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), check_content=False)
131
def _inventory_add_lines(self, inv_vf, revision_id, parents, lines,
134
for parent in parents:
136
final_parents.append(parent)
137
inv_vf.add_lines(revision_id, final_parents, lines,
138
check_content=check_content)
141
def add_revision(self, revision_id, rev, inv=None, config=None):
142
"""Add rev to the revision store as revision_id.
144
:param revision_id: the revision id to use.
145
:param rev: The revision object.
146
:param inv: The inventory for the revision. if None, it will be looked
147
up in the inventory storer
148
:param config: If None no digital signature will be created.
149
If supplied its signature_needed method will be used
150
to determine if a signature should be made.
152
revision_id = osutils.safe_revision_id(revision_id)
153
# TODO: jam 20070210 Shouldn't we check rev.revision_id and
155
_mod_revision.check_not_reserved_id(revision_id)
156
if config is not None and config.signature_needed():
158
inv = self.get_inventory(revision_id)
159
plaintext = Testament(rev, inv).as_short_text()
160
self.store_revision_signature(
161
gpg.GPGStrategy(config), plaintext, revision_id)
162
if not revision_id in self.get_inventory_weave():
164
raise errors.WeaveRevisionNotPresent(revision_id,
165
self.get_inventory_weave())
167
# yes, this is not suitable for adding with ghosts.
168
self.add_inventory(revision_id, inv, rev.parent_ids)
169
self._revision_store.add_revision(rev, self.get_transaction())
171
def _add_revision_text(self, revision_id, text):
172
revision = self._revision_store._serializer.read_revision_from_string(
174
self._revision_store._add_revision(revision, StringIO(text),
175
self.get_transaction())
178
def _all_possible_ids(self):
179
"""Return all the possible revisions that we could find."""
180
return self.get_inventory_weave().versions()
182
def all_revision_ids(self):
183
"""Returns a list of all the revision ids in the repository.
185
This is deprecated because code should generally work on the graph
186
reachable from a particular revision, and ignore any other revisions
187
that might be present. There is no direct replacement method.
189
return self._all_revision_ids()
192
def _all_revision_ids(self):
193
"""Returns a list of all the revision ids in the repository.
195
These are in as much topological order as the underlying store can
196
present: for weaves ghosts may lead to a lack of correctness until
197
the reweave updates the parents list.
199
if self._revision_store.text_store.listable():
200
return self._revision_store.all_revision_ids(self.get_transaction())
201
result = self._all_possible_ids()
202
# TODO: jam 20070210 Ensure that _all_possible_ids returns non-unicode
203
# ids. (It should, since _revision_store's API should change to
204
# return utf8 revision_ids)
205
return self._eliminate_revisions_not_present(result)
207
def break_lock(self):
208
"""Break a lock if one is present from another instance.
210
Uses the ui factory to ask for confirmation if the lock may be from
213
self.control_files.break_lock()
216
def _eliminate_revisions_not_present(self, revision_ids):
217
"""Check every revision id in revision_ids to see if we have it.
219
Returns a set of the present revisions.
222
for id in revision_ids:
223
if self.has_revision(id):
228
def create(a_bzrdir):
229
"""Construct the current default format repository in a_bzrdir."""
230
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
232
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
233
"""instantiate a Repository.
235
:param _format: The format of the repository on disk.
236
:param a_bzrdir: The BzrDir of the repository.
238
In the future we will have a single api for all stores for
239
getting file texts, inventories and revisions, then
240
this construct will accept instances of those things.
242
super(Repository, self).__init__()
243
self._format = _format
244
# the following are part of the public API for Repository:
245
self.bzrdir = a_bzrdir
246
self.control_files = control_files
247
self._revision_store = _revision_store
248
# backwards compatibility
249
self.weave_store = text_store
251
self._reconcile_does_inventory_gc = True
252
# not right yet - should be more semantically clear ?
254
self.control_store = control_store
255
self.control_weaves = control_store
256
# TODO: make sure to construct the right store classes, etc, depending
257
# on whether escaping is required.
258
self._warn_if_deprecated()
259
self._write_group = None
262
return '%s(%r)' % (self.__class__.__name__,
263
self.bzrdir.transport.base)
265
def has_same_location(self, other):
266
"""Returns a boolean indicating if this repository is at the same
267
location as another repository.
269
This might return False even when two repository objects are accessing
270
the same physical repository via different URLs.
272
if self.__class__ is not other.__class__:
274
return (self.control_files._transport.base ==
275
other.control_files._transport.base)
277
def is_in_write_group(self):
278
"""Return True if there is an open write group.
280
:seealso: start_write_group.
282
return self._write_group is not None
285
return self.control_files.is_locked()
287
def lock_write(self, token=None):
288
"""Lock this repository for writing.
290
This causes caching within the repository obejct to start accumlating
291
data during reads, and allows a 'write_group' to be obtained. Write
292
groups must be used for actual data insertion.
294
:param token: if this is already locked, then lock_write will fail
295
unless the token matches the existing lock.
296
:returns: a token if this instance supports tokens, otherwise None.
297
:raises TokenLockingNotSupported: when a token is given but this
298
instance doesn't support using token locks.
299
:raises MismatchedToken: if the specified token doesn't match the token
300
of the existing lock.
301
:seealso: start_write_group.
303
A token should be passed in if you know that you have locked the object
304
some other way, and need to synchronise this object's state with that
307
XXX: this docstring is duplicated in many places, e.g. lockable_files.py
309
result = self.control_files.lock_write(token=token)
314
self.control_files.lock_read()
317
def get_physical_lock_status(self):
318
return self.control_files.get_physical_lock_status()
320
def leave_lock_in_place(self):
321
"""Tell this repository not to release the physical lock when this
324
If lock_write doesn't return a token, then this method is not supported.
326
self.control_files.leave_in_place()
328
def dont_leave_lock_in_place(self):
329
"""Tell this repository to release the physical lock when this
330
object is unlocked, even if it didn't originally acquire it.
332
If lock_write doesn't return a token, then this method is not supported.
334
self.control_files.dont_leave_in_place()
337
def gather_stats(self, revid=None, committers=None):
338
"""Gather statistics from a revision id.
340
:param revid: The revision id to gather statistics from, if None, then
341
no revision specific statistics are gathered.
342
:param committers: Optional parameter controlling whether to grab
343
a count of committers from the revision specific statistics.
344
:return: A dictionary of statistics. Currently this contains:
345
committers: The number of committers if requested.
346
firstrev: A tuple with timestamp, timezone for the penultimate left
347
most ancestor of revid, if revid is not the NULL_REVISION.
348
latestrev: A tuple with timestamp, timezone for revid, if revid is
349
not the NULL_REVISION.
350
revisions: The total revision count in the repository.
351
size: An estimate disk size of the repository in bytes.
354
if revid and committers:
355
result['committers'] = 0
356
if revid and revid != _mod_revision.NULL_REVISION:
358
all_committers = set()
359
revisions = self.get_ancestry(revid)
360
# pop the leading None
362
first_revision = None
364
# ignore the revisions in the middle - just grab first and last
365
revisions = revisions[0], revisions[-1]
366
for revision in self.get_revisions(revisions):
367
if not first_revision:
368
first_revision = revision
370
all_committers.add(revision.committer)
371
last_revision = revision
373
result['committers'] = len(all_committers)
374
result['firstrev'] = (first_revision.timestamp,
375
first_revision.timezone)
376
result['latestrev'] = (last_revision.timestamp,
377
last_revision.timezone)
379
# now gather global repository information
380
if self.bzrdir.root_transport.listable():
381
c, t = self._revision_store.total_size(self.get_transaction())
382
result['revisions'] = c
387
def missing_revision_ids(self, other, revision_id=None):
388
"""Return the revision ids that other has that this does not.
