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 cStringIO import StringIO
19
17
from bzrlib.lazy_import import lazy_import
20
18
lazy_import(globals(), """
21
from binascii import hexlify
22
from copy import deepcopy
27
23
from bzrlib import (
38
38
revision as _mod_revision,
47
from bzrlib.osutils import (
44
from bzrlib.bundle import serializer
52
45
from bzrlib.revisiontree import RevisionTree
53
46
from bzrlib.store.versioned import VersionedFileStore
54
from bzrlib.store.text import TextStore
55
47
from bzrlib.testament import Testament
50
from bzrlib import registry
58
51
from bzrlib.decorators import needs_read_lock, needs_write_lock
59
52
from bzrlib.inter import InterObject
60
53
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
61
54
from bzrlib.symbol_versioning import (
65
from bzrlib.trace import mutter, note, warning
60
from bzrlib.trace import (
61
log_exception_quietly, note, mutter, mutter_callsite, warning)
68
64
# Old formats display a warning, but only once
69
65
_deprecation_warning_done = False
68
class CommitBuilder(object):
69
"""Provides an interface to build up a commit.
71
This allows describing a tree to be committed without needing to
72
know the internals of the format of the repository.
75
# all clients should supply tree roots.
76
record_root_entry = True
77
# the default CommitBuilder does not manage trees whose root is versioned.
78
_versioned_root = False
80
def __init__(self, repository, parents, config, timestamp=None,
81
timezone=None, committer=None, revprops=None,
83
"""Initiate a CommitBuilder.
85
:param repository: Repository to commit to.
86
:param parents: Revision ids of the parents of the new revision.
87
:param config: Configuration to use.
88
:param timestamp: Optional timestamp recorded for commit.
89
:param timezone: Optional timezone for timestamp.
90
:param committer: Optional committer to set for commit.
91
:param revprops: Optional dictionary of revision properties.
92
:param revision_id: Optional revision id.
97
self._committer = self._config.username()
99
self._committer = committer
101
self.new_inventory = Inventory(None)
102
self._new_revision_id = revision_id
103
self.parents = parents
104
self.repository = repository
107
if revprops is not None:
108
self._validate_revprops(revprops)
109
self._revprops.update(revprops)
111
if timestamp is None:
112
timestamp = time.time()
113
# Restrict resolution to 1ms
114
self._timestamp = round(timestamp, 3)
117
self._timezone = osutils.local_time_offset()
119
self._timezone = int(timezone)
121
self._generate_revision_if_needed()
122
self.__heads = graph.HeadsCache(repository.get_graph()).heads
123
self._basis_delta = []
124
# API compatibility, older code that used CommitBuilder did not call
125
# .record_delete(), which means the delta that is computed would not be
126
# valid. Callers that will call record_delete() should call
127
# .will_record_deletes() to indicate that.
128
self._recording_deletes = False
130
def _validate_unicode_text(self, text, context):
131
"""Verify things like commit messages don't have bogus characters."""
133
raise ValueError('Invalid value for %s: %r' % (context, text))
135
def _validate_revprops(self, revprops):
136
for key, value in revprops.iteritems():
137
# We know that the XML serializers do not round trip '\r'
138
# correctly, so refuse to accept them
139
if not isinstance(value, basestring):
140
raise ValueError('revision property (%s) is not a valid'
141
' (unicode) string: %r' % (key, value))
142
self._validate_unicode_text(value,
143
'revision property (%s)' % (key,))
145
def commit(self, message):
146
"""Make the actual commit.
148
:return: The revision id of the recorded revision.
150
self._validate_unicode_text(message, 'commit message')
151
rev = _mod_revision.Revision(
152
timestamp=self._timestamp,
153
timezone=self._timezone,
154
committer=self._committer,
156
inventory_sha1=self.inv_sha1,
157
revision_id=self._new_revision_id,
158
properties=self._revprops)
159
rev.parent_ids = self.parents
160
self.repository.add_revision(self._new_revision_id, rev,
161
self.new_inventory, self._config)
162
self.repository.commit_write_group()
163
return self._new_revision_id
166
"""Abort the commit that is being built.
168
self.repository.abort_write_group()
170
def revision_tree(self):
171
"""Return the tree that was just committed.
173
After calling commit() this can be called to get a RevisionTree
174
representing the newly committed tree. This is preferred to
175
calling Repository.revision_tree() because that may require
176
deserializing the inventory, while we already have a copy in
179
return RevisionTree(self.repository, self.new_inventory,
180
self._new_revision_id)
182
def finish_inventory(self):
183
"""Tell the builder that the inventory is finished."""
184
if self.new_inventory.root is None:
185
raise AssertionError('Root entry should be supplied to'
186
' record_entry_contents, as of bzr 0.10.')
187
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
188
self.new_inventory.revision_id = self._new_revision_id
189
self.inv_sha1 = self.repository.add_inventory(
190
self._new_revision_id,
195
def _gen_revision_id(self):
196
"""Return new revision-id."""
197
return generate_ids.gen_revision_id(self._config.username(),
200
def _generate_revision_if_needed(self):
201
"""Create a revision id if None was supplied.
203
If the repository can not support user-specified revision ids
204
they should override this function and raise CannotSetRevisionId
205
if _new_revision_id is not None.
207
:raises: CannotSetRevisionId
209
if self._new_revision_id is None:
210
self._new_revision_id = self._gen_revision_id()
211
self.random_revid = True
213
self.random_revid = False
215
def _heads(self, file_id, revision_ids):
216
"""Calculate the graph heads for revision_ids in the graph of file_id.
218
This can use either a per-file graph or a global revision graph as we
219
have an identity relationship between the two graphs.
221
return self.__heads(revision_ids)
223
def _check_root(self, ie, parent_invs, tree):
224
"""Helper for record_entry_contents.
226
:param ie: An entry being added.
227
:param parent_invs: The inventories of the parent revisions of the
229
:param tree: The tree that is being committed.
231
# In this revision format, root entries have no knit or weave When
232
# serializing out to disk and back in root.revision is always
234
ie.revision = self._new_revision_id
236
def _get_delta(self, ie, basis_inv, path):
237
"""Get a delta against the basis inventory for ie."""
238
if ie.file_id not in basis_inv:
240
result = (None, path, ie.file_id, ie)
241
self._basis_delta.append(result)
243
elif ie != basis_inv[ie.file_id]:
245
# TODO: avoid tis id2path call.
246
result = (basis_inv.id2path(ie.file_id), path, ie.file_id, ie)
247
self._basis_delta.append(result)
253
def get_basis_delta(self):
254
"""Return the complete inventory delta versus the basis inventory.
256
This has been built up with the calls to record_delete and
257
record_entry_contents. The client must have already called
258
will_record_deletes() to indicate that they will be generating a
261
:return: An inventory delta, suitable for use with apply_delta, or
262
Repository.add_inventory_by_delta, etc.
264
if not self._recording_deletes:
265
raise AssertionError("recording deletes not activated.")
266
return self._basis_delta
268
def record_delete(self, path, file_id):
269
"""Record that a delete occured against a basis tree.
271
This is an optional API - when used it adds items to the basis_delta
272
being accumulated by the commit builder. It cannot be called unless the
273
method will_record_deletes() has been called to inform the builder that
274
a delta is being supplied.
276
:param path: The path of the thing deleted.
277
:param file_id: The file id that was deleted.
279
if not self._recording_deletes:
280
raise AssertionError("recording deletes not activated.")
281
delta = (path, None, file_id, None)
282
self._basis_delta.append(delta)
285
def will_record_deletes(self):
286
"""Tell the commit builder that deletes are being notified.
288
This enables the accumulation of an inventory delta; for the resulting
289
commit to be valid, deletes against the basis MUST be recorded via
290
builder.record_delete().
292
self._recording_deletes = True
294
def record_entry_contents(self, ie, parent_invs, path, tree,
296
"""Record the content of ie from tree into the commit if needed.
298
Side effect: sets ie.revision when unchanged
300
:param ie: An inventory entry present in the commit.
301
:param parent_invs: The inventories of the parent revisions of the
303
:param path: The path the entry is at in the tree.
304
:param tree: The tree which contains this entry and should be used to
306
:param content_summary: Summary data from the tree about the paths
307
content - stat, length, exec, sha/link target. This is only
308
accessed when the entry has a revision of None - that is when it is
309
a candidate to commit.
310
:return: A tuple (change_delta, version_recorded, fs_hash).
311
change_delta is an inventory_delta change for this entry against
312
the basis tree of the commit, or None if no change occured against
314
version_recorded is True if a new version of the entry has been
315
recorded. For instance, committing a merge where a file was only
316
changed on the other side will return (delta, False).
317
fs_hash is either None, or the hash details for the path (currently
318
a tuple of the contents sha1 and the statvalue returned by
319
tree.get_file_with_stat()).
321
if self.new_inventory.root is None:
322
if ie.parent_id is not None:
323
raise errors.RootMissing()
324
self._check_root(ie, parent_invs, tree)
325
if ie.revision is None:
326
kind = content_summary[0]
328
# ie is carried over from a prior commit
330
# XXX: repository specific check for nested tree support goes here - if
331
# the repo doesn't want nested trees we skip it ?
332
if (kind == 'tree-reference' and
333
not self.repository._format.supports_tree_reference):
334
# mismatch between commit builder logic and repository:
335
# this needs the entry creation pushed down into the builder.
336
raise NotImplementedError('Missing repository subtree support.')
337
self.new_inventory.add(ie)
339
# TODO: slow, take it out of the inner loop.
341
basis_inv = parent_invs[0]
343
basis_inv = Inventory(root_id=None)
345
# ie.revision is always None if the InventoryEntry is considered
346
# for committing. We may record the previous parents revision if the
347
# content is actually unchanged against a sole head.
348
if ie.revision is not None:
349
if not self._versioned_root and path == '':
350
# repositories that do not version the root set the root's
351
# revision to the new commit even when no change occurs (more
352
# specifically, they do not record a revision on the root; and
353
# the rev id is assigned to the root during deserialisation -
354
# this masks when a change may have occurred against the basis.
355
# To match this we always issue a delta, because the revision
356
# of the root will always be changing.
357
if ie.file_id in basis_inv:
358
delta = (basis_inv.id2path(ie.file_id), path,
362
delta = (None, path, ie.file_id, ie)
363
self._basis_delta.append(delta)
364
return delta, False, None
366
# we don't need to commit this, because the caller already
367
# determined that an existing revision of this file is
368
# appropriate. If its not being considered for committing then
369
# it and all its parents to the root must be unaltered so
370
# no-change against the basis.
371
if ie.revision == self._new_revision_id:
372
raise AssertionError("Impossible situation, a skipped "
373
"inventory entry (%r) claims to be modified in this "
374
"commit (%r).", (ie, self._new_revision_id))
375
return None, False, None
376
# XXX: Friction: parent_candidates should return a list not a dict
377
# so that we don't have to walk the inventories again.
378
parent_candiate_entries = ie.parent_candidates(parent_invs)
379
head_set = self._heads(ie.file_id, parent_candiate_entries.keys())
381
for inv in parent_invs:
382
if ie.file_id in inv:
383
old_rev = inv[ie.file_id].revision
384
if old_rev in head_set:
385
heads.append(inv[ie.file_id].revision)
386
head_set.remove(inv[ie.file_id].revision)
389
# now we check to see if we need to write a new record to the
391
# We write a new entry unless there is one head to the ancestors, and
392
# the kind-derived content is unchanged.
394
# Cheapest check first: no ancestors, or more the one head in the
395
# ancestors, we write a new node.
399
# There is a single head, look it up for comparison
400
parent_entry = parent_candiate_entries[heads[0]]
401
# if the non-content specific data has changed, we'll be writing a
403
if (parent_entry.parent_id != ie.parent_id or
404
parent_entry.name != ie.name):
406
# now we need to do content specific checks:
408
# if the kind changed the content obviously has
409
if kind != parent_entry.kind:
411
# Stat cache fingerprint feedback for the caller - None as we usually
412
# don't generate one.
415
if content_summary[2] is None:
416
raise ValueError("Files must not have executable = None")
418
if (# if the file length changed we have to store:
419
parent_entry.text_size != content_summary[1] or
420
# if the exec bit has changed we have to store:
421
parent_entry.executable != content_summary[2]):
423
elif parent_entry.text_sha1 == content_summary[3]:
424
# all meta and content is unchanged (using a hash cache
425
# hit to check the sha)
426
ie.revision = parent_entry.revision
427
ie.text_size = parent_entry.text_size
428
ie.text_sha1 = parent_entry.text_sha1
429
ie.executable = parent_entry.executable
430
return self._get_delta(ie, basis_inv, path), False, None
432
# Either there is only a hash change(no hash cache entry,
433
# or same size content change), or there is no change on
435
# Provide the parent's hash to the store layer, so that the
436
# content is unchanged we will not store a new node.
437
nostore_sha = parent_entry.text_sha1
439
# We want to record a new node regardless of the presence or
440
# absence of a content change in the file.
442
ie.executable = content_summary[2]
443
file_obj, stat_value = tree.get_file_with_stat(ie.file_id, path)
445
lines = file_obj.readlines()
449
ie.text_sha1, ie.text_size = self._add_text_to_weave(
450
ie.file_id, lines, heads, nostore_sha)
451
# Let the caller know we generated a stat fingerprint.
452
fingerprint = (ie.text_sha1, stat_value)
453
except errors.ExistingContent:
454
# Turns out that the file content was unchanged, and we were
455
# only going to store a new node if it was changed. Carry over
457
ie.revision = parent_entry.revision
458
ie.text_size = parent_entry.text_size
459
ie.text_sha1 = parent_entry.text_sha1
460
ie.executable = parent_entry.executable
461
return self._get_delta(ie, basis_inv, path), False, None
462
elif kind == 'directory':
464
# all data is meta here, nothing specific to directory, so
466
ie.revision = parent_entry.revision
467
return self._get_delta(ie, basis_inv, path), False, None
469
self._add_text_to_weave(ie.file_id, lines, heads, None)
470
elif kind == 'symlink':
471
current_link_target = content_summary[3]
473
# symlink target is not generic metadata, check if it has
475
if current_link_target != parent_entry.symlink_target:
478
# unchanged, carry over.
479
ie.revision = parent_entry.revision
480
ie.symlink_target = parent_entry.symlink_target
481
return self._get_delta(ie, basis_inv, path), False, None
482
ie.symlink_target = current_link_target
484
self._add_text_to_weave(ie.file_id, lines, heads, None)
485
elif kind == 'tree-reference':
487
if content_summary[3] != parent_entry.reference_revision:
490
# unchanged, carry over.
491
ie.reference_revision = parent_entry.reference_revision
492
ie.revision = parent_entry.revision
493
return self._get_delta(ie, basis_inv, path), False, None
494
ie.reference_revision = content_summary[3]
496
self._add_text_to_weave(ie.file_id, lines, heads, None)
498
raise NotImplementedError('unknown kind')
499
ie.revision = self._new_revision_id
500
return self._get_delta(ie, basis_inv, path), True, fingerprint
502
def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
503
# Note: as we read the content directly from the tree, we know its not
504
# been turned into unicode or badly split - but a broken tree
505
# implementation could give us bad output from readlines() so this is
506
# not a guarantee of safety. What would be better is always checking
507
# the content during test suite execution. RBC 20070912
508
parent_keys = tuple((file_id, parent) for parent in parents)
509
return self.repository.texts.add_lines(
510
(file_id, self._new_revision_id), parent_keys, new_lines,
511
nostore_sha=nostore_sha, random_id=self.random_revid,
512
check_content=False)[0:2]
515
class RootCommitBuilder(CommitBuilder):
516
"""This commitbuilder actually records the root id"""
518
# the root entry gets versioned properly by this builder.
519
_versioned_root = True
521
def _check_root(self, ie, parent_invs, tree):
522
"""Helper for record_entry_contents.
