1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from cStringIO import StringIO
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
39
revision as _mod_revision,
44
from bzrlib.bundle import serializer
45
from bzrlib.revisiontree import RevisionTree
46
from bzrlib.store.versioned import VersionedFileStore
47
from bzrlib.store.text import TextStore
48
from bzrlib.testament import Testament
51
from bzrlib.decorators import needs_read_lock, needs_write_lock
52
from bzrlib.inter import InterObject
53
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
54
from bzrlib.symbol_versioning import (
57
from bzrlib.trace import mutter, mutter_callsite, note, warning
60
# Old formats display a warning, but only once
61
_deprecation_warning_done = False
64
######################################################################
67
class Repository(object):
68
"""Repository holding history for one or more branches.
70
The repository holds and retrieves historical information including
71
revisions and file history. It's normally accessed only by the Branch,
72
which views a particular line of development through that history.
74
The Repository builds on top of Stores and a Transport, which respectively
75
describe the disk data format and the way of accessing the (possibly
79
_file_ids_altered_regex = lazy_regex.lazy_compile(
80
r'file_id="(?P<file_id>[^"]+)"'
81
r'.*revision="(?P<revision_id>[^"]+)"'
84
def abort_write_group(self):
85
"""Commit the contents accrued within the current write group.
87
:seealso: start_write_group.
89
if self._write_group is not self.get_transaction():
90
# has an unlock or relock occured ?
91
raise errors.BzrError('mismatched lock context and write group.')
92
self._abort_write_group()
93
self._write_group = None
95
def _abort_write_group(self):
96
"""Template method for per-repository write group cleanup.
98
This is called during abort before the write group is considered to be
99
finished and should cleanup any internal state accrued during the write
100
group. There is no requirement that data handed to the repository be
101
*not* made available - this is not a rollback - but neither should any
102
attempt be made to ensure that data added is fully commited. Abort is
103
invoked when an error has occured so futher disk or network operations
104
may not be possible or may error and if possible should not be
109
def add_inventory(self, revision_id, inv, parents):
110
"""Add the inventory inv to the repository as revision_id.
112
:param parents: The revision ids of the parents that revision_id
113
is known to have and are in the repository already.
115
returns the sha1 of the serialized inventory.
117
revision_id = osutils.safe_revision_id(revision_id)
118
_mod_revision.check_not_reserved_id(revision_id)
119
assert inv.revision_id is None or inv.revision_id == revision_id, \
120
"Mismatch between inventory revision" \
121
" id and insertion revid (%r, %r)" % (inv.revision_id, revision_id)
122
assert inv.root is not None
123
inv_text = self.serialise_inventory(inv)
124
inv_sha1 = osutils.sha_string(inv_text)
125
inv_vf = self.control_weaves.get_weave('inventory',
126
self.get_transaction())
127
self._inventory_add_lines(inv_vf, revision_id, parents,
128
osutils.split_lines(inv_text))
131
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines):
133
for parent in parents:
135
final_parents.append(parent)
137
inv_vf.add_lines(revision_id, final_parents, lines)
140
def add_revision(self, revision_id, rev, inv=None, config=None):
141
"""Add rev to the revision store as revision_id.
143
:param revision_id: the revision id to use.
144
:param rev: The revision object.
145
:param inv: The inventory for the revision. if None, it will be looked
146
up in the inventory storer
147
:param config: If None no digital signature will be created.
148
If supplied its signature_needed method will be used
149
to determine if a signature should be made.
151
revision_id = osutils.safe_revision_id(revision_id)
152
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
154
_mod_revision.check_not_reserved_id(revision_id)
155
if config is not None and config.signature_needed():
157
inv = self.get_inventory(revision_id)
158
plaintext = Testament(rev, inv).as_short_text()
159
self.store_revision_signature(
160
gpg.GPGStrategy(config), plaintext, revision_id)
161
if not revision_id in self.get_inventory_weave():
163
raise errors.WeaveRevisionNotPresent(revision_id,
164
self.get_inventory_weave())
166
# yes, this is not suitable for adding with ghosts.
167
self.add_inventory(revision_id, inv, rev.parent_ids)
168
self._revision_store.add_revision(rev, self.get_transaction())
170
def _add_revision_text(self, revision_id, text):
171
revision = self._revision_store._serializer.read_revision_from_string(
173
self._revision_store._add_revision(revision, StringIO(text),
174
self.get_transaction())
177
def _all_possible_ids(self):
178
"""Return all the possible revisions that we could find."""
179
return self.get_inventory_weave().versions()
181
def all_revision_ids(self):
182
"""Returns a list of all the revision ids in the repository.
184
This is deprecated because code should generally work on the graph
185
reachable from a particular revision, and ignore any other revisions
186
that might be present. There is no direct replacement method.
188
return self._all_revision_ids()
191
def _all_revision_ids(self):
192
"""Returns a list of all the revision ids in the repository.
194
These are in as much topological order as the underlying store can
195
present: for weaves ghosts may lead to a lack of correctness until
196
the reweave updates the parents list.
198
if self._revision_store.text_store.listable():
199
return self._revision_store.all_revision_ids(self.get_transaction())
200
result = self._all_possible_ids()
201
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
202
# ids. (It should, since _revision_store's API should change to
203
# return utf8 revision_ids)
204
return self._eliminate_revisions_not_present(result)
206
def break_lock(self):
207
"""Break a lock if one is present from another instance.
209
Uses the ui factory to ask for confirmation if the lock may be from
212
self.control_files.break_lock()
215
def _eliminate_revisions_not_present(self, revision_ids):
216
"""Check every revision id in revision_ids to see if we have it.
218
Returns a set of the present revisions.
221
for id in revision_ids:
222
if self.has_revision(id):
227
def create(a_bzrdir):
228
"""Construct the current default format repository in a_bzrdir."""
229
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
231
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
232
"""instantiate a Repository.
234
:param _format: The format of the repository on disk.
235
:param a_bzrdir: The BzrDir of the repository.
237
In the future we will have a single api for all stores for
238
getting file texts, inventories and revisions, then
239
this construct will accept instances of those things.
241
super(Repository, self).__init__()
242
self._format = _format
243
# the following are part of the public API for Repository:
244
self.bzrdir = a_bzrdir
245
self.control_files = control_files
246
self._revision_store = _revision_store
247
# backwards compatibility
248
self.weave_store = text_store
250
self._reconcile_does_inventory_gc = True
251
# not right yet - should be more semantically clear ?
253
self.control_store = control_store
254
self.control_weaves = control_store
255
# TODO: make sure to construct the right store classes, etc, depending
256
# on whether escaping is required.
257
self._warn_if_deprecated()
258
self._write_group = None
261
return '%s(%r)' % (self.__class__.__name__,
262
self.bzrdir.transport.base)
264
def has_same_location(self, other):
265
"""Returns a boolean indicating if this repository is at the same
266
location as another repository.
268
This might return False even when two repository objects are accessing
269
the same physical repository via different URLs.
271
if self.__class__ is not other.__class__:
273
return (self.control_files._transport.base ==
274
other.control_files._transport.base)
276
def is_in_write_group(self):
277
"""Return True if there is an open write group.
279
:seealso: start_write_group.
281
return self._write_group is not None
284
return self.control_files.is_locked()
286
def lock_write(self, token=None):
287
"""Lock this repository for writing.
289
This causes caching within the repository obejct to start accumlating
290
data during reads, and allows a 'write_group' to be obtained. Write
291
groups must be used for actual data insertion.
293
:param token: if this is already locked, then lock_write will fail
294
unless the token matches the existing lock.
295
:returns: a token if this instance supports tokens, otherwise None.
296
:raises TokenLockingNotSupported: when a token is given but this
297
instance doesn't support using token locks.
298
:raises MismatchedToken: if the specified token doesn't match the token
299
of the existing lock.
300
:seealso: start_write_group.
302
A token should be passed in if you know that you have locked the object
303
some other way, and need to synchronise this object's state with that
306
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
308
result = self.control_files.lock_write(token=token)
313
self.control_files.lock_read()
316
def get_physical_lock_status(self):
317
return self.control_files.get_physical_lock_status()
319
def leave_lock_in_place(self):
320
"""Tell this repository not to release the physical lock when this
323
If lock_write doesn't return a token, then this method is not supported.
325
self.control_files.leave_in_place()
327
def dont_leave_lock_in_place(self):
328
"""Tell this repository to release the physical lock when this
329
object is unlocked, even if it didn't originally acquire it.
331
If lock_write doesn't return a token, then this method is not supported.
333
self.control_files.dont_leave_in_place()
336
def gather_stats(self, revid=None, committers=None):
337
"""Gather statistics from a revision id.
339
:param revid: The revision id to gather statistics from, if None, then
340
no revision specific statistics are gathered.
341
:param committers: Optional parameter controlling whether to grab
342
a count of committers from the revision specific statistics.