390
These are returned in topological order.
392
revision_id: only return revision ids included by revision_id.
394
revision_id = osutils.safe_revision_id(revision_id)
395
return InterRepository.get(other, self).missing_revision_ids(revision_id)
399
"""Open the repository rooted at base.
401
For instance, if the repository is at URL/.bzr/repository,
402
Repository.open(URL) -> a Repository instance.
404
control = bzrdir.BzrDir.open(base)
405
return control.open_repository()
407
def copy_content_into(self, destination, revision_id=None):
408
"""Make a complete copy of the content in self into destination.
410
This is a destructive operation! Do not use it on existing
413
revision_id = osutils.safe_revision_id(revision_id)
414
return InterRepository.get(self, destination).copy_content(revision_id)
416
def commit_write_group(self):
417
"""Commit the contents accrued within the current write group.
419
:seealso: start_write_group.
421
if self._write_group is not self.get_transaction():
422
# has an unlock or relock occured ?
423
raise errors.BzrError('mismatched lock context and write group.')
424
self._commit_write_group()
425
self._write_group = None
427
def _commit_write_group(self):
428
"""Template method for per-repository write group cleanup.
430
This is called before the write group is considered to be
431
finished and should ensure that all data handed to the repository
432
for writing during the write group is safely committed (to the
433
extent possible considering file system caching etc).
436
def fetch(self, source, revision_id=None, pb=None):
437
"""Fetch the content required to construct revision_id from source.
439
If revision_id is None all content is copied.
441
revision_id = osutils.safe_revision_id(revision_id)
442
inter = InterRepository.get(source, self)
444
return inter.fetch(revision_id=revision_id, pb=pb)
445
except NotImplementedError:
446
raise errors.IncompatibleRepositories(source, self)
448
def create_bundle(self, target, base, fileobj, format=None):
449
return serializer.write_bundle(self, target, base, fileobj, format)
451
def get_commit_builder(self, branch, parents, config, timestamp=None,
452
timezone=None, committer=None, revprops=None,
454
"""Obtain a CommitBuilder for this repository.
456
:param branch: Branch to commit to.
457
:param parents: Revision ids of the parents of the new revision.
458
:param config: Configuration to use.
459
:param timestamp: Optional timestamp recorded for commit.
460
:param timezone: Optional timezone for timestamp.
461
:param committer: Optional committer to set for commit.
462
:param revprops: Optional dictionary of revision properties.
463
:param revision_id: Optional revision id.
465
revision_id = osutils.safe_revision_id(revision_id)
466
result = CommitBuilder(self, parents, config, timestamp, timezone,
467
committer, revprops, revision_id)
468
self.start_write_group()
472
if (self.control_files._lock_count == 1 and
473
self.control_files._lock_mode == 'w'):
474
if self._write_group is not None:
475
raise errors.BzrError(
476
'Must end write groups before releasing write locks.')
477
self.control_files.unlock()
480
def clone(self, a_bzrdir, revision_id=None):
481
"""Clone this repository into a_bzrdir using the current format.
483
Currently no check is made that the format of this repository and
484
the bzrdir format are compatible. FIXME RBC 20060201.
486
:return: The newly created destination repository.
488
# TODO: deprecate after 0.16; cloning this with all its settings is
489
# probably not very useful -- mbp 20070423
490
dest_repo = self._create_sprouting_repo(a_bzrdir, shared=self.is_shared())
491
self.copy_content_into(dest_repo, revision_id)
494
def start_write_group(self):
495
"""Start a write group in the repository.
497
Write groups are used by repositories which do not have a 1:1 mapping
498
between file ids and backend store to manage the insertion of data from
499
both fetch and commit operations.
501
A write lock is required around the start_write_group/commit_write_group
502
for the support of lock-requiring repository formats.
504
One can only insert data into a repository inside a write group.
508
if not self.is_locked() or self.control_files._lock_mode != 'w':
509
raise errors.NotWriteLocked(self)
510
if self._write_group:
511
raise errors.BzrError('already in a write group')
512
self._start_write_group()
513
# so we can detect unlock/relock - the write group is now entered.
514
self._write_group = self.get_transaction()
516
def _start_write_group(self):
517
"""Template method for per-repository write group startup.
519
This is called before the write group is considered to be
524
def sprout(self, to_bzrdir, revision_id=None):
525
"""Create a descendent repository for new development.
527
Unlike clone, this does not copy the settings of the repository.
529
dest_repo = self._create_sprouting_repo(to_bzrdir, shared=False)
530
dest_repo.fetch(self, revision_id=revision_id)
533
def _create_sprouting_repo(self, a_bzrdir, shared):
534
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
535
# use target default format.
536
dest_repo = a_bzrdir.create_repository()
538
# Most control formats need the repository to be specifically
539
# created, but on some old all-in-one formats it's not needed
541
dest_repo = self._format.initialize(a_bzrdir, shared=shared)
542
except errors.UninitializableFormat:
543
dest_repo = a_bzrdir.open_repository()
547
def has_revision(self, revision_id):
548
"""True if this repository has a copy of the revision."""
549
if 'evil' in debug.debug_flags:
550
mutter_callsite(2, "has_revision is a LBYL symptom.")
551
revision_id = osutils.safe_revision_id(revision_id)
552
return self._revision_store.has_revision_id(revision_id,
553
self.get_transaction())
556
def get_revision_reconcile(self, revision_id):
557
"""'reconcile' helper routine that allows access to a revision always.
559
This variant of get_revision does not cross check the weave graph
560
against the revision one as get_revision does: but it should only
561
be used by reconcile, or reconcile-alike commands that are correcting
562
or testing the revision graph.
564
if not revision_id or not isinstance(revision_id, basestring):
565
raise errors.InvalidRevisionId(revision_id=revision_id,
567
return self.get_revisions([revision_id])[0]
570
def get_revisions(self, revision_ids):
571
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
572
revs = self._revision_store.get_revisions(revision_ids,
573
self.get_transaction())
575
assert not isinstance(rev.revision_id, unicode)
576
for parent_id in rev.parent_ids:
577
assert not isinstance(parent_id, unicode)
581
def get_revision_xml(self, revision_id):
582
# TODO: jam 20070210 This shouldn't be necessary since get_revision
583
# would have already do it.
584
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
585
revision_id = osutils.safe_revision_id(revision_id)
586
rev = self.get_revision(revision_id)
588
# the current serializer..
589
self._revision_store._serializer.write_revision(rev, rev_tmp)
591
return rev_tmp.getvalue()
594
def get_revision(self, revision_id):
595
"""Return the Revision object for a named revision"""
596
# TODO: jam 20070210 get_revision_reconcile should do this for us
597
revision_id = osutils.safe_revision_id(revision_id)
598
r = self.get_revision_reconcile(revision_id)
599
# weave corruption can lead to absent revision markers that should be
601
# the following test is reasonably cheap (it needs a single weave read)
602
# and the weave is cached in read transactions. In write transactions
603
# it is not cached but typically we only read a small number of
604
# revisions. For knits when they are introduced we will probably want
605
# to ensure that caching write transactions are in use.
606
inv = self.get_inventory_weave()
607
self._check_revision_parents(r, inv)
611
def get_deltas_for_revisions(self, revisions):
612
"""Produce a generator of revision deltas.
614
Note that the input is a sequence of REVISIONS, not revision_ids.
615
Trees will be held in memory until the generator exits.