524
:param ie: An entry being added.
525
:param parent_invs: The inventories of the parent revisions of the
527
:param tree: The tree that is being committed.
531
######################################################################
72
534
class Repository(object):
73
535
"""Repository holding history for one or more branches.
76
538
revisions and file history. It's normally accessed only by the Branch,
77
539
which views a particular line of development through that history.
79
The Repository builds on top of Stores and a Transport, which respectively
80
describe the disk data format and the way of accessing the (possibly
541
The Repository builds on top of some byte storage facilies (the revisions,
542
signatures, inventories and texts attributes) and a Transport, which
543
respectively provide byte storage and a means to access the (possibly
546
The byte storage facilities are addressed via tuples, which we refer to
547
as 'keys' throughout the code base. Revision_keys, inventory_keys and
548
signature_keys are all 1-tuples: (revision_id,). text_keys are two-tuples:
549
(file_id, revision_id). We use this interface because it allows low
550
friction with the underlying code that implements disk indices, network
551
encoding and other parts of bzrlib.
553
:ivar revisions: A bzrlib.versionedfile.VersionedFiles instance containing
554
the serialised revisions for the repository. This can be used to obtain
555
revision graph information or to access raw serialised revisions.
556
The result of trying to insert data into the repository via this store
557
is undefined: it should be considered read-only except for implementors
559
:ivar signatures: A bzrlib.versionedfile.VersionedFiles instance containing
560
the serialised signatures for the repository. This can be used to
561
obtain access to raw serialised signatures. The result of trying to
562
insert data into the repository via this store is undefined: it should
563
be considered read-only except for implementors of repositories.
564
:ivar inventories: A bzrlib.versionedfile.VersionedFiles instance containing
565
the serialised inventories for the repository. This can be used to
566
obtain unserialised inventories. The result of trying to insert data
567
into the repository via this store is undefined: it should be
568
considered read-only except for implementors of repositories.
569
:ivar texts: A bzrlib.versionedfile.VersionedFiles instance containing the
570
texts of files and directories for the repository. This can be used to
571
obtain file texts or file graphs. Note that Repository.iter_file_bytes
572
is usually a better interface for accessing file texts.
573
The result of trying to insert data into the repository via this store
574
is undefined: it should be considered read-only except for implementors
576
:ivar _transport: Transport for file access to repository, typically
577
pointing to .bzr/repository.
85
def add_inventory(self, revid, inv, parents):
86
"""Add the inventory inv to the repository as revid.
88
:param parents: The revision ids of the parents that revid
580
# What class to use for a CommitBuilder. Often its simpler to change this
581
# in a Repository class subclass rather than to override
582
# get_commit_builder.
583
_commit_builder_class = CommitBuilder
584
# The search regex used by xml based repositories to determine what things
585
# where changed in a single commit.
586
_file_ids_altered_regex = lazy_regex.lazy_compile(
587
r'file_id="(?P<file_id>[^"]+)"'
588
r'.* revision="(?P<revision_id>[^"]+)"'
591
def abort_write_group(self, suppress_errors=False):
592
"""Commit the contents accrued within the current write group.
594
:param suppress_errors: if true, abort_write_group will catch and log
595
unexpected errors that happen during the abort, rather than
596
allowing them to propagate. Defaults to False.
598
:seealso: start_write_group.
600
if self._write_group is not self.get_transaction():
601
# has an unlock or relock occured ?
602
raise errors.BzrError('mismatched lock context and write group.')
604
self._abort_write_group()
605
except Exception, exc:
606
self._write_group = None
607
if not suppress_errors:
609
mutter('abort_write_group failed')
610
log_exception_quietly()
611
note('bzr: ERROR (ignored): %s', exc)
612
self._write_group = None
614
def _abort_write_group(self):
615
"""Template method for per-repository write group cleanup.
617
This is called during abort before the write group is considered to be
618
finished and should cleanup any internal state accrued during the write
619
group. There is no requirement that data handed to the repository be
620
*not* made available - this is not a rollback - but neither should any
621
attempt be made to ensure that data added is fully commited. Abort is
622
invoked when an error has occured so futher disk or network operations
623
may not be possible or may error and if possible should not be
627
def add_fallback_repository(self, repository):
628
"""Add a repository to use for looking up data not held locally.
630
:param repository: A repository.
632
if not self._format.supports_external_lookups:
633
raise errors.UnstackableRepositoryFormat(self._format, self.base)
634
self._check_fallback_repository(repository)
635
self._fallback_repositories.append(repository)
636
self.texts.add_fallback_versioned_files(repository.texts)
637
self.inventories.add_fallback_versioned_files(repository.inventories)
638
self.revisions.add_fallback_versioned_files(repository.revisions)
639
self.signatures.add_fallback_versioned_files(repository.signatures)
640
self._fetch_order = 'topological'
642
def _check_fallback_repository(self, repository):
643
"""Check that this repository can fallback to repository safely.
645
Raise an error if not.
647
:param repository: A repository to fallback to.
649
return InterRepository._assert_same_model(self, repository)
651
def add_inventory(self, revision_id, inv, parents):
652
"""Add the inventory inv to the repository as revision_id.
654
:param parents: The revision ids of the parents that revision_id
89
655
is known to have and are in the repository already.
91
returns the sha1 of the serialized inventory.
93
assert inv.revision_id is None or inv.revision_id == revid, \
94
"Mismatch between inventory revision" \
95
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
96
assert inv.root is not None
97
inv_text = self.serialise_inventory(inv)
98
inv_sha1 = osutils.sha_string(inv_text)
99
inv_vf = self.control_weaves.get_weave('inventory',
100
self.get_transaction())
101
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
104
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
106
for parent in parents:
108
final_parents.append(parent)
110
inv_vf.add_lines(revid, final_parents, lines)
113
def add_revision(self, rev_id, rev, inv=None, config=None):
114
"""Add rev to the revision store as rev_id.
116
:param rev_id: the revision id to use.
657
:returns: The validator(which is a sha1 digest, though what is sha'd is
658
repository format specific) of the serialized inventory.
660
if not self.is_in_write_group():
661
raise AssertionError("%r not in write group" % (self,))
662
_mod_revision.check_not_reserved_id(revision_id)
663
if not (inv.revision_id is None or inv.revision_id == revision_id):
664
raise AssertionError(
665
"Mismatch between inventory revision"
666
" id and insertion revid (%r, %r)"
667
% (inv.revision_id, revision_id))
669
raise AssertionError()
670
inv_lines = self._serialise_inventory_to_lines(inv)
671
return self._inventory_add_lines(revision_id, parents,
672
inv_lines, check_content=False)
674
def add_inventory_by_delta(self, basis_revision_id, delta, new_revision_id,
676
"""Add a new inventory expressed as a delta against another revision.
678
:param basis_revision_id: The inventory id the delta was created
679
against. (This does not have to be a direct parent.)
680
:param delta: The inventory delta (see Inventory.apply_delta for
682
:param new_revision_id: The revision id that the inventory is being
684
:param parents: The revision ids of the parents that revision_id is
685
known to have and are in the repository already. These are supplied
686
for repositories that depend on the inventory graph for revision
687
graph access, as well as for those that pun ancestry with delta
690
:returns: (validator, new_inv)
691
The validator(which is a sha1 digest, though what is sha'd is
692
repository format specific) of the serialized inventory, and the
695
if not self.is_in_write_group():
696
raise AssertionError("%r not in write group" % (self,))
697
_mod_revision.check_not_reserved_id(new_revision_id)
698
basis_tree = self.revision_tree(basis_revision_id)
699
basis_tree.lock_read()
701
# Note that this mutates the inventory of basis_tree, which not all
702
# inventory implementations may support: A better idiom would be to
703
# return a new inventory, but as there is no revision tree cache in
704
# repository this is safe for now - RBC 20081013
705
basis_inv = basis_tree.inventory
706
basis_inv.apply_delta(delta)
707
basis_inv.revision_id = new_revision_id
708
return (self.add_inventory(new_revision_id, basis_inv, parents),
713
def _inventory_add_lines(self, revision_id, parents, lines,
715
"""Store lines in inv_vf and return the sha1 of the inventory."""
716
parents = [(parent,) for parent in parents]
717
return self.inventories.add_lines((revision_id,), parents, lines,
718
check_content=check_content)[0]
720
def add_revision(self, revision_id, rev, inv=None, config=None):
721
"""Add rev to the revision store as revision_id.
723
:param revision_id: the revision id to use.
117
724
:param rev: The revision object.
118
725
:param inv: The inventory for the revision. if None, it will be looked
119
726
up in the inventory storer
203
817
# the following are part of the public API for Repository:
204
818
self.bzrdir = a_bzrdir
205
819
self.control_files = control_files
206
self._revision_store = _revision_store
207
self.text_store = text_store
208
# backwards compatibility
209
self.weave_store = text_store
820
self._transport = control_files._transport
821
self.base = self._transport.base
823
self._reconcile_does_inventory_gc = True
824
self._reconcile_fixes_text_parents = False
825
self._reconcile_backsup_inventory = True
210
826
# not right yet - should be more semantically clear ?
212
self.control_store = control_store
213
self.control_weaves = control_store
214
828
# TODO: make sure to construct the right store classes, etc, depending
215
829
# on whether escaping is required.
216
830
self._warn_if_deprecated()
217
self._serializer = xml5.serializer_v5
831
self._write_group = None
832
# Additional places to query for data.
833
self._fallback_repositories = []
834
# What order should fetch operations request streams in?
835
# The default is unordered as that is the cheapest for an origin to
837
self._fetch_order = 'unordered'
838
# Does this repository use deltas that can be fetched as-deltas ?
839
# (E.g. knits, where the knit deltas can be transplanted intact.
840
# We default to False, which will ensure that enough data to get
841
# a full text out of any fetch stream will be grabbed.
842
self._fetch_uses_deltas = False
843
# Should fetch trigger a reconcile after the fetch? Only needed for
844
# some repository formats that can suffer internal inconsistencies.
845
self._fetch_reconcile = False
846
# An InventoryEntry cache, used during deserialization
847
self._inventory_entry_cache = fifo_cache.FIFOCache(10*1024)
219
849
def __repr__(self):
220
return '%s(%r)' % (self.__class__.__name__,
221
self.bzrdir.transport.base)
850
return '%s(%r)' % (self.__class__.__name__,
853
def has_same_location(self, other):
854
"""Returns a boolean indicating if this repository is at the same
855
location as another repository.
857
This might return False even when two repository objects are accessing
858
the same physical repository via different URLs.
860
if self.__class__ is not other.__class__:
862
return (self._transport.base == other._transport.base)
864
def is_in_write_group(self):
865
"""Return True if there is an open write group.
867
:seealso: start_write_group.
869
return self._write_group is not None
223
871
def is_locked(self):
224
872
return self.control_files.is_locked()
226
def lock_write(self):
227
self.control_files.lock_write()
874
def is_write_locked(self):
875
"""Return True if this object is write locked."""
876
return self.is_locked() and self.control_files._lock_mode == 'w'
878
def lock_write(self, token=None):
879
"""Lock this repository for writing.
881
This causes caching within the repository obejct to start accumlating
882
data during reads, and allows a 'write_group' to be obtained. Write
883
groups must be used for actual data insertion.
885
:param token: if this is already locked, then lock_write will fail
886
unless the token matches the existing lock.
887
:returns: a token if this instance supports tokens, otherwise None.
888
:raises TokenLockingNotSupported: when a token is given but this
889
instance doesn't support using token locks.
890
:raises MismatchedToken: if the specified token doesn't match the token
891
of the existing lock.
892
:seealso: start_write_group.
894
A token should be passed in if you know that you have locked the object
895
some other way, and need to synchronise this object's state with that
898
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
900
result = self.control_files.lock_write(token=token)
901
for repo in self._fallback_repositories:
902
# Writes don't affect fallback repos
229
907
def lock_read(self):
230
908
self.control_files.lock_read()
909
for repo in self._fallback_repositories:
232
913
def get_physical_lock_status(self):
233
914
return self.control_files.get_physical_lock_status()
236
def missing_revision_ids(self, other, revision_id=None):
237
"""Return the revision ids that other has that this does not.
239
These are returned in topological order.
241
revision_id: only return revision ids included by revision_id.
243
return InterRepository.get(other, self).missing_revision_ids(revision_id)
916
def leave_lock_in_place(self):
917
"""Tell this repository not to release the physical lock when this
920
If lock_write doesn't return a token, then this method is not supported.
922
self.control_files.leave_in_place()
924
def dont_leave_lock_in_place(self):
925
"""Tell this repository to release the physical lock when this
926
object is unlocked, even if it didn't originally acquire it.
928
If lock_write doesn't return a token, then this method is not supported.
930
self.control_files.dont_leave_in_place()
933
def gather_stats(self, revid=None, committers=None):
934
"""Gather statistics from a revision id.
936
:param revid: The revision id to gather statistics from, if None, then
937
no revision specific statistics are gathered.
938
:param committers: Optional parameter controlling whether to grab
939
a count of committers from the revision specific statistics.
940
:return: A dictionary of statistics. Currently this contains:
941
committers: The number of committers if requested.
942
firstrev: A tuple with timestamp, timezone for the penultimate left
943
most ancestor of revid, if revid is not the NULL_REVISION.
944
latestrev: A tuple with timestamp, timezone for revid, if revid is
945
not the NULL_REVISION.
946
revisions: The total revision count in the repository.
947
size: An estimate disk size of the repository in bytes.
950
if revid and committers:
951
result['committers'] = 0
952
if revid and revid != _mod_revision.NULL_REVISION:
954
all_committers = set()
955
revisions = self.get_ancestry(revid)
956
# pop the leading None
958
first_revision = None
960
# ignore the revisions in the middle - just grab first and last
961
revisions = revisions[0], revisions[-1]
962
for revision in self.get_revisions(revisions):
963
if not first_revision:
964
first_revision = revision
966
all_committers.add(revision.committer)
967
last_revision = revision
969
result['committers'] = len(all_committers)
970
result['firstrev'] = (first_revision.timestamp,
971
first_revision.timezone)
972
result['latestrev'] = (last_revision.timestamp,
973
last_revision.timezone)
975
# now gather global repository information
976
# XXX: This is available for many repos regardless of listability.
977
if self.bzrdir.root_transport.listable():
978
# XXX: do we want to __define len__() ?
979
# Maybe the versionedfiles object should provide a different
980
# method to get the number of keys.
981
result['revisions'] = len(self.revisions.keys())
985
def find_branches(self, using=False):
986
"""Find branches underneath this repository.
988
This will include branches inside other branches.
990
:param using: If True, list only branches using this repository.