343
:return: A dictionary of statistics. Currently this contains:
344
committers: The number of committers if requested.
345
firstrev: A tuple with timestamp, timezone for the penultimate left
346
most ancestor of revid, if revid is not the NULL_REVISION.
347
latestrev: A tuple with timestamp, timezone for revid, if revid is
348
not the NULL_REVISION.
349
revisions: The total revision count in the repository.
350
size: An estimate disk size of the repository in bytes.
353
if revid and committers:
354
result['committers'] = 0
355
if revid and revid != _mod_revision.NULL_REVISION:
357
all_committers = set()
358
revisions = self.get_ancestry(revid)
359
# pop the leading None
361
first_revision = None
363
# ignore the revisions in the middle - just grab first and last
364
revisions = revisions[0], revisions[-1]
365
for revision in self.get_revisions(revisions):
366
if not first_revision:
367
first_revision = revision
369
all_committers.add(revision.committer)
370
last_revision = revision
372
result['committers'] = len(all_committers)
373
result['firstrev'] = (first_revision.timestamp,
374
first_revision.timezone)
375
result['latestrev'] = (last_revision.timestamp,
376
last_revision.timezone)
378
# now gather global repository information
379
if self.bzrdir.root_transport.listable():
380
c, t = self._revision_store.total_size(self.get_transaction())
381
result['revisions'] = c
386
def missing_revision_ids(self, other, revision_id=None):
387
"""Return the revision ids that other has that this does not.
389
These are returned in topological order.
391
revision_id: only return revision ids included by revision_id.
393
revision_id = osutils.safe_revision_id(revision_id)
394
return InterRepository.get(other, self).missing_revision_ids(revision_id)
398
"""Open the repository rooted at base.
400
For instance, if the repository is at URL/.bzr/repository,
401
Repository.open(URL) -> a Repository instance.
403
control = bzrdir.BzrDir.open(base)
404
return control.open_repository()
406
def copy_content_into(self, destination, revision_id=None):
407
"""Make a complete copy of the content in self into destination.
409
This is a destructive operation! Do not use it on existing
412
revision_id = osutils.safe_revision_id(revision_id)
413
return InterRepository.get(self, destination).copy_content(revision_id)
415
def commit_write_group(self):
416
"""Commit the contents accrued within the current write group.
418
:seealso: start_write_group.
420
if self._write_group is not self.get_transaction():
421
# has an unlock or relock occured ?
422
raise errors.BzrError('mismatched lock context and write group.')
423
self._commit_write_group()
424
self._write_group = None
426
def _commit_write_group(self):
427
"""Template method for per-repository write group cleanup.
429
This is called before the write group is considered to be
430
finished and should ensure that all data handed to the repository
431
for writing during the write group is safely committed (to the
432
extent possible considering file system caching etc).
435
def fetch(self, source, revision_id=None, pb=None):
436
"""Fetch the content required to construct revision_id from source.
438
If revision_id is None all content is copied.
440
revision_id = osutils.safe_revision_id(revision_id)
441
inter = InterRepository.get(source, self)
443
return inter.fetch(revision_id=revision_id, pb=pb)
444
except NotImplementedError:
445
raise errors.IncompatibleRepositories(source, self)
447
def create_bundle(self, target, base, fileobj, format=None):
448
return serializer.write_bundle(self, target, base, fileobj, format)
450
def get_commit_builder(self, branch, parents, config, timestamp=None,
451
timezone=None, committer=None, revprops=None,
453
"""Obtain a CommitBuilder for this repository.
455
:param branch: Branch to commit to.
456
:param parents: Revision ids of the parents of the new revision.
457
:param config: Configuration to use.
458
:param timestamp: Optional timestamp recorded for commit.
459
:param timezone: Optional timezone for timestamp.
460
:param committer: Optional committer to set for commit.
461
:param revprops: Optional dictionary of revision properties.
462
:param revision_id: Optional revision id.
464
revision_id = osutils.safe_revision_id(revision_id)
465
result =_CommitBuilder(self, parents, config, timestamp, timezone,
466
committer, revprops, revision_id)
467
self.start_write_group()
471
if (self.control_files._lock_count == 1 and
472
self.control_files._lock_mode == 'w'):
473
if self._write_group is not None:
474
raise errors.BzrError(
475
'Must end write groups before releasing write locks.')
476
self.control_files.unlock()
479
def clone(self, a_bzrdir, revision_id=None):
480
"""Clone this repository into a_bzrdir using the current format.
482
Currently no check is made that the format of this repository and
483
the bzrdir format are compatible. FIXME RBC 20060201.
485
:return: The newly created destination repository.
487
# TODO: deprecate after 0.16; cloning this with all its settings is
488
# probably not very useful -- mbp 20070423
489
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
490
self.copy_content_into(dest_repo, revision_id)
493
def start_write_group(self):
494
"""Start a write group in the repository.
496
Write groups are used by repositories which do not have a 1:1 mapping
497
between file ids and backend store to manage the insertion of data from
498
both fetch and commit operations.
500
A write lock is required around the start_write_group/commit_write_group
501
for the support of lock-requiring repository formats.
503
One can only insert data into a repository inside a write group.
507
if not self.is_locked() or self.control_files._lock_mode != 'w':
508
raise errors.NotWriteLocked(self)
509
if self._write_group:
510
raise errors.BzrError('already in a write group')
511
self._start_write_group()
512
# so we can detect unlock/relock - the write group is now entered.
513
self._write_group = self.get_transaction()
515
def _start_write_group(self):
516
"""Template method for per-repository write group startup.
518
This is called before the write group is considered to be
523
def sprout(self, to_bzrdir, revision_id=None):
524
"""Create a descendent repository for new development.
526
Unlike clone, this does not copy the settings of the repository.
528
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
529
dest_repo.fetch(self, revision_id=revision_id)
532
def _create_sprouting_repo(self, a_bzrdir, shared):
533
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
534
# use target default format.
535
dest_repo = a_bzrdir.create_repository()
537
# Most control formats need the repository to be specifically
538
# created, but on some old all-in-one formats it's not needed
540
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
541
except errors.UninitializableFormat:
542
dest_repo = a_bzrdir.open_repository()
546
def has_revision(self, revision_id):
547
"""True if this repository has a copy of the revision."""
548
if 'evil' in debug.debug_flags:
549
mutter_callsite(2, "has_revision is a LBYL symptom.")
550
revision_id = osutils.safe_revision_id(revision_id)
551
return self._revision_store.has_revision_id(revision_id,
552
self.get_transaction())
555
def get_revision_reconcile(self, revision_id):
556
"""'reconcile' helper routine that allows access to a revision always.
558
This variant of get_revision does not cross check the weave graph
559
against the revision one as get_revision does: but it should only
560
be used by reconcile, or reconcile-alike commands that are correcting
561
or testing the revision graph.
563
if not revision_id or not isinstance(revision_id, basestring):
564
raise errors.InvalidRevisionId(revision_id=revision_id,
566
return self.get_revisions([revision_id])[0]
569
def get_revisions(self, revision_ids):
570
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
571
revs = self._revision_store.get_revisions(revision_ids,
572
self.get_transaction())
574
assert not isinstance(rev.revision_id, unicode)
575
for parent_id in rev.parent_ids:
576
assert not isinstance(parent_id, unicode)
580
def get_revision_xml(self, revision_id):
581
# TODO: jam 20070210 This shouldn't be necessary since get_revision
582
# would have already do it.
583
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
584
revision_id = osutils.safe_revision_id(revision_id)
585
rev = self.get_revision(revision_id)
587
# the current serializer..
588
self._revision_store._serializer.write_revision(rev, rev_tmp)
590
return rev_tmp.getvalue()
593
def get_revision(self, revision_id):
594
"""Return the Revision object for a named revision"""
595
# TODO: jam 20070210 get_revision_reconcile should do this for us
596
revision_id = osutils.safe_revision_id(revision_id)
597
r = self.get_revision_reconcile(revision_id)
598
# weave corruption can lead to absent revision markers that should be
600
# the following test is reasonably cheap (it needs a single weave read)
601
# and the weave is cached in read transactions. In write transactions
602
# it is not cached but typically we only read a small number of
603
# revisions. For knits when they are introduced we will probably want
604
# to ensure that caching write transactions are in use.
605
inv = self.get_inventory_weave()
606
self._check_revision_parents(r, inv)
610
def get_deltas_for_revisions(self, revisions):
611
"""Produce a generator of revision deltas.
613
Note that the input is a sequence of REVISIONS, not revision_ids.
614
Trees will be held in memory until the generator exits.
615
Each delta is relative to the revision's lefthand predecessor.
