, revision_id=None):
2503
"""Make a complete copy of the content in self into destination.
2505
This copies both the repository's revision data, and configuration information
2506
such as the make_working_trees setting.
2508
This is a destructive operation! Do not use it on existing
2511
:param revision_id: Only copy the content needed to construct
2512
revision_id and its parents.
2515
self.target.set_make_working_trees(self.source.make_working_trees())
2516
except NotImplementedError:
2518
# but don't bother fetching if we have the needed data now.
2519
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2520
self.target.has_revision(revision_id)):
2522
self.target.fetch(self.source, revision_id=revision_id)
2525
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2526
"""See InterRepository.fetch()."""
2527
from bzrlib.fetch import RepoFetcher
2528
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2529
self.source, self.source._format, self.target,
2530
self.target._format)
2531
f = RepoFetcher(to_repository=self.target,
2532
from_repository=self.source,
2533
last_revision=revision_id,
2534
pb=pb, find_ghosts=find_ghosts)
2535
return f.count_copied, f.failed_revisions
2538
class InterWeaveRepo(InterSameDataRepository):
2539
"""Optimised code paths between Weave based repositories.
2541
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2542
implemented lazy inter-object optimisation.
2546
def _get_repo_format_to_test(self):
2547
from bzrlib.repofmt import weaverepo
2548
return weaverepo.RepositoryFormat7()
2551
def is_compatible(source, target):
2552
"""Be compatible with known Weave formats.
2554
We don't test for the stores being of specific types because that
2555
could lead to confusing results, and there is no need to be
2558
from bzrlib.repofmt.weaverepo import (
2564
return (isinstance(source._format, (RepositoryFormat5,
2566
RepositoryFormat7)) and
2567
isinstance(target._format, (RepositoryFormat5,
2569
RepositoryFormat7)))
2570
except AttributeError:
2574
def copy_content(self, revision_id=None):
2575
"""See InterRepository.copy_content()."""
2576
# weave specific optimised path:
2578
self.target.set_make_working_trees(self.source.make_working_trees())
2579
except (errors.RepositoryUpgradeRequired, NotImplemented):
2581
# FIXME do not peek!
2582
if self.source._transport.listable():
2583
pb = ui.ui_factory.nested_progress_bar()
2585
self.target.texts.insert_record_stream(
2586
self.source.texts.get_record_stream(
2587
self.source.texts.keys(), 'topological', False))
2588
pb.update('copying inventory', 0, 1)
2589
self.target.inventories.insert_record_stream(
2590
self.source.inventories.get_record_stream(
2591
self.source.inventories.keys(), 'topological', False))
2592
self.target.signatures.insert_record_stream(
2593
self.source.signatures.get_record_stream(
2594
self.source.signatures.keys(),
2596
self.target.revisions.insert_record_stream(
2597
self.source.revisions.get_record_stream(
2598
self.source.revisions.keys(),
2599
'topological', True))
2603
self.target.fetch(self.source, revision_id=revision_id)
2606
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2607
"""See InterRepository.fetch()."""
2608
from bzrlib.fetch import RepoFetcher
2609
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2610
self.source, self.source._format, self.target, self.target._format)
2611
f = RepoFetcher(to_repository=self.target,
2612
from_repository=self.source,
2613
last_revision=revision_id,
2614
pb=pb, find_ghosts=find_ghosts)
2615
return f.count_copied, f.failed_revisions
2618
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2619
"""See InterRepository.missing_revision_ids()."""
2620
# we want all revisions to satisfy revision_id in source.
2621
# but we don't want to stat every file here and there.
2622
# we want then, all revisions other needs to satisfy revision_id
2623
# checked, but not those that we have locally.
2624
# so the first thing is to get a subset of the revisions to
2625
# satisfy revision_id in source, and then eliminate those that
2626
# we do already have.
2627
# this is slow on high latency connection to self, but as as this
2628
# disk format scales terribly for push anyway due to rewriting
2629
# inventory.weave, this is considered acceptable.
2631
if revision_id is not None:
2632
source_ids = self.source.get_ancestry(revision_id)
2633
if source_ids[0] is not None:
2634
raise AssertionError()
2637
source_ids = self.source._all_possible_ids()
2638
source_ids_set = set(source_ids)
