1
# Copyright (C) 2005, 2006, 2007, 2008 Canonical Ltd
1
# Copyright (C) 2005, 2006 Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from bzrlib.lazy_import import lazy_import
18
lazy_import(globals(), """
17
from copy import deepcopy
18
from cStringIO import StringIO
19
from unittest import TestSuite
20
import xml.sax.saxutils
37
revision as _mod_revision,
42
from bzrlib.bundle import serializer
43
from bzrlib.revisiontree import RevisionTree
44
from bzrlib.store.versioned import VersionedFileStore
22
import bzrlib.bzrdir as bzrdir
23
from bzrlib.decorators import needs_read_lock, needs_write_lock
24
import bzrlib.errors as errors
25
from bzrlib.errors import InvalidRevisionId
26
from bzrlib.lockable_files import LockableFiles, TransportLock
27
from bzrlib.lockdir import LockDir
28
from bzrlib.osutils import safe_unicode
29
from bzrlib.revision import NULL_REVISION
30
from bzrlib.store import copy_all
31
from bzrlib.store.weave import WeaveStore
32
from bzrlib.store.text import TextStore
33
from bzrlib.symbol_versioning import *
34
from bzrlib.trace import mutter
35
from bzrlib.tree import RevisionTree
45
36
from bzrlib.testament import Testament
48
from bzrlib import registry
49
from bzrlib.decorators import needs_read_lock, needs_write_lock
50
from bzrlib.inter import InterObject
51
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
52
from bzrlib.symbol_versioning import (
58
from bzrlib.trace import mutter, mutter_callsite, warning
61
# Old formats display a warning, but only once
62
_deprecation_warning_done = False
65
class CommitBuilder(object):
66
"""Provides an interface to build up a commit.
68
This allows describing a tree to be committed without needing to
69
know the internals of the format of the repository.
72
# all clients should supply tree roots.
73
record_root_entry = True
74
# the default CommitBuilder does not manage trees whose root is versioned.
75
_versioned_root = False
77
def __init__(self, repository, parents, config, timestamp=None,
78
timezone=None, committer=None, revprops=None,
80
"""Initiate a CommitBuilder.
82
:param repository: Repository to commit to.
83
:param parents: Revision ids of the parents of the new revision.
84
:param config: Configuration to use.
85
:param timestamp: Optional timestamp recorded for commit.
86
:param timezone: Optional timezone for timestamp.
87
:param committer: Optional committer to set for commit.
88
:param revprops: Optional dictionary of revision properties.
89
:param revision_id: Optional revision id.
94
self._committer = self._config.username()
96
self._committer = committer
98
self.new_inventory = Inventory(None)
99
self._new_revision_id = revision_id
100
self.parents = parents
101
self.repository = repository
104
if revprops is not None:
105
self._revprops.update(revprops)
107
if timestamp is None:
108
timestamp = time.time()
109
# Restrict resolution to 1ms
110
self._timestamp = round(timestamp, 3)
113
self._timezone = osutils.local_time_offset()
115
self._timezone = int(timezone)
117
self._generate_revision_if_needed()
118
self.__heads = graph.HeadsCache(repository.get_graph()).heads
120
def commit(self, message):
121
"""Make the actual commit.
123
:return: The revision id of the recorded revision.
125
rev = _mod_revision.Revision(
126
timestamp=self._timestamp,
127
timezone=self._timezone,
128
committer=self._committer,
130
inventory_sha1=self.inv_sha1,
131
revision_id=self._new_revision_id,
132
properties=self._revprops)
133
rev.parent_ids = self.parents
134
self.repository.add_revision(self._new_revision_id, rev,
135
self.new_inventory, self._config)
136
self.repository.commit_write_group()
137
return self._new_revision_id
140
"""Abort the commit that is being built.
142
self.repository.abort_write_group()
144
def revision_tree(self):
145
"""Return the tree that was just committed.
147
After calling commit() this can be called to get a RevisionTree
148
representing the newly committed tree. This is preferred to
149
calling Repository.revision_tree() because that may require
150
deserializing the inventory, while we already have a copy in
153
return RevisionTree(self.repository, self.new_inventory,
154
self._new_revision_id)
156
def finish_inventory(self):
157
"""Tell the builder that the inventory is finished."""
158
if self.new_inventory.root is None:
159
raise AssertionError('Root entry should be supplied to'
160
' record_entry_contents, as of bzr 0.10.')
161
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
162
self.new_inventory.revision_id = self._new_revision_id
163
self.inv_sha1 = self.repository.add_inventory(
164
self._new_revision_id,
169
def _gen_revision_id(self):
170
"""Return new revision-id."""
171
return generate_ids.gen_revision_id(self._config.username(),
174
def _generate_revision_if_needed(self):
175
"""Create a revision id if None was supplied.
177
If the repository can not support user-specified revision ids
178
they should override this function and raise CannotSetRevisionId
179
if _new_revision_id is not None.
181
:raises: CannotSetRevisionId
183
if self._new_revision_id is None:
184
self._new_revision_id = self._gen_revision_id()
185
self.random_revid = True
187
self.random_revid = False
189
def _heads(self, file_id, revision_ids):
190
"""Calculate the graph heads for revision_ids in the graph of file_id.
192
This can use either a per-file graph or a global revision graph as we
193
have an identity relationship between the two graphs.
195
return self.__heads(revision_ids)
197
def _check_root(self, ie, parent_invs, tree):
198
"""Helper for record_entry_contents.
200
:param ie: An entry being added.
201
:param parent_invs: The inventories of the parent revisions of the
203
:param tree: The tree that is being committed.
205
# In this revision format, root entries have no knit or weave When
206
# serializing out to disk and back in root.revision is always
208
ie.revision = self._new_revision_id
210
def _get_delta(self, ie, basis_inv, path):
211
"""Get a delta against the basis inventory for ie."""
212
if ie.file_id not in basis_inv:
214
return (None, path, ie.file_id, ie)
215
elif ie != basis_inv[ie.file_id]:
217
# TODO: avoid tis id2path call.
218
return (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
223
def record_entry_contents(self, ie, parent_invs, path, tree,
225
"""Record the content of ie from tree into the commit if needed.
227
Side effect: sets ie.revision when unchanged
229
:param ie: An inventory entry present in the commit.
230
:param parent_invs: The inventories of the parent revisions of the
232
:param path: The path the entry is at in the tree.
233
:param tree: The tree which contains this entry and should be used to
235
:param content_summary: Summary data from the tree about the paths
236
content - stat, length, exec, sha/link target. This is only
237
accessed when the entry has a revision of None - that is when it is
238
a candidate to commit.
239
:return: A tuple (change_delta, version_recorded, fs_hash).
240
change_delta is an inventory_delta change for this entry against
241
the basis tree of the commit, or None if no change occured against
243
version_recorded is True if a new version of the entry has been
244
recorded. For instance, committing a merge where a file was only
245
changed on the other side will return (delta, False).
246
fs_hash is either None, or the hash details for the path (currently
247
a tuple of the contents sha1 and the statvalue returned by
248
tree.get_file_with_stat()).
250
if self.new_inventory.root is None:
251
if ie.parent_id is not None:
252
raise errors.RootMissing()
253
self._check_root(ie, parent_invs, tree)
254
if ie.revision is None:
255
kind = content_summary[0]
257
# ie is carried over from a prior commit
259
# XXX: repository specific check for nested tree support goes here - if
260
# the repo doesn't want nested trees we skip it ?
261
if (kind == 'tree-reference' and
262
not self.repository._format.supports_tree_reference):
263
# mismatch between commit builder logic and repository:
264
# this needs the entry creation pushed down into the builder.
265
raise NotImplementedError('Missing repository subtree support.')
266
self.new_inventory.add(ie)
268
# TODO: slow, take it out of the inner loop.
270
basis_inv = parent_invs[0]
272
basis_inv = Inventory(root_id=None)
274
# ie.revision is always None if the InventoryEntry is considered
275
# for committing. We may record the previous parents revision if the
276
# content is actually unchanged against a sole head.
277
if ie.revision is not None:
278
if not self._versioned_root and path == '':
279
# repositories that do not version the root set the root's
280
# revision to the new commit even when no change occurs, and
281
# this masks when a change may have occurred against the basis,
282
# so calculate if one happened.
283
if ie.file_id in basis_inv:
284
delta = (basis_inv.id2path(ie.file_id), path,
288
delta = (None, path, ie.file_id, ie)
289
return delta, False, None
291
# we don't need to commit this, because the caller already
292
# determined that an existing revision of this file is
293
# appropriate. If its not being considered for committing then
294
# it and all its parents to the root must be unaltered so
295
# no-change against the basis.
296
if ie.revision == self._new_revision_id:
297
raise AssertionError("Impossible situation, a skipped "
298
"inventory entry (%r) claims to be modified in this "
299
"commit (%r).", (ie, self._new_revision_id))
300
return None, False, None
301
# XXX: Friction: parent_candidates should return a list not a dict
302
# so that we don't have to walk the inventories again.
303
parent_candiate_entries = ie.parent_candidates(parent_invs)
304
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
306
for inv in parent_invs:
307
if ie.file_id in inv:
308
old_rev = inv[ie.file_id].revision
309
if old_rev in head_set:
310
heads.append(inv[ie.file_id].revision)
311
head_set.remove(inv[ie.file_id].revision)
314
# now we check to see if we need to write a new record to the
316
# We write a new entry unless there is one head to the ancestors, and
317
# the kind-derived content is unchanged.
319
# Cheapest check first: no ancestors, or more the one head in the
320
# ancestors, we write a new node.
324
# There is a single head, look it up for comparison
325
parent_entry = parent_candiate_entries[heads[0]]
326
# if the non-content specific data has changed, we'll be writing a
328
if (parent_entry.parent_id != ie.parent_id or
329
parent_entry.name != ie.name):
331
# now we need to do content specific checks:
333
# if the kind changed the content obviously has
334
if kind != parent_entry.kind:
336
# Stat cache fingerprint feedback for the caller - None as we usually
337
# don't generate one.
340
if content_summary[2] is None:
341
raise ValueError("Files must not have executable = None")
343
if (# if the file length changed we have to store:
344
parent_entry.text_size != content_summary[1] or
345
# if the exec bit has changed we have to store:
346
parent_entry.executable != content_summary[2]):
348
elif parent_entry.text_sha1 == content_summary[3]:
349
# all meta and content is unchanged (using a hash cache
350
# hit to check the sha)
351
ie.revision = parent_entry.revision
352
ie.text_size = parent_entry.text_size
353
ie.text_sha1 = parent_entry.text_sha1
354
ie.executable = parent_entry.executable
355
return self._get_delta(ie, basis_inv, path), False, None
357
# Either there is only a hash change(no hash cache entry,
358
# or same size content change), or there is no change on
360
# Provide the parent's hash to the store layer, so that the
361
# content is unchanged we will not store a new node.
362
nostore_sha = parent_entry.text_sha1
364
# We want to record a new node regardless of the presence or
365
# absence of a content change in the file.
367
ie.executable = content_summary[2]
368
file_obj, stat_value = tree.get_file_with_stat(ie.file_id, path)
370
lines = file_obj.readlines()
374
ie.text_sha1, ie.text_size = self._add_text_to_weave(
375
ie.file_id, lines, heads, nostore_sha)
376
# Let the caller know we generated a stat fingerprint.
377
fingerprint = (ie.text_sha1, stat_value)
378
except errors.ExistingContent:
379
# Turns out that the file content was unchanged, and we were
380
# only going to store a new node if it was changed. Carry over
382
ie.revision = parent_entry.revision
383
ie.text_size = parent_entry.text_size
384
ie.text_sha1 = parent_entry.text_sha1
385
ie.executable = parent_entry.executable
386
return self._get_delta(ie, basis_inv, path), False, None
387
elif kind == 'directory':
389
# all data is meta here, nothing specific to directory, so
391
ie.revision = parent_entry.revision
392
return self._get_delta(ie, basis_inv, path), False, None
394
self._add_text_to_weave(ie.file_id, lines, heads, None)
395
elif kind == 'symlink':
396
current_link_target = content_summary[3]
398
# symlink target is not generic metadata, check if it has
400
if current_link_target != parent_entry.symlink_target:
403
# unchanged, carry over.
404
ie.revision = parent_entry.revision
405
ie.symlink_target = parent_entry.symlink_target
406
return self._get_delta(ie, basis_inv, path), False, None
407
ie.symlink_target = current_link_target
409
self._add_text_to_weave(ie.file_id, lines, heads, None)
410
elif kind == 'tree-reference':
412
if content_summary[3] != parent_entry.reference_revision:
415
# unchanged, carry over.
416
ie.reference_revision = parent_entry.reference_revision
417
ie.revision = parent_entry.revision
418
return self._get_delta(ie, basis_inv, path), False, None
419
ie.reference_revision = content_summary[3]
421
self._add_text_to_weave(ie.file_id, lines, heads, None)
423
raise NotImplementedError('unknown kind')
424
ie.revision = self._new_revision_id
425
return self._get_delta(ie, basis_inv, path), True, fingerprint
427
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
428
# Note: as we read the content directly from the tree, we know its not
429
# been turned into unicode or badly split - but a broken tree
430
# implementation could give us bad output from readlines() so this is
431
# not a guarantee of safety. What would be better is always checking
432
# the content during test suite execution. RBC 20070912
433
parent_keys = tuple((file_id, parent) for parent in parents)
434
return self.repository.texts.add_lines(
435
(file_id, self._new_revision_id), parent_keys, new_lines,
436
nostore_sha=nostore_sha, random_id=self.random_revid,
437
check_content=False)[0:2]
440
class RootCommitBuilder(CommitBuilder):
441
"""This commitbuilder actually records the root id"""
443
# the root entry gets versioned properly by this builder.
444
_versioned_root = True
446
def _check_root(self, ie, parent_invs, tree):
447
"""Helper for record_entry_contents.
449
:param ie: An entry being added.
450
:param parent_invs: The inventories of the parent revisions of the
452
:param tree: The tree that is being committed.