616
Each delta is relative to the revision's lefthand predecessor.
618
required_trees = set()
619
for revision in revisions:
620
required_trees.add(revision.revision_id)
621
required_trees.update(revision.parent_ids[:1])
622
trees = dict((t.get_revision_id(), t) for
623
t in self.revision_trees(required_trees))
624
for revision in revisions:
625
if not revision.parent_ids:
626
old_tree = self.revision_tree(None)
628
old_tree = trees[revision.parent_ids[0]]
629
yield trees[revision.revision_id].changes_from(old_tree)
632
def get_revision_delta(self, revision_id):
633
"""Return the delta for one revision.
635
The delta is relative to the left-hand predecessor of the
638
r = self.get_revision(revision_id)
639
return list(self.get_deltas_for_revisions([r]))[0]
641
def _check_revision_parents(self, revision, inventory):
642
"""Private to Repository and Fetch.
644
This checks the parentage of revision in an inventory weave for
645
consistency and is only applicable to inventory-weave-for-ancestry
646
using repository formats & fetchers.
648
weave_parents = inventory.get_parents(revision.revision_id)
649
weave_names = inventory.versions()
650
for parent_id in revision.parent_ids:
651
if parent_id in weave_names:
652
# this parent must not be a ghost.
653
if not parent_id in weave_parents:
655
raise errors.CorruptRepository(self)
658
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
659
revision_id = osutils.safe_revision_id(revision_id)
660
signature = gpg_strategy.sign(plaintext)
661
self._revision_store.add_revision_signature_text(revision_id,
663
self.get_transaction())
665
def fileids_altered_by_revision_ids(self, revision_ids):
666
"""Find the file ids and versions affected by revisions.
668
:param revisions: an iterable containing revision ids.
669
:return: a dictionary mapping altered file-ids to an iterable of
670
revision_ids. Each altered file-ids has the exact revision_ids that
671
altered it listed explicitly.
673
assert self._serializer.support_altered_by_hack, \
674
("fileids_altered_by_revision_ids only supported for branches "
675
"which store inventory as unnested xml, not on %r" % self)
676
selected_revision_ids = set(osutils.safe_revision_id(r)
677
for r in revision_ids)
678
w = self.get_inventory_weave()
681
# this code needs to read every new line in every inventory for the
682
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
683
# not present in one of those inventories is unnecessary but not
684
# harmful because we are filtering by the revision id marker in the
685
# inventory lines : we only select file ids altered in one of those
686
# revisions. We don't need to see all lines in the inventory because
687
# only those added in an inventory in rev X can contain a revision=X
689
unescape_revid_cache = {}
690
unescape_fileid_cache = {}
692
# jam 20061218 In a big fetch, this handles hundreds of thousands
693
# of lines, so it has had a lot of inlining and optimizing done.
694
# Sorry that it is a little bit messy.
695
# Move several functions to be local variables, since this is a long
697
search = self._file_ids_altered_regex.search
698
unescape = _unescape_xml
699
setdefault = result.setdefault
700
pb = ui.ui_factory.nested_progress_bar()
702
for line in w.iter_lines_added_or_present_in_versions(
703
selected_revision_ids, pb=pb):
707
# One call to match.group() returning multiple items is quite a
708
# bit faster than 2 calls to match.group() each returning 1
709
file_id, revision_id = match.group('file_id', 'revision_id')
711
# Inlining the cache lookups helps a lot when you make 170,000
712
# lines and 350k ids, versus 8.4 unique ids.
713
# Using a cache helps in 2 ways:
714
# 1) Avoids unnecessary decoding calls
715
# 2) Re-uses cached strings, which helps in future set and
717
# (2) is enough that removing encoding entirely along with
718
# the cache (so we are using plain strings) results in no
719
# performance improvement.
721
revision_id = unescape_revid_cache[revision_id]
723
unescaped = unescape(revision_id)
724
unescape_revid_cache[revision_id] = unescaped
725
revision_id = unescaped
727
if revision_id in selected_revision_ids:
729
file_id = unescape_fileid_cache[file_id]
731
unescaped = unescape(file_id)
732
unescape_fileid_cache[file_id] = unescaped
734
setdefault(file_id, set()).add(revision_id)
739
def iter_files_bytes(self, desired_files):
740
"""Iterate through file versions.
742
Files will not necessarily be returned in the order they occur in
743
desired_files. No specific order is guaranteed.
745
Yields pairs of identifier, bytes_iterator. identifier is an opaque
746
value supplied by the caller as part of desired_files. It should
747
uniquely identify the file version in the caller's context. (Examples:
748
an index number or a TreeTransform trans_id.)
750
bytes_iterator is an iterable of bytestrings for the file. The
751
kind of iterable and length of the bytestrings are unspecified, but for
752
this implementation, it is a list of lines produced by
753
VersionedFile.get_lines().
755
:param desired_files: a list of (file_id, revision_id, identifier)
758
transaction = self.get_transaction()
759
for file_id, revision_id, callable_data in desired_files:
761
weave = self.weave_store.get_weave(file_id, transaction)
762
except errors.NoSuchFile:
763
raise errors.NoSuchIdInRepository(self, file_id)
764
yield callable_data, weave.get_lines(revision_id)
766
def item_keys_introduced_by(self, revision_ids, _files_pb=None):
767
"""Get an iterable listing the keys of all the data introduced by a set
770
The keys will be ordered so that the corresponding items can be safely
771
fetched and inserted in that order.
773
:returns: An iterable producing tuples of (knit-kind, file-id,
774
versions). knit-kind is one of 'file', 'inventory', 'signatures',
775
'revisions'. file-id is None unless knit-kind is 'file'.
777
# XXX: it's a bit weird to control the inventory weave caching in this
778
# generator. Ideally the caching would be done in fetch.py I think. Or
779
# maybe this generator should explicitly have the contract that it
780
# should not be iterated until the previously yielded item has been
782
inv_w = self.get_inventory_weave()
785
# file ids that changed
786
file_ids = self.fileids_altered_by_revision_ids(revision_ids)
788
num_file_ids = len(file_ids)
789
for file_id, altered_versions in file_ids.iteritems():
790
if _files_pb is not None:
791
_files_pb.update("fetch texts", count, num_file_ids)
793
yield ("file", file_id, altered_versions)
794
# We're done with the files_pb. Note that it finished by the caller,
795
# just as it was created by the caller.
799
yield ("inventory", None, revision_ids)
803
revisions_with_signatures = set()
804
for rev_id in revision_ids:
806
self.get_signature_text(rev_id)
807
except errors.NoSuchRevision:
811
revisions_with_signatures.add(rev_id)
812
yield ("signatures", None, revisions_with_signatures)
815
yield ("revisions", None, revision_ids)
818
def get_inventory_weave(self):
819
return self.control_weaves.get_weave('inventory',
820
self.get_transaction())
823
def get_inventory(self, revision_id):
824
"""Get Inventory object by hash."""
825
# TODO: jam 20070210 Technically we don't need to sanitize, since all
826
# called functions must sanitize.
827
revision_id = osutils.safe_revision_id(revision_id)
828
return self.deserialise_inventory(
829
revision_id, self.get_inventory_xml(revision_id))
831
def deserialise_inventory(self, revision_id, xml):
832
"""Transform the xml into an inventory object.
834
:param revision_id: The expected revision id of the inventory.
835
:param xml: A serialised inventory.
837
revision_id = osutils.safe_revision_id(revision_id)
838
result = self._serializer.read_inventory_from_string(xml)
839
result.root.revision = revision_id
842
def serialise_inventory(self, inv):
843
return self._serializer.write_inventory_to_string(inv)
845
def get_serializer_format(self):
846
return self._serializer.format_num
849
def get_inventory_xml(self, revision_id):
850
"""Get inventory XML as a file object."""