2639
# source_ids is the worst possible case we may need to pull.
2640
# now we want to filter source_ids against what we actually
2641
# have in target, but don't try to check for existence where we know
2642
# we do not have a revision as that would be pointless.
2643
target_ids = set(self.target._all_possible_ids())
2644
possibly_present_revisions = target_ids.intersection(source_ids_set)
2645
actually_present_revisions = set(
2646
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2647
required_revisions = source_ids_set.difference(actually_present_revisions)
2648
if revision_id is not None:
2649
# we used get_ancestry to determine source_ids then we are assured all
2650
# revisions referenced are present as they are installed in topological order.
2651
# and the tip revision was validated by get_ancestry.
2652
result_set = required_revisions
2654
# if we just grabbed the possibly available ids, then
2655
# we only have an estimate of whats available and need to validate
2656
# that against the revision records.
2658
self.source._eliminate_revisions_not_present(required_revisions))
2659
return self.source.revision_ids_to_search_result(result_set)
2662
class InterKnitRepo(InterSameDataRepository):
2663
"""Optimised code paths between Knit based repositories."""
2666
def _get_repo_format_to_test(self):
2667
from bzrlib.repofmt import knitrepo
2668
return knitrepo.RepositoryFormatKnit1()
2671
def is_compatible(source, target):
2672
"""Be compatible with known Knit formats.
2674
We don't test for the stores being of specific types because that
2675
could lead to confusing results, and there is no need to be
2678
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2680
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2681
isinstance(target._format, RepositoryFormatKnit))
2682
except AttributeError:
2684
return are_knits and InterRepository._same_model(source, target)
2687
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2688
"""See InterRepository.fetch()."""
2689
from bzrlib.fetch import RepoFetcher
2690
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2691
self.source, self.source._format, self.target, self.target._format)
2692
f = RepoFetcher(to_repository=self.target,
2693
from_repository=self.source,
2694
last_revision=revision_id,
2695
pb=pb, find_ghosts=find_ghosts)
2696
return f.count_copied, f.failed_revisions
2699
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2700
"""See InterRepository.missing_revision_ids()."""
2701
if revision_id is not None:
2702
source_ids = self.source.get_ancestry(revision_id)
2703
if source_ids[0] is not None:
2704
raise AssertionError()
2707
source_ids = self.source.all_revision_ids()
2708
source_ids_set = set(source_ids)
2709
# source_ids is the worst possible case we may need to pull.
2710
# now we want to filter source_ids against what we actually
2711
# have in target, but don't try to check for existence where we know
2712
# we do not have a revision as that would be pointless.
2713
target_ids = set(self.target.all_revision_ids())
2714
possibly_present_revisions = target_ids.intersection(source_ids_set)
2715
actually_present_revisions = set(
2716
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2717
required_revisions = source_ids_set.difference(actually_present_revisions)
2718
if revision_id is not None:
2719
# we used get_ancestry to determine source_ids then we are assured all
2720
# revisions referenced are present as they are installed in topological order.
2721
# and the tip revision was validated by get_ancestry.
2722
result_set = required_revisions
2724
# if we just grabbed the possibly available ids, then
2725
# we only have an estimate of whats available and need to validate
2726
# that against the revision records.
2728
self.source._eliminate_revisions_not_present(required_revisions))
2729
return self.source.revision_ids_to_search_result(result_set)
2732
class InterPackRepo(InterSameDataRepository):
2733
"""Optimised code paths between Pack based repositories."""
2736
def _get_repo_format_to_test(self):
2737
from bzrlib.repofmt import pack_repo
2738
return pack_repo.RepositoryFormatKnitPack1()
2741
def is_compatible(source, target):
2742
"""Be compatible with known Pack formats.
2744
We don't test for the stores being of specific types because that
2745
could lead to confusing results, and there is no need to be
2748
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2750
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2751
isinstance(target._format, RepositoryFormatPack))
2752
except AttributeError:
2754
return are_packs and InterRepository._same_model(source, target)
2757
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2758
"""See InterRepository.fetch()."""