456
######################################################################
37
from bzrlib.tree import EmptyTree
459
42
class Repository(object):
460
43
"""Repository holding history for one or more branches.
463
46
revisions and file history. It's normally accessed only by the Branch,
464
47
which views a particular line of development through that history.
466
The Repository builds on top of some byte storage facilies (the revisions,
467
signatures, inventories and texts attributes) and a Transport, which
468
respectively provide byte storage and a means to access the (possibly
49
The Repository builds on top of Stores and a Transport, which respectively
50
describe the disk data format and the way of accessing the (possibly
471
The byte storage facilities are addressed via tuples, which we refer to
472
as 'keys' throughout the code base. Revision_keys, inventory_keys and
473
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
474
(file_id, revision_id). We use this interface because it allows low
475
friction with the underlying code that implements disk indices, network
476
encoding and other parts of bzrlib.
478
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
479
the serialised revisions for the repository. This can be used to obtain
480
revision graph information or to access raw serialised revisions.
481
The result of trying to insert data into the repository via this store
482
is undefined: it should be considered read-only except for implementors
484
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
485
the serialised signatures for the repository. This can be used to
486
obtain access to raw serialised signatures. The result of trying to
487
insert data into the repository via this store is undefined: it should
488
be considered read-only except for implementors of repositories.
489
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
490
the serialised inventories for the repository. This can be used to
491
obtain unserialised inventories. The result of trying to insert data
492
into the repository via this store is undefined: it should be
493
considered read-only except for implementors of repositories.
494
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
495
texts of files and directories for the repository. This can be used to
496
obtain file texts or file graphs. Note that Repository.iter_file_bytes
497
is usually a better interface for accessing file texts.
498
The result of trying to insert data into the repository via this store
499
is undefined: it should be considered read-only except for implementors
501
:ivar _transport: Transport for file access to repository, typically
502
pointing to .bzr/repository.
505
# What class to use for a CommitBuilder. Often its simpler to change this
506
# in a Repository class subclass rather than to override
507
# get_commit_builder.
508
_commit_builder_class = CommitBuilder
509
# The search regex used by xml based repositories to determine what things
510
# where changed in a single commit.
511
_file_ids_altered_regex = lazy_regex.lazy_compile(
512
r'file_id="(?P<file_id>[^"]+)"'
513
r'.* revision="(?P<revision_id>[^"]+)"'
516
def abort_write_group(self):
517
"""Commit the contents accrued within the current write group.
519
:seealso: start_write_group.
521
if self._write_group is not self.get_transaction():
522
# has an unlock or relock occured ?
523
raise errors.BzrError('mismatched lock context and write group.')
524
self._abort_write_group()
525
self._write_group = None
527
def _abort_write_group(self):
528
"""Template method for per-repository write group cleanup.
530
This is called during abort before the write group is considered to be
531
finished and should cleanup any internal state accrued during the write
532
group. There is no requirement that data handed to the repository be
533
*not* made available - this is not a rollback - but neither should any
534
attempt be made to ensure that data added is fully commited. Abort is
535
invoked when an error has occured so futher disk or network operations
536
may not be possible or may error and if possible should not be
540
def add_fallback_repository(self, repository):
541
"""Add a repository to use for looking up data not held locally.
543
:param repository: A repository.
545
if not self._format.supports_external_lookups:
546
raise errors.UnstackableRepositoryFormat(self._format, self.base)
547
self._check_fallback_repository(repository)
548
self._fallback_repositories.append(repository)
549
self.texts.add_fallback_versioned_files(repository.texts)
550
self.inventories.add_fallback_versioned_files(repository.inventories)
551
self.revisions.add_fallback_versioned_files(repository.revisions)
552
self.signatures.add_fallback_versioned_files(repository.signatures)
554
def _check_fallback_repository(self, repository):
555
"""Check that this repository can fallback to repository safely.
557
Raise an error if not.
559
:param repository: A repository to fallback to.
561
return InterRepository._assert_same_model(self, repository)
563
def add_inventory(self, revision_id, inv, parents):
564
"""Add the inventory inv to the repository as revision_id.
566
:param parents: The revision ids of the parents that revision_id
567
is known to have and are in the repository already.
569
:returns: The validator(which is a sha1 digest, though what is sha'd is
570
repository format specific) of the serialized inventory.
572
if not self.is_in_write_group():
573
raise AssertionError("%r not in write group" % (self,))
574
_mod_revision.check_not_reserved_id(revision_id)
575
if not (inv.revision_id is None or inv.revision_id == revision_id):
576
raise AssertionError(
577
"Mismatch between inventory revision"
578
" id and insertion revid (%r, %r)"
579
% (inv.revision_id, revision_id))
581
raise AssertionError()
582
inv_lines = self._serialise_inventory_to_lines(inv)
583
return self._inventory_add_lines(revision_id, parents,
584
inv_lines, check_content=False)
586
def _inventory_add_lines(self, revision_id, parents, lines,
588
"""Store lines in inv_vf and return the sha1 of the inventory."""
589
parents = [(parent,) for parent in parents]
590
return self.inventories.add_lines((revision_id,), parents, lines,
591
check_content=check_content)[0]
593
def add_revision(self, revision_id, rev, inv=None, config=None):
594
"""Add rev to the revision store as revision_id.
596
:param revision_id: the revision id to use.
597
:param rev: The revision object.
598
:param inv: The inventory for the revision. if None, it will be looked
599
up in the inventory storer
600
:param config: If None no digital signature will be created.
601
If supplied its signature_needed method will be used
602
to determine if a signature should be made.
604
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
606
_mod_revision.check_not_reserved_id(revision_id)
607
if config is not None and config.signature_needed():
609
inv = self.get_inventory(revision_id)
610
plaintext = Testament(rev, inv).as_short_text()
611
self.store_revision_signature(
612
gpg.GPGStrategy(config), plaintext, revision_id)
613
# check inventory present
614
if not self.inventories.get_parent_map([(revision_id,)]):
616
raise errors.WeaveRevisionNotPresent(revision_id,
619
# yes, this is not suitable for adding with ghosts.
620
rev.inventory_sha1 = self.add_inventory(revision_id, inv,
624
rev.inventory_sha1 = self.inventories.get_sha1s([key])[key]
625
self._add_revision(rev)
627
def _add_revision(self, revision):
628
text = self._serializer.write_revision_to_string(revision)
629
key = (revision.revision_id,)
630
parents = tuple((parent,) for parent in revision.parent_ids)
631
self.revisions.add_lines(key, parents, osutils.split_lines(text))
55
def _all_possible_ids(self):
56
"""Return all the possible revisions that we could find."""
57
return self.get_inventory_weave().names()
633
60
def all_revision_ids(self):
634
61
"""Returns a list of all the revision ids in the repository.
636
This is conceptually deprecated because code should generally work on
637
the graph reachable from a particular revision, and ignore any other
638
revisions that might be present. There is no direct replacement
641
if 'evil' in debug.debug_flags:
642
mutter_callsite(2, "all_revision_ids is linear with history.")
643
return self._all_revision_ids()
645
def _all_revision_ids(self):
646
"""Returns a list of all the revision ids in the repository.
648
63
These are in as much topological order as the underlying store can
651
raise NotImplementedError(self._all_revision_ids)
653
def break_lock(self):
654
"""Break a lock if one is present from another instance.
656
Uses the ui factory to ask for confirmation if the lock may be from
659
self.control_files.break_lock()
64
present: for weaves ghosts may lead to a lack of correctness until
65
the reweave updates the parents list.
67
result = self._all_possible_ids()
68
return self._eliminate_revisions_not_present(result)
662
71
def _eliminate_revisions_not_present(self, revision_ids):
665
74
Returns a set of the present revisions.
668
graph = self.get_graph()
669
parent_map = graph.get_parent_map(revision_ids)
670
# The old API returned a list, should this actually be a set?
671
return parent_map.keys()
77
for id in revision_ids:
78
if self.has_revision(id):
674
83
def create(a_bzrdir):
675
84
"""Construct the current default format repository in a_bzrdir."""
676
85
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
678
def __init__(self, _format, a_bzrdir, control_files):
679
"""instantiate a Repository.
681
:param _format: The format of the repository on disk.
682
:param a_bzrdir: The BzrDir of the repository.
684
In the future we will have a single api for all stores for
685
getting file texts, inventories and revisions, then
686
this construct will accept instances of those things.
688
super(Repository, self).__init__()
87
def __init__(self, _format, a_bzrdir):
89
if isinstance(_format, (RepositoryFormat4,
92
# legacy: use a common control files.
93
self.control_files = a_bzrdir._control_files
95
repo_transport = a_bzrdir.get_repository_transport(None)
96
self.control_files = LockableFiles(repo_transport,
100
dir_mode = self.control_files._dir_mode
101
file_mode = self.control_files._file_mode
689
102
self._format = _format
690
# the following are part of the public API for Repository:
691
103
self.bzrdir = a_bzrdir
692
self.control_files = control_files
693
self._transport = control_files._transport
694
self.base = self._transport.base
696
self._reconcile_does_inventory_gc = True
697
self._reconcile_fixes_text_parents = False
698
self._reconcile_backsup_inventory = True
699
# not right yet - should be more semantically clear ?
701
# TODO: make sure to construct the right store classes, etc, depending
702
# on whether escaping is required.
703
self._warn_if_deprecated()
704
self._write_group = None
705
# Additional places to query for data.
706
self._fallback_repositories = []
707
# What order should fetch operations request streams in?
708
# The default is unordered as that is the cheapest for an origin to
710
self._fetch_order = 'unordered'
711
# Does this repository use deltas that can be fetched as-deltas ?
712
# (E.g. knits, where the knit deltas can be transplanted intact.
713
# We default to False, which will ensure that enough data to get
714
# a full text out of any fetch stream will be grabbed.
715
self._fetch_uses_deltas = False
716
# Should fetch trigger a reconcile after the fetch? Only needed for
717
# some repository formats that can suffer internal inconsistencies.
718
self._fetch_reconcile = False
721
return '%s(%r)' % (self.__class__.__name__,
724
def has_same_location(self, other):
725
"""Returns a boolean indicating if this repository is at the same
726
location as another repository.
728
This might return False even when two repository objects are accessing
729
the same physical repository via different URLs.
731
if self.__class__ is not other.__class__:
733
return (self._transport.base == other._transport.base)
735
def is_in_write_group(self):
736
"""Return True if there is an open write group.
738
:seealso: start_write_group.
740
return self._write_group is not None
105
def get_weave(name, prefixed=False):
107
name = safe_unicode(name)
110
relpath = self.control_files._escape(name)
111
weave_transport = self.control_files._transport.clone(relpath)
112
ws = WeaveStore(weave_transport, prefixed=prefixed,
115
if self.control_files._transport.should_cache():
116
ws.enable_cache = True
119
def get_store(name, compressed=True, prefixed=False):
120
# FIXME: This approach of assuming stores are all entirely compressed
121
# or entirely uncompressed is tidy, but breaks upgrade from
122
# some existing branches where there's a mixture; we probably
123
# still want the option to look for both.
125
name = safe_unicode(name)
128
relpath = self.control_files._escape(name)
129
store = TextStore(self.control_files._transport.clone(relpath),
130
prefixed=prefixed, compressed=compressed,
133
#if self._transport.should_cache():
134
# cache_path = os.path.join(self.cache_root, name)
135
# os.mkdir(cache_path)
136
# store = bzrlib.store.CachedStore(store, cache_path)
139
if isinstance(self._format, RepositoryFormat4):
140
self.inventory_store = get_store('inventory-store')
141
self.text_store = get_store('text-store')
142
self.revision_store = get_store('revision-store')
143
elif isinstance(self._format, RepositoryFormat5):
144
self.control_weaves = get_weave('')
145
self.weave_store = get_weave('weaves')
146
self.revision_store = get_store('revision-store', compressed=False)
147
elif isinstance(self._format, RepositoryFormat6):
148
self.control_weaves = get_weave('')
149
self.weave_store = get_weave('weaves', prefixed=True)
150
self.revision_store = get_store('revision-store', compressed=False,
152
elif isinstance(self._format, RepositoryFormat7):
153
self.control_weaves = get_weave('')
154
self.weave_store = get_weave('weaves', prefixed=True)
155
self.revision_store = get_store('revision-store', compressed=False,
157
self.revision_store.register_suffix('sig')
159
def lock_write(self):
160
self.control_files.lock_write()
163
## self.control_files.lock_read()
164
# repositories are no longer locked
742
167
def is_locked(self):
743
168
return self.control_files.is_locked()
745
def is_write_locked(self):
746
"""Return True if this object is write locked."""
747
return self.is_locked() and self.control_files._lock_mode == 'w'
749
def lock_write(self, token=None):
750
"""Lock this repository for writing.
752
This causes caching within the repository obejct to start accumlating
753
data during reads, and allows a 'write_group' to be obtained. Write
754
groups must be used for actual data insertion.
756
:param token: if this is already locked, then lock_write will fail
757
unless the token matches the existing lock.
758
:returns: a token if this instance supports tokens, otherwise None.
759
:raises TokenLockingNotSupported: when a token is given but this
760
instance doesn't support using token locks.
761
:raises MismatchedToken: if the specified token doesn't match the token
762
of the existing lock.
763
:seealso: start_write_group.
765
A token should be passed in if you know that you have locked the object
766
some other way, and need to synchronise this object's state with that
769
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
771
result = self.control_files.lock_write(token=token)
772
for repo in self._fallback_repositories:
773
# Writes don't affect fallback repos
779
self.control_files.lock_read()
780
for repo in self._fallback_repositories:
784
def get_physical_lock_status(self):
785
return self.control_files.get_physical_lock_status()
787
def leave_lock_in_place(self):
788
"""Tell this repository not to release the physical lock when this
791
If lock_write doesn't return a token, then this method is not supported.
793
self.control_files.leave_in_place()
795
def dont_leave_lock_in_place(self):
796
"""Tell this repository to release the physical lock when this
797
object is unlocked, even if it didn't originally acquire it.