851
revision_id = osutils.safe_revision_id(revision_id)
853
assert isinstance(revision_id, str), type(revision_id)
854
iw = self.get_inventory_weave()
855
return iw.get_text(revision_id)
857
raise errors.HistoryMissing(self, 'inventory', revision_id)
860
def get_inventory_sha1(self, revision_id):
861
"""Return the sha1 hash of the inventory entry
863
# TODO: jam 20070210 Shouldn't this be deprecated / removed?
864
revision_id = osutils.safe_revision_id(revision_id)
865
return self.get_revision(revision_id).inventory_sha1
868
def get_revision_graph(self, revision_id=None):
869
"""Return a dictionary containing the revision graph.
871
:param revision_id: The revision_id to get a graph from. If None, then
872
the entire revision graph is returned. This is a deprecated mode of
873
operation and will be removed in the future.
874
:return: a dictionary of revision_id->revision_parents_list.
876
if 'evil' in debug.debug_flags:
878
"get_revision_graph scales with size of history.")
879
# special case NULL_REVISION
880
if revision_id == _mod_revision.NULL_REVISION:
882
revision_id = osutils.safe_revision_id(revision_id)
883
a_weave = self.get_inventory_weave()
884
all_revisions = self._eliminate_revisions_not_present(
886
entire_graph = dict([(node, tuple(a_weave.get_parents(node))) for
887
node in all_revisions])
888
if revision_id is None:
890
elif revision_id not in entire_graph:
891
raise errors.NoSuchRevision(self, revision_id)
893
# add what can be reached from revision_id
895
pending = set([revision_id])
896
while len(pending) > 0:
898
result[node] = entire_graph[node]
899
for revision_id in result[node]:
900
if revision_id not in result:
901
pending.add(revision_id)
905
def get_revision_graph_with_ghosts(self, revision_ids=None):
906
"""Return a graph of the revisions with ghosts marked as applicable.
908
:param revision_ids: an iterable of revisions to graph or None for all.
909
:return: a Graph object with the graph reachable from revision_ids.
911
if 'evil' in debug.debug_flags:
913
"get_revision_graph_with_ghosts scales with size of history.")
914
result = deprecated_graph.Graph()
916
pending = set(self.all_revision_ids())
919
pending = set(osutils.safe_revision_id(r) for r in revision_ids)
920
# special case NULL_REVISION
921
if _mod_revision.NULL_REVISION in pending:
922
pending.remove(_mod_revision.NULL_REVISION)
923
required = set(pending)
926
revision_id = pending.pop()
928
rev = self.get_revision(revision_id)
929
except errors.NoSuchRevision:
930
if revision_id in required:
933
result.add_ghost(revision_id)
935
for parent_id in rev.parent_ids:
936
# is this queued or done ?
937
if (parent_id not in pending and
938
parent_id not in done):
940
pending.add(parent_id)
941
result.add_node(revision_id, rev.parent_ids)
942
done.add(revision_id)
945
def _get_history_vf(self):
946
"""Get a versionedfile whose history graph reflects all revisions.
948
For weave repositories, this is the inventory weave.
950
return self.get_inventory_weave()
952
def iter_reverse_revision_history(self, revision_id):
953
"""Iterate backwards through revision ids in the lefthand history
955
:param revision_id: The revision id to start with. All its lefthand
956
ancestors will be traversed.
958
revision_id = osutils.safe_revision_id(revision_id)
959
if revision_id in (None, _mod_revision.NULL_REVISION):
961
next_id = revision_id
962
versionedfile = self._get_history_vf()
965
parents = versionedfile.get_parents(next_id)
966
if len(parents) == 0:
972
def get_revision_inventory(self, revision_id):
973
"""Return inventory of a past revision."""
974
# TODO: Unify this with get_inventory()
975
# bzr 0.0.6 and later imposes the constraint that the inventory_id
976
# must be the same as its revision, so this is trivial.
977
if revision_id is None:
978
# This does not make sense: if there is no revision,
979
# then it is the current tree inventory surely ?!
980
# and thus get_root_id() is something that looks at the last
981
# commit on the branch, and the get_root_id is an inventory check.
982
raise NotImplementedError
983
# return Inventory(self.get_root_id())
985
return self.get_inventory(revision_id)
989
"""Return True if this repository is flagged as a shared repository."""
990
raise NotImplementedError(self.is_shared)
993
def reconcile(self, other=None, thorough=False):
994
"""Reconcile this repository."""
995
from bzrlib.reconcile import RepoReconciler
996
reconciler = RepoReconciler(self, thorough=thorough)
997
reconciler.reconcile()
1000
def _refresh_data(self):
1001
"""Helper called from lock_* to ensure coherency with disk.
1003
The default implementation does nothing; it is however possible
1004
for repositories to maintain loaded indices across multiple locks
1005
by checking inside their implementation of this method to see
1006
whether their indices are still valid. This depends of course on
1007
the disk format being validatable in this manner.
1011
def revision_tree(self, revision_id):
1012
"""Return Tree for a revision on this branch.
1014
`revision_id` may be None for the empty tree revision.
1016
# TODO: refactor this to use an existing revision object
1017
# so we don't need to read it in twice.
1018
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
1019
return RevisionTree(self, Inventory(root_id=None),
1020
_mod_revision.NULL_REVISION)
1022
revision_id = osutils.safe_revision_id(revision_id)
1023
inv = self.get_revision_inventory(revision_id)
1024
return RevisionTree(self, inv, revision_id)
1027
def revision_trees(self, revision_ids):
1028
"""Return Tree for a revision on this branch.
1030
`revision_id` may not be None or 'null:'"""
1031
assert None not in revision_ids
1032
assert _mod_revision.NULL_REVISION not in revision_ids
1033
texts = self.get_inventory_weave().get_texts(revision_ids)
1034
for text, revision_id in zip(texts, revision_ids):
1035
inv = self.deserialise_inventory(revision_id, text)
1036
yield RevisionTree(self, inv, revision_id)
1039
def get_ancestry(self, revision_id, topo_sorted=True):
1040
"""Return a list of revision-ids integrated by a revision.
1042
The first element of the list is always None, indicating the origin
1043
revision. This might change when we have history horizons, or
1044
perhaps we should have a new API.
1046
This is topologically sorted.
1048
if _mod_revision.is_null(revision_id):
1050
revision_id = osutils.safe_revision_id(revision_id)
1051
if not self.has_revision(revision_id):
1052
raise errors.NoSuchRevision(self, revision_id)
1053
w = self.get_inventory_weave()
1054
candidates = w.get_ancestry(revision_id, topo_sorted)
1055
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
1058
"""Compress the data within the repository.
1060
This operation only makes sense for some repository types. For other
1061
types it should be a no-op that just returns.
1063
This stub method does not require a lock, but subclasses should use
1064
@needs_write_lock as this is a long running call its reasonable to
1065
implicitly lock for the user.
1069
def print_file(self, file, revision_id):
1070
"""Print `file` to stdout.
1072
FIXME RBC 20060125 as John Meinel points out this is a bad api
1073
- it writes to stdout, it assumes that that is valid etc. Fix
1074
by creating a new more flexible convenience function.