617
required_trees = set()
618
for revision in revisions:
619
required_trees.add(revision.revision_id)
620
required_trees.update(revision.parent_ids[:1])
621
trees = dict((t.get_revision_id(), t) for
622
t in self.revision_trees(required_trees))
623
for revision in revisions:
624
if not revision.parent_ids:
625
old_tree = self.revision_tree(None)
627
old_tree = trees[revision.parent_ids[0]]
628
yield trees[revision.revision_id].changes_from(old_tree)
631
def get_revision_delta(self, revision_id):
632
"""Return the delta for one revision.
634
The delta is relative to the left-hand predecessor of the
637
r = self.get_revision(revision_id)
638
return list(self.get_deltas_for_revisions([r]))[0]
640
def _check_revision_parents(self, revision, inventory):
641
"""Private to Repository and Fetch.
643
This checks the parentage of revision in an inventory weave for
644
consistency and is only applicable to inventory-weave-for-ancestry
645
using repository formats & fetchers.
647
weave_parents = inventory.get_parents(revision.revision_id)
648
weave_names = inventory.versions()
649
for parent_id in revision.parent_ids:
650
if parent_id in weave_names:
651
# this parent must not be a ghost.
652
if not parent_id in weave_parents:
654
raise errors.CorruptRepository(self)
657
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
658
revision_id = osutils.safe_revision_id(revision_id)
659
signature = gpg_strategy.sign(plaintext)
660
self._revision_store.add_revision_signature_text(revision_id,
662
self.get_transaction())
664
def fileids_altered_by_revision_ids(self, revision_ids):
665
"""Find the file ids and versions affected by revisions.
667
:param revisions: an iterable containing revision ids.
668
:return: a dictionary mapping altered file-ids to an iterable of
669
revision_ids. Each altered file-ids has the exact revision_ids that
670
altered it listed explicitly.
672
assert self._serializer.support_altered_by_hack, \
673
("fileids_altered_by_revision_ids only supported for branches "
674
"which store inventory as unnested xml, not on %r" % self)
675
selected_revision_ids = set(osutils.safe_revision_id(r)
676
for r in revision_ids)
677
w = self.get_inventory_weave()
680
# this code needs to read every new line in every inventory for the
681
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
682
# not present in one of those inventories is unnecessary but not
683
# harmful because we are filtering by the revision id marker in the
684
# inventory lines : we only select file ids altered in one of those
685
# revisions. We don't need to see all lines in the inventory because
686
# only those added in an inventory in rev X can contain a revision=X
688
unescape_revid_cache = {}
689
unescape_fileid_cache = {}
691
# jam 20061218 In a big fetch, this handles hundreds of thousands
692
# of lines, so it has had a lot of inlining and optimizing done.
693
# Sorry that it is a little bit messy.
694
# Move several functions to be local variables, since this is a long
696
search = self._file_ids_altered_regex.search
697
unescape = _unescape_xml
698
setdefault = result.setdefault
699
pb = ui.ui_factory.nested_progress_bar()
701
for line in w.iter_lines_added_or_present_in_versions(
702
selected_revision_ids, pb=pb):
706
# One call to match.group() returning multiple items is quite a
707
# bit faster than 2 calls to match.group() each returning 1
708
file_id, revision_id = match.group('file_id', 'revision_id')
710
# Inlining the cache lookups helps a lot when you make 170,000
711
# lines and 350k ids, versus 8.4 unique ids.
712
# Using a cache helps in 2 ways:
713
# 1) Avoids unnecessary decoding calls
714
# 2) Re-uses cached strings, which helps in future set and
716
# (2) is enough that removing encoding entirely along with
717
# the cache (so we are using plain strings) results in no
718
# performance improvement.
720
revision_id = unescape_revid_cache[revision_id]
722
unescaped = unescape(revision_id)
723
unescape_revid_cache[revision_id] = unescaped
724
revision_id = unescaped
726
if revision_id in selected_revision_ids:
728
file_id = unescape_fileid_cache[file_id]
730
unescaped = unescape(file_id)
731
unescape_fileid_cache[file_id] = unescaped
733
setdefault(file_id, set()).add(revision_id)
738
def iter_files_bytes(self, desired_files):
739
"""Iterate through file versions.
741
Files will not necessarily be returned in the order they occur in
742
desired_files. No specific order is guaranteed.
744
Yields pairs of identifier, bytes_iterator. identifier is an opaque
745
value supplied by the caller as part of desired_files. It should
746
uniquely identify the file version in the caller's context. (Examples:
747
an index number or a TreeTransform trans_id.)
749
bytes_iterator is an iterable of bytestrings for the file. The
750
kind of iterable and length of the bytestrings are unspecified, but for
751
this implementation, it is a list of lines produced by
752
VersionedFile.get_lines().
754
:param desired_files: a list of (file_id, revision_id, identifier)
757
transaction = self.get_transaction()
758
for file_id, revision_id, callable_data in desired_files:
760
weave = self.weave_store.get_weave(file_id, transaction)
761
except errors.NoSuchFile:
762
raise errors.NoSuchIdInRepository(self, file_id)
763
yield callable_data, weave.get_lines(revision_id)
765
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
766
"""Get an iterable listing the keys of all the data introduced by a set
769
The keys will be ordered so that the corresponding items can be safely
770
fetched and inserted in that order.
772
:returns: An iterable producing tuples of (knit-kind, file-id,
773
versions). knit-kind is one of 'file', 'inventory', 'signatures',
774
'revisions'. file-id is None unless knit-kind is 'file'.
776
# XXX: it's a bit weird to control the inventory weave caching in this
777
# generator. Ideally the caching would be done in fetch.py I think. Or
778
# maybe this generator should explicitly have the contract that it
779
# should not be iterated until the previously yielded item has been
781
inv_w = self.get_inventory_weave()
784
# file ids that changed
785
file_ids = self.fileids_altered_by_revision_ids(revision_ids)
787
num_file_ids = len(file_ids)
788
for file_id, altered_versions in file_ids.iteritems():
789
if _files_pb is not None:
790
_files_pb.update("fetch texts", count, num_file_ids)
792
yield ("file", file_id, altered_versions)
793
# We're done with the files_pb. Note that it finished by the caller,
794
# just as it was created by the caller.
798
yield ("inventory", None, revision_ids)
802
revisions_with_signatures = set()
803
for rev_id in revision_ids:
805
self.get_signature_text(rev_id)
806
except errors.NoSuchRevision:
810
revisions_with_signatures.add(rev_id)
811
yield ("signatures", None, revisions_with_signatures)
814
yield ("revisions", None, revision_ids)
817
def get_inventory_weave(self):
818
return self.control_weaves.get_weave('inventory',
819
self.get_transaction())
822
def get_inventory(self, revision_id):
823
"""Get Inventory object by hash."""
824
# TODO: jam 20070210 Technically we don't need to sanitize, since all
825
# called functions must sanitize.
826
revision_id = osutils.safe_revision_id(revision_id)
827
return self.deserialise_inventory(
828
revision_id, self.get_inventory_xml(revision_id))
830
def deserialise_inventory(self, revision_id, xml):
831
"""Transform the xml into an inventory object.
833
:param revision_id: The expected revision id of the inventory.
834
:param xml: A serialised inventory.
836
revision_id = osutils.safe_revision_id(revision_id)
837
result = self._serializer.read_inventory_from_string(xml)
838
result.root.revision = revision_id
841
def serialise_inventory(self, inv):
842
return self._serializer.write_inventory_to_string(inv)
844
def get_serializer_format(self):
845
return self._serializer.format_num
848
def get_inventory_xml(self, revision_id):
849
"""Get inventory XML as a file object."""
850
revision_id = osutils.safe_revision_id(revision_id)
852
assert isinstance(revision_id, str), type(revision_id)
853
iw = self.get_inventory_weave()
854
return iw.get_text(revision_id)
856
raise errors.HistoryMissing(self, 'inventory', revision_id)
859
def get_inventory_sha1(self, revision_id):
860
"""Return the sha1 hash of the inventory entry
862
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
863
revision_id = osutils.safe_revision_id(revision_id)
864
return self.get_revision(revision_id).inventory_sha1
867
def get_revision_graph(self, revision_id=None):
868
"""Return a dictionary containing the revision graph.
870
:param revision_id: The revision_id to get a graph from. If None, then
871
the entire revision graph is returned. This is a deprecated mode of
872
operation and will be removed in the future.
873
:return: a dictionary of revision_id->revision_parents_list.
875
if 'evil' in debug.debug_flags:
877
"get_revision_graph scales with size of history.")
878
# special case NULL_REVISION
879
if revision_id == _mod_revision.NULL_REVISION:
881
revision_id = osutils.safe_revision_id(revision_id)
882
a_weave = self.get_inventory_weave()
883
all_revisions = self._eliminate_revisions_not_present(
885
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
886
node in all_revisions])
887
if revision_id is None:
889
elif revision_id not in entire_graph:
890
raise errors.NoSuchRevision(self, revision_id)
892
# add what can be reached from revision_id
894
pending = set([revision_id])
895
while len(pending) > 0:
897
result[node] = entire_graph[node]
898
for revision_id in result[node]:
899
if revision_id not in result:
900
pending.add(revision_id)
904
def get_revision_graph_with_ghosts(self, revision_ids=None):
905
"""Return a graph of the revisions with ghosts marked as applicable.