992
if using and not self.is_shared():
994
return [self.bzrdir.open_branch()]
995
except errors.NotBranchError:
997
class Evaluator(object):
1000
self.first_call = True
1002
def __call__(self, bzrdir):
1003
# On the first call, the parameter is always the bzrdir
1004
# containing the current repo.
1005
if not self.first_call:
1007
repository = bzrdir.open_repository()
1008
except errors.NoRepositoryPresent:
1011
return False, (None, repository)
1012
self.first_call = False
1014
value = (bzrdir.open_branch(), None)
1015
except errors.NotBranchError:
1016
value = (None, None)
1020
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
1021
self.bzrdir.root_transport, evaluate=Evaluator()):
1022
if branch is not None:
1023
branches.append(branch)
1024
if not using and repository is not None:
1025
branches.extend(repository.find_branches())
1029
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
1030
"""Return the revision ids that other has that this does not.
1032
These are returned in topological order.
1034
revision_id: only return revision ids included by revision_id.
1036
return InterRepository.get(other, self).search_missing_revision_ids(
1037
revision_id, find_ghosts)
1039
@deprecated_method(one_two)
1041
def missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
1042
"""Return the revision ids that other has that this does not.
1044
These are returned in topological order.
1046
revision_id: only return revision ids included by revision_id.
1048
keys = self.search_missing_revision_ids(
1049
other, revision_id, find_ghosts).get_keys()
1052
parents = other.get_graph().get_parent_map(keys)
1055
return tsort.topo_sort(parents)
282
1134
:param revprops: Optional dictionary of revision properties.
283
1135
:param revision_id: Optional revision id.
285
return _CommitBuilder(self, parents, config, timestamp, timezone,
286
committer, revprops, revision_id)
1137
result = self._commit_builder_class(self, parents, config,
1138
timestamp, timezone, committer, revprops, revision_id)
1139
self.start_write_group()
288
1142
def unlock(self):
1143
if (self.control_files._lock_count == 1 and
1144
self.control_files._lock_mode == 'w'):
1145
if self._write_group is not None:
1146
self.abort_write_group()
1147
self.control_files.unlock()
1148
raise errors.BzrError(
1149
'Must end write groups before releasing write locks.')
289
1150
self.control_files.unlock()
1151
if self.control_files._lock_count == 0:
1152
self._inventory_entry_cache.clear()
1153
for repo in self._fallback_repositories:
291
1156
@needs_read_lock
292
def clone(self, a_bzrdir, revision_id=None, basis=None):
1157
def clone(self, a_bzrdir, revision_id=None):
293
1158
"""Clone this repository into a_bzrdir using the current format.
295
1160
Currently no check is made that the format of this repository and
296
1161
the bzrdir format are compatible. FIXME RBC 20060201.
1163
:return: The newly created destination repository.
1165
# TODO: deprecate after 0.16; cloning this with all its settings is
1166
# probably not very useful -- mbp 20070423
1167
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
1168
self.copy_content_into(dest_repo, revision_id)
1171
def start_write_group(self):
1172
"""Start a write group in the repository.
1174
Write groups are used by repositories which do not have a 1:1 mapping
1175
between file ids and backend store to manage the insertion of data from
1176
both fetch and commit operations.
1178
A write lock is required around the start_write_group/commit_write_group
1179
for the support of lock-requiring repository formats.
1181
One can only insert data into a repository inside a write group.
1185
if not self.is_write_locked():
1186
raise errors.NotWriteLocked(self)
1187
if self._write_group:
1188
raise errors.BzrError('already in a write group')
1189
self._start_write_group()
1190
# so we can detect unlock/relock - the write group is now entered.
1191
self._write_group = self.get_transaction()
1193
def _start_write_group(self):
1194
"""Template method for per-repository write group startup.
1196
This is called before the write group is considered to be
1201
def sprout(self, to_bzrdir, revision_id=None):
1202
"""Create a descendent repository for new development.
1204
Unlike clone, this does not copy the settings of the repository.
1206
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
1207
dest_repo.fetch(self, revision_id=revision_id)
1210
def _create_sprouting_repo(self, a_bzrdir, shared):
298
1211
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
299
1212
# use target default format.
300
result = a_bzrdir.create_repository()
301
# FIXME RBC 20060209 split out the repository type to avoid this check ?
302
elif isinstance(a_bzrdir._format,
303
(bzrdir.BzrDirFormat4,
304
bzrdir.BzrDirFormat5,
305
bzrdir.BzrDirFormat6)):
306
result = a_bzrdir.open_repository()
1213
dest_repo = a_bzrdir.create_repository()
308
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
309
self.copy_content_into(result, revision_id, basis)
1215
# Most control formats need the repository to be specifically
1216
# created, but on some old all-in-one formats it's not needed
1218
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
1219
except errors.UninitializableFormat:
1220
dest_repo = a_bzrdir.open_repository()
312
1223
@needs_read_lock
313
1224
def has_revision(self, revision_id):
314
1225
"""True if this repository has a copy of the revision."""
315
return self._revision_store.has_revision_id(revision_id,
316
self.get_transaction())
1226
return revision_id in self.has_revisions((revision_id,))
1229
def has_revisions(self, revision_ids):
1230
"""Probe to find out the presence of multiple revisions.
1232
:param revision_ids: An iterable of revision_ids.
1233
:return: A set of the revision_ids that were present.
1235
parent_map = self.revisions.get_parent_map(
1236
[(rev_id,) for rev_id in revision_ids])
1238
if _mod_revision.NULL_REVISION in revision_ids:
1239
result.add(_mod_revision.NULL_REVISION)
1240
result.update([key[0] for key in parent_map])
1244
def get_revision(self, revision_id):
1245
"""Return the Revision object for a named revision."""
1246
return self.get_revisions([revision_id])[0]
318
1248
@needs_read_lock
319
1249
def get_revision_reconcile(self, revision_id):
435
1378
# revisions. We don't need to see all lines in the inventory because
436
1379
# only those added in an inventory in rev X can contain a revision=X
438
pb = ui.ui_factory.nested_progress_bar()
440
for line in w.iter_lines_added_or_present_in_versions(
441
selected_revision_ids, pb=pb):
442
start = line.find('file_id="')+9
443
if start < 9: continue
444
end = line.find('"', start)
446
file_id = _unescape_xml(line[start:end])
448
start = line.find('revision="')+10
449
if start < 10: continue
450
end = line.find('"', start)
452
revision_id = _unescape_xml(line[start:end])
453
if revision_id in selected_revision_ids:
454
result.setdefault(file_id, set()).add(revision_id)
460
def get_inventory_weave(self):
461
return self.control_weaves.get_weave('inventory',
462
self.get_transaction())
1381
unescape_revid_cache = {}
1382
unescape_fileid_cache = {}
1384
# jam 20061218 In a big fetch, this handles hundreds of thousands
1385
# of lines, so it has had a lot of inlining and optimizing done.
1386
# Sorry that it is a little bit messy.
1387
# Move several functions to be local variables, since this is a long
1389
search = self._file_ids_altered_regex.search
1390
unescape = _unescape_xml
1391
setdefault = result.setdefault
1392
for line, line_key in line_iterator:
1393
match = search(line)
1396
# One call to match.group() returning multiple items is quite a
1397
# bit faster than 2 calls to match.group() each returning 1
1398
file_id, revision_id = match.group('file_id', 'revision_id')
1400
# Inlining the cache lookups helps a lot when you make 170,000
1401
# lines and 350k ids, versus 8.4 unique ids.
1402
# Using a cache helps in 2 ways:
1403
# 1) Avoids unnecessary decoding calls
1404
# 2) Re-uses cached strings, which helps in future set and
1406
# (2) is enough that removing encoding entirely along with
1407
# the cache (so we are using plain strings) results in no
1408
# performance improvement.
1410
revision_id = unescape_revid_cache[revision_id]
1412
unescaped = unescape(revision_id)
1413
unescape_revid_cache[revision_id] = unescaped
1414
revision_id = unescaped
1416
# Note that unconditionally unescaping means that we deserialise
1417
# every fileid, which for general 'pull' is not great, but we don't
1418
# really want to have some many fulltexts that this matters anyway.
1421
file_id = unescape_fileid_cache[file_id]
1423
unescaped = unescape(file_id)
1424
unescape_fileid_cache[file_id] = unescaped
1427
key = (file_id, revision_id)
1428
setdefault(key, False)
1429
if revision_id == line_key[-1]:
1433
def _find_file_ids_from_xml_inventory_lines(self, line_iterator,
1435
"""Helper routine for fileids_altered_by_revision_ids.
1437
This performs the translation of xml lines to revision ids.
1439
:param line_iterator: An iterator of lines, origin_version_id
1440
:param revision_ids: The revision ids to filter for. This should be a
1441
set or other type which supports efficient __contains__ lookups, as
1442
the revision id from each parsed line will be looked up in the
1443
revision_ids filter.
1444
:return: a dictionary mapping altered file-ids to an iterable of
1445
revision_ids. Each altered file-ids has the exact revision_ids that
1446
altered it listed explicitly.
1449
setdefault = result.setdefault
1451
self._find_text_key_references_from_xml_inventory_lines(
1452
line_iterator).iterkeys():
1453
# once data is all ensured-consistent; then this is
1454
# if revision_id == version_id
1455
if key[-1:] in revision_ids:
1456
setdefault(key[0], set()).add(key[-1])
1459
def fileids_altered_by_revision_ids(self, revision_ids, _inv_weave=None):
1460
"""Find the file ids and versions affected by revisions.
1462
:param revisions: an iterable containing revision ids.
1463
:param _inv_weave: The inventory weave from this repository or None.
1464
If None, the inventory weave will be opened automatically.
1465
:return: a dictionary mapping altered file-ids to an iterable of
1466
revision_ids. Each altered file-ids has the exact revision_ids that
1467
altered it listed explicitly.
1469
selected_keys = set((revid,) for revid in revision_ids)
1470
w = _inv_weave or self.inventories
1471
pb = ui.ui_factory.nested_progress_bar()
1473
return self._find_file_ids_from_xml_inventory_lines(
1474
w.iter_lines_added_or_present_in_keys(
1475
selected_keys, pb=pb),
1480
def iter_files_bytes(self, desired_files):
1481
"""Iterate through file versions.
1483
Files will not necessarily be returned in the order they occur in
1484
desired_files. No specific order is guaranteed.
1486
Yields pairs of identifier, bytes_iterator. identifier is an opaque
1487
value supplied by the caller as part of desired_files. It should
1488
uniquely identify the file version in the caller's context. (Examples:
1489
an index number or a TreeTransform trans_id.)
1491
bytes_iterator is an iterable of bytestrings for the file. The
1492
kind of iterable and length of the bytestrings are unspecified, but for
1493
this implementation, it is a list of bytes produced by
1494
VersionedFile.get_record_stream().
1496
:param desired_files: a list of (file_id, revision_id, identifier)
1500
for file_id, revision_id, callable_data in desired_files:
1501
text_keys[(file_id, revision_id)] = callable_data
1502
for record in self.texts.get_record_stream(text_keys, 'unordered', True):
1503
if record.storage_kind == 'absent':
1504
raise errors.RevisionNotPresent(record.key, self)
1505
yield text_keys[record.key], record.get_bytes_as('fulltext')
1507
def _generate_text_key_index(self, text_key_references=None,
1509
"""Generate a new text key index for the repository.
1511
This is an expensive function that will take considerable time to run.
1513
:return: A dict mapping text keys ((file_id, revision_id) tuples) to a
1514
list of parents, also text keys. When a given key has no parents,
1515
the parents list will be [NULL_REVISION].
1517
# All revisions, to find inventory parents.
1518
if ancestors is None:
1519
graph = self.get_graph()
1520
ancestors = graph.get_parent_map(self.all_revision_ids())
1521
if text_key_references is None:
1522
text_key_references = self.find_text_key_references()
1523
pb = ui.ui_factory.nested_progress_bar()
1525
return self._do_generate_text_key_index(ancestors,
1526
text_key_references, pb)
1530
def _do_generate_text_key_index(self, ancestors, text_key_references, pb):
1531
"""Helper for _generate_text_key_index to avoid deep nesting."""
1532
revision_order = tsort.topo_sort(ancestors)
1533
invalid_keys = set()
1535
for revision_id in revision_order:
1536
revision_keys[revision_id] = set()
1537
text_count = len(text_key_references)
1538
# a cache of the text keys to allow reuse; costs a dict of all the
1539
# keys, but saves a 2-tuple for every child of a given key.
1541
for text_key, valid in text_key_references.iteritems():
1543
invalid_keys.add(text_key)
1545
revision_keys[text_key[1]].add(text_key)
1546
text_key_cache[text_key] = text_key
1547
del text_key_references
1549
text_graph = graph.Graph(graph.DictParentsProvider(text_index))
1550
NULL_REVISION = _mod_revision.NULL_REVISION
1551
# Set a cache with a size of 10 - this suffices for bzr.dev but may be
1552
# too small for large or very branchy trees. However, for 55K path
1553
# trees, it would be easy to use too much memory trivially. Ideally we
1554
# could gauge this by looking at available real memory etc, but this is
1555
# always a tricky proposition.
1556
inventory_cache = lru_cache.LRUCache(10)
1557
batch_size = 10 # should be ~150MB on a 55K path tree
1558
batch_count = len(revision_order) / batch_size + 1
1560
pb.update("Calculating text parents.", processed_texts, text_count)
1561
for offset in xrange(batch_count):
1562
to_query = revision_order[offset * batch_size:(offset + 1) *
1566
for rev_tree in self.revision_trees(to_query):
1567
revision_id = rev_tree.get_revision_id()
1568
parent_ids = ancestors[revision_id]
1569
for text_key in revision_keys[revision_id]:
1570
pb.update("Calculating text parents.", processed_texts)
1571
processed_texts += 1
1572
candidate_parents = []
1573
for parent_id in parent_ids:
1574
parent_text_key = (text_key[0], parent_id)
1576
check_parent = parent_text_key not in \
1577
revision_keys[parent_id]
1579
# the parent parent_id is a ghost:
1580
check_parent = False
1581
# truncate the derived graph against this ghost.
1582
parent_text_key = None
1584
# look at the parent commit details inventories to
1585
# determine possible candidates in the per file graph.
1588
inv = inventory_cache[parent_id]
1590
inv = self.revision_tree(parent_id).inventory
1591
inventory_cache[parent_id] = inv
1592
parent_entry = inv._byid.get(text_key[0], None)
1593
if parent_entry is not None:
1595
text_key[0], parent_entry.revision)
1597
parent_text_key = None
1598
if parent_text_key is not None:
1599
candidate_parents.append(
1600
text_key_cache[parent_text_key])
1601
parent_heads = text_graph.heads(candidate_parents)
1602
new_parents = list(parent_heads)
1603
new_parents.sort(key=lambda x:candidate_parents.index(x))
1604
if new_parents == []:
1605
new_parents = [NULL_REVISION]
1606
text_index[text_key] = new_parents
1608
for text_key in invalid_keys:
1609
text_index[text_key] = [NULL_REVISION]
1612
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
1613
"""Get an iterable listing the keys of all the data introduced by a set
1616
The keys will be ordered so that the corresponding items can be safely
1617
fetched and inserted in that order.