799
If lock_write doesn't return a token, then this method is not supported.
801
self.control_files.dont_leave_in_place()
804
def gather_stats(self, revid=None, committers=None):
805
"""Gather statistics from a revision id.
807
:param revid: The revision id to gather statistics from, if None, then
808
no revision specific statistics are gathered.
809
:param committers: Optional parameter controlling whether to grab
810
a count of committers from the revision specific statistics.
811
:return: A dictionary of statistics. Currently this contains:
812
committers: The number of committers if requested.
813
firstrev: A tuple with timestamp, timezone for the penultimate left
814
most ancestor of revid, if revid is not the NULL_REVISION.
815
latestrev: A tuple with timestamp, timezone for revid, if revid is
816
not the NULL_REVISION.
817
revisions: The total revision count in the repository.
818
size: An estimate disk size of the repository in bytes.
821
if revid and committers:
822
result['committers'] = 0
823
if revid and revid != _mod_revision.NULL_REVISION:
825
all_committers = set()
826
revisions = self.get_ancestry(revid)
827
# pop the leading None
829
first_revision = None
831
# ignore the revisions in the middle - just grab first and last
832
revisions = revisions[0], revisions[-1]
833
for revision in self.get_revisions(revisions):
834
if not first_revision:
835
first_revision = revision
837
all_committers.add(revision.committer)
838
last_revision = revision
840
result['committers'] = len(all_committers)
841
result['firstrev'] = (first_revision.timestamp,
842
first_revision.timezone)
843
result['latestrev'] = (last_revision.timestamp,
844
last_revision.timezone)
846
# now gather global repository information
847
# XXX: This is available for many repos regardless of listability.
848
if self.bzrdir.root_transport.listable():
849
# XXX: do we want to __define len__() ?
850
# Maybe the versionedfiles object should provide a different
851
# method to get the number of keys.
852
result['revisions'] = len(self.revisions.keys())
856
def find_branches(self, using=False):
857
"""Find branches underneath this repository.
859
This will include branches inside other branches.
861
:param using: If True, list only branches using this repository.
863
if using and not self.is_shared():
865
return [self.bzrdir.open_branch()]
866
except errors.NotBranchError:
868
class Evaluator(object):
871
self.first_call = True
873
def __call__(self, bzrdir):
874
# On the first call, the parameter is always the bzrdir
875
# containing the current repo.
876
if not self.first_call:
878
repository = bzrdir.open_repository()
879
except errors.NoRepositoryPresent:
882
return False, (None, repository)
883
self.first_call = False
885
value = (bzrdir.open_branch(), None)
886
except errors.NotBranchError:
891
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
892
self.bzrdir.root_transport, evaluate=Evaluator()):
893
if branch is not None:
894
branches.append(branch)
895
if not using and repository is not None:
896
branches.extend(repository.find_branches())
900
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
901
"""Return the revision ids that other has that this does not.
903
These are returned in topological order.
905
revision_id: only return revision ids included by revision_id.
907
return InterRepository.get(other, self).search_missing_revision_ids(
908
revision_id, find_ghosts)
910
@deprecated_method(one_two)
912
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
913
"""Return the revision ids that other has that this does not.
915
These are returned in topological order.
917
revision_id: only return revision ids included by revision_id.
919
keys = self.search_missing_revision_ids(
920
other, revision_id, find_ghosts).get_keys()
923
parents = other.get_graph().get_parent_map(keys)
926
return tsort.topo_sort(parents)
171
def missing_revision_ids(self, other, revision_id=None):
172
"""Return the revision ids that other has that this does not.
174
These are returned in topological order.
176
revision_id: only return revision ids included by revision_id.
178
return InterRepository.get(other, self).missing_revision_ids(revision_id)
935
187
control = bzrdir.BzrDir.open(base)
936
188
return control.open_repository()
938
def copy_content_into(self, destination, revision_id=None):
190
def copy_content_into(self, destination, revision_id=None, basis=None):
939
191
"""Make a complete copy of the content in self into destination.
941
193
This is a destructive operation! Do not use it on existing
944
return InterRepository.get(self, destination).copy_content(revision_id)
946
def commit_write_group(self):
947
"""Commit the contents accrued within the current write group.
949
:seealso: start_write_group.
951
if self._write_group is not self.get_transaction():
952
# has an unlock or relock occured ?
953
raise errors.BzrError('mismatched lock context %r and '
955
(self.get_transaction(), self._write_group))
956
self._commit_write_group()
957
self._write_group = None
959
def _commit_write_group(self):
960
"""Template method for per-repository write group cleanup.
962
This is called before the write group is considered to be
963
finished and should ensure that all data handed to the repository
964
for writing during the write group is safely committed (to the
965
extent possible considering file system caching etc).
968
def fetch(self, source, revision_id=None, pb=None, find_ghosts=False):
196
return InterRepository.get(self, destination).copy_content(revision_id, basis)
198
def fetch(self, source, revision_id=None, pb=None):
969
199
"""Fetch the content required to construct revision_id from source.
971
201
If revision_id is None all content is copied.
972
:param find_ghosts: Find and copy revisions in the source that are
973
ghosts in the target (and not reachable directly by walking out to
974
the first-present revision in target from revision_id).
976
# fast path same-url fetch operations
977
if self.has_same_location(source):
978
# check that last_revision is in 'from' and then return a
980
if (revision_id is not None and
981
not _mod_revision.is_null(revision_id)):
982
self.get_revision(revision_id)
984
# if there is no specific appropriate InterRepository, this will get
985
# the InterRepository base class, which raises an
986
# IncompatibleRepositories when asked to fetch.
987
inter = InterRepository.get(source, self)
988
return inter.fetch(revision_id=revision_id, pb=pb,
989
find_ghosts=find_ghosts)
991
def create_bundle(self, target, base, fileobj, format=None):
992
return serializer.write_bundle(self, target, base, fileobj, format)
994
def get_commit_builder(self, branch, parents, config, timestamp=None,
995
timezone=None, committer=None, revprops=None,
997
"""Obtain a CommitBuilder for this repository.
999
:param branch: Branch to commit to.
1000
:param parents: Revision ids of the parents of the new revision.
1001
:param config: Configuration to use.
1002
:param timestamp: Optional timestamp recorded for commit.
1003
:param timezone: Optional timezone for timestamp.
1004
:param committer: Optional committer to set for commit.
1005
:param revprops: Optional dictionary of revision properties.
1006
:param revision_id: Optional revision id.
1008
result = self._commit_builder_class(self, parents, config,
1009
timestamp, timezone, committer, revprops, revision_id)
1010
self.start_write_group()
203
return InterRepository.get(source, self).fetch(revision_id=revision_id,
1013
206
def unlock(self):
1014
if (self.control_files._lock_count == 1 and
1015
self.control_files._lock_mode == 'w'):
1016
if self._write_group is not None:
1017
self.abort_write_group()
1018
self.control_files.unlock()
1019
raise errors.BzrError(
1020
'Must end write groups before releasing write locks.')
1021
207
self.control_files.unlock()
1022
for repo in self._fallback_repositories:
1025
209
@needs_read_lock
1026
def clone(self, a_bzrdir, revision_id=None):
210
def clone(self, a_bzrdir, revision_id=None, basis=None):
1027
211
"""Clone this repository into a_bzrdir using the current format.
1029
213
Currently no check is made that the format of this repository and
1030
214
the bzrdir format are compatible. FIXME RBC 20060201.
1032
:return: The newly created destination repository.
1034
# TODO: deprecate after 0.16; cloning this with all its settings is
1035
# probably not very useful -- mbp 20070423
1036
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1037
self.copy_content_into(dest_repo, revision_id)
1040
def start_write_group(self):
1041
"""Start a write group in the repository.
1043
Write groups are used by repositories which do not have a 1:1 mapping
1044
between file ids and backend store to manage the insertion of data from
1045
both fetch and commit operations.
1047
A write lock is required around the start_write_group/commit_write_group
1048
for the support of lock-requiring repository formats.
1050
One can only insert data into a repository inside a write group.
1054
if not self.is_write_locked():
1055
raise errors.NotWriteLocked(self)
1056
if self._write_group:
1057
raise errors.BzrError('already in a write group')
1058
self._start_write_group()
1059
# so we can detect unlock/relock - the write group is now entered.
1060
self._write_group = self.get_transaction()
1062
def _start_write_group(self):
1063
"""Template method for per-repository write group startup.
1065
This is called before the write group is considered to be
1070
def sprout(self, to_bzrdir, revision_id=None):
1071
"""Create a descendent repository for new development.
1073
Unlike clone, this does not copy the settings of the repository.
1075
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1076
dest_repo.fetch(self, revision_id=revision_id)
1079
def _create_sprouting_repo(self, a_bzrdir, shared):
1080
216
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
1081
217
# use target default format.
1082
dest_repo = a_bzrdir.create_repository()
218
result = a_bzrdir.create_repository()
219
# FIXME RBC 20060209 split out the repository type to avoid this check ?
220
elif isinstance(a_bzrdir._format,
221
(bzrdir.BzrDirFormat4,
222
bzrdir.BzrDirFormat5,
223
bzrdir.BzrDirFormat6)):
224
result = a_bzrdir.open_repository()
1084
# Most control formats need the repository to be specifically
1085
# created, but on some old all-in-one formats it's not needed
1087
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1088
except errors.UninitializableFormat:
1089
dest_repo = a_bzrdir.open_repository()
226
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
227
self.copy_content_into(result, revision_id, basis)
1093
230
def has_revision(self, revision_id):
1094
"""True if this repository has a copy of the revision."""
1095
return revision_id in self.has_revisions((revision_id,))
1098
def has_revisions(self, revision_ids):
1099
"""Probe to find out the presence of multiple revisions.
1101
:param revision_ids: An iterable of revision_ids.
1102
:return: A set of the revision_ids that were present.
1104
parent_map = self.revisions.get_parent_map(
1105
[(rev_id,) for rev_id in revision_ids])
1107
if _mod_revision.NULL_REVISION in revision_ids:
1108
result.add(_mod_revision.NULL_REVISION)
1109
result.update([key[0] for key in parent_map])
231
"""True if this branch has a copy of the revision.
233
This does not necessarily imply the revision is merge
234
or on the mainline."""
235
return (revision_id is None
236
or self.revision_store.has_id(revision_id))
239
def get_revision_xml_file(self, revision_id):
240
"""Return XML file object for revision object."""
241
if not revision_id or not isinstance(revision_id, basestring):
242
raise InvalidRevisionId(revision_id=revision_id, branch=self)
244
return self.revision_store.get(revision_id)
245
except (IndexError, KeyError):
246
raise bzrlib.errors.NoSuchRevision(self, revision_id)
249
def get_revision_xml(self, revision_id):
250
return self.get_revision_xml_file(revision_id).read()
1112
252
@needs_read_lock
1113
253
def get_revision(self, revision_id):
1114
"""Return the Revision object for a named revision."""
1115
return self.get_revisions([revision_id])[0]
1118
def get_revision_reconcile(self, revision_id):
1119
"""'reconcile' helper routine that allows access to a revision always.
1121
This variant of get_revision does not cross check the weave graph
1122
against the revision one as get_revision does: but it should only
1123
be used by reconcile, or reconcile-alike commands that are correcting
1124
or testing the revision graph.
1126
return self._get_revisions([revision_id])[0]
1129
def get_revisions(self, revision_ids):
1130
"""Get many revisions at once."""
1131
return self._get_revisions(revision_ids)
1134
def _get_revisions(self, revision_ids):
1135
"""Core work logic to get many revisions without sanity checks."""
1136
for rev_id in revision_ids:
1137
if not rev_id or not isinstance(rev_id, basestring):
1138
raise errors.InvalidRevisionId(revision_id=rev_id, branch=self)
1139
keys = [(key,) for key in revision_ids]
1140
stream = self.revisions.get_record_stream(keys, 'unordered', True)
1142
for record in stream:
1143
if record.storage_kind == 'absent':
1144
raise errors.NoSuchRevision(self, record.key[0])
1145
text = record.get_bytes_as('fulltext')
1146
rev = self._serializer.read_revision_from_string(text)
1147
revs[record.key[0]] = rev
1148
return [revs[revid] for revid in revision_ids]
1151
def get_revision_xml(self, revision_id):
1152
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1153
# would have already do it.
1154
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1155
rev = self.get_revision(revision_id)
1156
rev_tmp = cStringIO.StringIO()
1157
# the current serializer..
1158
self._serializer.write_revision(rev, rev_tmp)
1160
return rev_tmp.getvalue()
1162
def get_deltas_for_revisions(self, revisions):
1163
"""Produce a generator of revision deltas.
1165
Note that the input is a sequence of REVISIONS, not revision_ids.
1166
Trees will be held in memory until the generator exits.
1167
Each delta is relative to the revision's lefthand predecessor.
1169
required_trees = set()
1170
for revision in revisions:
1171
required_trees.add(revision.revision_id)
1172
required_trees.update(revision.parent_ids[:1])
1173
trees = dict((t.get_revision_id(), t) for
1174
t in self.revision_trees(required_trees))
1175
for revision in revisions:
1176
if not revision.parent_ids:
1177
old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
1179
old_tree = trees[revision.parent_ids[0]]
1180
yield trees[revision.revision_id].changes_from(old_tree)
1183
def get_revision_delta(self, revision_id):
1184
"""Return the delta for one revision.
1186
The delta is relative to the left-hand predecessor of the
1189
r = self.get_revision(revision_id)
1190
return list(self.get_deltas_for_revisions([r]))[0]
254
"""Return the Revision object for a named revision"""
255
xml_file = self.get_revision_xml_file(revision_id)
258
r = bzrlib.xml5.serializer_v5.read_revision(xml_file)
259
except SyntaxError, e:
260
raise bzrlib.errors.BzrError('failed to unpack revision_xml',
264
assert r.revision_id == revision_id
268
def get_revision_sha1(self, revision_id):
269
"""Hash the stored value of a revision, and return it."""