907
:param revision_ids: an iterable of revisions to graph or None for all.
908
:return: a Graph object with the graph reachable from revision_ids.
910
if 'evil' in debug.debug_flags:
912
"get_revision_graph_with_ghosts scales with size of history.")
913
result = deprecated_graph.Graph()
915
pending = set(self.all_revision_ids())
918
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
919
# special case NULL_REVISION
920
if _mod_revision.NULL_REVISION in pending:
921
pending.remove(_mod_revision.NULL_REVISION)
922
required = set(pending)
925
revision_id = pending.pop()
927
rev = self.get_revision(revision_id)
928
except errors.NoSuchRevision:
929
if revision_id in required:
932
result.add_ghost(revision_id)
934
for parent_id in rev.parent_ids:
935
# is this queued or done ?
936
if (parent_id not in pending and
937
parent_id not in done):
939
pending.add(parent_id)
940
result.add_node(revision_id, rev.parent_ids)
941
done.add(revision_id)
944
def _get_history_vf(self):
945
"""Get a versionedfile whose history graph reflects all revisions.
947
For weave repositories, this is the inventory weave.
949
return self.get_inventory_weave()
951
def iter_reverse_revision_history(self, revision_id):
952
"""Iterate backwards through revision ids in the lefthand history
954
:param revision_id: The revision id to start with. All its lefthand
955
ancestors will be traversed.
957
revision_id = osutils.safe_revision_id(revision_id)
958
if revision_id in (None, _mod_revision.NULL_REVISION):
960
next_id = revision_id
961
versionedfile = self._get_history_vf()
964
parents = versionedfile.get_parents(next_id)
965
if len(parents) == 0:
971
def get_revision_inventory(self, revision_id):
972
"""Return inventory of a past revision."""
973
# TODO: Unify this with get_inventory()
974
# bzr 0.0.6 and later imposes the constraint that the inventory_id
975
# must be the same as its revision, so this is trivial.
976
if revision_id is None:
977
# This does not make sense: if there is no revision,
978
# then it is the current tree inventory surely ?!
979
# and thus get_root_id() is something that looks at the last
980
# commit on the branch, and the get_root_id is an inventory check.
981
raise NotImplementedError
982
# return Inventory(self.get_root_id())
984
return self.get_inventory(revision_id)
988
"""Return True if this repository is flagged as a shared repository."""
989
raise NotImplementedError(self.is_shared)
992
def reconcile(self, other=None, thorough=False):
993
"""Reconcile this repository."""
994
from bzrlib.reconcile import RepoReconciler
995
reconciler = RepoReconciler(self, thorough=thorough)
996
reconciler.reconcile()
999
def _refresh_data(self):
1000
"""Helper called from lock_* to ensure coherency with disk.
1002
The default implementation does nothing; it is however possible
1003
for repositories to maintain loaded indices across multiple locks
1004
by checking inside their implementation of this method to see
1005
whether their indices are still valid. This depends of course on
1006
the disk format being validatable in this manner.
1010
def revision_tree(self, revision_id):
1011
"""Return Tree for a revision on this branch.
1013
`revision_id` may be None for the empty tree revision.
1015
# TODO: refactor this to use an existing revision object
1016
# so we don't need to read it in twice.
1017
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1018
return RevisionTree(self, Inventory(root_id=None),
1019
_mod_revision.NULL_REVISION)
1021
revision_id = osutils.safe_revision_id(revision_id)
1022
inv = self.get_revision_inventory(revision_id)
1023
return RevisionTree(self, inv, revision_id)
1026
def revision_trees(self, revision_ids):
1027
"""Return Tree for a revision on this branch.
1029
`revision_id` may not be None or 'null:'"""
1030
assert None not in revision_ids
1031
assert _mod_revision.NULL_REVISION not in revision_ids
1032
texts = self.get_inventory_weave().get_texts(revision_ids)
1033
for text, revision_id in zip(texts, revision_ids):
1034
inv = self.deserialise_inventory(revision_id, text)
1035
yield RevisionTree(self, inv, revision_id)
1038
def get_ancestry(self, revision_id, topo_sorted=True):
1039
"""Return a list of revision-ids integrated by a revision.
1041
The first element of the list is always None, indicating the origin
1042
revision. This might change when we have history horizons, or
1043
perhaps we should have a new API.
1045
This is topologically sorted.
1047
if _mod_revision.is_null(revision_id):
1049
revision_id = osutils.safe_revision_id(revision_id)
1050
if not self.has_revision(revision_id):
1051
raise errors.NoSuchRevision(self, revision_id)
1052
w = self.get_inventory_weave()
1053
candidates = w.get_ancestry(revision_id, topo_sorted)
1054
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1057
"""Compress the data within the repository.
1059
This operation only makes sense for some repository types. For other
1060
types it should be a no-op that just returns.
1062
This stub method does not require a lock, but subclasses should use
1063
@needs_write_lock as this is a long running call its reasonable to
1064
implicitly lock for the user.
1068
def print_file(self, file, revision_id):
1069
"""Print `file` to stdout.
1071
FIXME RBC 20060125 as John Meinel points out this is a bad api
1072
- it writes to stdout, it assumes that that is valid etc. Fix
1073
by creating a new more flexible convenience function.
1075
revision_id = osutils.safe_revision_id(revision_id)
1076
tree = self.revision_tree(revision_id)
1077
# use inventory as it was in that revision
1078
file_id = tree.inventory.path2id(file)
1080
# TODO: jam 20060427 Write a test for this code path
1081
# it had a bug in it, and was raising the wrong
1083
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1084
tree.print_file(file_id)
1086
def get_transaction(self):
1087
return self.control_files.get_transaction()
1089
def revision_parents(self, revision_id):
1090
revision_id = osutils.safe_revision_id(revision_id)
1091
return self.get_inventory_weave().parent_names(revision_id)
1093
def get_parents(self, revision_ids):
1094
"""See StackedParentsProvider.get_parents"""
1096
for revision_id in revision_ids:
1097
if revision_id == _mod_revision.NULL_REVISION:
1101
parents = self.get_revision(revision_id).parent_ids
1102
except errors.NoSuchRevision:
1105
if len(parents) == 0:
1106
parents = [_mod_revision.NULL_REVISION]
1107
parents_list.append(parents)
1110
def _make_parents_provider(self):
1113
def get_graph(self, other_repository=None):
1114
"""Return the graph walker for this repository format"""
1115
parents_provider = self._make_parents_provider()
1116
if (other_repository is not None and
1117
other_repository.bzrdir.transport.base !=
1118
self.bzrdir.transport.base):
1119
parents_provider = graph._StackedParentsProvider(
1120
[parents_provider, other_repository._make_parents_provider()])
1121
return graph.Graph(parents_provider)
1124
def set_make_working_trees(self, new_value):
1125
"""Set the policy flag for making working trees when creating branches.
1127
This only applies to branches that use this repository.
1129
The default is 'True'.
1130
:param new_value: True to restore the default, False to disable making
1133
raise NotImplementedError(self.set_make_working_trees)
1135
def make_working_trees(self):
1136
"""Returns the policy for making working trees on new branches."""
1137
raise NotImplementedError(self.make_working_trees)
1140
def sign_revision(self, revision_id, gpg_strategy):
1141
revision_id = osutils.safe_revision_id(revision_id)
1142
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1143
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1146
def has_signature_for_revision_id(self, revision_id):
1147
"""Query for a revision signature for revision_id in the repository."""
1148
revision_id = osutils.safe_revision_id(revision_id)
1149
return self._revision_store.has_signature(revision_id,
1150
self.get_transaction())
1153
def get_signature_text(self, revision_id):
1154
"""Return the text for a signature."""
1155
revision_id = osutils.safe_revision_id(revision_id)
1156
return self._revision_store.get_signature_text(revision_id,
1157
self.get_transaction())
1160
def check(self, revision_ids):
1161
"""Check consistency of all history of given revision_ids.
1163
Different repository implementations should override _check().
1165
:param revision_ids: A non-empty list of revision_ids whose ancestry
1166
will be checked. Typically the last revision_id of a branch.
1168
if not revision_ids:
1169
raise ValueError("revision_ids must be non-empty in %s.check"
1171
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1172
return self._check(revision_ids)
1174
def _check(self, revision_ids):
1175
result = check.Check(self)
1179
def _warn_if_deprecated(self):
1180
global _deprecation_warning_done
1181
if _deprecation_warning_done:
1183
_deprecation_warning_done = True
1184
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1185
% (self._format, self.bzrdir.transport.base))
1187
def supports_rich_root(self):
1188
return self._format.rich_root_data
1190
def _check_ascii_revisionid(self, revision_id, method):
1191
"""Private helper for ascii-only repositories."""