1619
:returns: An iterable producing tuples of (knit-kind, file-id,
1620
versions). knit-kind is one of 'file', 'inventory', 'signatures',
1621
'revisions'. file-id is None unless knit-kind is 'file'.
1623
# XXX: it's a bit weird to control the inventory weave caching in this
1624
# generator. Ideally the caching would be done in fetch.py I think. Or
1625
# maybe this generator should explicitly have the contract that it
1626
# should not be iterated until the previously yielded item has been
1628
inv_w = self.inventories
1630
# file ids that changed
1631
file_ids = self.fileids_altered_by_revision_ids(revision_ids, inv_w)
1633
num_file_ids = len(file_ids)
1634
for file_id, altered_versions in file_ids.iteritems():
1635
if _files_pb is not None:
1636
_files_pb.update("fetch texts", count, num_file_ids)
1638
yield ("file", file_id, altered_versions)
1639
# We're done with the files_pb. Note that it finished by the caller,
1640
# just as it was created by the caller.
1644
yield ("inventory", None, revision_ids)
1647
# XXX: Note ATM no callers actually pay attention to this return
1648
# instead they just use the list of revision ids and ignore
1649
# missing sigs. Consider removing this work entirely
1650
revisions_with_signatures = set(self.signatures.get_parent_map(
1651
[(r,) for r in revision_ids]))
1652
revisions_with_signatures = set(
1653
[r for (r,) in revisions_with_signatures])
1654
revisions_with_signatures.intersection_update(revision_ids)
1655
yield ("signatures", None, revisions_with_signatures)
1658
yield ("revisions", None, revision_ids)
464
1660
@needs_read_lock
465
1661
def get_inventory(self, revision_id):
466
"""Get Inventory object by hash."""
467
return self.deserialise_inventory(
468
revision_id, self.get_inventory_xml(revision_id))
1662
"""Get Inventory object by revision id."""
1663
return self.iter_inventories([revision_id]).next()
1665
def iter_inventories(self, revision_ids):
1666
"""Get many inventories by revision_ids.
1668
This will buffer some or all of the texts used in constructing the
1669
inventories in memory, but will only parse a single inventory at a
1672
:return: An iterator of inventories.
1674
if ((None in revision_ids)
1675
or (_mod_revision.NULL_REVISION in revision_ids)):
1676
raise ValueError('cannot get null revision inventory')
1677
return self._iter_inventories(revision_ids)
1679
def _iter_inventories(self, revision_ids):
1680
"""single-document based inventory iteration."""
1681
for text, revision_id in self._iter_inventory_xmls(revision_ids):
1682
yield self.deserialise_inventory(revision_id, text)
1684
def _iter_inventory_xmls(self, revision_ids):
1685
keys = [(revision_id,) for revision_id in revision_ids]
1686
stream = self.inventories.get_record_stream(keys, 'unordered', True)
1688
for record in stream:
1689
if record.storage_kind != 'absent':
1690
text_chunks[record.key] = record.get_bytes_as('chunked')
1692
raise errors.NoSuchRevision(self, record.key)
1694
chunks = text_chunks.pop(key)
1695
yield ''.join(chunks), key[-1]
470
1697
def deserialise_inventory(self, revision_id, xml):
471
1698
"""Transform the xml into an inventory object.
728
2004
def supports_rich_root(self):
729
2005
return self._format.rich_root_data
732
class AllInOneRepository(Repository):
733
"""Legacy support - the repository behaviour for all-in-one branches."""
735
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
736
# we reuse one control files instance.
737
dir_mode = a_bzrdir._control_files._dir_mode
738
file_mode = a_bzrdir._control_files._file_mode
740
def get_store(name, compressed=True, prefixed=False):
741
# FIXME: This approach of assuming stores are all entirely compressed
742
# or entirely uncompressed is tidy, but breaks upgrade from
743
# some existing branches where there's a mixture; we probably
744
# still want the option to look for both.
745
relpath = a_bzrdir._control_files._escape(name)
746
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
747
prefixed=prefixed, compressed=compressed,
750
#if self._transport.should_cache():
751
# cache_path = os.path.join(self.cache_root, name)
752
# os.mkdir(cache_path)
753
# store = bzrlib.store.CachedStore(store, cache_path)
756
# not broken out yet because the controlweaves|inventory_store
757
# and text_store | weave_store bits are still different.
758
if isinstance(_format, RepositoryFormat4):
759
# cannot remove these - there is still no consistent api
760
# which allows access to this old info.
761
self.inventory_store = get_store('inventory-store')
762
text_store = get_store('text-store')
763
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
767
"""AllInOne repositories cannot be shared."""
771
def set_make_working_trees(self, new_value):
772
"""Set the policy flag for making working trees when creating branches.
774
This only applies to branches that use this repository.
776
The default is 'True'.
777
:param new_value: True to restore the default, False to disable making
2007
def _check_ascii_revisionid(self, revision_id, method):
2008
"""Private helper for ascii-only repositories."""
2009
# weave repositories refuse to store revisionids that are non-ascii.
2010
if revision_id is not None:
2011
# weaves require ascii revision ids.
2012
if isinstance(revision_id, unicode):
2014
revision_id.encode('ascii')
2015
except UnicodeEncodeError:
2016
raise errors.NonAsciiRevisionId(method, self)
2019
revision_id.decode('ascii')
2020
except UnicodeDecodeError:
2021
raise errors.NonAsciiRevisionId(method, self)
2023
def revision_graph_can_have_wrong_parents(self):
2024
"""Is it possible for this repository to have a revision graph with
2027
If True, then this repository must also implement
2028
_find_inconsistent_revision_parents so that check and reconcile can
2029
check for inconsistencies before proceeding with other checks that may
2030
depend on the revision index being consistent.
780
raise NotImplementedError(self.set_make_working_trees)
782
def make_working_trees(self):
783
"""Returns the policy for making working trees on new branches."""
2032
raise NotImplementedError(self.revision_graph_can_have_wrong_parents)
2035
# remove these delegates a while after bzr 0.15
2036
def __make_delegated(name, from_module):
2037
def _deprecated_repository_forwarder():
2038
symbol_versioning.warn('%s moved to %s in bzr 0.15'
2039
% (name, from_module),
2042
m = __import__(from_module, globals(), locals(), [name])
2044
return getattr(m, name)
2045
except AttributeError:
2046
raise AttributeError('module %s has no name %s'
2048
globals()[name] = _deprecated_repository_forwarder
2051
'AllInOneRepository',
2052
'WeaveMetaDirRepository',
2053
'PreSplitOutRepositoryFormat',
2054
'RepositoryFormat4',
2055
'RepositoryFormat5',
2056
'RepositoryFormat6',
2057
'RepositoryFormat7',
2059
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
2063
'RepositoryFormatKnit',
2064
'RepositoryFormatKnit1',
2066
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
787
2069
def install_revision(repository, rev, revision_tree):
788
2070
"""Install all revision data into a repository."""
2071
install_revisions(repository, [(rev, revision_tree, None)])
2074
def install_revisions(repository, iterable, num_revisions=None, pb=None):
2075
"""Install all revision data into a repository.
2077
Accepts an iterable of revision, tree, signature tuples. The signature
2080
repository.start_write_group()
2082
for n, (revision, revision_tree, signature) in enumerate(iterable):
2083
_install_revision(repository, revision, revision_tree, signature)
2085
pb.update('Transferring revisions', n + 1, num_revisions)
2087
repository.abort_write_group()
2090
repository.commit_write_group()
2093
def _install_revision(repository, rev, revision_tree, signature):
2094
"""Install all revision data into a repository."""
789
2095
present_parents = []
790
2096
parent_trees = {}
791
2097
for p_id in rev.parent_ids:
793
2099
present_parents.append(p_id)
794
2100
parent_trees[p_id] = repository.revision_tree(p_id)
796
parent_trees[p_id] = repository.revision_tree(None)
2102
parent_trees[p_id] = repository.revision_tree(
2103
_mod_revision.NULL_REVISION)
798
2105
inv = revision_tree.inventory
799
2106
entries = inv.iter_entries()
800
# backwards compatability hack: skip the root id.
2107
# backwards compatibility hack: skip the root id.
801
2108
if not repository.supports_rich_root():
802
2109
path, root = entries.next()
803
2110
if root.revision != rev.revision_id:
804
2111
raise errors.IncompatibleRevision(repr(repository))
2113
for path, ie in entries:
2114
text_keys[(ie.file_id, ie.revision)] = ie
2115
text_parent_map = repository.texts.get_parent_map(text_keys)
2116
missing_texts = set(text_keys) - set(text_parent_map)
805
2117
# Add the texts that are not already present
806
for path, ie in entries:
807
w = repository.weave_store.get_weave_or_empty(ie.file_id,
808
repository.get_transaction())
809
if ie.revision not in w:
811
# FIXME: TODO: The following loop *may* be overlapping/duplicate
812
# with InventoryEntry.find_previous_heads(). if it is, then there
813
# is a latent bug here where the parents may have ancestors of each
815
for revision, tree in parent_trees.iteritems():
816
if ie.file_id not in tree:
818
parent_id = tree.inventory[ie.file_id].revision
819
if parent_id in text_parents:
821
text_parents.append(parent_id)
823
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
824
repository.get_transaction())
825
lines = revision_tree.get_file(ie.file_id).readlines()
826
vfile.add_lines(rev.revision_id, text_parents, lines)
2118
for text_key in missing_texts:
2119
ie = text_keys[text_key]
2121
# FIXME: TODO: The following loop overlaps/duplicates that done by
2122
# commit to determine parents. There is a latent/real bug here where
2123
# the parents inserted are not those commit would do - in particular
2124
# they are not filtered by heads(). RBC, AB
2125
for revision, tree in parent_trees.iteritems():
2126
if ie.file_id not in tree:
2128
parent_id = tree.inventory[ie.file_id].revision
2129
if parent_id in text_parents:
2131
text_parents.append((ie.file_id, parent_id))
2132
lines = revision_tree.get_file(ie.file_id).readlines()
2133
repository.texts.add_lines(text_key, text_parents, lines)
828
2135
# install the inventory
829
2136
repository.add_inventory(rev.revision_id, inv, present_parents)
830
2137
except errors.RevisionAlreadyPresent:
2139
if signature is not None:
2140
repository.add_signature_text(rev.revision_id, signature)
832
2141
repository.add_revision(rev.revision_id, rev, inv)
835
2144
class MetaDirRepository(Repository):
836
"""Repositories in the new meta-dir layout."""
838
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
839
super(MetaDirRepository, self).__init__(_format,
845
dir_mode = self.control_files._dir_mode
846
file_mode = self.control_files._file_mode
2145
"""Repositories in the new meta-dir layout.
2147
:ivar _transport: Transport for access to repository control files,
2148
typically pointing to .bzr/repository.
2151
def __init__(self, _format, a_bzrdir, control_files):
2152
super(MetaDirRepository, self).__init__(_format, a_bzrdir, control_files)
2153
self._transport = control_files._transport
849
2155
def is_shared(self):
850
2156
"""Return True if this repository is flagged as a shared repository."""
851
return self.control_files._transport.has('shared-storage')
2157
return self._transport.has('shared-storage')
853
2159
@needs_write_lock
854
2160
def set_make_working_trees(self, new_value):
865
self.control_files._transport.delete('no-working-trees')
2171
self._transport.delete('no-working-trees')
866
2172
except errors.NoSuchFile:
869
self.control_files.put_utf8('no-working-trees', '')
2175
self._transport.put_bytes('no-working-trees', '',
2176
mode=self.bzrdir._get_file_mode())
871
2178
def make_working_trees(self):
872
2179
"""Returns the policy for making working trees on new branches."""
873
return not self.control_files._transport.has('no-working-trees')
876
class KnitRepository(MetaDirRepository):
877
"""Knit format repository."""
879
def _warn_if_deprecated(self):
880
# This class isn't deprecated
883
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
884
inv_vf.add_lines_with_ghosts(revid, parents, lines)
887
def _all_revision_ids(self):
888
"""See Repository.all_revision_ids()."""
889
# Knits get the revision graph from the index of the revision knit, so
890
# it's always possible even if they're on an unlistable transport.
891
return self._revision_store.all_revision_ids(self.get_transaction())
893
def fileid_involved_between_revs(self, from_revid, to_revid):
894
"""Find file_id(s) which are involved in the changes between revisions.
896
This determines the set of revisions which are involved, and then
897
finds all file ids affected by those revisions.
899
vf = self._get_revision_vf()
900
from_set = set(vf.get_ancestry(from_revid))
901
to_set = set(vf.get_ancestry(to_revid))
902
changed = to_set.difference(from_set)
903
return self._fileid_involved_by_set(changed)
905
def fileid_involved(self, last_revid=None):
906
"""Find all file_ids modified in the ancestry of last_revid.
908
:param last_revid: If None, last_revision() will be used.
911
changed = set(self.all_revision_ids())
913
changed = set(self.get_ancestry(last_revid))
916
return self._fileid_involved_by_set(changed)
919
def get_ancestry(self, revision_id):
920
"""Return a list of revision-ids integrated by a revision.
922
This is topologically sorted.
924
if revision_id is None:
926
vf = self._get_revision_vf()
928
return [None] + vf.get_ancestry(revision_id)
929
except errors.RevisionNotPresent:
930
raise errors.NoSuchRevision(self, revision_id)
933
def get_revision(self, revision_id):
934
"""Return the Revision object for a named revision"""
935
return self.get_revision_reconcile(revision_id)
938
def get_revision_graph(self, revision_id=None):
939
"""Return a dictionary containing the revision graph.
941
:param revision_id: The revision_id to get a graph from. If None, then
942
the entire revision graph is returned. This is a deprecated mode of
943
operation and will be removed in the future.
944
:return: a dictionary of revision_id->revision_parents_list.
946
# special case NULL_REVISION
947
if revision_id == _mod_revision.NULL_REVISION:
949
a_weave = self._get_revision_vf()
950
entire_graph = a_weave.get_graph()
951
if revision_id is None:
952
return a_weave.get_graph()
953
elif revision_id not in a_weave:
954
raise errors.NoSuchRevision(self, revision_id)
956
# add what can be reached from revision_id
958
pending = set([revision_id])
959
while len(pending) > 0:
961
result[node] = a_weave.get_parents(node)
962
for revision_id in result[node]:
963
if revision_id not in result:
964
pending.add(revision_id)
968
def get_revision_graph_with_ghosts(self, revision_ids=None):
969
"""Return a graph of the revisions with ghosts marked as applicable.
971
:param revision_ids: an iterable of revisions to graph or None for all.
972
:return: a Graph object with the graph reachable from revision_ids.
974
result = graph.Graph()
975
vf = self._get_revision_vf()
976
versions = set(vf.versions())
978
pending = set(self.all_revision_ids())
981
pending = set(revision_ids)
982
# special case NULL_REVISION
983
if _mod_revision.NULL_REVISION in pending:
984
pending.remove(_mod_revision.NULL_REVISION)
985
required = set(pending)
988
revision_id = pending.pop()
989
if not revision_id in versions:
990
if revision_id in required:
991
raise errors.NoSuchRevision(self, revision_id)
993
result.add_ghost(revision_id)
994
# mark it as done so we don't try for it again.
995
done.add(revision_id)
997
parent_ids = vf.get_parents_with_ghosts(revision_id)
998
for parent_id in parent_ids:
999
# is this queued or done ?
1000
if (parent_id not in pending and
1001
parent_id not in done):
1003
pending.add(parent_id)
1004
result.add_node(revision_id, parent_ids)
1005
done.add(revision_id)
1008
def _get_revision_vf(self):
1009
""":return: a versioned file containing the revisions."""