270
# In the future, revision entries will be signed. At that
271
# point, it is probably best *not* to include the signature
272
# in the revision hash. Because that lets you re-sign
273
# the revision, (add signatures/remove signatures) and still
274
# have all hash pointers stay consistent.
275
# But for now, just hash the contents.
276
return bzrlib.osutils.sha_file(self.get_revision_xml_file(revision_id))
1192
278
@needs_write_lock
1193
279
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
1194
signature = gpg_strategy.sign(plaintext)
1195
self.add_signature_text(revision_id, signature)
1198
def add_signature_text(self, revision_id, signature):
1199
self.signatures.add_lines((revision_id,), (),
1200
osutils.split_lines(signature))
1202
def find_text_key_references(self):
1203
"""Find the text key references within the repository.
1205
:return: a dictionary mapping (file_id, revision_id) tuples to altered file-ids to an iterable of
1206
revision_ids. Each altered file-ids has the exact revision_ids that
1207
altered it listed explicitly.
1208
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1209
to whether they were referred to by the inventory of the
1210
revision_id that they contain. The inventory texts from all present
1211
revision ids are assessed to generate this report.
1213
revision_keys = self.revisions.keys()
1214
w = self.inventories
1215
pb = ui.ui_factory.nested_progress_bar()
1217
return self._find_text_key_references_from_xml_inventory_lines(
1218
w.iter_lines_added_or_present_in_keys(revision_keys, pb=pb))
1222
def _find_text_key_references_from_xml_inventory_lines(self,
1224
"""Core routine for extracting references to texts from inventories.
1226
This performs the translation of xml lines to revision ids.
1228
:param line_iterator: An iterator of lines, origin_version_id
1229
:return: A dictionary mapping text keys ((fileid, revision_id) tuples)
1230
to whether they were referred to by the inventory of the
1231
revision_id that they contain. Note that if that revision_id was
1232
not part of the line_iterator's output then False will be given -
1233
even though it may actually refer to that key.
1235
if not self._serializer.support_altered_by_hack:
1236
raise AssertionError(
1237
"_find_text_key_references_from_xml_inventory_lines only "
1238
"supported for branches which store inventory as unnested xml"
1239
", not on %r" % self)
1242
# this code needs to read every new line in every inventory for the
1243
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
1244
# not present in one of those inventories is unnecessary but not
1245
# harmful because we are filtering by the revision id marker in the
1246
# inventory lines : we only select file ids altered in one of those
1247
# revisions. We don't need to see all lines in the inventory because
1248
# only those added in an inventory in rev X can contain a revision=X
1250
unescape_revid_cache = {}
1251
unescape_fileid_cache = {}
1253
# jam 20061218 In a big fetch, this handles hundreds of thousands
1254
# of lines, so it has had a lot of inlining and optimizing done.
1255
# Sorry that it is a little bit messy.
1256
# Move several functions to be local variables, since this is a long
1258
search = self._file_ids_altered_regex.search
1259
unescape = _unescape_xml
1260
setdefault = result.setdefault
1261
for line, line_key in line_iterator:
1262
match = search(line)
1265
# One call to match.group() returning multiple items is quite a
1266
# bit faster than 2 calls to match.group() each returning 1
1267
file_id, revision_id = match.group('file_id', 'revision_id')
1269
# Inlining the cache lookups helps a lot when you make 170,000
1270
# lines and 350k ids, versus 8.4 unique ids.
1271
# Using a cache helps in 2 ways:
1272
# 1) Avoids unnecessary decoding calls
1273
# 2) Re-uses cached strings, which helps in future set and
1275
# (2) is enough that removing encoding entirely along with
1276
# the cache (so we are using plain strings) results in no
1277
# performance improvement.
1279
revision_id = unescape_revid_cache[revision_id]
1281
unescaped = unescape(revision_id)
1282
unescape_revid_cache[revision_id] = unescaped
1283
revision_id = unescaped
1285
# Note that unconditionally unescaping means that we deserialise
1286
# every fileid, which for general 'pull' is not great, but we don't
1287
# really want to have some many fulltexts that this matters anyway.
1290
file_id = unescape_fileid_cache[file_id]
1292
unescaped = unescape(file_id)
1293
unescape_fileid_cache[file_id] = unescaped
1296
key = (file_id, revision_id)
1297
setdefault(key, False)
1298
if revision_id == line_key[-1]:
1302
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1304
"""Helper routine for fileids_altered_by_revision_ids.
1306
This performs the translation of xml lines to revision ids.
1308
:param line_iterator: An iterator of lines, origin_version_id
1309
:param revision_ids: The revision ids to filter for. This should be a
1310
set or other type which supports efficient __contains__ lookups, as
1311
the revision id from each parsed line will be looked up in the
1312
revision_ids filter.
1313
:return: a dictionary mapping altered file-ids to an iterable of
1314
revision_ids. Each altered file-ids has the exact revision_ids that
1315
altered it listed explicitly.
1318
setdefault = result.setdefault
1320
self._find_text_key_references_from_xml_inventory_lines(
1321
line_iterator).iterkeys():
1322
# once data is all ensured-consistent; then this is
1323
# if revision_id == version_id
1324
if key[-1:] in revision_ids:
1325
setdefault(key[0], set()).add(key[-1])
1328
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1329
"""Find the file ids and versions affected by revisions.
1331
:param revisions: an iterable containing revision ids.
1332
:param _inv_weave: The inventory weave from this repository or None.
1333
If None, the inventory weave will be opened automatically.
1334
:return: a dictionary mapping altered file-ids to an iterable of
1335
revision_ids. Each altered file-ids has the exact revision_ids that
1336
altered it listed explicitly.
1338
selected_keys = set((revid,) for revid in revision_ids)
1339
w = _inv_weave or self.inventories
1340
pb = ui.ui_factory.nested_progress_bar()
1342
return self._find_file_ids_from_xml_inventory_lines(
1343
w.iter_lines_added_or_present_in_keys(
1344
selected_keys, pb=pb),
1349
def iter_files_bytes(self, desired_files):
1350
"""Iterate through file versions.
1352
Files will not necessarily be returned in the order they occur in
1353
desired_files. No specific order is guaranteed.
1355
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1356
value supplied by the caller as part of desired_files. It should
1357
uniquely identify the file version in the caller's context. (Examples:
1358
an index number or a TreeTransform trans_id.)
1360
bytes_iterator is an iterable of bytestrings for the file. The
1361
kind of iterable and length of the bytestrings are unspecified, but for
1362
this implementation, it is a list of bytes produced by
1363
VersionedFile.get_record_stream().
1365
:param desired_files: a list of (file_id, revision_id, identifier)
1368
transaction = self.get_transaction()
1370
for file_id, revision_id, callable_data in desired_files:
1371
text_keys[(file_id, revision_id)] = callable_data
1372
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1373
if record.storage_kind == 'absent':
1374
raise errors.RevisionNotPresent(record.key, self)
1375
yield text_keys[record.key], record.get_bytes_as('fulltext')
1377
def _generate_text_key_index(self, text_key_references=None,
1379
"""Generate a new text key index for the repository.
1381
This is an expensive function that will take considerable time to run.
1383
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1384
list of parents, also text keys. When a given key has no parents,
1385
the parents list will be [NULL_REVISION].
1387
# All revisions, to find inventory parents.
1388
if ancestors is None:
1389
graph = self.get_graph()
1390
ancestors = graph.get_parent_map(self.all_revision_ids())
1391
if text_key_references is None:
1392
text_key_references = self.find_text_key_references()
1393
pb = ui.ui_factory.nested_progress_bar()
1395
return self._do_generate_text_key_index(ancestors,
1396
text_key_references, pb)
1400
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1401
"""Helper for _generate_text_key_index to avoid deep nesting."""
1402
revision_order = tsort.topo_sort(ancestors)
1403
invalid_keys = set()
1405
for revision_id in revision_order:
1406
revision_keys[revision_id] = set()
1407
text_count = len(text_key_references)
1408
# a cache of the text keys to allow reuse; costs a dict of all the
1409
# keys, but saves a 2-tuple for every child of a given key.
1411
for text_key, valid in text_key_references.iteritems():
1413
invalid_keys.add(text_key)
1415
revision_keys[text_key[1]].add(text_key)
1416
text_key_cache[text_key] = text_key
1417
del text_key_references
1419
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1420
NULL_REVISION = _mod_revision.NULL_REVISION
1421
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1422
# too small for large or very branchy trees. However, for 55K path
1423
# trees, it would be easy to use too much memory trivially. Ideally we
1424
# could gauge this by looking at available real memory etc, but this is
1425
# always a tricky proposition.
1426
inventory_cache = lru_cache.LRUCache(10)
1427
batch_size = 10 # should be ~150MB on a 55K path tree
1428
batch_count = len(revision_order) / batch_size + 1
1430
pb.update("Calculating text parents.", processed_texts, text_count)
1431
for offset in xrange(batch_count):
1432
to_query = revision_order[offset * batch_size:(offset + 1) *
1436
for rev_tree in self.revision_trees(to_query):
1437
revision_id = rev_tree.get_revision_id()
1438
parent_ids = ancestors[revision_id]
1439
for text_key in revision_keys[revision_id]:
1440
pb.update("Calculating text parents.", processed_texts)
1441
processed_texts += 1
1442
candidate_parents = []
1443
for parent_id in parent_ids:
1444
parent_text_key = (text_key[0], parent_id)
1446
check_parent = parent_text_key not in \
1447
revision_keys[parent_id]
1449
# the parent parent_id is a ghost:
1450
check_parent = False
1451
# truncate the derived graph against this ghost.
1452
parent_text_key = None
1454
# look at the parent commit details inventories to
1455
# determine possible candidates in the per file graph.
1458
inv = inventory_cache[parent_id]
1460
inv = self.revision_tree(parent_id).inventory
1461
inventory_cache[parent_id] = inv
1462
parent_entry = inv._byid.get(text_key[0], None)
1463
if parent_entry is not None:
1465
text_key[0], parent_entry.revision)
1467
parent_text_key = None
1468
if parent_text_key is not None:
1469
candidate_parents.append(
1470
text_key_cache[parent_text_key])
1471
parent_heads = text_graph.heads(candidate_parents)
1472
new_parents = list(parent_heads)
1473
new_parents.sort(key=lambda x:candidate_parents.index(x))
1474
if new_parents == []:
1475
new_parents = [NULL_REVISION]
1476
text_index[text_key] = new_parents
1478
for text_key in invalid_keys:
1479
text_index[text_key] = [NULL_REVISION]
1482
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1483
"""Get an iterable listing the keys of all the data introduced by a set
1486
The keys will be ordered so that the corresponding items can be safely
1487
fetched and inserted in that order.
1489
:returns: An iterable producing tuples of (knit-kind, file-id,
1490
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1491
'revisions'. file-id is None unless knit-kind is 'file'.
1493
# XXX: it's a bit weird to control the inventory weave caching in this
1494
# generator. Ideally the caching would be done in fetch.py I think. Or
1495
# maybe this generator should explicitly have the contract that it
1496
# should not be iterated until the previously yielded item has been
1498
inv_w = self.inventories
1500
# file ids that changed
1501
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1503
num_file_ids = len(file_ids)
1504
for file_id, altered_versions in file_ids.iteritems():
1505
if _files_pb is not None:
1506
_files_pb.update("fetch texts", count, num_file_ids)
1508
yield ("file", file_id, altered_versions)
1509
# We're done with the files_pb. Note that it finished by the caller,
1510
# just as it was created by the caller.
1514
yield ("inventory", None, revision_ids)
1517
revisions_with_signatures = set()
1518
for rev_id in revision_ids:
1520
self.get_signature_text(rev_id)
1521
except errors.NoSuchRevision:
1525
revisions_with_signatures.add(rev_id)
1526
yield ("signatures", None, revisions_with_signatures)
1529
yield ("revisions", None, revision_ids)
280
self.revision_store.add(StringIO(gpg_strategy.sign(plaintext)),
283
def fileid_involved_between_revs(self, from_revid, to_revid):
284
"""Find file_id(s) which are involved in the changes between revisions.
286
This determines the set of revisions which are involved, and then
287
finds all file ids affected by those revisions.
289
# TODO: jam 20060119 This code assumes that w.inclusions will
290
# always be correct. But because of the presence of ghosts
291
# it is possible to be wrong.
292
# One specific example from Robert Collins:
293
# Two branches, with revisions ABC, and AD
294
# C is a ghost merge of D.
295
# Inclusions doesn't recognize D as an ancestor.
296
# If D is ever merged in the future, the weave
297
# won't be fixed, because AD never saw revision C
298
# to cause a conflict which would force a reweave.
299
w = self.get_inventory_weave()
300
from_set = set(w.inclusions([w.lookup(from_revid)]))
301
to_set = set(w.inclusions([w.lookup(to_revid)]))
302
included = to_set.difference(from_set)
303
changed = map(w.idx_to_name, included)
304
return self._fileid_involved_by_set(changed)
306
def fileid_involved(self, last_revid=None):
307
"""Find all file_ids modified in the ancestry of last_revid.
309
:param last_revid: If None, last_revision() will be used.
311
w = self.get_inventory_weave()
313
changed = set(w._names)
315
included = w.inclusions([w.lookup(last_revid)])
316
changed = map(w.idx_to_name, included)
317
return self._fileid_involved_by_set(changed)
319
def fileid_involved_by_set(self, changes):
320
"""Find all file_ids modified by the set of revisions passed in.
322
:param changes: A set() of revision ids
324
# TODO: jam 20060119 This line does *nothing*, remove it.
325
# or better yet, change _fileid_involved_by_set so
326
# that it takes the inventory weave, rather than
327
# pulling it out by itself.
328
return self._fileid_involved_by_set(changes)
330
def _fileid_involved_by_set(self, changes):
331
"""Find the set of file-ids affected by the set of revisions.
333
:param changes: A set() of revision ids.
334
:return: A set() of file ids.