1076
revision_id = osutils.safe_revision_id(revision_id)
1077
tree = self.revision_tree(revision_id)
1078
# use inventory as it was in that revision
1079
file_id = tree.inventory.path2id(file)
1081
# TODO: jam 20060427 Write a test for this code path
1082
# it had a bug in it, and was raising the wrong
1084
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
1085
tree.print_file(file_id)
1087
def get_transaction(self):
1088
return self.control_files.get_transaction()
1090
def revision_parents(self, revision_id):
1091
revision_id = osutils.safe_revision_id(revision_id)
1092
return self.get_inventory_weave().parent_names(revision_id)
1094
def get_parents(self, revision_ids):
1095
"""See StackedParentsProvider.get_parents"""
1097
for revision_id in revision_ids:
1098
if revision_id == _mod_revision.NULL_REVISION:
1102
parents = self.get_revision(revision_id).parent_ids
1103
except errors.NoSuchRevision:
1106
if len(parents) == 0:
1107
parents = [_mod_revision.NULL_REVISION]
1108
parents_list.append(parents)
1111
def _make_parents_provider(self):
1114
def get_graph(self, other_repository=None):
1115
"""Return the graph walker for this repository format"""
1116
parents_provider = self._make_parents_provider()
1117
if (other_repository is not None and
1118
other_repository.bzrdir.transport.base !=
1119
self.bzrdir.transport.base):
1120
parents_provider = graph._StackedParentsProvider(
1121
[parents_provider, other_repository._make_parents_provider()])
1122
return graph.Graph(parents_provider)
1125
def set_make_working_trees(self, new_value):
1126
"""Set the policy flag for making working trees when creating branches.
1128
This only applies to branches that use this repository.
1130
The default is 'True'.
1131
:param new_value: True to restore the default, False to disable making
1134
raise NotImplementedError(self.set_make_working_trees)
1136
def make_working_trees(self):
1137
"""Returns the policy for making working trees on new branches."""
1138
raise NotImplementedError(self.make_working_trees)
1141
def sign_revision(self, revision_id, gpg_strategy):
1142
revision_id = osutils.safe_revision_id(revision_id)
1143
plaintext = Testament.from_revision(self, revision_id).as_short_text()
1144
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
1147
def has_signature_for_revision_id(self, revision_id):
1148
"""Query for a revision signature for revision_id in the repository."""
1149
revision_id = osutils.safe_revision_id(revision_id)
1150
return self._revision_store.has_signature(revision_id,
1151
self.get_transaction())
1154
def get_signature_text(self, revision_id):
1155
"""Return the text for a signature."""
1156
revision_id = osutils.safe_revision_id(revision_id)
1157
return self._revision_store.get_signature_text(revision_id,
1158
self.get_transaction())
1161
def check(self, revision_ids):
1162
"""Check consistency of all history of given revision_ids.
1164
Different repository implementations should override _check().
1166
:param revision_ids: A non-empty list of revision_ids whose ancestry
1167
will be checked. Typically the last revision_id of a branch.
1169
if not revision_ids:
1170
raise ValueError("revision_ids must be non-empty in %s.check"
1172
revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1173
return self._check(revision_ids)
1175
def _check(self, revision_ids):
1176
result = check.Check(self)
1180
def _warn_if_deprecated(self):
1181
global _deprecation_warning_done
1182
if _deprecation_warning_done:
1184
_deprecation_warning_done = True
1185
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
1186
% (self._format, self.bzrdir.transport.base))
1188
def supports_rich_root(self):
1189
return self._format.rich_root_data
1191
def _check_ascii_revisionid(self, revision_id, method):
1192
"""Private helper for ascii-only repositories."""
1193
# weave repositories refuse to store revisionids that are non-ascii.
1194
if revision_id is not None:
1195
# weaves require ascii revision ids.
1196
if isinstance(revision_id, unicode):
1198
revision_id.encode('ascii')
1199
except UnicodeEncodeError:
1200
raise errors.NonAsciiRevisionId(method, self)
1203
revision_id.decode('ascii')
1204
except UnicodeDecodeError:
1205
raise errors.NonAsciiRevisionId(method, self)
1209
# remove these delegates a while after bzr 0.15
1210
def __make_delegated(name, from_module):
1211
def _deprecated_repository_forwarder():
1212
symbol_versioning.warn('%s moved to %s in bzr 0.15'
1213
% (name, from_module),
1216
m = __import__(from_module, globals(), locals(), [name])
1218
return getattr(m, name)
1219
except AttributeError:
1220
raise AttributeError('module %s has no name %s'
1222
globals()[name] = _deprecated_repository_forwarder
1225
'AllInOneRepository',
1226
'WeaveMetaDirRepository',
1227
'PreSplitOutRepositoryFormat',
1228
'RepositoryFormat4',
1229
'RepositoryFormat5',
1230
'RepositoryFormat6',
1231
'RepositoryFormat7',
1233
__make_delegated(_name, 'bzrlib.repofmt.weaverepo')
1237
'RepositoryFormatKnit',
1238
'RepositoryFormatKnit1',
1240
__make_delegated(_name, 'bzrlib.repofmt.knitrepo')
1243
def install_revision(repository, rev, revision_tree):
1244
"""Install all revision data into a repository."""
1245
present_parents = []
1247
for p_id in rev.parent_ids:
1248
if repository.has_revision(p_id):
1249
present_parents.append(p_id)
1250
parent_trees[p_id] = repository.revision_tree(p_id)
1252
parent_trees[p_id] = repository.revision_tree(None)
1254
inv = revision_tree.inventory
1255
entries = inv.iter_entries()
1256
# backwards compatibility hack: skip the root id.
1257
if not repository.supports_rich_root():
1258
path, root = entries.next()
1259
if root.revision != rev.revision_id:
1260
raise errors.IncompatibleRevision(repr(repository))
1261
# Add the texts that are not already present
1262
for path, ie in entries:
1263
w = repository.weave_store.get_weave_or_empty(ie.file_id,
1264
repository.get_transaction())
1265
if ie.revision not in w:
1267
# FIXME: TODO: The following loop *may* be overlapping/duplicate
1268
# with InventoryEntry.find_previous_heads(). if it is, then there
1269
# is a latent bug here where the parents may have ancestors of each
1271
for revision, tree in parent_trees.iteritems():
1272
if ie.file_id not in tree:
1274
parent_id = tree.inventory[ie.file_id].revision
1275
if parent_id in text_parents:
1277
text_parents.append(parent_id)
1279
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
1280
repository.get_transaction())
1281
lines = revision_tree.get_file(ie.file_id).readlines()
1282
vfile.add_lines(rev.revision_id, text_parents, lines)
1284
# install the inventory
1285
repository.add_inventory(rev.revision_id, inv, present_parents)
1286
except errors.RevisionAlreadyPresent:
1288
repository.add_revision(rev.revision_id, rev, inv)
1291
class MetaDirRepository(Repository):
1292
"""Repositories in the new meta-dir layout."""
1294
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
1295
super(MetaDirRepository, self).__init__(_format,
1301
dir_mode = self.control_files._dir_mode
1302
file_mode = self.control_files._file_mode
1305
def is_shared(self):
1306
"""Return True if this repository is flagged as a shared repository."""
1307
return self.control_files._transport.has('shared-storage')
1310
def set_make_working_trees(self, new_value):
1311
"""Set the policy flag for making working trees when creating branches.
1313
This only applies to branches that use this repository.
1315
The default is 'True'.
1316
:param new_value: True to restore the default, False to disable making
1321
self.control_files._transport.delete('no-working-trees')
1322
except errors.NoSuchFile:
1325
self.control_files.put_utf8('no-working-trees', '')
1327
def make_working_trees(self):
1328
"""Returns the policy for making working trees on new branches."""
1329
return not self.control_files._transport.has('no-working-trees')
1332
class RepositoryFormatRegistry(registry.Registry):
1333
"""Registry of RepositoryFormats.