1010
vf = self._revision_store.get_revision_file(self.get_transaction())
1014
def reconcile(self, other=None, thorough=False):
1015
"""Reconcile this repository."""
1016
from bzrlib.reconcile import KnitReconciler
1017
reconciler = KnitReconciler(self, thorough=thorough)
1018
reconciler.reconcile()
2180
return not self._transport.has('no-working-trees')
2183
class MetaDirVersionedFileRepository(MetaDirRepository):
2184
"""Repositories in a meta-dir, that work via versioned file objects."""
2186
def __init__(self, _format, a_bzrdir, control_files):
2187
super(MetaDirVersionedFileRepository, self).__init__(_format, a_bzrdir,
2191
class RepositoryFormatRegistry(registry.Registry):
2192
"""Registry of RepositoryFormats."""
2194
def get(self, format_string):
2195
r = registry.Registry.get(self, format_string)
1021
def revision_parents(self, revision_id):
1022
return self._get_revision_vf().get_parents(revision_id)
1025
class KnitRepository2(KnitRepository):
1027
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1028
control_store, text_store):
1029
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1030
_revision_store, control_store, text_store)
1031
self._serializer = xml6.serializer_v6
1033
def deserialise_inventory(self, revision_id, xml):
1034
"""Transform the xml into an inventory object.
1036
:param revision_id: The expected revision id of the inventory.
1037
:param xml: A serialised inventory.
1039
result = self._serializer.read_inventory_from_string(xml)
1040
assert result.root.revision is not None
1043
def serialise_inventory(self, inv):
1044
"""Transform the inventory object into XML text.
1046
:param revision_id: The expected revision id of the inventory.
1047
:param xml: A serialised inventory.
1049
assert inv.revision_id is not None
1050
assert inv.root.revision is not None
1051
return KnitRepository.serialise_inventory(self, inv)
1053
def get_commit_builder(self, branch, parents, config, timestamp=None,
1054
timezone=None, committer=None, revprops=None,
1056
"""Obtain a CommitBuilder for this repository.
1058
:param branch: Branch to commit to.
1059
:param parents: Revision ids of the parents of the new revision.
1060
:param config: Configuration to use.
1061
:param timestamp: Optional timestamp recorded for commit.
1062
:param timezone: Optional timezone for timestamp.
1063
:param committer: Optional committer to set for commit.
1064
:param revprops: Optional dictionary of revision properties.
1065
:param revision_id: Optional revision id.
1067
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1068
committer, revprops, revision_id)
2201
format_registry = RepositoryFormatRegistry()
2202
"""Registry of formats, indexed by their identifying format string.
2204
This can contain either format instances themselves, or classes/factories that
2205
can be called to obtain one.
2209
#####################################################################
2210
# Repository Formats
1071
2212
class RepositoryFormat(object):
1072
2213
"""A repository format.
1209
2353
raise NotImplementedError(self.open)
1212
def register_format(klass, format):
1213
klass._formats[format.get_format_string()] = format
1216
def set_default_format(klass, format):
1217
klass._default_format = format
1220
def unregister_format(klass, format):
1221
assert klass._formats[format.get_format_string()] is format
1222
del klass._formats[format.get_format_string()]
1225
class PreSplitOutRepositoryFormat(RepositoryFormat):
1226
"""Base class for the pre split out repository formats."""
1228
rich_root_data = False
1230
def initialize(self, a_bzrdir, shared=False, _internal=False):
1231
"""Create a weave repository.
1233
TODO: when creating split out bzr branch formats, move this to a common
1234
base for Format5, Format6. or something like that.
1237
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1240
# always initialized when the bzrdir is.
1241
return self.open(a_bzrdir, _found=True)
1243
# Create an empty weave
1245
weavefile.write_weave_v5(weave.Weave(), sio)
1246
empty_weave = sio.getvalue()
1248
mutter('creating repository in %s.', a_bzrdir.transport.base)
1249
dirs = ['revision-store', 'weaves']
1250
files = [('inventory.weave', StringIO(empty_weave)),
1253
# FIXME: RBC 20060125 don't peek under the covers
1254
# NB: no need to escape relative paths that are url safe.
1255
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1256
'branch-lock', lockable_files.TransportLock)
1257
control_files.create_lock()
1258
control_files.lock_write()
1259
control_files._transport.mkdir_multi(dirs,
1260
mode=control_files._dir_mode)
1262
for file, content in files:
1263
control_files.put(file, content)
1265
control_files.unlock()
1266
return self.open(a_bzrdir, _found=True)
1268
def _get_control_store(self, repo_transport, control_files):
1269
"""Return the control store for this repository."""
1270
return self._get_versioned_file_store('',
1275
def _get_text_store(self, transport, control_files):
1276
"""Get a store for file texts for this format."""
1277
raise NotImplementedError(self._get_text_store)
1279
def open(self, a_bzrdir, _found=False):
1280
"""See RepositoryFormat.open()."""
1282
# we are being called directly and must probe.
1283
raise NotImplementedError
1285
repo_transport = a_bzrdir.get_repository_transport(None)
1286
control_files = a_bzrdir._control_files
1287
text_store = self._get_text_store(repo_transport, control_files)
1288
control_store = self._get_control_store(repo_transport, control_files)
1289
_revision_store = self._get_revision_store(repo_transport, control_files)
1290
return AllInOneRepository(_format=self,
1292
_revision_store=_revision_store,
1293
control_store=control_store,
1294
text_store=text_store)
1296
def check_conversion_target(self, target_format):
1300
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1301
"""Bzr repository format 4.
1303
This repository format has:
1305
- TextStores for texts, inventories,revisions.
1307
This format is deprecated: it indexes texts using a text id which is
1308
removed in format 5; initialization and write support for this format
1313
super(RepositoryFormat4, self).__init__()
1314
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1316
def get_format_description(self):
1317
"""See RepositoryFormat.get_format_description()."""
1318
return "Repository format 4"
1320
def initialize(self, url, shared=False, _internal=False):
1321
"""Format 4 branches cannot be created."""
1322
raise errors.UninitializableFormat(self)
1324
def is_supported(self):
1325
"""Format 4 is not supported.
1327
It is not supported because the model changed from 4 to 5 and the
1328
conversion logic is expensive - so doing it on the fly was not
1333
def _get_control_store(self, repo_transport, control_files):
1334
"""Format 4 repositories have no formal control store at this point.
1336
This will cause any control-file-needing apis to fail - this is desired.
1340
def _get_revision_store(self, repo_transport, control_files):
1341
"""See RepositoryFormat._get_revision_store()."""
1342
from bzrlib.xml4 import serializer_v4
1343
return self._get_text_rev_store(repo_transport,
1346
serializer=serializer_v4)
1348
def _get_text_store(self, transport, control_files):
1349
"""See RepositoryFormat._get_text_store()."""
1352
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1353
"""Bzr control format 5.
1355
This repository format has:
1356
- weaves for file texts and inventory
1358
- TextStores for revisions and signatures.
1362
super(RepositoryFormat5, self).__init__()
1363
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1365
def get_format_description(self):
1366
"""See RepositoryFormat.get_format_description()."""
1367
return "Weave repository format 5"
1369
def _get_revision_store(self, repo_transport, control_files):
1370
"""See RepositoryFormat._get_revision_store()."""
1371
"""Return the revision store object for this a_bzrdir."""
1372
return self._get_text_rev_store(repo_transport,
1377
def _get_text_store(self, transport, control_files):
1378
"""See RepositoryFormat._get_text_store()."""
1379
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1382
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1383
"""Bzr control format 6.
1385
This repository format has:
1386
- weaves for file texts and inventory
1387
- hash subdirectory based stores.
1388
- TextStores for revisions and signatures.
1392
super(RepositoryFormat6, self).__init__()
1393
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1395
def get_format_description(self):
1396
"""See RepositoryFormat.get_format_description()."""
1397
return "Weave repository format 6"
1399
def _get_revision_store(self, repo_transport, control_files):
1400
"""See RepositoryFormat._get_revision_store()."""
1401
return self._get_text_rev_store(repo_transport,
1407
def _get_text_store(self, transport, control_files):
1408
"""See RepositoryFormat._get_text_store()."""
1409
return self._get_versioned_file_store('weaves', transport, control_files)
1412
2356
class MetaDirRepositoryFormat(RepositoryFormat):
1413
2357
"""Common base class for the new repositories using the metadir layout."""
1415
2359
rich_root_data = False
2360
supports_tree_reference = False
2361
supports_external_lookups = False
2364
def _matchingbzrdir(self):
2365
matching = bzrdir.BzrDirMetaFormat1()
2366
matching.repository_format = self
1417
2369
def __init__(self):
1418
2370
super(MetaDirRepositoryFormat, self).__init__()
1419
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1421
2372
def _create_control_files(self, a_bzrdir):
1422
2373
"""Create the required files and the initial control_files object."""
1432
2383
"""Upload the initial blank content."""
1433
2384
control_files = self._create_control_files(a_bzrdir)
1434
2385
control_files.lock_write()
2386
transport = control_files._transport
2388
utf8_files += [('shared-storage', '')]
1436
control_files._transport.mkdir_multi(dirs,
1437
mode=control_files._dir_mode)
1438
for file, content in files:
1439
control_files.put(file, content)
1440
for file, content in utf8_files:
1441
control_files.put_utf8(file, content)
1443
control_files.put_utf8('shared-storage', '')
2390
transport.mkdir_multi(dirs, mode=a_bzrdir._get_dir_mode())
2391
for (filename, content_stream) in files:
2392
transport.put_file(filename, content_stream,
2393
mode=a_bzrdir._get_file_mode())
2394
for (filename, content_bytes) in utf8_files:
2395
transport.put_bytes_non_atomic(filename, content_bytes,
2396
mode=a_bzrdir._get_file_mode())
1445
2398
control_files.unlock()
1448
class RepositoryFormat7(MetaDirRepositoryFormat):
1449
"""Bzr repository 7.
1451
This repository format has:
1452
- weaves for file texts and inventory
1453
- hash subdirectory based stores.
1454
- TextStores for revisions and signatures.
1455
- a format marker of its own
1456
- an optional 'shared-storage' flag
1457
- an optional 'no-working-trees' flag
1460
def _get_control_store(self, repo_transport, control_files):
1461
"""Return the control store for this repository."""
1462
return self._get_versioned_file_store('',
1467
def get_format_string(self):
1468
"""See RepositoryFormat.get_format_string()."""
1469
return "Bazaar-NG Repository format 7"
1471
def get_format_description(self):
1472
"""See RepositoryFormat.get_format_description()."""
1473
return "Weave repository format 7"
1475
def check_conversion_target(self, target_format):
1478
def _get_revision_store(self, repo_transport, control_files):
1479
"""See RepositoryFormat._get_revision_store()."""
1480
return self._get_text_rev_store(repo_transport,
1487
def _get_text_store(self, transport, control_files):
1488
"""See RepositoryFormat._get_text_store()."""
1489
return self._get_versioned_file_store('weaves',
1493
def initialize(self, a_bzrdir, shared=False):
1494
"""Create a weave repository.
1496
:param shared: If true the repository will be initialized as a shared
1499
# Create an empty weave
1501
weavefile.write_weave_v5(weave.Weave(), sio)
1502
empty_weave = sio.getvalue()
1504
mutter('creating repository in %s.', a_bzrdir.transport.base)
1505
dirs = ['revision-store', 'weaves']
1506
files = [('inventory.weave', StringIO(empty_weave)),
1508
utf8_files = [('format', self.get_format_string())]
1510
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1511
return self.open(a_bzrdir=a_bzrdir, _found=True)
1513
def open(self, a_bzrdir, _found=False, _override_transport=None):
1514
"""See RepositoryFormat.open().
1516
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1517
repository at a slightly different url
1518
than normal. I.e. during 'upgrade'.
1521
format = RepositoryFormat.find_format(a_bzrdir)
1522
assert format.__class__ == self.__class__
1523
if _override_transport is not None:
1524
repo_transport = _override_transport
1526
repo_transport = a_bzrdir.get_repository_transport(None)
1527
control_files = lockable_files.LockableFiles(repo_transport,
1528
'lock', lockdir.LockDir)
1529
text_store = self._get_text_store(repo_transport, control_files)
1530
control_store = self._get_control_store(repo_transport, control_files)
1531
_revision_store = self._get_revision_store(repo_transport, control_files)
1532
return MetaDirRepository(_format=self,
1534
control_files=control_files,
1535
_revision_store=_revision_store,
1536
control_store=control_store,
1537
text_store=text_store)
1540
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1541
"""Bzr repository knit format (generalized).
1543
This repository format has:
1544
- knits for file texts and inventory
1545
- hash subdirectory based stores.
1546
- knits for revisions and signatures
1547
- TextStores for revisions and signatures.
1548
- a format marker of its own
1549
- an optional 'shared-storage' flag
1550
- an optional 'no-working-trees' flag
1554
def _get_control_store(self, repo_transport, control_files):
1555
"""Return the control store for this repository."""
1556
return VersionedFileStore(
1559
file_mode=control_files._file_mode,
1560
versionedfile_class=knit.KnitVersionedFile,
1561
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1564
def _get_revision_store(self, repo_transport, control_files):
1565
"""See RepositoryFormat._get_revision_store()."""
1566
from bzrlib.store.revision.knit import KnitRevisionStore
1567
versioned_file_store = VersionedFileStore(
1569
file_mode=control_files._file_mode,
1572
versionedfile_class=knit.KnitVersionedFile,
1573
versionedfile_kwargs={'delta':False,
1574
'factory':knit.KnitPlainFactory(),
1578
return KnitRevisionStore(versioned_file_store)
1580
def _get_text_store(self, transport, control_files):
1581
"""See RepositoryFormat._get_text_store()."""
1582
return self._get_versioned_file_store('knits',
1585
versionedfile_class=knit.KnitVersionedFile,
1586
versionedfile_kwargs={
1587
'create_parent_dir':True,
1588
'delay_create':True,
1589
'dir_mode':control_files._dir_mode,
1593
def initialize(self, a_bzrdir, shared=False):
1594
"""Create a knit format 1 repository.
1596
:param a_bzrdir: bzrdir to contain the new repository; must already
1598
:param shared: If true the repository will be initialized as a shared
1601
mutter('creating repository in %s.', a_bzrdir.transport.base)
1602
dirs = ['revision-store', 'knits']
1604
utf8_files = [('format', self.get_format_string())]
1606
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1607
repo_transport = a_bzrdir.get_repository_transport(None)
1608
control_files = lockable_files.LockableFiles(repo_transport,
1609
'lock', lockdir.LockDir)
1610
control_store = self._get_control_store(repo_transport, control_files)
1611
transaction = transactions.WriteTransaction()
1612
# trigger a write of the inventory store.
1613
control_store.get_weave_or_empty('inventory', transaction)
1614
_revision_store = self._get_revision_store(repo_transport, control_files)
1615
_revision_store.has_revision_id('A', transaction)
1616
_revision_store.get_signature_file(transaction)
1617
return self.open(a_bzrdir=a_bzrdir, _found=True)
1619
def open(self, a_bzrdir, _found=False, _override_transport=None):
1620
"""See RepositoryFormat.open().