336
This peaks at the Weave, interpreting each line, looking to
337
see if it mentions one of the revisions. And if so, includes
338
the file id mentioned.
339
This expects both the Weave format, and the serialization
340
to have a single line per file/directory, and to have
341
fileid="" and revision="" on that line.
343
assert isinstance(self._format, (RepositoryFormat5,
345
RepositoryFormat7)), \
346
"fileid_involved only supported for branches which store inventory as unnested xml"
348
w = self.get_inventory_weave()
350
for line in w._weave:
352
# it is ugly, but it is due to the weave structure
353
if not isinstance(line, basestring): continue
355
start = line.find('file_id="')+9
356
if start < 9: continue
357
end = line.find('"', start)
359
file_id = xml.sax.saxutils.unescape(line[start:end])
361
# check if file_id is already present
362
if file_id in file_ids: continue
364
start = line.find('revision="')+10
365
if start < 10: continue
366
end = line.find('"', start)
368
revision_id = xml.sax.saxutils.unescape(line[start:end])
370
if revision_id in changes:
371
file_ids.add(file_id)
375
def get_inventory_weave(self):
376
return self.control_weaves.get_weave('inventory',
377
self.get_transaction())
1531
379
@needs_read_lock
1532
380
def get_inventory(self, revision_id):
1533
"""Get Inventory object by revision id."""
1534
return self.iter_inventories([revision_id]).next()
1536
def iter_inventories(self, revision_ids):
1537
"""Get many inventories by revision_ids.
1539
This will buffer some or all of the texts used in constructing the
1540
inventories in memory, but will only parse a single inventory at a
1543
:return: An iterator of inventories.
1545
if ((None in revision_ids)
1546
or (_mod_revision.NULL_REVISION in revision_ids)):
1547
raise ValueError('cannot get null revision inventory')
1548
return self._iter_inventories(revision_ids)
1550
def _iter_inventories(self, revision_ids):
1551
"""single-document based inventory iteration."""
1552
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1553
yield self.deserialise_inventory(revision_id, text)
1555
def _iter_inventory_xmls(self, revision_ids):
1556
keys = [(revision_id,) for revision_id in revision_ids]
1557
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1559
for record in stream:
1560
if record.storage_kind != 'absent':
1561
texts[record.key] = record.get_bytes_as('fulltext')
1563
raise errors.NoSuchRevision(self, record.key)
1565
yield texts[key], key[-1]
1567
def deserialise_inventory(self, revision_id, xml):
1568
"""Transform the xml into an inventory object.
1570
:param revision_id: The expected revision id of the inventory.
1571
:param xml: A serialised inventory.
1573
result = self._serializer.read_inventory_from_string(xml, revision_id)
1574
if result.revision_id != revision_id:
1575
raise AssertionError('revision id mismatch %s != %s' % (
1576
result.revision_id, revision_id))
1579
def serialise_inventory(self, inv):
1580
return self._serializer.write_inventory_to_string(inv)
1582
def _serialise_inventory_to_lines(self, inv):
1583
return self._serializer.write_inventory_to_lines(inv)
1585
def get_serializer_format(self):
1586
return self._serializer.format_num
381
"""Get Inventory object by hash."""
382
xml = self.get_inventory_xml(revision_id)
383
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
1588
385
@needs_read_lock
1589
386
def get_inventory_xml(self, revision_id):
1590
387
"""Get inventory XML as a file object."""
1591
texts = self._iter_inventory_xmls([revision_id])
1593
text, revision_id = texts.next()
1594
except StopIteration:
1595
raise errors.HistoryMissing(self, 'inventory', revision_id)
389
assert isinstance(revision_id, basestring), type(revision_id)
390
iw = self.get_inventory_weave()
391
return iw.get_text(iw.lookup(revision_id))
393
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
1598
395
@needs_read_lock
1599
396
def get_inventory_sha1(self, revision_id):
1816
486
:param new_value: True to restore the default, False to disable making
1819
raise NotImplementedError(self.set_make_working_trees)
489
# FIXME: split out into a new class/strategy ?
490
if isinstance(self._format, (RepositoryFormat4,
493
raise NotImplementedError(self.set_make_working_trees)
496
self.control_files._transport.delete('no-working-trees')
497
except errors.NoSuchFile:
500
self.control_files.put_utf8('no-working-trees', '')
1821
502
def make_working_trees(self):
1822
503
"""Returns the policy for making working trees on new branches."""
1823
raise NotImplementedError(self.make_working_trees)
504
# FIXME: split out into a new class/strategy ?
505
if isinstance(self._format, (RepositoryFormat4,
509
return not self.control_files._transport.has('no-working-trees')
1825
511
@needs_write_lock
1826
512
def sign_revision(self, revision_id, gpg_strategy):
1827
513
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1828
514
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1831
def has_signature_for_revision_id(self, revision_id):
1832
"""Query for a revision signature for revision_id in the repository."""
1833
if not self.has_revision(revision_id):
1834
raise errors.NoSuchRevision(self, revision_id)
1835
sig_present = (1 == len(
1836
self.signatures.get_parent_map([(revision_id,)])))
1840
def get_signature_text(self, revision_id):
1841
"""Return the text for a signature."""
1842
stream = self.signatures.get_record_stream([(revision_id,)],
1844
record = stream.next()
1845
if record.storage_kind == 'absent':
1846
raise errors.NoSuchRevision(self, revision_id)
1847
return record.get_bytes_as('fulltext')
1850
def check(self, revision_ids=None):
1851
"""Check consistency of all history of given revision_ids.
1853
Different repository implementations should override _check().
1855
:param revision_ids: A non-empty list of revision_ids whose ancestry
1856
will be checked. Typically the last revision_id of a branch.
1858
return self._check(revision_ids)
1860
def _check(self, revision_ids):
1861
result = check.Check(self)
1865
def _warn_if_deprecated(self):
1866
global _deprecation_warning_done
1867
if _deprecation_warning_done:
1869
_deprecation_warning_done = True
1870
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1871
% (self._format, self.bzrdir.transport.base))
1873
def supports_rich_root(self):
1874
return self._format.rich_root_data
1876
def _check_ascii_revisionid(self, revision_id, method):
1877
"""Private helper for ascii-only repositories."""
1878
# weave repositories refuse to store revisionids that are non-ascii.
1879
if revision_id is not None:
1880
# weaves require ascii revision ids.
1881
if isinstance(revision_id, unicode):
1883
revision_id.encode('ascii')
1884
except UnicodeEncodeError:
1885
raise errors.NonAsciiRevisionId(method, self)
1888
revision_id.decode('ascii')
1889
except UnicodeDecodeError:
1890
raise errors.NonAsciiRevisionId(method, self)
1892
def revision_graph_can_have_wrong_parents(self):
1893
"""Is it possible for this repository to have a revision graph with
1896
If True, then this repository must also implement
1897
_find_inconsistent_revision_parents so that check and reconcile can
1898
check for inconsistencies before proceeding with other checks that may
1899
depend on the revision index being consistent.
1901
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
1904
# remove these delegates a while after bzr 0.15
1905
def __make_delegated(name, from_module):
1906
def _deprecated_repository_forwarder():
1907
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1908
% (name, from_module),
1911
m = __import__(from_module, globals(), locals(), [name])
1913
return getattr(m, name)
1914
except AttributeError:
1915
raise AttributeError('module %s has no name %s'
1917
globals()[name] = _deprecated_repository_forwarder
1920
'AllInOneRepository',
1921
'WeaveMetaDirRepository',
1922
'PreSplitOutRepositoryFormat',
1923
'RepositoryFormat4',
1924
'RepositoryFormat5',
1925
'RepositoryFormat6',
1926
'RepositoryFormat7',
1928
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1932
'RepositoryFormatKnit',
1933
'RepositoryFormatKnit1',
1935
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1938
def install_revision(repository, rev, revision_tree):
1939
"""Install all revision data into a repository."""
1940
install_revisions(repository, [(rev, revision_tree, None)])
1943
def install_revisions(repository, iterable, num_revisions=None, pb=None):
1944
"""Install all revision data into a repository.
1946
Accepts an iterable of revision, tree, signature tuples. The signature
1949
repository.start_write_group()
1951
for n, (revision, revision_tree, signature) in enumerate(iterable):
1952
_install_revision(repository, revision, revision_tree, signature)
1954
pb.update('Transferring revisions', n + 1, num_revisions)
1956
repository.abort_write_group()
1959
repository.commit_write_group()
1962
def _install_revision(repository, rev, revision_tree, signature):
1963
"""Install all revision data into a repository."""
1964
present_parents = []
1966
for p_id in rev.parent_ids:
1967
if repository.has_revision(p_id):
1968
present_parents.append(p_id)
1969
parent_trees[p_id] = repository.revision_tree(p_id)
1971
parent_trees[p_id] = repository.revision_tree(
1972
_mod_revision.NULL_REVISION)
1974
inv = revision_tree.inventory
1975
entries = inv.iter_entries()
1976
# backwards compatibility hack: skip the root id.
1977
if not repository.supports_rich_root():
1978
path, root = entries.next()
1979
if root.revision != rev.revision_id:
1980
raise errors.IncompatibleRevision(repr(repository))
1982
for path, ie in entries:
1983
text_keys[(ie.file_id, ie.revision)] = ie
1984
text_parent_map = repository.texts.get_parent_map(text_keys)
1985
missing_texts = set(text_keys) - set(text_parent_map)
1986
# Add the texts that are not already present
1987
for text_key in missing_texts:
1988
ie = text_keys[text_key]
1990
# FIXME: TODO: The following loop overlaps/duplicates that done by
1991
# commit to determine parents. There is a latent/real bug here where
1992
# the parents inserted are not those commit would do - in particular
1993
# they are not filtered by heads(). RBC, AB
1994
for revision, tree in parent_trees.iteritems():
1995
if ie.file_id not in tree:
1997
parent_id = tree.inventory[ie.file_id].revision
1998
if parent_id in text_parents:
2000
text_parents.append((ie.file_id, parent_id))
2001
lines = revision_tree.get_file(ie.file_id).readlines()
2002
repository.texts.add_lines(text_key, text_parents, lines)
2004
# install the inventory
2005
repository.add_inventory(rev.revision_id, inv, present_parents)
2006
except errors.RevisionAlreadyPresent:
2008
if signature is not None:
2009
repository.add_signature_text(rev.revision_id, signature)
2010
repository.add_revision(rev.revision_id, rev, inv)
2013
class MetaDirRepository(Repository):
2014
"""Repositories in the new meta-dir layout.
2016
:ivar _transport: Transport for access to repository control files,
2017
typically pointing to .bzr/repository.
2020
def __init__(self, _format, a_bzrdir, control_files):
2021
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2022
self._transport = control_files._transport
2024
def is_shared(self):
2025
"""Return True if this repository is flagged as a shared repository."""
2026
return self._transport.has('shared-storage')
2029
def set_make_working_trees(self, new_value):
2030
"""Set the policy flag for making working trees when creating branches.
2032
This only applies to branches that use this repository.
2034
The default is 'True'.
2035
:param new_value: True to restore the default, False to disable making
2040
self._transport.delete('no-working-trees')
2041
except errors.NoSuchFile:
2044
self._transport.put_bytes('no-working-trees', '',
2045
mode=self.bzrdir._get_file_mode())
2047
def make_working_trees(self):
2048
"""Returns the policy for making working trees on new branches."""
2049
return not self._transport.has('no-working-trees')
2052
class MetaDirVersionedFileRepository(MetaDirRepository):
2053
"""Repositories in a meta-dir, that work via versioned file objects."""
2055
def __init__(self, _format, a_bzrdir, control_files):
2056
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2060
class RepositoryFormatRegistry(registry.Registry):
2061
"""Registry of RepositoryFormats."""
2063
def get(self, format_string):
2064
r = registry.Registry.get(self, format_string)
2070
format_registry = RepositoryFormatRegistry()
2071
"""Registry of formats, indexed by their identifying format string.
2073
This can contain either format instances themselves, or classes/factories that
2074
can be called to obtain one.
2078
#####################################################################
2079
# Repository Formats
2081
517
class RepositoryFormat(object):
2082
518
"""A repository format.
2214
def check_conversion_target(self, target_format):
2215
raise NotImplementedError(self.check_conversion_target)
2217
591
def open(self, a_bzrdir, _found=False):
2218
592
"""Return an instance of this format for the bzrdir a_bzrdir.
2220
594
_found is a private parameter, do not use it.
2222
raise NotImplementedError(self.open)
2225
class MetaDirRepositoryFormat(RepositoryFormat):
2226
"""Common base class for the new repositories using the metadir layout."""
2228
rich_root_data = False
2229
supports_tree_reference = False
2230
supports_external_lookups = False
2231
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2234
super(MetaDirRepositoryFormat, self).__init__()
2236
def _create_control_files(self, a_bzrdir):
2237
"""Create the required files and the initial control_files object."""
2238
# FIXME: RBC 20060125 don't peek under the covers
597
# we are being called directly and must probe.
598
raise NotImplementedError
599
return Repository(_format=self, a_bzrdir=a_bzrdir)
602
def register_format(klass, format):
603
klass._formats[format.get_format_string()] = format
606
def set_default_format(klass, format):
607
klass._default_format = format
610
def unregister_format(klass, format):
611
assert klass._formats[format.get_format_string()] is format
612
del klass._formats[format.get_format_string()]
615
class PreSplitOutRepositoryFormat(RepositoryFormat):
616
"""Base class for the pre split out repository formats."""
618
def initialize(self, a_bzrdir, shared=False, _internal=False):
619
"""Create a weave repository.
621
TODO: when creating split out bzr branch formats, move this to a common
622
base for Format5, Format6. or something like that.