1336
def get(self, format_string):
1337
r = registry.Registry.get(self, format_string)
1343
format_registry = RepositoryFormatRegistry()
1344
"""Registry of formats, indexed by their identifying format string.
1346
This can contain either format instances themselves, or classes/factories that
1347
can be called to obtain one.
1351
#####################################################################
1352
# Repository Formats
1354
class RepositoryFormat(object):
1355
"""A repository format.
1357
Formats provide three things:
1358
* An initialization routine to construct repository data on disk.
1359
* a format string which is used when the BzrDir supports versioned
1361
* an open routine which returns a Repository instance.
1363
Formats are placed in an dict by their format string for reference
1364
during opening. These should be subclasses of RepositoryFormat
1367
Once a format is deprecated, just deprecate the initialize and open
1368
methods on the format class. Do not deprecate the object, as the
1369
object will be created every system load.
1371
Common instance attributes:
1372
_matchingbzrdir - the bzrdir format that the repository format was
1373
originally written to work with. This can be used if manually
1374
constructing a bzrdir and repository, or more commonly for test suite
1379
return "<%s>" % self.__class__.__name__
1381
def __eq__(self, other):
1382
# format objects are generally stateless
1383
return isinstance(other, self.__class__)
1385
def __ne__(self, other):
1386
return not self == other
1389
def find_format(klass, a_bzrdir):
1390
"""Return the format for the repository object in a_bzrdir.
1392
This is used by bzr native formats that have a "format" file in
1393
the repository. Other methods may be used by different types of
1397
transport = a_bzrdir.get_repository_transport(None)
1398
format_string = transport.get("format").read()
1399
return format_registry.get(format_string)
1400
except errors.NoSuchFile:
1401
raise errors.NoRepositoryPresent(a_bzrdir)
1403
raise errors.UnknownFormatError(format=format_string)
1406
def register_format(klass, format):
1407
format_registry.register(format.get_format_string(), format)
1410
def unregister_format(klass, format):
1411
format_registry.remove(format.get_format_string())
1414
def get_default_format(klass):
1415
"""Return the current default format."""
1416
from bzrlib import bzrdir
1417
return bzrdir.format_registry.make_bzrdir('default').repository_format
1419
def _get_control_store(self, repo_transport, control_files):
1420
"""Return the control store for this repository."""
1421
raise NotImplementedError(self._get_control_store)
1423
def get_format_string(self):
1424
"""Return the ASCII format string that identifies this format.
1426
Note that in pre format ?? repositories the format string is
1427
not permitted nor written to disk.
1429
raise NotImplementedError(self.get_format_string)
1431
def get_format_description(self):
1432
"""Return the short description for this format."""
1433
raise NotImplementedError(self.get_format_description)
1435
def _get_revision_store(self, repo_transport, control_files):
1436
"""Return the revision store object for this a_bzrdir."""
1437
raise NotImplementedError(self._get_revision_store)
1439
def _get_text_rev_store(self,
1446
"""Common logic for getting a revision store for a repository.
1448
see self._get_revision_store for the subclass-overridable method to
1449
get the store for a repository.
1451
from bzrlib.store.revision.text import TextRevisionStore
1452
dir_mode = control_files._dir_mode
1453
file_mode = control_files._file_mode
1454
text_store = TextStore(transport.clone(name),
1456
compressed=compressed,
1458
file_mode=file_mode)
1459
_revision_store = TextRevisionStore(text_store, serializer)
1460
return _revision_store
1462
# TODO: this shouldn't be in the base class, it's specific to things that
1463
# use weaves or knits -- mbp 20070207
1464
def _get_versioned_file_store(self,
1469
versionedfile_class=None,
1470
versionedfile_kwargs={},
1472
if versionedfile_class is None:
1473
versionedfile_class = self._versionedfile_class
1474
weave_transport = control_files._transport.clone(name)
1475
dir_mode = control_files._dir_mode
1476
file_mode = control_files._file_mode
1477
return VersionedFileStore(weave_transport, prefixed=prefixed,
1479
file_mode=file_mode,
1480
versionedfile_class=versionedfile_class,
1481
versionedfile_kwargs=versionedfile_kwargs,
1484
def initialize(self, a_bzrdir, shared=False):
1485
"""Initialize a repository of this format in a_bzrdir.
1487
:param a_bzrdir: The bzrdir to put the new repository in it.
1488
:param shared: The repository should be initialized as a sharable one.
1489
:returns: The new repository object.
1491
This may raise UninitializableFormat if shared repository are not
1492
compatible the a_bzrdir.
1494
raise NotImplementedError(self.initialize)
1496
def is_supported(self):
1497
"""Is this format supported?
1499
Supported formats must be initializable and openable.
1500
Unsupported formats may not support initialization or committing or
1501
some other features depending on the reason for not being supported.
1505
def check_conversion_target(self, target_format):
1506
raise NotImplementedError(self.check_conversion_target)
1508
def open(self, a_bzrdir, _found=False):
1509
"""Return an instance of this format for the bzrdir a_bzrdir.
1511
_found is a private parameter, do not use it.
1513
raise NotImplementedError(self.open)
1516
class MetaDirRepositoryFormat(RepositoryFormat):
1517
"""Common base class for the new repositories using the metadir layout."""
1519
rich_root_data = False
1520
supports_tree_reference = False
1521
_matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1524
super(MetaDirRepositoryFormat, self).__init__()
1526
def _create_control_files(self, a_bzrdir):
1527
"""Create the required files and the initial control_files object."""
1528
# FIXME: RBC 20060125 don't peek under the covers
1529
# NB: no need to escape relative paths that are url safe.
1530
repository_transport = a_bzrdir.get_repository_transport(self)
1531
control_files = lockable_files.LockableFiles(repository_transport,
1532
'lock', lockdir.LockDir)
1533
control_files.create_lock()
1534
return control_files
1536
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1537
"""Upload the initial blank content."""
1538
control_files = self._create_control_files(a_bzrdir)
1539
control_files.lock_write()
1541
control_files._transport.mkdir_multi(dirs,
1542
mode=control_files._dir_mode)
1543
for file, content in files:
1544
control_files.put(file, content)
1545
for file, content in utf8_files:
1546
control_files.put_utf8(file, content)
1548
control_files.put_utf8('shared-storage', '')
1550
control_files.unlock()
1553
# formats which have no format string are not discoverable
1554
# and not independently creatable, so are not registered. They're
1555
# all in bzrlib.repofmt.weaverepo now. When an instance of one of these is
1556
# needed, it's constructed directly by the BzrDir. Non-native formats where
1557
# the repository is not separately opened are similar.
1559
format_registry.register_lazy(
1560
'Bazaar-NG Repository format 7',
1561
'bzrlib.repofmt.weaverepo',
1564
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1565
# default control directory format
1567
format_registry.register_lazy(
1568
'Bazaar-NG Knit Repository Format 1',
1569
'bzrlib.repofmt.knitrepo',
1570
'RepositoryFormatKnit1',
1572
format_registry.default_key = 'Bazaar-NG Knit Repository Format 1'
1574
format_registry.register_lazy(
1575
'Bazaar Knit Repository Format 3 (bzr 0.15)\n',
1576
'bzrlib.repofmt.knitrepo',
1577
'RepositoryFormatKnit3',
1581
class InterRepository(InterObject):
1582
"""This class represents operations taking place between two repositories.
1584
Its instances have methods like copy_content and fetch, and contain
1585
references to the source and target repositories these operations can be
1588
Often we will provide convenience methods on 'repository' which carry out
1589
operations with another repository - they will always forward to
1590
InterRepository.get(other).method_name(parameters).