1622
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1623
repository at a slightly different url
1624
than normal. I.e. during 'upgrade'.
1627
format = RepositoryFormat.find_format(a_bzrdir)
1628
assert format.__class__ == self.__class__
1629
if _override_transport is not None:
1630
repo_transport = _override_transport
1632
repo_transport = a_bzrdir.get_repository_transport(None)
1633
control_files = lockable_files.LockableFiles(repo_transport,
1634
'lock', lockdir.LockDir)
1635
text_store = self._get_text_store(repo_transport, control_files)
1636
control_store = self._get_control_store(repo_transport, control_files)
1637
_revision_store = self._get_revision_store(repo_transport, control_files)
1638
return KnitRepository(_format=self,
1640
control_files=control_files,
1641
_revision_store=_revision_store,
1642
control_store=control_store,
1643
text_store=text_store)
1646
class RepositoryFormatKnit1(RepositoryFormatKnit):
1647
"""Bzr repository knit format 1.
1649
This repository format has:
1650
- knits for file texts and inventory
1651
- hash subdirectory based stores.
1652
- knits for revisions and signatures
1653
- TextStores for revisions and signatures.
1654
- a format marker of its own
1655
- an optional 'shared-storage' flag
1656
- an optional 'no-working-trees' flag
1659
This format was introduced in bzr 0.8.
1661
def get_format_string(self):
1662
"""See RepositoryFormat.get_format_string()."""
1663
return "Bazaar-NG Knit Repository Format 1"
1665
def get_format_description(self):
1666
"""See RepositoryFormat.get_format_description()."""
1667
return "Knit repository format 1"
1669
def check_conversion_target(self, target_format):
1673
class RepositoryFormatKnit2(RepositoryFormatKnit):
1674
"""Bzr repository knit format 2.
1676
THIS FORMAT IS EXPERIMENTAL
1677
This repository format has:
1678
- knits for file texts and inventory
1679
- hash subdirectory based stores.
1680
- knits for revisions and signatures
1681
- TextStores for revisions and signatures.
1682
- a format marker of its own
1683
- an optional 'shared-storage' flag
1684
- an optional 'no-working-trees' flag
1686
- Support for recording full info about the tree root
1690
rich_root_data = True
1692
def get_format_string(self):
1693
"""See RepositoryFormat.get_format_string()."""
1694
return "Bazaar Knit Repository Format 2\n"
1696
def get_format_description(self):
1697
"""See RepositoryFormat.get_format_description()."""
1698
return "Knit repository format 2"
1700
def check_conversion_target(self, target_format):
1701
if not target_format.rich_root_data:
1702
raise errors.BadConversionTarget(
1703
'Does not support rich root data.', target_format)
1705
def open(self, a_bzrdir, _found=False, _override_transport=None):
1706
"""See RepositoryFormat.open().
1708
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1709
repository at a slightly different url
1710
than normal. I.e. during 'upgrade'.
1713
format = RepositoryFormat.find_format(a_bzrdir)
1714
assert format.__class__ == self.__class__
1715
if _override_transport is not None:
1716
repo_transport = _override_transport
1718
repo_transport = a_bzrdir.get_repository_transport(None)
1719
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1721
text_store = self._get_text_store(repo_transport, control_files)
1722
control_store = self._get_control_store(repo_transport, control_files)
1723
_revision_store = self._get_revision_store(repo_transport, control_files)
1724
return KnitRepository2(_format=self,
1726
control_files=control_files,
1727
_revision_store=_revision_store,
1728
control_store=control_store,
1729
text_store=text_store)
1733
2401
# formats which have no format string are not discoverable
1734
# and not independently creatable, so are not registered.
1735
RepositoryFormat.register_format(RepositoryFormat7())
1736
_default_format = RepositoryFormatKnit1()
1737
RepositoryFormat.register_format(_default_format)
1738
RepositoryFormat.register_format(RepositoryFormatKnit2())
1739
RepositoryFormat.set_default_format(_default_format)
1740
_legacy_formats = [RepositoryFormat4(),
1741
RepositoryFormat5(),
1742
RepositoryFormat6()]
2402
# and not independently creatable, so are not registered. They're
2403
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
2404
# needed, it's constructed directly by the BzrDir. Non-native formats where
2405
# the repository is not separately opened are similar.
2407
format_registry.register_lazy(
2408
'Bazaar-NG Repository format 7',
2409
'bzrlib.repofmt.weaverepo',
2413
format_registry.register_lazy(
2414
'Bazaar-NG Knit Repository Format 1',
2415
'bzrlib.repofmt.knitrepo',
2416
'RepositoryFormatKnit1',
2419
format_registry.register_lazy(
2420
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
2421
'bzrlib.repofmt.knitrepo',
2422
'RepositoryFormatKnit3',
2425
format_registry.register_lazy(
2426
'Bazaar Knit Repository Format 4 (bzr 1.0)\n',
2427
'bzrlib.repofmt.knitrepo',
2428
'RepositoryFormatKnit4',
2431
# Pack-based formats. There is one format for pre-subtrees, and one for
2432
# post-subtrees to allow ease of testing.
2433
# NOTE: These are experimental in 0.92. Stable in 1.0 and above
2434
format_registry.register_lazy(
2435
'Bazaar pack repository format 1 (needs bzr 0.92)\n',
2436
'bzrlib.repofmt.pack_repo',
2437
'RepositoryFormatKnitPack1',
2439
format_registry.register_lazy(
2440
'Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n',
2441
'bzrlib.repofmt.pack_repo',
2442
'RepositoryFormatKnitPack3',
2444
format_registry.register_lazy(
2445
'Bazaar pack repository format 1 with rich root (needs bzr 1.0)\n',
2446
'bzrlib.repofmt.pack_repo',
2447
'RepositoryFormatKnitPack4',
2449
format_registry.register_lazy(
2450
'Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n',
2451
'bzrlib.repofmt.pack_repo',
2452
'RepositoryFormatKnitPack5',
2454
format_registry.register_lazy(
2455
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n',
2456
'bzrlib.repofmt.pack_repo',
2457
'RepositoryFormatKnitPack5RichRoot',
2459
format_registry.register_lazy(
2460
'Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n',
2461
'bzrlib.repofmt.pack_repo',
2462
'RepositoryFormatKnitPack5RichRootBroken',
2464
format_registry.register_lazy(
2465
'Bazaar RepositoryFormatKnitPack6 (bzr 1.9)\n',
2466
'bzrlib.repofmt.pack_repo',
2467
'RepositoryFormatKnitPack6',
2469
format_registry.register_lazy(
2470
'Bazaar RepositoryFormatKnitPack6RichRoot (bzr 1.9)\n',
2471
'bzrlib.repofmt.pack_repo',
2472
'RepositoryFormatKnitPack6RichRoot',
2475
# Development formats.
2476
# 1.7->1.8 go below here
2477
format_registry.register_lazy(
2478
"Bazaar development format 2 (needs bzr.dev from before 1.8)\n",
2479
'bzrlib.repofmt.pack_repo',
2480
'RepositoryFormatPackDevelopment2',
2482
format_registry.register_lazy(
2483
("Bazaar development format 2 with subtree support "
2484
"(needs bzr.dev from before 1.8)\n"),
2485
'bzrlib.repofmt.pack_repo',
2486
'RepositoryFormatPackDevelopment2Subtree',
1745
2490
class InterRepository(InterObject):
1770
2523
:param pb: optional progress bar to use for progress reports. If not
1771
2524
provided a default one will be created.
1773
Returns the copied revision count and the failed revisions in a tuple:
1776
raise NotImplementedError(self.fetch)
2526
:returns: (copied_revision_count, failures).
2528
# Normally we should find a specific InterRepository subclass to do
2529
# the fetch; if nothing else then at least InterSameDataRepository.
2530
# If none of them is suitable it looks like fetching is not possible;
2531
# we try to give a good message why. _assert_same_model will probably
2532
# give a helpful message; otherwise a generic one.
2533
self._assert_same_model(self.source, self.target)
2534
raise errors.IncompatibleRepositories(self.source, self.target,
2535
"no suitableInterRepository found")
2537
def _walk_to_common_revisions(self, revision_ids):
2538
"""Walk out from revision_ids in source to revisions target has.
2540
:param revision_ids: The start point for the search.
2541
:return: A set of revision ids.
2543
target_graph = self.target_get_graph()
2544
revision_ids = frozenset(revision_ids)
2545
# Fast path for the case where all the revisions are already in the
2547
# (Although this does incur an extra round trip for the
2548
# fairly common case where the target doesn't already have the revision
2550
if set(target_graph.get_parent_map(revision_ids)) == revision_ids:
2551
return graph.SearchResult(revision_ids, set(), 0, set())
2552
missing_revs = set()
2553
source_graph = self.source.get_graph()
2554
# ensure we don't pay silly lookup costs.
2555
searcher = source_graph._make_breadth_first_searcher(revision_ids)
2556
null_set = frozenset([_mod_revision.NULL_REVISION])
2557
searcher_exhausted = False
2561
# Iterate the searcher until we have enough next_revs
2562
while len(next_revs) < self._walk_to_common_revisions_batch_size:
2564
next_revs_part, ghosts_part = searcher.next_with_ghosts()
2565
next_revs.update(next_revs_part)
2566
ghosts.update(ghosts_part)
2567
except StopIteration:
2568
searcher_exhausted = True
2570
# If there are ghosts in the source graph, and the caller asked for
2571
# them, make sure that they are present in the target.
2572
# We don't care about other ghosts as we can't fetch them and
2573
# haven't been asked to.
2574
ghosts_to_check = set(revision_ids.intersection(ghosts))
2575
revs_to_get = set(next_revs).union(ghosts_to_check)
2577
have_revs = set(target_graph.get_parent_map(revs_to_get))
2578
# we always have NULL_REVISION present.
2579
have_revs = have_revs.union(null_set)
2580
# Check if the target is missing any ghosts we need.
2581
ghosts_to_check.difference_update(have_revs)
2583
# One of the caller's revision_ids is a ghost in both the
2584
# source and the target.
2585
raise errors.NoSuchRevision(
2586
self.source, ghosts_to_check.pop())
2587
missing_revs.update(next_revs - have_revs)
2588
# Because we may have walked past the original stop point, make
2589
# sure everything is stopped
2590
stop_revs = searcher.find_seen_ancestors(have_revs)
2591
searcher.stop_searching_any(stop_revs)
2592
if searcher_exhausted:
2594
return searcher.get_result()
2596
@deprecated_method(one_two)
1778
2597
@needs_read_lock
1779
def missing_revision_ids(self, revision_id=None):
2598
def missing_revision_ids(self, revision_id=None, find_ghosts=True):
1780
2599
"""Return the revision ids that source has that target does not.
1782
2601
These are returned in topological order.
1784
2603
:param revision_id: only return revision ids included by this
2605
:param find_ghosts: If True find missing revisions in deep history
2606
rather than just finding the surface difference.
2608
return list(self.search_missing_revision_ids(
2609
revision_id, find_ghosts).get_keys())
2612
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2613
"""Return the revision ids that source has that target does not.
2615
:param revision_id: only return revision ids included by this
2617
:param find_ghosts: If True find missing revisions in deep history
2618
rather than just finding the surface difference.
2619
:return: A bzrlib.graph.SearchResult.
2621
# stop searching at found target revisions.
2622
if not find_ghosts and revision_id is not None:
2623
return self._walk_to_common_revisions([revision_id])
1787
2624
# generic, possibly worst case, slow code path.
1788
2625
target_ids = set(self.target.all_revision_ids())
1789
2626
if revision_id is not None:
1790
2627
source_ids = self.source.get_ancestry(revision_id)
1791
assert source_ids[0] is None
2628
if source_ids[0] is not None:
2629
raise AssertionError()
1792
2630
source_ids.pop(0)
1794
2632
source_ids = self.source.all_revision_ids()
1795
2633
result_set = set(source_ids).difference(target_ids)
1796
# this may look like a no-op: its not. It preserves the ordering
1797
# other_ids had while only returning the members from other_ids
1798
# that we've decided we need.
1799
return [rev_id for rev_id in source_ids if rev_id in result_set]
2634
return self.source.revision_ids_to_search_result(result_set)
2637
def _same_model(source, target):
2638
"""True if source and target have the same data representation.
2640
Note: this is always called on the base class; overriding it in a
2641
subclass will have no effect.
2644
InterRepository._assert_same_model(source, target)
2646
except errors.IncompatibleRepositories, e:
2650
def _assert_same_model(source, target):
2651
"""Raise an exception if two repositories do not use the same model.
2653
if source.supports_rich_root() != target.supports_rich_root():
2654
raise errors.IncompatibleRepositories(source, target,
2655
"different rich-root support")
2656
if source._serializer != target._serializer:
2657
raise errors.IncompatibleRepositories(source, target,
2658
"different serializers")
1802
2661
class InterSameDataRepository(InterRepository):
1884
2753
@needs_write_lock
1885
def copy_content(self, revision_id=None, basis=None):
2754
def copy_content(self, revision_id=None):
1886
2755
"""See InterRepository.copy_content()."""
1887
2756
# weave specific optimised path:
1888
if basis is not None:
1889
# copy the basis in, then fetch remaining data.
1890
basis.copy_content_into(self.target, revision_id)
1891
# the basis copy_content_into could miss-set this.
2758
self.target.set_make_working_trees(self.source.make_working_trees())
2759
except (errors.RepositoryUpgradeRequired, NotImplemented):
2761
# FIXME do not peek!
2762
if self.source._transport.listable():
2763
pb = ui.ui_factory.nested_progress_bar()
1893
self.target.set_make_working_trees(self.source.make_working_trees())
1894
except NotImplementedError:
2765
self.target.texts.insert_record_stream(
2766
self.source.texts.get_record_stream(
2767
self.source.texts.keys(), 'topological', False))
2768
pb.update('copying inventory', 0, 1)
2769
self.target.inventories.insert_record_stream(
2770
self.source.inventories.get_record_stream(
2771
self.source.inventories.keys(), 'topological', False))
2772
self.target.signatures.insert_record_stream(
2773
self.source.signatures.get_record_stream(
2774
self.source.signatures.keys(),
2776
self.target.revisions.insert_record_stream(
2777
self.source.revisions.get_record_stream(
2778
self.source.revisions.keys(),
2779
'topological', True))
1896
2783
self.target.fetch(self.source, revision_id=revision_id)
1899
self.target.set_make_working_trees(self.source.make_working_trees())
1900
except NotImplementedError:
1902
# FIXME do not peek!
1903
if self.source.control_files._transport.listable():
1904
pb = ui.ui_factory.nested_progress_bar()
1906
self.target.weave_store.copy_all_ids(
1907
self.source.weave_store,
1909
from_transaction=self.source.get_transaction(),
1910
to_transaction=self.target.get_transaction())
1911
pb.update('copying inventory', 0, 1)
1912
self.target.control_weaves.copy_multi(
1913
self.source.control_weaves, ['inventory'],
1914
from_transaction=self.source.get_transaction(),
1915
to_transaction=self.target.get_transaction())
1916
self.target._revision_store.text_store.copy_all_ids(
1917
self.source._revision_store.text_store,
1922
self.target.fetch(self.source, revision_id=revision_id)
1924
2785
@needs_write_lock
1925
def fetch(self, revision_id=None, pb=None):
2786
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
1926
2787
"""See InterRepository.fetch()."""