624
from bzrlib.weavefile import write_weave_v5
625
from bzrlib.weave import Weave
628
raise errors.IncompatibleFormat(self, a_bzrdir._format)
631
# always initialized when the bzrdir is.
632
return Repository(_format=self, a_bzrdir=a_bzrdir)
634
# Create an empty weave
636
bzrlib.weavefile.write_weave_v5(Weave(), sio)
637
empty_weave = sio.getvalue()
639
mutter('creating repository in %s.', a_bzrdir.transport.base)
640
dirs = ['revision-store', 'weaves']
641
lock_file = 'branch-lock'
642
files = [('inventory.weave', StringIO(empty_weave)),
645
# FIXME: RBC 20060125 dont peek under the covers
646
# NB: no need to escape relative paths that are url safe.
647
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
649
control_files.lock_write()
650
control_files._transport.mkdir_multi(dirs,
651
mode=control_files._dir_mode)
653
for file, content in files:
654
control_files.put(file, content)
656
control_files.unlock()
657
return Repository(_format=self, a_bzrdir=a_bzrdir)
660
class RepositoryFormat4(PreSplitOutRepositoryFormat):
661
"""Bzr repository format 4.
663
This repository format has:
665
- TextStores for texts, inventories,revisions.
667
This format is deprecated: it indexes texts using a text id which is
668
removed in format 5; initializationa and write support for this format
673
super(RepositoryFormat4, self).__init__()
674
self._matchingbzrdir = bzrdir.BzrDirFormat4()
676
def initialize(self, url, shared=False, _internal=False):
677
"""Format 4 branches cannot be created."""
678
raise errors.UninitializableFormat(self)
680
def is_supported(self):
681
"""Format 4 is not supported.
683
It is not supported because the model changed from 4 to 5 and the
684
conversion logic is expensive - so doing it on the fly was not
690
class RepositoryFormat5(PreSplitOutRepositoryFormat):
691
"""Bzr control format 5.
693
This repository format has:
694
- weaves for file texts and inventory
696
- TextStores for revisions and signatures.
700
super(RepositoryFormat5, self).__init__()
701
self._matchingbzrdir = bzrdir.BzrDirFormat5()
704
class RepositoryFormat6(PreSplitOutRepositoryFormat):
705
"""Bzr control format 6.
707
This repository format has:
708
- weaves for file texts and inventory
709
- hash subdirectory based stores.
710
- TextStores for revisions and signatures.
714
super(RepositoryFormat6, self).__init__()
715
self._matchingbzrdir = bzrdir.BzrDirFormat6()
718
class RepositoryFormat7(RepositoryFormat):
721
This repository format has:
722
- weaves for file texts and inventory
723
- hash subdirectory based stores.
724
- TextStores for revisions and signatures.
725
- a format marker of its own
726
- an optional 'shared-storage' flag
729
def get_format_string(self):
730
"""See RepositoryFormat.get_format_string()."""
731
return "Bazaar-NG Repository format 7"
733
def initialize(self, a_bzrdir, shared=False):
734
"""Create a weave repository.
736
:param shared: If true the repository will be initialized as a shared
739
from bzrlib.weavefile import write_weave_v5
740
from bzrlib.weave import Weave
742
# Create an empty weave
744
bzrlib.weavefile.write_weave_v5(Weave(), sio)
745
empty_weave = sio.getvalue()
747
mutter('creating repository in %s.', a_bzrdir.transport.base)
748
dirs = ['revision-store', 'weaves']
749
files = [('inventory.weave', StringIO(empty_weave)),
751
utf8_files = [('format', self.get_format_string())]
753
# FIXME: RBC 20060125 dont peek under the covers
2239
754
# NB: no need to escape relative paths that are url safe.
2240
755
repository_transport = a_bzrdir.get_repository_transport(self)
2241
control_files = lockable_files.LockableFiles(repository_transport,
2242
'lock', lockdir.LockDir)
2243
control_files.create_lock()
2244
return control_files
2246
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
2247
"""Upload the initial blank content."""
2248
control_files = self._create_control_files(a_bzrdir)
756
control_files = LockableFiles(repository_transport, 'lock',
2249
758
control_files.lock_write()
2250
transport = control_files._transport
2252
utf8_files += [('shared-storage', '')]
2254
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2255
for (filename, content_stream) in files:
2256
transport.put_file(filename, content_stream,
2257
mode=a_bzrdir._get_file_mode())
2258
for (filename, content_bytes) in utf8_files:
2259
transport.put_bytes_non_atomic(filename, content_bytes,
2260
mode=a_bzrdir._get_file_mode())
760
control_files._transport.mkdir_multi(dirs,
761
mode=control_files._dir_mode)
762
for file, content in files:
763
control_files.put(file, content)
764
for file, content in utf8_files:
765
control_files.put_utf8(file, content)
767
control_files.put_utf8('shared-storage', '')
2262
769
control_files.unlock()
770
return Repository(_format=self, a_bzrdir=a_bzrdir)
773
super(RepositoryFormat7, self).__init__()
774
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
2265
777
# formats which have no format string are not discoverable
2266
# and not independently creatable, so are not registered. They're
2267
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2268
# needed, it's constructed directly by the BzrDir. Non-native formats where
2269
# the repository is not separately opened are similar.
2271
format_registry.register_lazy(
2272
'Bazaar-NG Repository format 7',
2273
'bzrlib.repofmt.weaverepo',
2277
format_registry.register_lazy(
2278
'Bazaar-NG Knit Repository Format 1',
2279
'bzrlib.repofmt.knitrepo',
2280
'RepositoryFormatKnit1',
2283
format_registry.register_lazy(
2284
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2285
'bzrlib.repofmt.knitrepo',
2286
'RepositoryFormatKnit3',
2289
format_registry.register_lazy(
2290
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2291
'bzrlib.repofmt.knitrepo',
2292
'RepositoryFormatKnit4',
2295
# Pack-based formats. There is one format for pre-subtrees, and one for
2296
# post-subtrees to allow ease of testing.
2297
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2298
format_registry.register_lazy(
2299
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2300
'bzrlib.repofmt.pack_repo',
2301
'RepositoryFormatKnitPack1',
2303
format_registry.register_lazy(
2304
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2305
'bzrlib.repofmt.pack_repo',
2306
'RepositoryFormatKnitPack3',
2308
format_registry.register_lazy(
2309
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2310
'bzrlib.repofmt.pack_repo',
2311
'RepositoryFormatKnitPack4',
2313
format_registry.register_lazy(
2314
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2315
'bzrlib.repofmt.pack_repo',
2316
'RepositoryFormatKnitPack5',
2318
format_registry.register_lazy(
2319
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2320
'bzrlib.repofmt.pack_repo',
2321
'RepositoryFormatKnitPack5RichRoot',
2323
format_registry.register_lazy(
2324
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2325
'bzrlib.repofmt.pack_repo',
2326
'RepositoryFormatKnitPack5RichRootBroken',
2329
# Development formats.
2330
# 1.7->1.8 go below here
2331
format_registry.register_lazy(
2332
"Bazaar development format 2 (needs bzr.dev from before 1.8)\n",
2333
'bzrlib.repofmt.pack_repo',
2334
'RepositoryFormatPackDevelopment2',
2336
format_registry.register_lazy(
2337
("Bazaar development format 2 with subtree support "
2338
"(needs bzr.dev from before 1.8)\n"),
2339
'bzrlib.repofmt.pack_repo',
2340
'RepositoryFormatPackDevelopment2Subtree',
2344
class InterRepository(InterObject):
778
# and not independently creatable, so are not registered.
779
_default_format = RepositoryFormat7()
780
RepositoryFormat.register_format(_default_format)
781
RepositoryFormat.set_default_format(_default_format)
782
_legacy_formats = [RepositoryFormat4(),
787
class InterRepository(object):
2345
788
"""This class represents operations taking place between two repositories.
2347
790
Its instances have methods like copy_content and fetch, and contain
2352
795
operations with another repository - they will always forward to
2353
796
InterRepository.get(other).method_name(parameters).
798
# XXX: FIXME: FUTURE: robertc
799
# testing of these probably requires a factory in optimiser type, and
800
# then a test adapter to test each type thoroughly.
2357
804
"""The available optimised InterRepository types."""
2359
def copy_content(self, revision_id=None):
2360
raise NotImplementedError(self.copy_content)
2362
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
806
def __init__(self, source, target):
807
"""Construct a default InterRepository instance. Please use 'get'.
809
Only subclasses of InterRepository should call
810
InterRepository.__init__ - clients should call InterRepository.get
811
instead which will create an optimised InterRepository if possible.
817
def copy_content(self, revision_id=None, basis=None):
818
"""Make a complete copy of the content in self into destination.
820
This is a destructive operation! Do not use it on existing
823
:param revision_id: Only copy the content needed to construct
824
revision_id and its parents.
825
:param basis: Copy the needed data preferentially from basis.
828
self.target.set_make_working_trees(self.source.make_working_trees())
829
except NotImplementedError:
831
# grab the basis available data
832
if basis is not None:
833
self.target.fetch(basis, revision_id=revision_id)
834
# but dont both fetching if we have the needed data now.
835
if (revision_id not in (None, NULL_REVISION) and
836
self.target.has_revision(revision_id)):
838
self.target.fetch(self.source, revision_id=revision_id)
840
def _double_lock(self, lock_source, lock_target):
841
"""Take out too locks, rolling back the first if the second throws."""
846
# we want to ensure that we don't leave source locked by mistake.
847
# and any error on target should not confuse source.
852
def fetch(self, revision_id=None, pb=None):
2363
853
"""Fetch the content required to construct revision_id.
2365
The content is copied from self.source to self.target.
855
The content is copied from source to target.
2367
857
:param revision_id: if None all content is copied, if NULL_REVISION no
2368
858
content is copied.
2369
859
:param pb: optional progress bar to use for progress reports. If not
2370
860
provided a default one will be created.
2372
:returns: (copied_revision_count, failures).
2374
# Normally we should find a specific InterRepository subclass to do
2375
# the fetch; if nothing else then at least InterSameDataRepository.
2376
# If none of them is suitable it looks like fetching is not possible;
2377
# we try to give a good message why. _assert_same_model will probably
2378
# give a helpful message; otherwise a generic one.
2379
self._assert_same_model(self.source, self.target)
2380
raise errors.IncompatibleRepositories(self.source, self.target,
2381
"no suitableInterRepository found")
2383
def _walk_to_common_revisions(self, revision_ids):
2384
"""Walk out from revision_ids in source to revisions target has.
2386
:param revision_ids: The start point for the search.
2387
:return: A set of revision ids.
2389
target_graph = self.target.get_graph()
2390
revision_ids = frozenset(revision_ids)
2391
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2392
return graph.SearchResult(revision_ids, set(), 0, set())
2393
missing_revs = set()
2394
source_graph = self.source.get_graph()
2395
# ensure we don't pay silly lookup costs.
2396
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2397
null_set = frozenset([_mod_revision.NULL_REVISION])
2400
next_revs, ghosts = searcher.next_with_ghosts()
2401
except StopIteration:
2403
if revision_ids.intersection(ghosts):
2404
absent_ids = set(revision_ids.intersection(ghosts))
2405
# If all absent_ids are present in target, no error is needed.
2406
absent_ids.difference_update(
2407
set(target_graph.get_parent_map(absent_ids)))
2409
raise errors.NoSuchRevision(self.source, absent_ids.pop())
2410
# we don't care about other ghosts as we can't fetch them and
2411
# haven't been asked to.
2412
next_revs = set(next_revs)
2413
# we always have NULL_REVISION present.
2414
have_revs = set(target_graph.get_parent_map(next_revs)).union(null_set)
2415
missing_revs.update(next_revs - have_revs)
2416
searcher.stop_searching_any(have_revs)
2417
return searcher.get_result()
2419
@deprecated_method(one_two)
862
Returns the copied revision count and the failed revisions in a tuple:
865
from bzrlib.fetch import RepoFetcher
866
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
867
self.source, self.source._format, self.target, self.target._format)
868
f = RepoFetcher(to_repository=self.target,
869
from_repository=self.source,
870
last_revision=revision_id,
872
return f.count_copied, f.failed_revisions
875
def get(klass, repository_source, repository_target):
876
"""Retrieve a InterRepository worker object for these repositories.
878
:param repository_source: the repository to be the 'source' member of
879
the InterRepository instance.
880
:param repository_target: the repository to be the 'target' member of
881
the InterRepository instance.
882
If an optimised InterRepository worker exists it will be used otherwise
883
a default InterRepository instance will be created.
885
for provider in klass._optimisers:
886
if provider.is_compatible(repository_source, repository_target):
887
return provider(repository_source, repository_target)
888
return InterRepository(repository_source, repository_target)
891
"""Take out a logical read lock.
893
This will lock the source branch and the target branch. The source gets
894
a read lock and the target a read lock.
896
self._double_lock(self.source.lock_read, self.target.lock_read)
898
def lock_write(self):
899
"""Take out a logical write lock.
901
This will lock the source branch and the target branch. The source gets
902
a read lock and the target a write lock.
904
self._double_lock(self.source.lock_read, self.target.lock_write)
2420
906
@needs_read_lock
2421
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
907
def missing_revision_ids(self, revision_id=None):
2422
908
"""Return the revision ids that source has that target does not.
2424
910
These are returned in topological order.
2426
912
:param revision_id: only return revision ids included by this
2428
:param find_ghosts: If True find missing revisions in deep history
2429
rather than just finding the surface difference.
2431
return list(self.search_missing_revision_ids(
2432
revision_id, find_ghosts).get_keys())
2435
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2436
"""Return the revision ids that source has that target does not.
2438
:param revision_id: only return revision ids included by this
2440
:param find_ghosts: If True find missing revisions in deep history
2441
rather than just finding the surface difference.
2442
:return: A bzrlib.graph.SearchResult.
2444
# stop searching at found target revisions.
2445
if not find_ghosts and revision_id is not None:
2446
return self._walk_to_common_revisions([revision_id])
2447
915
# generic, possibly worst case, slow code path.
2448
target_ids = set(self.target.all_revision_ids())
916
target_ids = set(self.source.all_revision_ids())
2449
917
if revision_id is not None:
2450
source_ids = self.source.get_ancestry(revision_id)
2451
if source_ids[0] is not None:
2452
raise AssertionError()
918
source_ids = self.target.get_ancestry(revision_id)
919
assert source_ids.pop(0) == None
2455
source_ids = self.source.all_revision_ids()
921
source_ids = self.target.all_revision_ids()
2456
922
result_set = set(source_ids).difference(target_ids)
2457
return self.source.revision_ids_to_search_result(result_set)
2460
def _same_model(source, target):
2461
"""True if source and target have the same data representation.