1192
# weave repositories refuse to store revisionids that are non-ascii.
1193
if revision_id is not None:
1194
# weaves require ascii revision ids.
1195
if isinstance(revision_id, unicode):
1197
revision_id.encode('ascii')
1198
except UnicodeEncodeError:
1199
raise errors.NonAsciiRevisionId(method, self)
1202
revision_id.decode('ascii')
1203
except UnicodeDecodeError:
1204
raise errors.NonAsciiRevisionId(method, self)
1208
# remove these delegates a while after bzr 0.15
1209
def __make_delegated(name, from_module):
1210
def _deprecated_repository_forwarder():
1211
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1212
% (name, from_module),
1215
m = __import__(from_module, globals(), locals(), [name])
1217
return getattr(m, name)
1218
except AttributeError:
1219
raise AttributeError('module %s has no name %s'
1221
globals()[name] = _deprecated_repository_forwarder
1224
'AllInOneRepository',
1225
'WeaveMetaDirRepository',
1226
'PreSplitOutRepositoryFormat',
1227
'RepositoryFormat4',
1228
'RepositoryFormat5',
1229
'RepositoryFormat6',
1230
'RepositoryFormat7',
1232
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1236
'RepositoryFormatKnit',
1237
'RepositoryFormatKnit1',
1239
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1242
def install_revision(repository, rev, revision_tree):
1243
"""Install all revision data into a repository."""
1244
present_parents = []
1246
for p_id in rev.parent_ids:
1247
if repository.has_revision(p_id):
1248
present_parents.append(p_id)
1249
parent_trees[p_id] = repository.revision_tree(p_id)
1251
parent_trees[p_id] = repository.revision_tree(None)
1253
inv = revision_tree.inventory
1254
entries = inv.iter_entries()
1255
# backwards compatibility hack: skip the root id.
1256
if not repository.supports_rich_root():
1257
path, root = entries.next()
1258
if root.revision != rev.revision_id:
1259
raise errors.IncompatibleRevision(repr(repository))
1260
# Add the texts that are not already present
1261
for path, ie in entries:
1262
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1263
repository.get_transaction())
1264
if ie.revision not in w:
1266
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1267
# with InventoryEntry.find_previous_heads(). if it is, then there
1268
# is a latent bug here where the parents may have ancestors of each
1270
for revision, tree in parent_trees.iteritems():
1271
if ie.file_id not in tree:
1273
parent_id = tree.inventory[ie.file_id].revision
1274
if parent_id in text_parents:
1276
text_parents.append(parent_id)
1278
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1279
repository.get_transaction())
1280
lines = revision_tree.get_file(ie.file_id).readlines()
1281
vfile.add_lines(rev.revision_id, text_parents, lines)
1283
# install the inventory
1284
repository.add_inventory(rev.revision_id, inv, present_parents)
1285
except errors.RevisionAlreadyPresent:
1287
repository.add_revision(rev.revision_id, rev, inv)
1290
class MetaDirRepository(Repository):
1291
"""Repositories in the new meta-dir layout."""
1293
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1294
super(MetaDirRepository, self).__init__(_format,
1300
dir_mode = self.control_files._dir_mode
1301
file_mode = self.control_files._file_mode
1304
def is_shared(self):
1305
"""Return True if this repository is flagged as a shared repository."""
1306
return self.control_files._transport.has('shared-storage')
1309
def set_make_working_trees(self, new_value):
1310
"""Set the policy flag for making working trees when creating branches.
1312
This only applies to branches that use this repository.
1314
The default is 'True'.
1315
:param new_value: True to restore the default, False to disable making
1320
self.control_files._transport.delete('no-working-trees')
1321
except errors.NoSuchFile:
1324
self.control_files.put_utf8('no-working-trees', '')
1326
def make_working_trees(self):
1327
"""Returns the policy for making working trees on new branches."""
1328
return not self.control_files._transport.has('no-working-trees')
1331
class RepositoryFormatRegistry(registry.Registry):
1332
"""Registry of RepositoryFormats.
1335
def get(self, format_string):
1336
r = registry.Registry.get(self, format_string)
1342
format_registry = RepositoryFormatRegistry()
1343
"""Registry of formats, indexed by their identifying format string.
1345
This can contain either format instances themselves, or classes/factories that
1346
can be called to obtain one.
1350
#####################################################################
1351
# Repository Formats
1353
class RepositoryFormat(object):
1354
"""A repository format.
1356
Formats provide three things:
1357
* An initialization routine to construct repository data on disk.
1358
* a format string which is used when the BzrDir supports versioned
1360
* an open routine which returns a Repository instance.
1362
Formats are placed in an dict by their format string for reference
1363
during opening. These should be subclasses of RepositoryFormat
1366
Once a format is deprecated, just deprecate the initialize and open
1367
methods on the format class. Do not deprecate the object, as the
1368
object will be created every system load.
1370
Common instance attributes:
1371
_matchingbzrdir - the bzrdir format that the repository format was
1372
originally written to work with. This can be used if manually
1373
constructing a bzrdir and repository, or more commonly for test suite
1378
return "<%s>" % self.__class__.__name__
1380
def __eq__(self, other):
1381
# format objects are generally stateless
1382
return isinstance(other, self.__class__)
1384
def __ne__(self, other):
1385
return not self == other
1388
def find_format(klass, a_bzrdir):
1389
"""Return the format for the repository object in a_bzrdir.
1391
This is used by bzr native formats that have a "format" file in
1392
the repository. Other methods may be used by different types of
1396
transport = a_bzrdir.get_repository_transport(None)
1397
format_string = transport.get("format").read()
1398
return format_registry.get(format_string)
1399
except errors.NoSuchFile:
1400
raise errors.NoRepositoryPresent(a_bzrdir)
1402
raise errors.UnknownFormatError(format=format_string)
1405
def register_format(klass, format):
1406
format_registry.register(format.get_format_string(), format)
1409
def unregister_format(klass, format):
1410
format_registry.remove(format.get_format_string())
1413
def get_default_format(klass):
1414
"""Return the current default format."""
1415
from bzrlib import bzrdir
1416
return bzrdir.format_registry.make_bzrdir('default').repository_format
1418
def _get_control_store(self, repo_transport, control_files):
1419
"""Return the control store for this repository."""
1420
raise NotImplementedError(self._get_control_store)
1422
def get_format_string(self):
1423
"""Return the ASCII format string that identifies this format.
1425
Note that in pre format ?? repositories the format string is
1426
not permitted nor written to disk.
1428
raise NotImplementedError(self.get_format_string)
1430
def get_format_description(self):
1431
"""Return the short description for this format."""
1432
raise NotImplementedError(self.get_format_description)
1434
def _get_revision_store(self, repo_transport, control_files):
1435
"""Return the revision store object for this a_bzrdir."""
1436
raise NotImplementedError(self._get_revision_store)
1438
def _get_text_rev_store(self,
1445
"""Common logic for getting a revision store for a repository.
1447
see self._get_revision_store for the subclass-overridable method to
1448
get the store for a repository.
1450
from bzrlib.store.revision.text import TextRevisionStore
1451
dir_mode = control_files._dir_mode
1452
file_mode = control_files._file_mode
1453
text_store = TextStore(transport.clone(name),
1455
compressed=compressed,
1457
file_mode=file_mode)
1458
_revision_store = TextRevisionStore(text_store, serializer)
1459
return _revision_store
1461
# TODO: this shouldn't be in the base class, it's specific to things that
1462
# use weaves or knits -- mbp 20070207
1463
def _get_versioned_file_store(self,
1468
versionedfile_class=None,
1469
versionedfile_kwargs={},
1471
if versionedfile_class is None:
1472
versionedfile_class = self._versionedfile_class
1473
weave_transport = control_files._transport.clone(name)
1474
dir_mode = control_files._dir_mode
1475
file_mode = control_files._file_mode
1476
return VersionedFileStore(weave_transport, prefixed=prefixed,
1478
file_mode=file_mode,
1479
versionedfile_class=versionedfile_class,
1480
versionedfile_kwargs=versionedfile_kwargs,
1483
def initialize(self, a_bzrdir, shared=False):
1484
"""Initialize a repository of this format in a_bzrdir.
1486
:param a_bzrdir: The bzrdir to put the new repository in it.
1487
:param shared: The repository should be initialized as a sharable one.
1488
:returns: The new repository object.
1490
This may raise UninitializableFormat if shared repository are not
1491
compatible the a_bzrdir.
1493
raise NotImplementedError(self.initialize)
1495
def is_supported(self):
1496
"""Is this format supported?
1498
Supported formats must be initializable and openable.
1499
Unsupported formats may not support initialization or committing or
1500
some other features depending on the reason for not being supported.