1927
from bzrlib.fetch import GenericRepoFetcher
2788
from bzrlib.fetch import RepoFetcher
1928
2789
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1929
2790
self.source, self.source._format, self.target, self.target._format)
1930
f = GenericRepoFetcher(to_repository=self.target,
2791
f = RepoFetcher(to_repository=self.target,
1931
2792
from_repository=self.source,
1932
2793
last_revision=revision_id,
2794
pb=pb, find_ghosts=find_ghosts)
1934
2795
return f.count_copied, f.failed_revisions
1936
2797
@needs_read_lock
1937
def missing_revision_ids(self, revision_id=None):
2798
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
1938
2799
"""See InterRepository.missing_revision_ids()."""
1939
2800
# we want all revisions to satisfy revision_id in source.
1940
2801
# but we don't want to stat every file here and there.
1989
2855
could lead to confusing results, and there is no need to be
1990
2856
overly general.
2858
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit
1993
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1994
isinstance(target._format, (RepositoryFormatKnit1)))
2860
are_knits = (isinstance(source._format, RepositoryFormatKnit) and
2861
isinstance(target._format, RepositoryFormatKnit))
1995
2862
except AttributeError:
2864
return are_knits and InterRepository._same_model(source, target)
1998
2866
@needs_write_lock
1999
def fetch(self, revision_id=None, pb=None):
2867
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2000
2868
"""See InterRepository.fetch()."""
2001
from bzrlib.fetch import KnitRepoFetcher
2869
from bzrlib.fetch import RepoFetcher
2002
2870
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2003
2871
self.source, self.source._format, self.target, self.target._format)
2004
f = KnitRepoFetcher(to_repository=self.target,
2872
f = RepoFetcher(to_repository=self.target,
2005
2873
from_repository=self.source,
2006
2874
last_revision=revision_id,
2875
pb=pb, find_ghosts=find_ghosts)
2008
2876
return f.count_copied, f.failed_revisions
2010
2878
@needs_read_lock
2011
def missing_revision_ids(self, revision_id=None):
2879
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
2012
2880
"""See InterRepository.missing_revision_ids()."""
2013
2881
if revision_id is not None:
2014
2882
source_ids = self.source.get_ancestry(revision_id)
2015
assert source_ids[0] is None
2883
if source_ids[0] is not None:
2884
raise AssertionError()
2016
2885
source_ids.pop(0)
2018
source_ids = self.source._all_possible_ids()
2887
source_ids = self.source.all_revision_ids()
2019
2888
source_ids_set = set(source_ids)
2020
2889
# source_ids is the worst possible case we may need to pull.
2021
2890
# now we want to filter source_ids against what we actually
2022
2891
# have in target, but don't try to check for existence where we know
2023
2892
# we do not have a revision as that would be pointless.
2024
target_ids = set(self.target._all_possible_ids())
2893
target_ids = set(self.target.all_revision_ids())
2025
2894
possibly_present_revisions = target_ids.intersection(source_ids_set)
2026
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2895
actually_present_revisions = set(
2896
self.target._eliminate_revisions_not_present(possibly_present_revisions))
2027
2897
required_revisions = source_ids_set.difference(actually_present_revisions)
2028
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2029
2898
if revision_id is not None:
2030
2899
# we used get_ancestry to determine source_ids then we are assured all
2031
2900
# revisions referenced are present as they are installed in topological order.
2032
2901
# and the tip revision was validated by get_ancestry.
2033
return required_topo_revisions
2902
result_set = required_revisions
2035
2904
# if we just grabbed the possibly available ids, then
2036
2905
# we only have an estimate of whats available and need to validate
2037
2906
# that against the revision records.
2038
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2908
self.source._eliminate_revisions_not_present(required_revisions))
2909
return self.source.revision_ids_to_search_result(result_set)
2912
class InterPackRepo(InterSameDataRepository):
2913
"""Optimised code paths between Pack based repositories."""
2916
def _get_repo_format_to_test(self):
2917
from bzrlib.repofmt import pack_repo
2918
return pack_repo.RepositoryFormatKnitPack1()
2921
def is_compatible(source, target):
2922
"""Be compatible with known Pack formats.
2924
We don't test for the stores being of specific types because that
2925
could lead to confusing results, and there is no need to be
2928
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
2930
are_packs = (isinstance(source._format, RepositoryFormatPack) and
2931
isinstance(target._format, RepositoryFormatPack))
2932
except AttributeError:
2934
return are_packs and InterRepository._same_model(source, target)
2937
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2938
"""See InterRepository.fetch()."""
2939
if (len(self.source._fallback_repositories) > 0 or
2940
len(self.target._fallback_repositories) > 0):
2941
# The pack layer is not aware of fallback repositories, so when
2942
# fetching from a stacked repository or into a stacked repository
2943
# we use the generic fetch logic which uses the VersionedFiles
2944
# attributes on repository.
2945
from bzrlib.fetch import RepoFetcher
2946
fetcher = RepoFetcher(self.target, self.source, revision_id,
2948
return fetcher.count_copied, fetcher.failed_revisions
2949
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2950
self.source, self.source._format, self.target, self.target._format)
2951
self.count_copied = 0
2952
if revision_id is None:
2954
# everything to do - use pack logic
2955
# to fetch from all packs to one without
2956
# inventory parsing etc, IFF nothing to be copied is in the target.
2958
source_revision_ids = frozenset(self.source.all_revision_ids())
2959
revision_ids = source_revision_ids - \
2960
frozenset(self.target_get_parent_map(source_revision_ids))
2961
revision_keys = [(revid,) for revid in revision_ids]
2962
target_pack_collection = self._get_target_pack_collection()
2963
index = target_pack_collection.revision_index.combined_index
2964
present_revision_ids = set(item[1][0] for item in
2965
index.iter_entries(revision_keys))
2966
revision_ids = set(revision_ids) - present_revision_ids
2967
# implementing the TODO will involve:
2968
# - detecting when all of a pack is selected
2969
# - avoiding as much as possible pre-selection, so the
2970
# more-core routines such as create_pack_from_packs can filter in
2971
# a just-in-time fashion. (though having a HEADS list on a
2972
# repository might make this a lot easier, because we could
2973
# sensibly detect 'new revisions' without doing a full index scan.
2974
elif _mod_revision.is_null(revision_id):
2979
revision_ids = self.search_missing_revision_ids(revision_id,
2980
find_ghosts=find_ghosts).get_keys()
2981
except errors.NoSuchRevision:
2982
raise errors.InstallFailed([revision_id])
2983
if len(revision_ids) == 0:
2985
return self._pack(self.source, self.target, revision_ids)
2987
def _pack(self, source, target, revision_ids):
2988
from bzrlib.repofmt.pack_repo import Packer
2989
target_pack_collection = self._get_target_pack_collection()
2990
packs = source._pack_collection.all_packs()
2991
pack = Packer(target_pack_collection, packs, '.fetch',
2992
revision_ids).pack()
2993
if pack is not None:
2994
target_pack_collection._save_pack_names()
2995
copied_revs = pack.get_revision_count()
2996
# Trigger an autopack. This may duplicate effort as we've just done
2997
# a pack creation, but for now it is simpler to think about as
2998
# 'upload data, then repack if needed'.
3000
return (copied_revs, [])
3004
def _autopack(self):
3005
self.target._pack_collection.autopack()
3007
def _get_target_pack_collection(self):
3008
return self.target._pack_collection
3011
def search_missing_revision_ids(self, revision_id=None, find_ghosts=True):
3012
"""See InterRepository.missing_revision_ids().
3014
:param find_ghosts: Find ghosts throughout the ancestry of
3017
if not find_ghosts and revision_id is not None:
3018
return self._walk_to_common_revisions([revision_id])
3019
elif revision_id is not None:
3020
# Find ghosts: search for revisions pointing from one repository to
3021
# the other, and vice versa, anywhere in the history of revision_id.
3022
graph = self.target_get_graph(other_repository=self.source)
3023
searcher = graph._make_breadth_first_searcher([revision_id])
3027
next_revs, ghosts = searcher.next_with_ghosts()
3028
except StopIteration:
3030
if revision_id in ghosts:
3031
raise errors.NoSuchRevision(self.source, revision_id)
3032
found_ids.update(next_revs)
3033
found_ids.update(ghosts)
3034
found_ids = frozenset(found_ids)
3035
# Double query here: should be able to avoid this by changing the
3036
# graph api further.
3037
result_set = found_ids - frozenset(
3038
self.target_get_parent_map(found_ids))
3040
source_ids = self.source.all_revision_ids()
3041
# source_ids is the worst possible case we may need to pull.
3042
# now we want to filter source_ids against what we actually
3043
# have in target, but don't try to check for existence where we know
3044
# we do not have a revision as that would be pointless.
3045
target_ids = set(self.target.all_revision_ids())
3046
result_set = set(source_ids).difference(target_ids)
3047
return self.source.revision_ids_to_search_result(result_set)
2041
3050
class InterModel1and2(InterRepository):
2043
_matching_repo_format = None
3053
def _get_repo_format_to_test(self):
2046
3057
def is_compatible(source, target):
2047
if not isinstance(source, Repository):
2049
if not isinstance(target, Repository):
2051
if not source._format.rich_root_data and target._format.rich_root_data:
3058
if not source.supports_rich_root() and target.supports_rich_root():
2056
3063
@needs_write_lock
2057
def fetch(self, revision_id=None, pb=None):
3064
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
2058
3065
"""See InterRepository.fetch()."""
2059
3066
from bzrlib.fetch import Model1toKnit2Fetcher
2060
3067
f = Model1toKnit2Fetcher(to_repository=self.target,
2061
3068
from_repository=self.source,
2062
3069
last_revision=revision_id,
3070
pb=pb, find_ghosts=find_ghosts)
2064
3071
return f.count_copied, f.failed_revisions
2066
3073
@needs_write_lock
2067
def copy_content(self, revision_id=None, basis=None):
3074
def copy_content(self, revision_id=None):
2068
3075
"""Make a complete copy of the content in self into destination.
2070
3077
This is a destructive operation! Do not use it on existing
2111
3160
f = Knit1to2Fetcher(to_repository=self.target,
2112
3161
from_repository=self.source,
2113
3162
last_revision=revision_id,
3163
pb=pb, find_ghosts=find_ghosts)
2115
3164
return f.count_copied, f.failed_revisions
3167
class InterDifferingSerializer(InterKnitRepo):
3170
def _get_repo_format_to_test(self):
3174
def is_compatible(source, target):
3175
"""Be compatible with Knit2 source and Knit3 target"""
3176
if source.supports_rich_root() != target.supports_rich_root():
3178
# Ideally, we'd support fetching if the source had no tree references
3179
# even if it supported them...
3180
if (getattr(source, '_format.supports_tree_reference', False) and
3181
not getattr(target, '_format.supports_tree_reference', False)):
3185
def _fetch_batch(self, revision_ids, basis_id, basis_tree):
3186
"""Fetch across a few revisions.
3188
:param revision_ids: The revisions to copy
3189
:param basis_id: The revision_id of basis_tree
3190
:param basis_tree: A tree that is not in revision_ids which should
3191
already exist in the target.
3192
:return: (basis_id, basis_tree) A new basis to use now that these trees
3195
# Walk though all revisions; get inventory deltas, copy referenced
3196
# texts that delta references, insert the delta, revision and
3200
pending_revisions = []
3201
for tree in self.source.revision_trees(revision_ids):
3202
current_revision_id = tree.get_revision_id()
3203
delta = tree.inventory._make_delta(basis_tree.inventory)
3204
for old_path, new_path, file_id, entry in delta:
3205
if new_path is not None:
3206
if not (new_path or self.target.supports_rich_root()):
3207
# We leave the inventory delta in, because that
3208
# will have the deserialised inventory root
3212
# "if entry.revision == current_revision_id" ?
3213
if entry.revision == current_revision_id:
3214
text_keys.add((file_id, entry.revision))
3215
revision = self.source.get_revision(current_revision_id)
3216
pending_deltas.append((basis_id, delta,
3217
current_revision_id, revision.parent_ids))
3218
pending_revisions.append(revision)
3219
basis_id = current_revision_id
3222
from_texts = self.source.texts
3223
to_texts = self.target.texts
3224
to_texts.insert_record_stream(from_texts.get_record_stream(
3225
text_keys, self.target._fetch_order,
3226
not self.target._fetch_uses_deltas))
3228
for delta in pending_deltas:
3229
self.target.add_inventory_by_delta(*delta)
3230
# insert signatures and revisions
3231
for revision in pending_revisions:
3233
signature = self.source.get_signature_text(
3234
revision.revision_id)
3235
self.target.add_signature_text(revision.revision_id,
3237
except errors.NoSuchRevision:
3239
self.target.add_revision(revision.revision_id, revision)
3240
return basis_id, basis_tree
3242
def _fetch_all_revisions(self, revision_ids, pb):
3243
"""Fetch everything for the list of revisions.
3245
:param revision_ids: The list of revisions to fetch. Must be in
3247
:param pb: A ProgressBar
3250
basis_id, basis_tree = self._get_basis(revision_ids[0])
3252
for offset in range(0, len(revision_ids), batch_size):
3253
self.target.start_write_group()
3255
pb.update('Transferring revisions', offset,
3257
batch = revision_ids[offset:offset+batch_size]
3258
basis_id, basis_tree = self._fetch_batch(batch,
3259
basis_id, basis_tree)
3261
self.target.abort_write_group()
3264
self.target.commit_write_group()
3265
pb.update('Transferring revisions', len(revision_ids),
3269
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3270
"""See InterRepository.fetch()."""
3271
revision_ids = self.target.search_missing_revision_ids(self.source,
3272
revision_id, find_ghosts=find_ghosts).get_keys()
3273
if not revision_ids:
3275
revision_ids = tsort.topo_sort(
3276
self.source.get_graph().get_parent_map(revision_ids))
3278
my_pb = ui.ui_factory.nested_progress_bar()
3283
self._fetch_all_revisions(revision_ids, pb)
3285
if my_pb is not None:
3287
return len(revision_ids), 0
3289
def _get_basis(self, first_revision_id):
3290
"""Get a revision and tree which exists in the target.
3292
This assumes that first_revision_id is selected for transmission
3293
because all other ancestors are already present. If we can't find an
3294
ancestor we fall back to NULL_REVISION since we know that is safe.
3296
:return: (basis_id, basis_tree)
3298
first_rev = self.source.get_revision(first_revision_id)
3300
basis_id = first_rev.parent_ids[0]
3301
# only valid as a basis if the target has it
3302
self.target.get_revision(basis_id)
3303
# Try to get a basis tree - if its a ghost it will hit the
3304
# NoSuchRevision case.
3305
basis_tree = self.source.revision_tree(basis_id)
3306
except (IndexError, errors.NoSuchRevision):
3307
basis_id = _mod_revision.NULL_REVISION
3308
basis_tree = self.source.revision_tree(basis_id)
3309
return basis_id, basis_tree
3312
class InterOtherToRemote(InterRepository):
3313
"""An InterRepository that simply delegates to the 'real' InterRepository
3314
calculated for (source, target._real_repository).