2463
Note: this is always called on the base class; overriding it in a
2464
subclass will have no effect.
2467
InterRepository._assert_same_model(source, target)
2469
except errors.IncompatibleRepositories, e:
2473
def _assert_same_model(source, target):
2474
"""Raise an exception if two repositories do not use the same model.
2476
if source.supports_rich_root() != target.supports_rich_root():
2477
raise errors.IncompatibleRepositories(source, target,
2478
"different rich-root support")
2479
if source._serializer != target._serializer:
2480
raise errors.IncompatibleRepositories(source, target,
2481
"different serializers")
2484
class InterSameDataRepository(InterRepository):
2485
"""Code for converting between repositories that represent the same data.
2487
Data format and model must match for this to work.
2491
def _get_repo_format_to_test(self):
2492
"""Repository format for testing with.
2494
InterSameData can pull from subtree to subtree and from non-subtree to
2495
non-subtree, so we test this with the richest repository format.
2497
from bzrlib.repofmt import knitrepo
2498
return knitrepo.RepositoryFormatKnit3()
2501
def is_compatible(source, target):
2502
return InterRepository._same_model(source, target)
2505
def copy_content(self, revision_id=None):
2506
"""Make a complete copy of the content in self into destination.
2508
This copies both the repository's revision data, and configuration information
2509
such as the make_working_trees setting.
2511
This is a destructive operation! Do not use it on existing
2514
:param revision_id: Only copy the content needed to construct
2515
revision_id and its parents.
2518
self.target.set_make_working_trees(self.source.make_working_trees())
2519
except NotImplementedError:
2521
# but don't bother fetching if we have the needed data now.
2522
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2523
self.target.has_revision(revision_id)):
2525
self.target.fetch(self.source, revision_id=revision_id)
2528
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2529
"""See InterRepository.fetch()."""
2530
from bzrlib.fetch import RepoFetcher
2531
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2532
self.source, self.source._format, self.target,
2533
self.target._format)
2534
f = RepoFetcher(to_repository=self.target,
2535
from_repository=self.source,
2536
last_revision=revision_id,
2537
pb=pb, find_ghosts=find_ghosts)
2538
return f.count_copied, f.failed_revisions
2541
class InterWeaveRepo(InterSameDataRepository):
2542
"""Optimised code paths between Weave based repositories.
2544
This should be in bzrlib/repofmt/weaverepo.py but we have not yet
2545
implemented lazy inter-object optimisation.
2549
def _get_repo_format_to_test(self):
2550
from bzrlib.repofmt import weaverepo
2551
return weaverepo.RepositoryFormat7()
923
# this may look like a no-op: its not. It preserves the ordering
924
# other_ids had while only returning the members from other_ids
925
# that we've decided we need.
926
return [rev_id for rev_id in other_ids if rev_id in result_set]
929
def register_optimiser(klass, optimiser):
930
"""Register an InterRepository optimiser."""
931
klass._optimisers.add(optimiser)
934
"""Release the locks on source and target."""
941
def unregister_optimiser(klass, optimiser):
942
"""Unregister an InterRepository optimiser."""
943
klass._optimisers.remove(optimiser)
946
class InterWeaveRepo(InterRepository):
947
"""Optimised code paths between Weave based repositories."""
949
_matching_repo_format = _default_format
950
"""Repository format for testing with."""
2554
953
def is_compatible(source, target):
2555
954
"""Be compatible with known Weave formats.
2557
We don't test for the stores being of specific types because that
956
We dont test for the stores being of specific types becase that
2558
957
could lead to confusing results, and there is no need to be
2561
from bzrlib.repofmt.weaverepo import (
2567
961
return (isinstance(source._format, (RepositoryFormat5,
2568
962
RepositoryFormat6,
2633
1026
# - RBC 20060209
2634
1027
if revision_id is not None:
2635
1028
source_ids = self.source.get_ancestry(revision_id)
2636
if source_ids[0] is not None:
2637
raise AssertionError()
1029
assert source_ids.pop(0) == None
2640
1031
source_ids = self.source._all_possible_ids()
2641
1032
source_ids_set = set(source_ids)
2642
1033
# source_ids is the worst possible case we may need to pull.
2643
1034
# now we want to filter source_ids against what we actually
2644
# have in target, but don't try to check for existence where we know
1035
# have in target, but dont try to check for existence where we know
2645
1036
# we do not have a revision as that would be pointless.
2646
1037
target_ids = set(self.target._all_possible_ids())
2647
1038
possibly_present_revisions = target_ids.intersection(source_ids_set)
2648
actually_present_revisions = set(
2649
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2650
required_revisions = source_ids_set.difference(actually_present_revisions)
2651
if revision_id is not None:
2652
# we used get_ancestry to determine source_ids then we are assured all
2653
# revisions referenced are present as they are installed in topological order.
2654
# and the tip revision was validated by get_ancestry.
2655
result_set = required_revisions
2657
# if we just grabbed the possibly available ids, then
2658
# we only have an estimate of whats available and need to validate
2659
# that against the revision records.
2661
self.source._eliminate_revisions_not_present(required_revisions))
2662
return self.source.revision_ids_to_search_result(result_set)
2665
class InterKnitRepo(InterSameDataRepository):
2666
"""Optimised code paths between Knit based repositories."""
2669
def _get_repo_format_to_test(self):
2670
from bzrlib.repofmt import knitrepo
2671
return knitrepo.RepositoryFormatKnit1()
2674
def is_compatible(source, target):
2675
"""Be compatible with known Knit formats.
2677
We don't test for the stores being of specific types because that
2678
could lead to confusing results, and there is no need to be
2681
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
2683
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2684
isinstance(target._format, RepositoryFormatKnit))
2685
except AttributeError:
2687
return are_knits and InterRepository._same_model(source, target)
2690
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2691
"""See InterRepository.fetch()."""
2692
from bzrlib.fetch import RepoFetcher
2693
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2694
self.source, self.source._format, self.target, self.target._format)
2695
f = RepoFetcher(to_repository=self.target,
2696
from_repository=self.source,
2697
last_revision=revision_id,
2698
pb=pb, find_ghosts=find_ghosts)
2699
return f.count_copied, f.failed_revisions
2702
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2703
"""See InterRepository.missing_revision_ids()."""
2704
if revision_id is not None:
2705
source_ids = self.source.get_ancestry(revision_id)
2706
if source_ids[0] is not None:
2707
raise AssertionError()
2710
source_ids = self.source.all_revision_ids()
2711
source_ids_set = set(source_ids)
2712
# source_ids is the worst possible case we may need to pull.
2713
# now we want to filter source_ids against what we actually
2714
# have in target, but don't try to check for existence where we know
2715
# we do not have a revision as that would be pointless.
2716
target_ids = set(self.target.all_revision_ids())
2717
possibly_present_revisions = target_ids.intersection(source_ids_set)
2718
actually_present_revisions = set(
2719
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2720
required_revisions = source_ids_set.difference(actually_present_revisions)
2721
if revision_id is not None:
2722
# we used get_ancestry to determine source_ids then we are assured all
2723
# revisions referenced are present as they are installed in topological order.
2724
# and the tip revision was validated by get_ancestry.
2725
result_set = required_revisions
2727
# if we just grabbed the possibly available ids, then
2728
# we only have an estimate of whats available and need to validate
2729
# that against the revision records.
2731
self.source._eliminate_revisions_not_present(required_revisions))
2732
return self.source.revision_ids_to_search_result(result_set)
2735
class InterPackRepo(InterSameDataRepository):
2736
"""Optimised code paths between Pack based repositories."""
2739
def _get_repo_format_to_test(self):
2740
from bzrlib.repofmt import pack_repo
2741
return pack_repo.RepositoryFormatKnitPack1()
2744
def is_compatible(source, target):
2745
"""Be compatible with known Pack formats.
2747
We don't test for the stores being of specific types because that
2748
could lead to confusing results, and there is no need to be
2751
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2753
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2754
isinstance(target._format, RepositoryFormatPack))
2755
except AttributeError:
2757
return are_packs and InterRepository._same_model(source, target)
2760
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2761
"""See InterRepository.fetch()."""
2762
if (len(self.source._fallback_repositories) > 0 or
2763
len(self.target._fallback_repositories) > 0):
2764
# The pack layer is not aware of fallback repositories, so when
2765
# fetching from a stacked repository or into a stacked repository
2766
# we use the generic fetch logic which uses the VersionedFiles
2767
# attributes on repository.
2768
from bzrlib.fetch import RepoFetcher
2769
fetcher = RepoFetcher(self.target, self.source, revision_id,
2771
return fetcher.count_copied, fetcher.failed_revisions
2772
from bzrlib.repofmt.pack_repo import Packer
2773
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2774
self.source, self.source._format, self.target, self.target._format)
2775
self.count_copied = 0
2776
if revision_id is None:
2778
# everything to do - use pack logic
2779
# to fetch from all packs to one without
2780
# inventory parsing etc, IFF nothing to be copied is in the target.
2782
source_revision_ids = frozenset(self.source.all_revision_ids())
2783
revision_ids = source_revision_ids - \
2784
frozenset(self.target.get_parent_map(source_revision_ids))
2785
revision_keys = [(revid,) for revid in revision_ids]
2786
index = self.target._pack_collection.revision_index.combined_index
2787
present_revision_ids = set(item[1][0] for item in
2788
index.iter_entries(revision_keys))
2789
revision_ids = set(revision_ids) - present_revision_ids
2790
# implementing the TODO will involve:
2791
# - detecting when all of a pack is selected
2792
# - avoiding as much as possible pre-selection, so the
2793
# more-core routines such as create_pack_from_packs can filter in
2794
# a just-in-time fashion. (though having a HEADS list on a
2795
# repository might make this a lot easier, because we could
2796
# sensibly detect 'new revisions' without doing a full index scan.
2797
elif _mod_revision.is_null(revision_id):
2802
revision_ids = self.search_missing_revision_ids(revision_id,
2803
find_ghosts=find_ghosts).get_keys()
2804
except errors.NoSuchRevision:
2805
raise errors.InstallFailed([revision_id])
2806
if len(revision_ids) == 0:
2808
packs = self.source._pack_collection.all_packs()
2809
pack = Packer(self.target._pack_collection, packs, '.fetch',
2810
revision_ids).pack()
2811
if pack is not None:
2812
self.target._pack_collection._save_pack_names()
2813
# Trigger an autopack. This may duplicate effort as we've just done
2814
# a pack creation, but for now it is simpler to think about as
2815
# 'upload data, then repack if needed'.
2816
self.target._pack_collection.autopack()
2817
return (pack.get_revision_count(), [])
2822
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2823
"""See InterRepository.missing_revision_ids().
2825
:param find_ghosts: Find ghosts throughout the ancestry of
2828
if not find_ghosts and revision_id is not None:
2829
return self._walk_to_common_revisions([revision_id])
2830
elif revision_id is not None:
2831
# Find ghosts: search for revisions pointing from one repository to
2832
# the other, and vice versa, anywhere in the history of revision_id.
2833
graph = self.target.get_graph(other_repository=self.source)
2834
searcher = graph._make_breadth_first_searcher([revision_id])
2838
next_revs, ghosts = searcher.next_with_ghosts()
2839
except StopIteration:
2841
if revision_id in ghosts:
2842
raise errors.NoSuchRevision(self.source, revision_id)
2843
found_ids.update(next_revs)
2844
found_ids.update(ghosts)
2845
found_ids = frozenset(found_ids)
2846
# Double query here: should be able to avoid this by changing the
2847
# graph api further.
2848
result_set = found_ids - frozenset(
2849
self.target.get_parent_map(found_ids))
2851
source_ids = self.source.all_revision_ids()
2852
# source_ids is the worst possible case we may need to pull.
2853
# now we want to filter source_ids against what we actually
2854
# have in target, but don't try to check for existence where we know
2855
# we do not have a revision as that would be pointless.
2856
target_ids = set(self.target.all_revision_ids())
2857
result_set = set(source_ids).difference(target_ids)
2858
return self.source.revision_ids_to_search_result(result_set)
2861
class InterModel1and2(InterRepository):
2864
def _get_repo_format_to_test(self):
2868
def is_compatible(source, target):
2869
if not source.supports_rich_root() and target.supports_rich_root():
2875
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2876
"""See InterRepository.fetch()."""
2877
from bzrlib.fetch import Model1toKnit2Fetcher
2878
f = Model1toKnit2Fetcher(to_repository=self.target,
2879
from_repository=self.source,
2880
last_revision=revision_id,
2881
pb=pb, find_ghosts=find_ghosts)
2882
return f.count_copied, f.failed_revisions
2885
def copy_content(self, revision_id=None):
2886
"""Make a complete copy of the content in self into destination.
2888
This is a destructive operation! Do not use it on existing
2891
:param revision_id: Only copy the content needed to construct
2892
revision_id and its parents.