1504
def check_conversion_target(self, target_format):
1505
raise NotImplementedError(self.check_conversion_target)
1507
def open(self, a_bzrdir, _found=False):
1508
"""Return an instance of this format for the bzrdir a_bzrdir.
1510
_found is a private parameter, do not use it.
1512
raise NotImplementedError(self.open)
1515
class MetaDirRepositoryFormat(RepositoryFormat):
1516
"""Common base class for the new repositories using the metadir layout."""
1518
rich_root_data = False
1519
supports_tree_reference = False
1520
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1523
super(MetaDirRepositoryFormat, self).__init__()
1525
def _create_control_files(self, a_bzrdir):
1526
"""Create the required files and the initial control_files object."""
1527
# FIXME: RBC 20060125 don't peek under the covers
1528
# NB: no need to escape relative paths that are url safe.
1529
repository_transport = a_bzrdir.get_repository_transport(self)
1530
control_files = lockable_files.LockableFiles(repository_transport,
1531
'lock', lockdir.LockDir)
1532
control_files.create_lock()
1533
return control_files
1535
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1536
"""Upload the initial blank content."""
1537
control_files = self._create_control_files(a_bzrdir)
1538
control_files.lock_write()
1540
control_files._transport.mkdir_multi(dirs,
1541
mode=control_files._dir_mode)
1542
for file, content in files:
1543
control_files.put(file, content)
1544
for file, content in utf8_files:
1545
control_files.put_utf8(file, content)
1547
control_files.put_utf8('shared-storage', '')
1549
control_files.unlock()
1552
# formats which have no format string are not discoverable
1553
# and not independently creatable, so are not registered. They're
1554
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1555
# needed, it's constructed directly by the BzrDir. Non-native formats where
1556
# the repository is not separately opened are similar.
1558
format_registry.register_lazy(
1559
'Bazaar-NG Repository format 7',
1560
'bzrlib.repofmt.weaverepo',
1563
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1564
# default control directory format
1566
format_registry.register_lazy(
1567
'Bazaar-NG Knit Repository Format 1',
1568
'bzrlib.repofmt.knitrepo',
1569
'RepositoryFormatKnit1',
1571
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1573
format_registry.register_lazy(
1574
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1575
'bzrlib.repofmt.knitrepo',
1576
'RepositoryFormatKnit3',
1580
class InterRepository(InterObject):
1581
"""This class represents operations taking place between two repositories.
1583
Its instances have methods like copy_content and fetch, and contain
1584
references to the source and target repositories these operations can be
1587
Often we will provide convenience methods on 'repository' which carry out
1588
operations with another repository - they will always forward to
1589
InterRepository.get(other).method_name(parameters).
1593
"""The available optimised InterRepository types."""
1595
def copy_content(self, revision_id=None):
1596
raise NotImplementedError(self.copy_content)
1598
def fetch(self, revision_id=None, pb=None):
1599
"""Fetch the content required to construct revision_id.
1601
The content is copied from self.source to self.target.
1603
:param revision_id: if None all content is copied, if NULL_REVISION no
1605
:param pb: optional progress bar to use for progress reports. If not
1606
provided a default one will be created.
1608
Returns the copied revision count and the failed revisions in a tuple:
1611
raise NotImplementedError(self.fetch)
1614
def missing_revision_ids(self, revision_id=None):
1615
"""Return the revision ids that source has that target does not.
1617
These are returned in topological order.
1619
:param revision_id: only return revision ids included by this
1622
# generic, possibly worst case, slow code path.
1623
target_ids = set(self.target.all_revision_ids())
1624
if revision_id is not None:
1625
# TODO: jam 20070210 InterRepository is internal enough that it
1626
# should assume revision_ids are already utf-8
1627
revision_id = osutils.safe_revision_id(revision_id)
1628
source_ids = self.source.get_ancestry(revision_id)
1629
assert source_ids[0] is None
1632
source_ids = self.source.all_revision_ids()
1633
result_set = set(source_ids).difference(target_ids)
1634
# this may look like a no-op: its not. It preserves the ordering
1635
# other_ids had while only returning the members from other_ids
1636
# that we've decided we need.
1637
return [rev_id for rev_id in source_ids if rev_id in result_set]
1640
class InterSameDataRepository(InterRepository):
1641
"""Code for converting between repositories that represent the same data.
1643
Data format and model must match for this to work.
1647
def _get_repo_format_to_test(self):
1648
"""Repository format for testing with."""
1649
return RepositoryFormat.get_default_format()
1652
def is_compatible(source, target):
1653
if source.supports_rich_root() != target.supports_rich_root():
1655
if source._serializer != target._serializer:
1660
def copy_content(self, revision_id=None):
1661
"""Make a complete copy of the content in self into destination.
1663
This copies both the repository's revision data, and configuration information
1664
such as the make_working_trees setting.
1666
This is a destructive operation! Do not use it on existing
1669
:param revision_id: Only copy the content needed to construct
1670
revision_id and its parents.
1673
self.target.set_make_working_trees(self.source.make_working_trees())
1674
except NotImplementedError:
1676
# TODO: jam 20070210 This is fairly internal, so we should probably
1677
# just assert that revision_id is not unicode.
1678
revision_id = osutils.safe_revision_id(revision_id)
1679
# but don't bother fetching if we have the needed data now.
1680
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1681
self.target.has_revision(revision_id)):
1683
self.target.fetch(self.source, revision_id=revision_id)
1686
def fetch(self, revision_id=None, pb=None):
1687
"""See InterRepository.fetch()."""
1688
from bzrlib.fetch import GenericRepoFetcher
1689
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1690
self.source, self.source._format, self.target,
1691
self.target._format)
1692
# TODO: jam 20070210 This should be an assert, not a translate
1693
revision_id = osutils.safe_revision_id(revision_id)
1694
f = GenericRepoFetcher(to_repository=self.target,
1695
from_repository=self.source,
1696
last_revision=revision_id,
1698
return f.count_copied, f.failed_revisions
1701
class InterWeaveRepo(InterSameDataRepository):
1702
"""Optimised code paths between Weave based repositories."""
1705
def _get_repo_format_to_test(self):
1706
from bzrlib.repofmt import weaverepo
1707
return weaverepo.RepositoryFormat7()
1710
def is_compatible(source, target):
1711
"""Be compatible with known Weave formats.
1713
We don't test for the stores being of specific types because that
1714
could lead to confusing results, and there is no need to be
1717
from bzrlib.repofmt.weaverepo import (
1723
return (isinstance(source._format, (RepositoryFormat5,
1725
RepositoryFormat7)) and
1726
isinstance(target._format, (RepositoryFormat5,
1728
RepositoryFormat7)))
1729
except AttributeError:
1733
def copy_content(self, revision_id=None):
1734
"""See InterRepository.copy_content()."""
1735
# weave specific optimised path:
1736
# TODO: jam 20070210 Internal, should be an assert, not translate
1737
revision_id = osutils.safe_revision_id(revision_id)
1739
self.target.set_make_working_trees(self.source.make_working_trees())
1740
except NotImplementedError:
1742
# FIXME do not peek!
1743
if self.source.control_files._transport.listable():
1744
pb = ui.ui_factory.nested_progress_bar()
1746
self.target.weave_store.copy_all_ids(
1747
self.source.weave_store,
1749
from_transaction=self.source.get_transaction(),
1750
to_transaction=self.target.get_transaction())
1751
pb.update('copying inventory', 0, 1)
1752
self.target.control_weaves.copy_multi(
1753
self.source.control_weaves, ['inventory'],
1754
from_transaction=self.source.get_transaction(),
1755
to_transaction=self.target.get_transaction())
1756
self.target._revision_store.text_store.copy_all_ids(
1757
self.source._revision_store.text_store,
1762
self.target.fetch(self.source, revision_id=revision_id)
1765
def fetch(self, revision_id=None, pb=None):
1766
"""See InterRepository.fetch()."""
1767
from bzrlib.fetch import GenericRepoFetcher
1768
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1769
self.source, self.source._format, self.target, self.target._format)
1770
# TODO: jam 20070210 This should be an assert, not a translate
1771
revision_id = osutils.safe_revision_id(revision_id)
1772
f = GenericRepoFetcher(to_repository=self.target,
1773
from_repository=self.source,
1774
last_revision=revision_id,
1776
return f.count_copied, f.failed_revisions
1779
def missing_revision_ids(self, revision_id=None):
1780
"""See InterRepository.missing_revision_ids()."""