1594
"""The available optimised InterRepository types."""
1596
def copy_content(self, revision_id=None):
1597
raise NotImplementedError(self.copy_content)
1599
def fetch(self, revision_id=None, pb=None):
1600
"""Fetch the content required to construct revision_id.
1602
The content is copied from self.source to self.target.
1604
:param revision_id: if None all content is copied, if NULL_REVISION no
1606
:param pb: optional progress bar to use for progress reports. If not
1607
provided a default one will be created.
1609
Returns the copied revision count and the failed revisions in a tuple:
1612
raise NotImplementedError(self.fetch)
1615
def missing_revision_ids(self, revision_id=None):
1616
"""Return the revision ids that source has that target does not.
1618
These are returned in topological order.
1620
:param revision_id: only return revision ids included by this
1623
# generic, possibly worst case, slow code path.
1624
target_ids = set(self.target.all_revision_ids())
1625
if revision_id is not None:
1626
# TODO: jam 20070210 InterRepository is internal enough that it
1627
# should assume revision_ids are already utf-8
1628
revision_id = osutils.safe_revision_id(revision_id)
1629
source_ids = self.source.get_ancestry(revision_id)
1630
assert source_ids[0] is None
1633
source_ids = self.source.all_revision_ids()
1634
result_set = set(source_ids).difference(target_ids)
1635
# this may look like a no-op: its not. It preserves the ordering
1636
# other_ids had while only returning the members from other_ids
1637
# that we've decided we need.
1638
return [rev_id for rev_id in source_ids if rev_id in result_set]
1641
class InterSameDataRepository(InterRepository):
1642
"""Code for converting between repositories that represent the same data.
1644
Data format and model must match for this to work.
1648
def _get_repo_format_to_test(self):
1649
"""Repository format for testing with.
1651
InterSameData can pull from subtree to subtree and from non-subtree to
1652
non-subtree, so we test this with the richest repository format.
1654
from bzrlib.repofmt import knitrepo
1655
return knitrepo.RepositoryFormatKnit3()
1658
def is_compatible(source, target):
1659
if source.supports_rich_root() != target.supports_rich_root():
1661
if source._serializer != target._serializer:
1666
def copy_content(self, revision_id=None):
1667
"""Make a complete copy of the content in self into destination.
1669
This copies both the repository's revision data, and configuration information
1670
such as the make_working_trees setting.
1672
This is a destructive operation! Do not use it on existing
1675
:param revision_id: Only copy the content needed to construct
1676
revision_id and its parents.
1679
self.target.set_make_working_trees(self.source.make_working_trees())
1680
except NotImplementedError:
1682
# TODO: jam 20070210 This is fairly internal, so we should probably
1683
# just assert that revision_id is not unicode.
1684
revision_id = osutils.safe_revision_id(revision_id)
1685
# but don't bother fetching if we have the needed data now.
1686
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1687
self.target.has_revision(revision_id)):
1689
self.target.fetch(self.source, revision_id=revision_id)
1692
def fetch(self, revision_id=None, pb=None):
1693
"""See InterRepository.fetch()."""
1694
from bzrlib.fetch import GenericRepoFetcher
1695
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1696
self.source, self.source._format, self.target,
1697
self.target._format)
1698
# TODO: jam 20070210 This should be an assert, not a translate
1699
revision_id = osutils.safe_revision_id(revision_id)
1700
f = GenericRepoFetcher(to_repository=self.target,
1701
from_repository=self.source,
1702
last_revision=revision_id,
1704
return f.count_copied, f.failed_revisions
1707
class InterWeaveRepo(InterSameDataRepository):
1708
"""Optimised code paths between Weave based repositories."""
1711
def _get_repo_format_to_test(self):
1712
from bzrlib.repofmt import weaverepo
1713
return weaverepo.RepositoryFormat7()
1716
def is_compatible(source, target):
1717
"""Be compatible with known Weave formats.
1719
We don't test for the stores being of specific types because that
1720
could lead to confusing results, and there is no need to be
1723
from bzrlib.repofmt.weaverepo import (
1729
return (isinstance(source._format, (RepositoryFormat5,
1731
RepositoryFormat7)) and
1732
isinstance(target._format, (RepositoryFormat5,
1734
RepositoryFormat7)))
1735
except AttributeError:
1739
def copy_content(self, revision_id=None):
1740
"""See InterRepository.copy_content()."""
1741
# weave specific optimised path:
1742
# TODO: jam 20070210 Internal, should be an assert, not translate
1743
revision_id = osutils.safe_revision_id(revision_id)
1745
self.target.set_make_working_trees(self.source.make_working_trees())
1746
except NotImplementedError:
1748
# FIXME do not peek!
1749
if self.source.control_files._transport.listable():
1750
pb = ui.ui_factory.nested_progress_bar()
1752
self.target.weave_store.copy_all_ids(
1753
self.source.weave_store,
1755
from_transaction=self.source.get_transaction(),
1756
to_transaction=self.target.get_transaction())
1757
pb.update('copying inventory', 0, 1)
1758
self.target.control_weaves.copy_multi(
1759
self.source.control_weaves, ['inventory'],
1760
from_transaction=self.source.get_transaction(),
1761
to_transaction=self.target.get_transaction())
1762
self.target._revision_store.text_store.copy_all_ids(
1763
self.source._revision_store.text_store,
1768
self.target.fetch(self.source, revision_id=revision_id)
1771
def fetch(self, revision_id=None, pb=None):
1772
"""See InterRepository.fetch()."""
1773
from bzrlib.fetch import GenericRepoFetcher
1774
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1775
self.source, self.source._format, self.target, self.target._format)
1776
# TODO: jam 20070210 This should be an assert, not a translate
1777
revision_id = osutils.safe_revision_id(revision_id)
1778
f = GenericRepoFetcher(to_repository=self.target,
1779
from_repository=self.source,
1780
last_revision=revision_id,
1782
return f.count_copied, f.failed_revisions
1785
def missing_revision_ids(self, revision_id=None):
1786
"""See InterRepository.missing_revision_ids()."""
1787
# we want all revisions to satisfy revision_id in source.
1788
# but we don't want to stat every file here and there.
1789
# we want then, all revisions other needs to satisfy revision_id
1790
# checked, but not those that we have locally.
1791
# so the first thing is to get a subset of the revisions to
1792
# satisfy revision_id in source, and then eliminate those that
1793
# we do already have.
1794
# this is slow on high latency connection to self, but as as this
1795
# disk format scales terribly for push anyway due to rewriting
1796
# inventory.weave, this is considered acceptable.
1798
if revision_id is not None:
1799
source_ids = self.source.get_ancestry(revision_id)
1800
assert source_ids[0] is None
1803
source_ids = self.source._all_possible_ids()
1804
source_ids_set = set(source_ids)
1805
# source_ids is the worst possible case we may need to pull.
1806
# now we want to filter source_ids against what we actually
1807
# have in target, but don't try to check for existence where we know
1808
# we do not have a revision as that would be pointless.
1809
target_ids = set(self.target._all_possible_ids())
1810
possibly_present_revisions = target_ids.intersection(source_ids_set)
1811
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1812
required_revisions = source_ids_set.difference(actually_present_revisions)
1813
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1814
if revision_id is not None:
1815
# we used get_ancestry to determine source_ids then we are assured all
1816
# revisions referenced are present as they are installed in topological order.
1817
# and the tip revision was validated by get_ancestry.
1818
return required_topo_revisions
1820
# if we just grabbed the possibly available ids, then
1821
# we only have an estimate of whats available and need to validate
1822
# that against the revision records.
1823
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1826
class InterKnitRepo(InterSameDataRepository):
1827
"""Optimised code paths between Knit based repositories."""