3317
_walk_to_common_revisions_batch_size = 50
3319
def __init__(self, source, target):
3320
InterRepository.__init__(self, source, target)
3321
self._real_inter = None
3324
def is_compatible(source, target):
3325
if isinstance(target, remote.RemoteRepository):
3329
def _ensure_real_inter(self):
3330
if self._real_inter is None:
3331
self.target._ensure_real()
3332
real_target = self.target._real_repository
3333
self._real_inter = InterRepository.get(self.source, real_target)
3334
# Make _real_inter use the RemoteRepository for get_parent_map
3335
self._real_inter.target_get_graph = self.target.get_graph
3336
self._real_inter.target_get_parent_map = self.target.get_parent_map
3338
def copy_content(self, revision_id=None):
3339
self._ensure_real_inter()
3340
self._real_inter.copy_content(revision_id=revision_id)
3342
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3343
self._ensure_real_inter()
3344
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3345
find_ghosts=find_ghosts)
3348
def _get_repo_format_to_test(self):
3352
class InterRemoteToOther(InterRepository):
3354
def __init__(self, source, target):
3355
InterRepository.__init__(self, source, target)
3356
self._real_inter = None
3359
def is_compatible(source, target):
3360
if not isinstance(source, remote.RemoteRepository):
3362
# Is source's model compatible with target's model?
3363
source._ensure_real()
3364
real_source = source._real_repository
3365
if isinstance(real_source, remote.RemoteRepository):
3366
raise NotImplementedError(
3367
"We don't support remote repos backed by remote repos yet.")
3368
return InterRepository._same_model(real_source, target)
3370
def _ensure_real_inter(self):
3371
if self._real_inter is None:
3372
self.source._ensure_real()
3373
real_source = self.source._real_repository
3374
self._real_inter = InterRepository.get(real_source, self.target)
3376
def fetch(self, revision_id=None, pb=None, find_ghosts=False):
3377
self._ensure_real_inter()
3378
return self._real_inter.fetch(revision_id=revision_id, pb=pb,
3379
find_ghosts=find_ghosts)
3381
def copy_content(self, revision_id=None):
3382
self._ensure_real_inter()
3383
self._real_inter.copy_content(revision_id=revision_id)
3386
def _get_repo_format_to_test(self):
3391
class InterPackToRemotePack(InterPackRepo):
3392
"""A specialisation of InterPackRepo for a target that is a
3395
This will use the get_parent_map RPC rather than plain readvs, and also
3396
uses an RPC for autopacking.
3399
_walk_to_common_revisions_batch_size = 50
3402
def is_compatible(source, target):
3403
from bzrlib.repofmt.pack_repo import RepositoryFormatPack
3404
if isinstance(source._format, RepositoryFormatPack):
3405
if isinstance(target, remote.RemoteRepository):
3406
target._ensure_real()
3407
if isinstance(target._real_repository._format,
3408
RepositoryFormatPack):
3409
if InterRepository._same_model(source, target):
3413
def _autopack(self):
3414
self.target.autopack()
3416
def _get_target_pack_collection(self):
3417
return self.target._real_repository._pack_collection
3420
def _get_repo_format_to_test(self):
3424
InterRepository.register_optimiser(InterDifferingSerializer)
2118
3425
InterRepository.register_optimiser(InterSameDataRepository)
2119
3426
InterRepository.register_optimiser(InterWeaveRepo)
2120
3427
InterRepository.register_optimiser(InterKnitRepo)
2121
3428
InterRepository.register_optimiser(InterModel1and2)
2122
3429
InterRepository.register_optimiser(InterKnit1and2)
2125
class RepositoryTestProviderAdapter(object):
2126
"""A tool to generate a suite testing multiple repository formats at once.
2128
This is done by copying the test once for each transport and injecting
2129
the transport_server, transport_readonly_server, and bzrdir_format and
2130
repository_format classes into each copy. Each copy is also given a new id()
2131
to make it easy to identify.
2134
def __init__(self, transport_server, transport_readonly_server, formats):
2135
self._transport_server = transport_server
2136
self._transport_readonly_server = transport_readonly_server
2137
self._formats = formats
2139
def adapt(self, test):
2140
result = unittest.TestSuite()
2141
for repository_format, bzrdir_format in self._formats:
2142
new_test = deepcopy(test)
2143
new_test.transport_server = self._transport_server
2144
new_test.transport_readonly_server = self._transport_readonly_server
2145
new_test.bzrdir_format = bzrdir_format
2146
new_test.repository_format = repository_format
2147
def make_new_test_id():
2148
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2149
return lambda: new_id
2150
new_test.id = make_new_test_id()
2151
result.addTest(new_test)
2155
class InterRepositoryTestProviderAdapter(object):
2156
"""A tool to generate a suite testing multiple inter repository formats.
2158
This is done by copying the test once for each interrepo provider and injecting
2159
the transport_server, transport_readonly_server, repository_format and
2160
repository_to_format classes into each copy.
2161
Each copy is also given a new id() to make it easy to identify.
2164
def __init__(self, transport_server, transport_readonly_server, formats):
2165
self._transport_server = transport_server
2166
self._transport_readonly_server = transport_readonly_server
2167
self._formats = formats
2169
def adapt(self, test):
2170
result = unittest.TestSuite()
2171
for interrepo_class, repository_format, repository_format_to in self._formats:
2172
new_test = deepcopy(test)
2173
new_test.transport_server = self._transport_server
2174
new_test.transport_readonly_server = self._transport_readonly_server
2175
new_test.interrepo_class = interrepo_class
2176
new_test.repository_format = repository_format
2177
new_test.repository_format_to = repository_format_to
2178
def make_new_test_id():
2179
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2180
return lambda: new_id
2181
new_test.id = make_new_test_id()
2182
result.addTest(new_test)
2186
def default_test_list():
2187
"""Generate the default list of interrepo permutations to test."""
2189
# test the default InterRepository between format 6 and the current
2191
# XXX: robertc 20060220 reinstate this when there are two supported
2192
# formats which do not have an optimal code path between them.
2193
#result.append((InterRepository,
2194
# RepositoryFormat6(),
2195
# RepositoryFormatKnit1()))
2196
for optimiser in InterRepository._optimisers:
2197
if optimiser._matching_repo_format is not None:
2198
result.append((optimiser,
2199
optimiser._matching_repo_format,
2200
optimiser._matching_repo_format
2202
# if there are specific combinations we want to use, we can add them
2204
result.append((InterModel1and2, RepositoryFormat5(),
2205
RepositoryFormatKnit2()))
2206
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2207
RepositoryFormatKnit2()))
3430
InterRepository.register_optimiser(InterPackRepo)
3431
InterRepository.register_optimiser(InterOtherToRemote)
3432
InterRepository.register_optimiser(InterRemoteToOther)
3433
InterRepository.register_optimiser(InterPackToRemotePack)
2211
3436
class CopyConverter(object):
2260
3485
self.pb.update(message, self.count, self.total)
2263
class CommitBuilder(object):
2264
"""Provides an interface to build up a commit.
2266
This allows describing a tree to be committed without needing to
2267
know the internals of the format of the repository.
2270
record_root_entry = False
2271
def __init__(self, repository, parents, config, timestamp=None,
2272
timezone=None, committer=None, revprops=None,
2274
"""Initiate a CommitBuilder.
2276
:param repository: Repository to commit to.
2277
:param parents: Revision ids of the parents of the new revision.
2278
:param config: Configuration to use.
2279
:param timestamp: Optional timestamp recorded for commit.
2280
:param timezone: Optional timezone for timestamp.
2281
:param committer: Optional committer to set for commit.
2282
:param revprops: Optional dictionary of revision properties.
2283
:param revision_id: Optional revision id.
2285
self._config = config
2287
if committer is None:
2288
self._committer = self._config.username()
2290
assert isinstance(committer, basestring), type(committer)
2291
self._committer = committer
2293
self.new_inventory = Inventory(None)
2294
self._new_revision_id = revision_id
2295
self.parents = parents
2296
self.repository = repository
2299
if revprops is not None:
2300
self._revprops.update(revprops)
2302
if timestamp is None:
2303
timestamp = time.time()
2304
# Restrict resolution to 1ms
2305
self._timestamp = round(timestamp, 3)
2307
if timezone is None:
2308
self._timezone = local_time_offset()
2310
self._timezone = int(timezone)
2312
self._generate_revision_if_needed()
2314
def commit(self, message):
2315
"""Make the actual commit.
2317
:return: The revision id of the recorded revision.
2319
rev = _mod_revision.Revision(
2320
timestamp=self._timestamp,
2321
timezone=self._timezone,
2322
committer=self._committer,
2324
inventory_sha1=self.inv_sha1,
2325
revision_id=self._new_revision_id,
2326
properties=self._revprops)
2327
rev.parent_ids = self.parents
2328
self.repository.add_revision(self._new_revision_id, rev,
2329
self.new_inventory, self._config)
2330
return self._new_revision_id
2332
def revision_tree(self):
2333
"""Return the tree that was just committed.
2335
After calling commit() this can be called to get a RevisionTree
2336
representing the newly committed tree. This is preferred to
2337
calling Repository.revision_tree() because that may require
2338
deserializing the inventory, while we already have a copy in
2341
return RevisionTree(self.repository, self.new_inventory,
2342
self._new_revision_id)
2344
def finish_inventory(self):
2345
"""Tell the builder that the inventory is finished."""
2346
if self.new_inventory.root is None:
2347
symbol_versioning.warn('Root entry should be supplied to'
2348
' record_entry_contents, as of bzr 0.10.',
2349
DeprecationWarning, stacklevel=2)
2350
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2351
self.new_inventory.revision_id = self._new_revision_id
2352
self.inv_sha1 = self.repository.add_inventory(
2353
self._new_revision_id,
2358
def _gen_revision_id(self):
2359
"""Return new revision-id."""
2360
s = '%s-%s-' % (self._config.user_email(),
2361
compact_date(self._timestamp))
2362
s += hexlify(rand_bytes(8))
2365
def _generate_revision_if_needed(self):
2366
"""Create a revision id if None was supplied.
2368
If the repository can not support user-specified revision ids
2369
they should override this function and raise UnsupportedOperation
2370
if _new_revision_id is not None.
2372
:raises: UnsupportedOperation
2374
if self._new_revision_id is None:
2375
self._new_revision_id = self._gen_revision_id()
2377
def record_entry_contents(self, ie, parent_invs, path, tree):
2378
"""Record the content of ie from tree into the commit if needed.
2380
Side effect: sets ie.revision when unchanged
2382
:param ie: An inventory entry present in the commit.
2383
:param parent_invs: The inventories of the parent revisions of the
2385
:param path: The path the entry is at in the tree.
2386
:param tree: The tree which contains this entry and should be used to
2389
if self.new_inventory.root is None and ie.parent_id is not None:
2390
symbol_versioning.warn('Root entry should be supplied to'
2391
' record_entry_contents, as of bzr 0.10.',
2392
DeprecationWarning, stacklevel=2)
2393
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2395
self.new_inventory.add(ie)
2397
# ie.revision is always None if the InventoryEntry is considered
2398
# for committing. ie.snapshot will record the correct revision
2399
# which may be the sole parent if it is untouched.
2400
if ie.revision is not None:
2403
# In this revision format, root entries have no knit or weave
2404
if ie is self.new_inventory.root:
2405
# When serializing out to disk and back in
2406
# root.revision is always _new_revision_id
2407
ie.revision = self._new_revision_id
2409
previous_entries = ie.find_previous_heads(
2411
self.repository.weave_store,
2412
self.repository.get_transaction())
2413
# we are creating a new revision for ie in the history store
2415
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2417
def modified_directory(self, file_id, file_parents):
2418
"""Record the presence of a symbolic link.
2420
:param file_id: The file_id of the link to record.
2421
:param file_parents: The per-file parent revision ids.
2423
self._add_text_to_weave(file_id, [], file_parents.keys())
2425
def modified_file_text(self, file_id, file_parents,
2426
get_content_byte_lines, text_sha1=None,
2428
"""Record the text of file file_id
2430
:param file_id: The file_id of the file to record the text of.
2431
:param file_parents: The per-file parent revision ids.
2432
:param get_content_byte_lines: A callable which will return the byte
2434
:param text_sha1: Optional SHA1 of the file contents.
2435
:param text_size: Optional size of the file contents.
2437
# mutter('storing text of file {%s} in revision {%s} into %r',
2438
# file_id, self._new_revision_id, self.repository.weave_store)
2439
# special case to avoid diffing on renames or
2441
if (len(file_parents) == 1
2442
and text_sha1 == file_parents.values()[0].text_sha1
2443
and text_size == file_parents.values()[0].text_size):
2444
previous_ie = file_parents.values()[0]
2445
versionedfile = self.repository.weave_store.get_weave(file_id,
2446
self.repository.get_transaction())
2447
versionedfile.clone_text(self._new_revision_id,
2448
previous_ie.revision, file_parents.keys())
2449
return text_sha1, text_size
2451
new_lines = get_content_byte_lines()
2452
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2453
# should return the SHA1 and size
2454
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2455
return osutils.sha_strings(new_lines), \
2456
sum(map(len, new_lines))
2458
def modified_link(self, file_id, file_parents, link_target):
2459
"""Record the presence of a symbolic link.
2461
:param file_id: The file_id of the link to record.
2462
:param file_parents: The per-file parent revision ids.
2463
:param link_target: Target location of this link.
2465
self._add_text_to_weave(file_id, [], file_parents.keys())
2467
def _add_text_to_weave(self, file_id, new_lines, parents):
2468
versionedfile = self.repository.weave_store.get_weave_or_empty(
2469
file_id, self.repository.get_transaction())
2470
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2471
versionedfile.clear_cache()
2474
class _CommitBuilder(CommitBuilder):
2475
"""Temporary class so old CommitBuilders are detected properly
2477
Note: CommitBuilder works whether or not root entry is recorded.
2480
record_root_entry = True
2483
class RootCommitBuilder(CommitBuilder):
2484
"""This commitbuilder actually records the root id"""
2486
record_root_entry = True
2488
def record_entry_contents(self, ie, parent_invs, path, tree):
2489
"""Record the content of ie from tree into the commit if needed.
2491
Side effect: sets ie.revision when unchanged
2493
:param ie: An inventory entry present in the commit.
2494
:param parent_invs: The inventories of the parent revisions of the
2496
:param path: The path the entry is at in the tree.
2497
:param tree: The tree which contains this entry and should be used to
2500
assert self.new_inventory.root is not None or ie.parent_id is None
2501
self.new_inventory.add(ie)
2503
# ie.revision is always None if the InventoryEntry is considered
2504
# for committing. ie.snapshot will record the correct revision
2505
# which may be the sole parent if it is untouched.
2506
if ie.revision is not None:
2509
previous_entries = ie.find_previous_heads(
2511
self.repository.weave_store,
2512
self.repository.get_transaction())
2513
# we are creating a new revision for ie in the history store
2515
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2518
3488
_unescape_map = {