2759
if (len(self.source._fallback_repositories) > 0 or
2760
len(self.target._fallback_repositories) > 0):
2761
# The pack layer is not aware of fallback repositories, so when
2762
# fetching from a stacked repository or into a stacked repository
2763
# we use the generic fetch logic which uses the VersionedFiles
2764
# attributes on repository.
2765
from bzrlib.fetch import RepoFetcher
2766
fetcher = RepoFetcher(self.target, self.source, revision_id,
2768
return fetcher.count_copied, fetcher.failed_revisions
2769
from bzrlib.repofmt.pack_repo import Packer
2770
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2771
self.source, self.source._format, self.target, self.target._format)
2772
self.count_copied = 0
2773
if revision_id is None:
2775
# everything to do - use pack logic
2776
# to fetch from all packs to one without
2777
# inventory parsing etc, IFF nothing to be copied is in the target.
2779
source_revision_ids = frozenset(self.source.all_revision_ids())
2780
revision_ids = source_revision_ids - \
2781
frozenset(self.target.get_parent_map(source_revision_ids))
2782
revision_keys = [(revid,) for revid in revision_ids]
2783
index = self.target._pack_collection.revision_index.combined_index
2784
present_revision_ids = set(item[1][0] for item in
2785
index.iter_entries(revision_keys))
2786
revision_ids = set(revision_ids) - present_revision_ids
2787
# implementing the TODO will involve:
2788
# - detecting when all of a pack is selected
2789
# - avoiding as much as possible pre-selection, so the
2790
# more-core routines such as create_pack_from_packs can filter in
2791
# a just-in-time fashion. (though having a HEADS list on a
2792
# repository might make this a lot easier, because we could
2793
# sensibly detect 'new revisions' without doing a full index scan.
2794
elif _mod_revision.is_null(revision_id):
2799
revision_ids = self.search_missing_revision_ids(revision_id,
2800
find_ghosts=find_ghosts).get_keys()
2801
except errors.NoSuchRevision:
2802
raise errors.InstallFailed([revision_id])
2803
if len(revision_ids) == 0:
2805
packs = self.source._pack_collection.all_packs()
2806
pack = Packer(self.target._pack_collection, packs, '.fetch',
2807
revision_ids).pack()
2808
if pack is not None:
2809
self.target._pack_collection._save_pack_names()
2810
# Trigger an autopack. This may duplicate effort as we've just done
2811
# a pack creation, but for now it is simpler to think about as
2812
# 'upload data, then repack if needed'.
2813
self.target._pack_collection.autopack()
2814
return (pack.get_revision_count(), [])
2819
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2820
"""See InterRepository.missing_revision_ids().
2822
:param find_ghosts: Find ghosts throughout the ancestry of
2825
if not find_ghosts and revision_id is not None:
2826
return self._walk_to_common_revisions([revision_id])
2827
elif revision_id is not None:
2828
# Find ghosts: search for revisions pointing from one repository to
2829
# the other, and vice versa, anywhere in the history of revision_id.
2830
graph = self.target.get_graph(other_repository=self.source)
2831
searcher = graph._make_breadth_first_searcher([revision_id])
2835
next_revs, ghosts = searcher.next_with_ghosts()
2836
except StopIteration:
2838
if revision_id in ghosts:
2839
raise errors.NoSuchRevision(self.source, revision_id)
2840
found_ids.update(next_revs)
2841
found_ids.update(ghosts)
2842
found_ids = frozenset(found_ids)
2843
# Double query here: should be able to avoid this by changing the
2844
# graph api further.
2845
result_set = found_ids - frozenset(
2846
self.target.get_parent_map(found_ids))
2848
source_ids = self.source.all_revision_ids()
2849
# source_ids is the worst possible case we may need to pull.
2850
# now we want to filter source_ids against what we actually
2851
# have in target, but don't try to check for existence where we know
2852
# we do not have a revision as that would be pointless.
2853
target_ids = set(self.target.all_revision_ids())
2854
result_set = set(source_ids).difference(target_ids)
2855
return self.source.revision_ids_to_search_result(result_set)
2858
class InterModel1and2(InterRepository):
2861
def _get_repo_format_to_test(self):
2865
def is_compatible(source, target):
2866
if not source.supports_rich_root() and target.supports_rich_root():
2872
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2873
"""See InterRepository.fetch()."""