1781
# we want all revisions to satisfy revision_id in source.
1782
# but we don't want to stat every file here and there.
1783
# we want then, all revisions other needs to satisfy revision_id
1784
# checked, but not those that we have locally.
1785
# so the first thing is to get a subset of the revisions to
1786
# satisfy revision_id in source, and then eliminate those that
1787
# we do already have.
1788
# this is slow on high latency connection to self, but as as this
1789
# disk format scales terribly for push anyway due to rewriting
1790
# inventory.weave, this is considered acceptable.
1792
if revision_id is not None:
1793
source_ids = self.source.get_ancestry(revision_id)
1794
assert source_ids[0] is None
1797
source_ids = self.source._all_possible_ids()
1798
source_ids_set = set(source_ids)
1799
# source_ids is the worst possible case we may need to pull.
1800
# now we want to filter source_ids against what we actually
1801
# have in target, but don't try to check for existence where we know
1802
# we do not have a revision as that would be pointless.
1803
target_ids = set(self.target._all_possible_ids())
1804
possibly_present_revisions = target_ids.intersection(source_ids_set)
1805
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1806
required_revisions = source_ids_set.difference(actually_present_revisions)
1807
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1808
if revision_id is not None:
1809
# we used get_ancestry to determine source_ids then we are assured all
1810
# revisions referenced are present as they are installed in topological order.
1811
# and the tip revision was validated by get_ancestry.
1812
return required_topo_revisions
1814
# if we just grabbed the possibly available ids, then
1815
# we only have an estimate of whats available and need to validate
1816
# that against the revision records.
1817
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1820
class InterKnitRepo(InterSameDataRepository):
1821
"""Optimised code paths between Knit based repositories."""
1824
def _get_repo_format_to_test(self):
1825
from bzrlib.repofmt import knitrepo
1826
return knitrepo.RepositoryFormatKnit1()
1829
def is_compatible(source, target):
1830
"""Be compatible with known Knit formats.
1832
We don't test for the stores being of specific types because that
1833
could lead to confusing results, and there is no need to be
1836
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1838
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1839
isinstance(target._format, (RepositoryFormatKnit1)))
1840
except AttributeError:
1844
def fetch(self, revision_id=None, pb=None):
1845
"""See InterRepository.fetch()."""
1846
from bzrlib.fetch import KnitRepoFetcher
1847
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1848
self.source, self.source._format, self.target, self.target._format)
1849
# TODO: jam 20070210 This should be an assert, not a translate
1850
revision_id = osutils.safe_revision_id(revision_id)
1851
f = KnitRepoFetcher(to_repository=self.target,
1852
from_repository=self.source,
1853
last_revision=revision_id,
1855
return f.count_copied, f.failed_revisions
1858
def missing_revision_ids(self, revision_id=None):
1859
"""See InterRepository.missing_revision_ids()."""
1860
if revision_id is not None:
1861
source_ids = self.source.get_ancestry(revision_id)
1862
assert source_ids[0] is None
1865
source_ids = self.source._all_possible_ids()
1866
source_ids_set = set(source_ids)
1867
# source_ids is the worst possible case we may need to pull.
1868
# now we want to filter source_ids against what we actually
1869
# have in target, but don't try to check for existence where we know
1870
# we do not have a revision as that would be pointless.
1871
target_ids = set(self.target._all_possible_ids())
1872
possibly_present_revisions = target_ids.intersection(source_ids_set)
1873
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1874
required_revisions = source_ids_set.difference(actually_present_revisions)
1875
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1876
if revision_id is not None:
1877
# we used get_ancestry to determine source_ids then we are assured all
1878
# revisions referenced are present as they are installed in topological order.
1879
# and the tip revision was validated by get_ancestry.
1880
return required_topo_revisions
1882
# if we just grabbed the possibly available ids, then
1883
# we only have an estimate of whats available and need to validate
1884
# that against the revision records.
1885
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1888
class InterModel1and2(InterRepository):
1891
def _get_repo_format_to_test(self):
1895
def is_compatible(source, target):
1896
if not source.supports_rich_root() and target.supports_rich_root():
1902
def fetch(self, revision_id=None, pb=None):
1903
"""See InterRepository.fetch()."""
1904
from bzrlib.fetch import Model1toKnit2Fetcher
1905
# TODO: jam 20070210 This should be an assert, not a translate
1906
revision_id = osutils.safe_revision_id(revision_id)
1907
f = Model1toKnit2Fetcher(to_repository=self.target,
1908
from_repository=self.source,
1909
last_revision=revision_id,
1911
return f.count_copied, f.failed_revisions
1914
def copy_content(self, revision_id=None):
1915
"""Make a complete copy of the content in self into destination.
1917
This is a destructive operation! Do not use it on existing
1920
:param revision_id: Only copy the content needed to construct
1921
revision_id and its parents.
1924
self.target.set_make_working_trees(self.source.make_working_trees())
1925
except NotImplementedError:
1927
# TODO: jam 20070210 Internal, assert, don't translate
1928
revision_id = osutils.safe_revision_id(revision_id)
1929
# but don't bother fetching if we have the needed data now.
1930
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1931
self.target.has_revision(revision_id)):
1933
self.target.fetch(self.source, revision_id=revision_id)
1936
class InterKnit1and2(InterKnitRepo):
1939
def _get_repo_format_to_test(self):
1943
def is_compatible(source, target):
1944
"""Be compatible with Knit1 source and Knit3 target"""
1945
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1947
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1948
RepositoryFormatKnit3
1949
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1950
isinstance(target._format, (RepositoryFormatKnit3)))
1951
except AttributeError:
1955
def fetch(self, revision_id=None, pb=None):
1956
"""See InterRepository.fetch()."""
1957
from bzrlib.fetch import Knit1to2Fetcher
1958
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1959
self.source, self.source._format, self.target,
1960
self.target._format)
1961
# TODO: jam 20070210 This should be an assert, not a translate
1962
revision_id = osutils.safe_revision_id(revision_id)
1963
f = Knit1to2Fetcher(to_repository=self.target,
1964
from_repository=self.source,
1965
last_revision=revision_id,
1967
return f.count_copied, f.failed_revisions
1970
class InterRemoteRepository(InterRepository):
1971
"""Code for converting between RemoteRepository objects.
1973
This just gets an non-remote repository from the RemoteRepository, and calls
1974
InterRepository.get again.
1977
def __init__(self, source, target):
1978
if isinstance(source, remote.RemoteRepository):
1979
source._ensure_real()
1980
real_source = source._real_repository
1982
real_source = source
1983
if isinstance(target, remote.RemoteRepository):
1984
target._ensure_real()
1985
real_target = target._real_repository
1987
real_target = target
1988
self.real_inter = InterRepository.get(real_source, real_target)
1991
def is_compatible(source, target):
1992
if isinstance(source, remote.RemoteRepository):
1994
if isinstance(target, remote.RemoteRepository):
1998
def copy_content(self, revision_id=None):
1999
self.real_inter.copy_content(revision_id=revision_id)
2001
def fetch(self, revision_id=None, pb=None):
2002
self.real_inter.fetch(revision_id=revision_id, pb=pb)
2005
def _get_repo_format_to_test(self):
2009
InterRepository.register_optimiser(InterSameDataRepository)
2010
InterRepository.register_optimiser(InterWeaveRepo)
2011
InterRepository.register_optimiser(InterKnitRepo)
2012
InterRepository.register_optimiser(InterModel1and2)
2013
InterRepository.register_optimiser(InterKnit1and2)
2014
InterRepository.register_optimiser(InterRemoteRepository)
2017
class CopyConverter(object):
2018
"""A repository conversion tool which just performs a copy of the content.
2020
This is slow but quite reliable.
2023
def __init__(self, target_format):
2024
"""Create a CopyConverter.
2026
:param target_format: The format the resulting repository should be.
2028
self.target_format = target_format
2030
def convert(self, repo, pb):
2031
"""Perform the conversion of to_convert, giving feedback via pb.
2033
:param to_convert: The disk object to convert.
2034
:param pb: a progress bar to use for progress information.
2039
# this is only useful with metadir layouts - separated repo content.
2040
# trigger an assertion if not such
2041
repo._format.get_format_string()
2042
self.repo_dir = repo.bzrdir
2043
self.step('Moving repository to repository.backup')
2044
self.repo_dir.transport.move('repository', 'repository.backup')
2045
backup_transport = self.repo_dir.transport.clone('repository.backup')
2046
repo._format.check_conversion_target(self.target_format)
2047
self.source_repo = repo._format.open(self.repo_dir,
2049
_override_transport=backup_transport)
2050
self.step('Creating new repository')
2051
converted = self.target_format.initialize(self.repo_dir,
2052
self.source_repo.is_shared())
2053
converted.lock_write()
2055
self.step('Copying content into repository.')
2056
self.source_repo.copy_content_into(converted)
2059
self.step('Deleting old repository content.')
2060
self.repo_dir.transport.delete_tree('repository.backup')
2061
self.pb.note('repository converted')
2063
def step(self, message):
2064
"""Update the pb by a step."""
2066
self.pb.update(message, self.count, self.total)
2069
class CommitBuilder(object):
2070
"""Provides an interface to build up a commit.
2072
This allows describing a tree to be committed without needing to
2073
know the internals of the format of the repository.