2895
self.target.set_make_working_trees(self.source.make_working_trees())
2896
except NotImplementedError:
2898
# but don't bother fetching if we have the needed data now.
2899
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2900
self.target.has_revision(revision_id)):
2902
self.target.fetch(self.source, revision_id=revision_id)
2905
class InterKnit1and2(InterKnitRepo):
2908
def _get_repo_format_to_test(self):
2912
def is_compatible(source, target):
2913
"""Be compatible with Knit1 source and Knit3 target"""
2915
from bzrlib.repofmt.knitrepo import (
2916
RepositoryFormatKnit1,
2917
RepositoryFormatKnit3,
2919
from bzrlib.repofmt.pack_repo import (
2920
RepositoryFormatKnitPack1,
2921
RepositoryFormatKnitPack3,
2922
RepositoryFormatKnitPack4,
2923
RepositoryFormatKnitPack5,
2924
RepositoryFormatKnitPack5RichRoot,
2925
RepositoryFormatPackDevelopment2,
2926
RepositoryFormatPackDevelopment2Subtree,
2929
RepositoryFormatKnit1, # no rr, no subtree
2930
RepositoryFormatKnitPack1, # no rr, no subtree
2931
RepositoryFormatPackDevelopment2, # no rr, no subtree
2932
RepositoryFormatKnitPack5, # no rr, no subtree
2935
RepositoryFormatKnit3, # rr, subtree
2936
RepositoryFormatKnitPack3, # rr, subtree
2937
RepositoryFormatKnitPack4, # rr, no subtree
2938
RepositoryFormatKnitPack5RichRoot,# rr, no subtree
2939
RepositoryFormatPackDevelopment2Subtree, # rr, subtree
2941
for format in norichroot:
2942
if format.rich_root_data:
2943
raise AssertionError('Format %s is a rich-root format'
2944
' but is included in the non-rich-root list'
2946
for format in richroot:
2947
if not format.rich_root_data:
2948
raise AssertionError('Format %s is not a rich-root format'
2949
' but is included in the rich-root list'
2951
# TODO: One alternative is to just check format.rich_root_data,
2952
# instead of keeping membership lists. However, the formats
2953
# *also* have to use the same 'Knit' style of storage
2954
# (line-deltas, fulltexts, etc.)
2955
return (isinstance(source._format, norichroot) and
2956
isinstance(target._format, richroot))
2957
except AttributeError:
2961
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2962
"""See InterRepository.fetch()."""
2963
from bzrlib.fetch import Knit1to2Fetcher
2964
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2965
self.source, self.source._format, self.target,
2966
self.target._format)
2967
f = Knit1to2Fetcher(to_repository=self.target,
2968
from_repository=self.source,
2969
last_revision=revision_id,
2970
pb=pb, find_ghosts=find_ghosts)
2971
return f.count_copied, f.failed_revisions
2974
class InterDifferingSerializer(InterKnitRepo):
2977
def _get_repo_format_to_test(self):
2981
def is_compatible(source, target):
2982
"""Be compatible with Knit2 source and Knit3 target"""
2983
if source.supports_rich_root() != target.supports_rich_root():
2985
# Ideally, we'd support fetching if the source had no tree references
2986
# even if it supported them...
2987
if (getattr(source, '_format.supports_tree_reference', False) and
2988
not getattr(target, '_format.supports_tree_reference', False)):
2993
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2994
"""See InterRepository.fetch()."""
2995
revision_ids = self.target.search_missing_revision_ids(self.source,
2996
revision_id, find_ghosts=find_ghosts).get_keys()
2997
revision_ids = tsort.topo_sort(
2998
self.source.get_graph().get_parent_map(revision_ids))
2999
def revisions_iterator():
3000
for current_revision_id in revision_ids:
3001
revision = self.source.get_revision(current_revision_id)
3002
tree = self.source.revision_tree(current_revision_id)
3004
signature = self.source.get_signature_text(
3005
current_revision_id)
3006
except errors.NoSuchRevision:
3008
yield revision, tree, signature
3010
my_pb = ui.ui_factory.nested_progress_bar()
3015
install_revisions(self.target, revisions_iterator(),
3016
len(revision_ids), pb)
3018
if my_pb is not None:
3020
return len(revision_ids), 0
3023
class InterOtherToRemote(InterRepository):
3025
def __init__(self, source, target):
3026
InterRepository.__init__(self, source, target)
3027
self._real_inter = None
3030
def is_compatible(source, target):
3031
if isinstance(target, remote.RemoteRepository):
3035
def _ensure_real_inter(self):
3036
if self._real_inter is None:
3037
self.target._ensure_real()
3038
real_target = self.target._real_repository
3039
self._real_inter = InterRepository.get(self.source, real_target)
3041
def copy_content(self, revision_id=None):
3042
self._ensure_real_inter()
3043
self._real_inter.copy_content(revision_id=revision_id)
3045
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3046
self._ensure_real_inter()
3047
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3048
find_ghosts=find_ghosts)
3051
def _get_repo_format_to_test(self):
3055
class InterRemoteToOther(InterRepository):
3057
def __init__(self, source, target):
3058
InterRepository.__init__(self, source, target)
3059
self._real_inter = None
3062
def is_compatible(source, target):
3063
if not isinstance(source, remote.RemoteRepository):
3065
# Is source's model compatible with target's model?
3066
source._ensure_real()
3067
real_source = source._real_repository
3068
if isinstance(real_source, remote.RemoteRepository):
3069
raise NotImplementedError(
3070
"We don't support remote repos backed by remote repos yet.")
3071
return InterRepository._same_model(real_source, target)
3073
def _ensure_real_inter(self):
3074
if self._real_inter is None:
3075
self.source._ensure_real()
3076
real_source = self.source._real_repository
3077
self._real_inter = InterRepository.get(real_source, self.target)
3079
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3080
self._ensure_real_inter()
3081
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3082
find_ghosts=find_ghosts)
3084
def copy_content(self, revision_id=None):
3085
self._ensure_real_inter()
3086
self._real_inter.copy_content(revision_id=revision_id)
3089
def _get_repo_format_to_test(self):
3094
InterRepository.register_optimiser(InterDifferingSerializer)
3095
InterRepository.register_optimiser(InterSameDataRepository)
1039
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1040
required_revisions = source_ids_set.difference(actually_present_revisions)
1041
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1042
if revision_id is not None:
1043
# we used get_ancestry to determine source_ids then we are assured all
1044
# revisions referenced are present as they are installed in topological order.
1045
# and the tip revision was validated by get_ancestry.
1046
return required_topo_revisions
1048
# if we just grabbed the possibly available ids, then
1049
# we only have an estimate of whats available and need to validate
1050
# that against the revision records.
1051
return self.source._eliminate_revisions_not_present(required_topo_revisions)
3096
1054
InterRepository.register_optimiser(InterWeaveRepo)
3097
InterRepository.register_optimiser(InterKnitRepo)
3098
InterRepository.register_optimiser(InterModel1and2)
3099
InterRepository.register_optimiser(InterKnit1and2)
3100
InterRepository.register_optimiser(InterPackRepo)
3101
InterRepository.register_optimiser(InterOtherToRemote)
3102
InterRepository.register_optimiser(InterRemoteToOther)
3105
class CopyConverter(object):
3106
"""A repository conversion tool which just performs a copy of the content.
3108
This is slow but quite reliable.
3111
def __init__(self, target_format):
3112
"""Create a CopyConverter.
3114
:param target_format: The format the resulting repository should be.
3116
self.target_format = target_format
3118
def convert(self, repo, pb):
3119
"""Perform the conversion of to_convert, giving feedback via pb.
3121
:param to_convert: The disk object to convert.
3122
:param pb: a progress bar to use for progress information.
3127
# this is only useful with metadir layouts - separated repo content.
3128
# trigger an assertion if not such
3129
repo._format.get_format_string()
3130
self.repo_dir = repo.bzrdir
3131
self.step('Moving repository to repository.backup')
3132
self.repo_dir.transport.move('repository', 'repository.backup')
3133
backup_transport = self.repo_dir.transport.clone('repository.backup')
3134
repo._format.check_conversion_target(self.target_format)
3135
self.source_repo = repo._format.open(self.repo_dir,
3137
_override_transport=backup_transport)
3138
self.step('Creating new repository')
3139
converted = self.target_format.initialize(self.repo_dir,
3140
self.source_repo.is_shared())
3141
converted.lock_write()
3143
self.step('Copying content into repository.')
3144
self.source_repo.copy_content_into(converted)
3147
self.step('Deleting old repository content.')
3148
self.repo_dir.transport.delete_tree('repository.backup')
3149
self.pb.note('repository converted')
3151
def step(self, message):
3152
"""Update the pb by a step."""
3154
self.pb.update(message, self.count, self.total)
3166
def _unescaper(match, _map=_unescape_map):
3167
code = match.group(1)
3171
if not code.startswith('#'):
3173
return unichr(int(code[1:])).encode('utf8')
3179
def _unescape_xml(data):
3180
"""Unescape predefined XML entities in a string of data."""
3182
if _unescape_re is None:
3183
_unescape_re = re.compile('\&([^;]*);')
3184
return _unescape_re.sub(_unescaper, data)
3187
class _VersionedFileChecker(object):
3189
def __init__(self, repository):
3190
self.repository = repository
3191
self.text_index = self.repository._generate_text_key_index()
3193
def calculate_file_version_parents(self, text_key):
3194
"""Calculate the correct parents for a file version according to
3197
parent_keys = self.text_index[text_key]
3198
if parent_keys == [_mod_revision.NULL_REVISION]:
3200
return tuple(parent_keys)
3202
def check_file_version_parents(self, texts, progress_bar=None):
3203
"""Check the parents stored in a versioned file are correct.
3205
It also detects file versions that are not referenced by their
3206
corresponding revision's inventory.
3208
:returns: A tuple of (wrong_parents, dangling_file_versions).
3209
wrong_parents is a dict mapping {revision_id: (stored_parents,
3210
correct_parents)} for each revision_id where the stored parents
3211
are not correct. dangling_file_versions is a set of (file_id,
3212
revision_id) tuples for versions that are present in this versioned
3213
file, but not used by the corresponding inventory.
3216
self.file_ids = set([file_id for file_id, _ in
3217
self.text_index.iterkeys()])
3218
# text keys is now grouped by file_id
3219
n_weaves = len(self.file_ids)
3220
files_in_revisions = {}
3221
revisions_of_files = {}
3222
n_versions = len(self.text_index)
3223
progress_bar.update('loading text store', 0, n_versions)
3224
parent_map = self.repository.texts.get_parent_map(self.text_index)
3225
# On unlistable transports this could well be empty/error...
3226
text_keys = self.repository.texts.keys()
3227
unused_keys = frozenset(text_keys) - set(self.text_index)
3228
for num, key in enumerate(self.text_index.iterkeys()):
3229
if progress_bar is not None:
3230
progress_bar.update('checking text graph', num, n_versions)
3231
correct_parents = self.calculate_file_version_parents(key)
3233
knit_parents = parent_map[key]
3234
except errors.RevisionNotPresent:
3237
if correct_parents != knit_parents:
3238
wrong_parents[key] = (knit_parents, correct_parents)
3239
return wrong_parents, unused_keys
3242
def _old_get_graph(repository, revision_id):
3243
"""DO NOT USE. That is all. I'm serious."""
3244
graph = repository.get_graph()
3245
revision_graph = dict(((key, value) for key, value in
3246
graph.iter_ancestry([revision_id]) if value is not None))
3247
return _strip_NULL_ghosts(revision_graph)
3250
def _strip_NULL_ghosts(revision_graph):
3251
"""Also don't use this. more compatibility code for unmigrated clients."""
3252
# Filter ghosts, and null:
3253
if _mod_revision.NULL_REVISION in revision_graph:
3254
del revision_graph[_mod_revision.NULL_REVISION]
3255
for key, parents in revision_graph.items():
3256
revision_graph[key] = tuple(parent for parent in parents if parent
3258
return revision_graph
1057
class RepositoryTestProviderAdapter(object):
1058
"""A tool to generate a suite testing multiple repository formats at once.
1060
This is done by copying the test once for each transport and injecting
1061
the transport_server, transport_readonly_server, and bzrdir_format and
1062
repository_format classes into each copy. Each copy is also given a new id()
1063
to make it easy to identify.
1066
def __init__(self, transport_server, transport_readonly_server, formats):
1067
self._transport_server = transport_server
1068
self._transport_readonly_server = transport_readonly_server
1069
self._formats = formats
1071
def adapt(self, test):
1072
result = TestSuite()
1073
for repository_format, bzrdir_format in self._formats:
1074
new_test = deepcopy(test)
1075
new_test.transport_server = self._transport_server
1076
new_test.transport_readonly_server = self._transport_readonly_server
1077
new_test.bzrdir_format = bzrdir_format
1078
new_test.repository_format = repository_format
1079
def make_new_test_id():
1080
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
1081
return lambda: new_id
1082
new_test.id = make_new_test_id()
1083
result.addTest(new_test)
1087
class InterRepositoryTestProviderAdapter(object):
1088
"""A tool to generate a suite testing multiple inter repository formats.
1090
This is done by copying the test once for each interrepo provider and injecting
1091
the transport_server, transport_readonly_server, repository_format and
1092
repository_to_format classes into each copy.
1093
Each copy is also given a new id() to make it easy to identify.
1096
def __init__(self, transport_server, transport_readonly_server, formats):
1097
self._transport_server = transport_server
1098
self._transport_readonly_server = transport_readonly_server
1099
self._formats = formats
1101
def adapt(self, test):
1102
result = TestSuite()
1103
for interrepo_class, repository_format, repository_format_to in self._formats:
1104
new_test = deepcopy(test)
1105
new_test.transport_server = self._transport_server
1106
new_test.transport_readonly_server = self._transport_readonly_server
1107
new_test.interrepo_class = interrepo_class
1108
new_test.repository_format = repository_format
1109
new_test.repository_format_to = repository_format_to
1110
def make_new_test_id():
1111
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1112
return lambda: new_id
1113
new_test.id = make_new_test_id()
1114
result.addTest(new_test)
1118
def default_test_list():
1119
"""Generate the default list of interrepo permutations to test."""
1121
# test the default InterRepository between format 6 and the current
1123
# XXX: robertc 20060220 reinstate this when there are two supported
1124
# formats which do not have an optimal code path between them.
1125
#result.append((InterRepository, RepositoryFormat6(),
1126
# RepositoryFormat.get_default_format()))
1127
for optimiser in InterRepository._optimisers:
1128
result.append((optimiser,
1129
optimiser._matching_repo_format,
1130
optimiser._matching_repo_format
1132
# if there are specific combinations we want to use, we can add them
added
removed
bzrlib/repository.py