2874
from bzrlib.fetch import Model1toKnit2Fetcher
2875
f = Model1toKnit2Fetcher(to_repository=self.target,
2876
from_repository=self.source,
2877
last_revision=revision_id,
2878
pb=pb, find_ghosts=find_ghosts)
2879
return f.count_copied, f.failed_revisions
2882
def copy_content(self, revision_id=None):
2883
"""Make a complete copy of the content in self into destination.
2885
This is a destructive operation! Do not use it on existing
2888
:param revision_id: Only copy the content needed to construct
2889
revision_id and its parents.
2892
self.target.set_make_working_trees(self.source.make_working_trees())
2893
except NotImplementedError:
2895
# but don't bother fetching if we have the needed data now.
2896
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2897
self.target.has_revision(revision_id)):
2899
self.target.fetch(self.source, revision_id=revision_id)
2902
class InterKnit1and2(InterKnitRepo):
2905
def _get_repo_format_to_test(self):
2909
def is_compatible(source, target):
2910
"""Be compatible with Knit1 source and Knit3 target"""
2912
from bzrlib.repofmt.knitrepo import (
2913
RepositoryFormatKnit1,
2914
RepositoryFormatKnit3,
2916
from bzrlib.repofmt.pack_repo import (
2917
RepositoryFormatKnitPack1,
2918
RepositoryFormatKnitPack3,
2919
RepositoryFormatKnitPack4,
2920
RepositoryFormatKnitPack5,
2921
RepositoryFormatKnitPack5RichRoot,
2922
RepositoryFormatPackDevelopment1,
2923
RepositoryFormatPackDevelopment1Subtree,
2926
RepositoryFormatKnit1, # no rr, no subtree
2927
RepositoryFormatKnitPack1, # no rr, no subtree
2928
RepositoryFormatPackDevelopment1, # no rr, no subtree
2929
RepositoryFormatKnitPack5, # no rr, no subtree
2932
RepositoryFormatKnit3, # rr, subtree
2933
RepositoryFormatKnitPack3, # rr, subtree
2934
RepositoryFormatKnitPack4, # rr, no subtree
2935
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
2936
RepositoryFormatPackDevelopment1Subtree, # rr, subtree
2938
for format in norichroot:
2939
if format.rich_root_data:
2940
raise AssertionError('Format %s is a rich-root format'
2941
' but is included in the non-rich-root list'
2943
for format in richroot:
2944
if not format.rich_root_data:
2945
raise AssertionError('Format %s is not a rich-root format'
2946
' but is included in the rich-root list'
2948
# TODO: One alternative is to just check format.rich_root_data,
2949
# instead of keeping membership lists. However, the formats
2950
# *also* have to use the same 'Knit' style of storage
2951
# (line-deltas, fulltexts, etc.)
2952
return (isinstance(source._format, norichroot) and
2953
isinstance(target._format, richroot))
2954
except AttributeError:
2958
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2959
"""See InterRepository.fetch()."""
2960
from bzrlib.fetch import Knit1to2Fetcher
2961
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2962
self.source, self.source._format, self.target,
2963
self.target._format)
2964
f = Knit1to2Fetcher(to_repository=self.target,
2965
from_repository=self.source,
2966
last_revision=revision_id,
2967
pb=pb, find_ghosts=find_ghosts)
2968
return f.count_copied, f.failed_revisions
2971
class InterDifferingSerializer(InterKnitRepo):
2974
def _get_repo_format_to_test(self):
2978
def is_compatible(source, target):
2979
"""Be compatible with Knit2 source and Knit3 target"""
2980
if source.supports_rich_root() != target.supports_rich_root():
2982
# Ideally, we'd support fetching if the source had no tree references
2983
# even if it supported them...
2984
if (getattr(source, '_format.supports_tree_reference', False) and
2985
not getattr(target, '_format.supports_tree_reference', False)):
2990
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2991
"""See InterRepository.fetch()."""