2076
record_root_entry = False
2077
def __init__(self, repository, parents, config, timestamp=None,
2078
timezone=None, committer=None, revprops=None,
2080
"""Initiate a CommitBuilder.
2082
:param repository: Repository to commit to.
2083
:param parents: Revision ids of the parents of the new revision.
2084
:param config: Configuration to use.
2085
:param timestamp: Optional timestamp recorded for commit.
2086
:param timezone: Optional timezone for timestamp.
2087
:param committer: Optional committer to set for commit.
2088
:param revprops: Optional dictionary of revision properties.
2089
:param revision_id: Optional revision id.
2091
self._config = config
2093
if committer is None:
2094
self._committer = self._config.username()
2096
assert isinstance(committer, basestring), type(committer)
2097
self._committer = committer
2099
self.new_inventory = Inventory(None)
2100
self._new_revision_id = osutils.safe_revision_id(revision_id)
2101
self.parents = parents
2102
self.repository = repository
2105
if revprops is not None:
2106
self._revprops.update(revprops)
2108
if timestamp is None:
2109
timestamp = time.time()
2110
# Restrict resolution to 1ms
2111
self._timestamp = round(timestamp, 3)
2113
if timezone is None:
2114
self._timezone = osutils.local_time_offset()
2116
self._timezone = int(timezone)
2118
self._generate_revision_if_needed()
2120
def commit(self, message):
2121
"""Make the actual commit.
2123
:return: The revision id of the recorded revision.
2125
rev = _mod_revision.Revision(
2126
timestamp=self._timestamp,
2127
timezone=self._timezone,
2128
committer=self._committer,
2130
inventory_sha1=self.inv_sha1,
2131
revision_id=self._new_revision_id,
2132
properties=self._revprops)
2133
rev.parent_ids = self.parents
2134
self.repository.add_revision(self._new_revision_id, rev,
2135
self.new_inventory, self._config)
2136
self.repository.commit_write_group()
2137
return self._new_revision_id
2140
"""Abort the commit that is being built.
2142
self.repository.abort_write_group()
2144
def revision_tree(self):
2145
"""Return the tree that was just committed.
2147
After calling commit() this can be called to get a RevisionTree
2148
representing the newly committed tree. This is preferred to
2149
calling Repository.revision_tree() because that may require
2150
deserializing the inventory, while we already have a copy in
2153
return RevisionTree(self.repository, self.new_inventory,
2154
self._new_revision_id)
2156
def finish_inventory(self):
2157
"""Tell the builder that the inventory is finished."""
2158
if self.new_inventory.root is None:
2159
symbol_versioning.warn('Root entry should be supplied to'
2160
' record_entry_contents, as of bzr 0.10.',
2161
DeprecationWarning, stacklevel=2)
2162
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2163
self.new_inventory.revision_id = self._new_revision_id
2164
self.inv_sha1 = self.repository.add_inventory(
2165
self._new_revision_id,
2170
def _gen_revision_id(self):
2171
"""Return new revision-id."""
2172
return generate_ids.gen_revision_id(self._config.username(),
2175
def _generate_revision_if_needed(self):
2176
"""Create a revision id if None was supplied.
2178
If the repository can not support user-specified revision ids
2179
they should override this function and raise CannotSetRevisionId
2180
if _new_revision_id is not None.
2182
:raises: CannotSetRevisionId
2184
if self._new_revision_id is None:
2185
self._new_revision_id = self._gen_revision_id()
2187
def record_entry_contents(self, ie, parent_invs, path, tree):
2188
"""Record the content of ie from tree into the commit if needed.
2190
Side effect: sets ie.revision when unchanged
2192
:param ie: An inventory entry present in the commit.
2193
:param parent_invs: The inventories of the parent revisions of the
2195
:param path: The path the entry is at in the tree.
2196
:param tree: The tree which contains this entry and should be used to
2199
if self.new_inventory.root is None and ie.parent_id is not None:
2200
symbol_versioning.warn('Root entry should be supplied to'
2201
' record_entry_contents, as of bzr 0.10.',
2202
DeprecationWarning, stacklevel=2)
2203
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2205
self.new_inventory.add(ie)
2207
# ie.revision is always None if the InventoryEntry is considered
2208
# for committing. ie.snapshot will record the correct revision
2209
# which may be the sole parent if it is untouched.
2210
if ie.revision is not None:
2213
# In this revision format, root entries have no knit or weave
2214
if ie is self.new_inventory.root:
2215
# When serializing out to disk and back in
2216
# root.revision is always _new_revision_id
2217
ie.revision = self._new_revision_id
2219
previous_entries = ie.find_previous_heads(
2221
self.repository.weave_store,
2222
self.repository.get_transaction())
2223
# we are creating a new revision for ie in the history store
2225
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2227
def modified_directory(self, file_id, file_parents):
2228
"""Record the presence of a symbolic link.
2230
:param file_id: The file_id of the link to record.
2231
:param file_parents: The per-file parent revision ids.
2233
self._add_text_to_weave(file_id, [], file_parents.keys())
2235
def modified_reference(self, file_id, file_parents):
2236
"""Record the modification of a reference.
2238
:param file_id: The file_id of the link to record.
2239
:param file_parents: The per-file parent revision ids.
2241
self._add_text_to_weave(file_id, [], file_parents.keys())
2243
def modified_file_text(self, file_id, file_parents,
2244
get_content_byte_lines, text_sha1=None,
2246
"""Record the text of file file_id
2248
:param file_id: The file_id of the file to record the text of.
2249
:param file_parents: The per-file parent revision ids.
2250
:param get_content_byte_lines: A callable which will return the byte
2252
:param text_sha1: Optional SHA1 of the file contents.
2253
:param text_size: Optional size of the file contents.
2255
# mutter('storing text of file {%s} in revision {%s} into %r',
2256
# file_id, self._new_revision_id, self.repository.weave_store)
2257
# special case to avoid diffing on renames or
2259
if (len(file_parents) == 1
2260
and text_sha1 == file_parents.values()[0].text_sha1
2261
and text_size == file_parents.values()[0].text_size):
2262
previous_ie = file_parents.values()[0]
2263
versionedfile = self.repository.weave_store.get_weave(file_id,
2264
self.repository.get_transaction())
2265
versionedfile.clone_text(self._new_revision_id,
2266
previous_ie.revision, file_parents.keys())
2267
return text_sha1, text_size
2269
new_lines = get_content_byte_lines()
2270
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2271
# should return the SHA1 and size
2272
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2273
return osutils.sha_strings(new_lines), \
2274
sum(map(len, new_lines))
2276
def modified_link(self, file_id, file_parents, link_target):
2277
"""Record the presence of a symbolic link.
2279
:param file_id: The file_id of the link to record.
2280
:param file_parents: The per-file parent revision ids.
2281
:param link_target: Target location of this link.
2283
self._add_text_to_weave(file_id, [], file_parents.keys())
2285
def _add_text_to_weave(self, file_id, new_lines, parents):
2286
versionedfile = self.repository.weave_store.get_weave_or_empty(
2287
file_id, self.repository.get_transaction())
2288
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2289
versionedfile.clear_cache()
2292
class _CommitBuilder(CommitBuilder):
2293
"""Temporary class so old CommitBuilders are detected properly
2295
Note: CommitBuilder works whether or not root entry is recorded.
2298
record_root_entry = True
2301
class RootCommitBuilder(CommitBuilder):
2302
"""This commitbuilder actually records the root id"""
2304
record_root_entry = True
2306
def record_entry_contents(self, ie, parent_invs, path, tree):
2307
"""Record the content of ie from tree into the commit if needed.
2309
Side effect: sets ie.revision when unchanged
2311
:param ie: An inventory entry present in the commit.
2312
:param parent_invs: The inventories of the parent revisions of the
2314
:param path: The path the entry is at in the tree.
2315
:param tree: The tree which contains this entry and should be used to
2318
assert self.new_inventory.root is not None or ie.parent_id is None
2319
self.new_inventory.add(ie)
2321
# ie.revision is always None if the InventoryEntry is considered
2322
# for committing. ie.snapshot will record the correct revision
2323
# which may be the sole parent if it is untouched.
2324
if ie.revision is not None:
2327
previous_entries = ie.find_previous_heads(
2329
self.repository.weave_store,
2330
self.repository.get_transaction())
2331
# we are creating a new revision for ie in the history store
2333
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2345
def _unescaper(match, _map=_unescape_map):
2346
code = match.group(1)
2350
if not code.startswith('#'):
2352
return unichr(int(code[1:])).encode('utf8')
2358
def _unescape_xml(data):
2359
"""Unescape predefined XML entities in a string of data."""
2361
if _unescape_re is None:
2362
_unescape_re = re.compile('\&([^;]*);')
2363
return _unescape_re.sub(_unescaper, data)
added
removed
bzrlib/repository.py