2992
revision_ids = self.target.search_missing_revision_ids(self.source,
2993
revision_id, find_ghosts=find_ghosts).get_keys()
2994
revision_ids = tsort.topo_sort(
2995
self.source.get_graph().get_parent_map(revision_ids))
2996
def revisions_iterator():
2997
for current_revision_id in revision_ids:
2998
revision = self.source.get_revision(current_revision_id)
2999
tree = self.source.revision_tree(current_revision_id)
3001
signature = self.source.get_signature_text(
3002
current_revision_id)
3003
except errors.NoSuchRevision:
3005
yield revision, tree, signature
3007
my_pb = ui.ui_factory.nested_progress_bar()
3012
install_revisions(self.target, revisions_iterator(),
3013
len(revision_ids), pb)
3015
if my_pb is not None:
3017
return len(revision_ids), 0
3020
class InterOtherToRemote(InterRepository):
3022
def __init__(self, source, target):
3023
InterRepository.__init__(self, source, target)
3024
self._real_inter = None
3027
def is_compatible(source, target):
3028
if isinstance(target, remote.RemoteRepository):
3032
def _ensure_real_inter(self):
3033
if self._real_inter is None:
3034
self.target._ensure_real()
3035
real_target = self.target._real_repository
3036
self._real_inter = InterRepository.get(self.source, real_target)
3038
def copy_content(self, revision_id=None):
3039
self._ensure_real_inter()
3040
self._real_inter.copy_content(revision_id=revision_id)
3042
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3043
self._ensure_real_inter()
3044
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3045
find_ghosts=find_ghosts)
3048
def _get_repo_format_to_test(self):
3052
class InterRemoteToOther(InterRepository):
3054
def __init__(self, source, target):
3055
InterRepository.__init__(self, source, target)
3056
self._real_inter = None
3059
def is_compatible(source, target):
3060
if not isinstance(source, remote.RemoteRepository):
3062
# Is source's model compatible with target's model?
3063
source._ensure_real()
3064
real_source = source._real_repository
3065
if isinstance(real_source, remote.RemoteRepository):
3066
raise NotImplementedError(
3067
"We don't support remote repos backed by remote repos yet.")
3068
return InterRepository._same_model(real_source, target)
3070
def _ensure_real_inter(self):
3071
if self._real_inter is None:
3072
self.source._ensure_real()
3073
real_source = self.source._real_repository
3074
self._real_inter = InterRepository.get(real_source, self.target)
3076
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3077
self._ensure_real_inter()
3078
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3079
find_ghosts=find_ghosts)
3081
def copy_content(self, revision_id=None):
3082
self._ensure_real_inter()
3083
self._real_inter.copy_content(revision_id=revision_id)
3086
def _get_repo_format_to_test(self):
3091
InterRepository.register_optimiser(InterDifferingSerializer)
3092
InterRepository.register_optimiser(InterSameDataRepository)
3093
InterRepository.register_optimiser(InterWeaveRepo)
3094
InterRepository.register_optimiser(InterKnitRepo)
3095
InterRepository.register_optimiser(InterModel1and2)
3096
InterRepository.register_optimiser(InterKnit1and2)
3097
InterRepository.register_optimiser(InterPackRepo)
3098
InterRepository.register_optimiser(InterOtherToRemote)
3099
InterRepository.register_optimiser(InterRemoteToOther)
3102
class CopyConverter(object):
3103
"""A repository conversion tool which just performs a copy of the content.
3105
This is slow but quite reliable.
3108
def __init__(self, target_format):
3109
"""Create a CopyConverter.
3111
:param target_format: The format the resulting repository should be.
3113
self.target_format = target_format
3115
def convert(self, repo, pb):
3116
"""Perform the conversion of to_convert, giving feedback via pb.
3118
:param to_convert: The disk object to convert.
3119
:param pb: a progress bar to use for progress information.
3124
# this is only useful with metadir layouts - separated repo content.
3125
# trigger an assertion if not such
3126
repo._format.get_format_string()
3127
self.repo_dir = repo.bzrdir
3128
self.step('Moving repository to repository.backup')
3129
self.repo_dir.transport.move('repository', 'repository.backup')
3130
backup_transport = self.repo_dir.transport.clone('repository.backup')
3131
repo._format.check_conversion_target(self.target_format)
3132
self.source_repo = repo._format.open(self.repo_dir,
3134
_override_transport=backup_transport)
3135
self.step('Creating new repository')
3136
converted = self.target_format.initialize(self.repo_dir,
3137
self.source_repo.is_shared())
3138
converted.lock_write()
3140
self.step('Copying content into repository.')
3141
self.source_repo.copy_content_into(converted)
3144
self.step('Deleting old repository content.')
3145
self.repo_dir.transport.delete_tree('repository.backup')
3146
self.pb.note('repository converted')
3148
def step(self, message):
3149
"""Update the pb by a step."""
3151
self.pb.update(message, self.count, self.total)
3163
def _unescaper(match, _map=_unescape_map):
3164
code = match.group(1)
3168
if not code.startswith('#'):
3170
return unichr(int(code[1:])).encode('utf8')
3176
def _unescape_xml(data):
3177
"""Unescape predefined XML entities in a string of data."""
3179
if _unescape_re is None:
3180
_unescape_re = re.compile('\&([^;]*);')
3181
return _unescape_re.sub(_unescaper, data)
3184
class _VersionedFileChecker(object):
3186
def __init__(self, repository):
3187
self.repository = repository
3188
self.text_index = self.repository._generate_text_key_index()
3190
def calculate_file_version_parents(self, text_key):
3191
"""Calculate the correct parents for a file version according to
3194
parent_keys = self.text_index[text_key]
3195
if parent_keys == [_mod_revision.NULL_REVISION]:
3197
return tuple(parent_keys)
3199
def check_file_version_parents(self, texts, progress_bar=None):
3200
"""Check the parents stored in a versioned file are correct.
3202
It also detects file versions that are not referenced by their
3203
corresponding revision's inventory.
3205
:returns: A tuple of (wrong_parents, dangling_file_versions).
3206
wrong_parents is a dict mapping {revision_id: (stored_parents,
3207
correct_parents)} for each revision_id where the stored parents
3208
are not correct. dangling_file_versions is a set of (file_id,
3209
revision_id) tuples for versions that are present in this versioned
3210
file, but not used by the corresponding inventory.
3213
self.file_ids = set([file_id for file_id, _ in
3214
self.text_index.iterkeys()])
3215
# text keys is now grouped by file_id
3216
n_weaves = len(self.file_ids)
3217
files_in_revisions = {}
3218
revisions_of_files = {}
3219
n_versions = len(self.text_index)
3220
progress_bar.update('loading text store', 0, n_versions)
3221
parent_map = self.repository.texts.get_parent_map(self.text_index)
3222
# On unlistable transports this could well be empty/error...
3223
text_keys = self.repository.texts.keys()
3224
unused_keys = frozenset(text_keys) - set(self.text_index)
3225
for num, key in enumerate(self.text_index.iterkeys()):
3226
if progress_bar is not None:
3227
progress_bar.update('checking text graph', num, n_versions)
3228
correct_parents = self.calculate_file_version_parents(key)
3230
knit_parents = parent_map[key]
3231
except errors.RevisionNotPresent:
3234
if correct_parents != knit_parents:
3235
wrong_parents[key] = (knit_parents, correct_parents)
3236
return wrong_parents, unused_keys
3239
def _old_get_graph(repository, revision_id):
3240
"""DO NOT USE. That is all. I'm serious."""
3241
graph = repository.get_graph()
3242
revision_graph = dict(((key, value) for key, value in
3243
graph.iter_ancestry([revision_id]) if value is not None))
3244
return _strip_NULL_ghosts(revision_graph)
3247
def _strip_NULL_ghosts(revision_graph):
3248
"""Also don't use this. more compatibility code for unmigrated clients."""
3249
# Filter ghosts, and null:
3250
if _mod_revision.NULL_REVISION in revision_graph:
3251
del revision_graph[_mod_revision.NULL_REVISION]
3252
for key, parents in revision_graph.items():
3253
revision_graph[key] = tuple(parent for parent in parents if parent
3255
return revision_graph
added
removed
bzrlib/repository.py