1830
def _get_repo_format_to_test(self):
1831
from bzrlib.repofmt import knitrepo
1832
return knitrepo.RepositoryFormatKnit1()
1835
def is_compatible(source, target):
1836
"""Be compatible with known Knit formats.
1838
We don't test for the stores being of specific types because that
1839
could lead to confusing results, and there is no need to be
1842
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1
1844
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1845
isinstance(target._format, (RepositoryFormatKnit1)))
1846
except AttributeError:
1850
def fetch(self, revision_id=None, pb=None):
1851
"""See InterRepository.fetch()."""
1852
from bzrlib.fetch import KnitRepoFetcher
1853
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1854
self.source, self.source._format, self.target, self.target._format)
1855
# TODO: jam 20070210 This should be an assert, not a translate
1856
revision_id = osutils.safe_revision_id(revision_id)
1857
f = KnitRepoFetcher(to_repository=self.target,
1858
from_repository=self.source,
1859
last_revision=revision_id,
1861
return f.count_copied, f.failed_revisions
1864
def missing_revision_ids(self, revision_id=None):
1865
"""See InterRepository.missing_revision_ids()."""
1866
if revision_id is not None:
1867
source_ids = self.source.get_ancestry(revision_id)
1868
assert source_ids[0] is None
1871
source_ids = self.source._all_possible_ids()
1872
source_ids_set = set(source_ids)
1873
# source_ids is the worst possible case we may need to pull.
1874
# now we want to filter source_ids against what we actually
1875
# have in target, but don't try to check for existence where we know
1876
# we do not have a revision as that would be pointless.
1877
target_ids = set(self.target._all_possible_ids())
1878
possibly_present_revisions = target_ids.intersection(source_ids_set)
1879
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1880
required_revisions = source_ids_set.difference(actually_present_revisions)
1881
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1882
if revision_id is not None:
1883
# we used get_ancestry to determine source_ids then we are assured all
1884
# revisions referenced are present as they are installed in topological order.
1885
# and the tip revision was validated by get_ancestry.
1886
return required_topo_revisions
1888
# if we just grabbed the possibly available ids, then
1889
# we only have an estimate of whats available and need to validate
1890
# that against the revision records.
1891
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1894
class InterModel1and2(InterRepository):
1897
def _get_repo_format_to_test(self):
1901
def is_compatible(source, target):
1902
if not source.supports_rich_root() and target.supports_rich_root():
1908
def fetch(self, revision_id=None, pb=None):
1909
"""See InterRepository.fetch()."""
1910
from bzrlib.fetch import Model1toKnit2Fetcher
1911
# TODO: jam 20070210 This should be an assert, not a translate
1912
revision_id = osutils.safe_revision_id(revision_id)
1913
f = Model1toKnit2Fetcher(to_repository=self.target,
1914
from_repository=self.source,
1915
last_revision=revision_id,
1917
return f.count_copied, f.failed_revisions
1920
def copy_content(self, revision_id=None):
1921
"""Make a complete copy of the content in self into destination.
1923
This is a destructive operation! Do not use it on existing
1926
:param revision_id: Only copy the content needed to construct
1927
revision_id and its parents.
1930
self.target.set_make_working_trees(self.source.make_working_trees())
1931
except NotImplementedError:
1933
# TODO: jam 20070210 Internal, assert, don't translate
1934
revision_id = osutils.safe_revision_id(revision_id)
1935
# but don't bother fetching if we have the needed data now.
1936
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1937
self.target.has_revision(revision_id)):
1939
self.target.fetch(self.source, revision_id=revision_id)
1942
class InterKnit1and2(InterKnitRepo):
1945
def _get_repo_format_to_test(self):
1949
def is_compatible(source, target):
1950
"""Be compatible with Knit1 source and Knit3 target"""
1951
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit3
1953
from bzrlib.repofmt.knitrepo import RepositoryFormatKnit1, \
1954
RepositoryFormatKnit3
1955
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1956
isinstance(target._format, (RepositoryFormatKnit3)))
1957
except AttributeError:
1961
def fetch(self, revision_id=None, pb=None):
1962
"""See InterRepository.fetch()."""
1963
from bzrlib.fetch import Knit1to2Fetcher
1964
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1965
self.source, self.source._format, self.target,
1966
self.target._format)
1967
# TODO: jam 20070210 This should be an assert, not a translate
1968
revision_id = osutils.safe_revision_id(revision_id)
1969
f = Knit1to2Fetcher(to_repository=self.target,
1970
from_repository=self.source,
1971
last_revision=revision_id,
1973
return f.count_copied, f.failed_revisions
1976
class InterRemoteRepository(InterRepository):
1977
"""Code for converting between RemoteRepository objects.
1979
This just gets an non-remote repository from the RemoteRepository, and calls
1980
InterRepository.get again.
1983
def __init__(self, source, target):
1984
if isinstance(source, remote.RemoteRepository):
1985
source._ensure_real()
1986
real_source = source._real_repository
1988
real_source = source
1989
if isinstance(target, remote.RemoteRepository):
1990
target._ensure_real()
1991
real_target = target._real_repository
1993
real_target = target
1994
self.real_inter = InterRepository.get(real_source, real_target)
1997
def is_compatible(source, target):
1998
if isinstance(source, remote.RemoteRepository):
2000
if isinstance(target, remote.RemoteRepository):
2004
def copy_content(self, revision_id=None):
2005
self.real_inter.copy_content(revision_id=revision_id)
2007
def fetch(self, revision_id=None, pb=None):
2008
self.real_inter.fetch(revision_id=revision_id, pb=pb)
2011
def _get_repo_format_to_test(self):
2015
InterRepository.register_optimiser(InterSameDataRepository)
2016
InterRepository.register_optimiser(InterWeaveRepo)
2017
InterRepository.register_optimiser(InterKnitRepo)
2018
InterRepository.register_optimiser(InterModel1and2)
2019
InterRepository.register_optimiser(InterKnit1and2)
2020
InterRepository.register_optimiser(InterRemoteRepository)
2023
class CopyConverter(object):
2024
"""A repository conversion tool which just performs a copy of the content.
2026
This is slow but quite reliable.
2029
def __init__(self, target_format):
2030
"""Create a CopyConverter.
2032
:param target_format: The format the resulting repository should be.
2034
self.target_format = target_format
2036
def convert(self, repo, pb):
2037
"""Perform the conversion of to_convert, giving feedback via pb.
2039
:param to_convert: The disk object to convert.
2040
:param pb: a progress bar to use for progress information.
2045
# this is only useful with metadir layouts - separated repo content.
2046
# trigger an assertion if not such
2047
repo._format.get_format_string()
2048
self.repo_dir = repo.bzrdir
2049
self.step('Moving repository to repository.backup')
2050
self.repo_dir.transport.move('repository', 'repository.backup')
2051
backup_transport = self.repo_dir.transport.clone('repository.backup')
2052
repo._format.check_conversion_target(self.target_format)
2053
self.source_repo = repo._format.open(self.repo_dir,
2055
_override_transport=backup_transport)
2056
self.step('Creating new repository')
2057
converted = self.target_format.initialize(self.repo_dir,
2058
self.source_repo.is_shared())
2059
converted.lock_write()
2061
self.step('Copying content into repository.')
2062
self.source_repo.copy_content_into(converted)
2065
self.step('Deleting old repository content.')
2066
self.repo_dir.transport.delete_tree('repository.backup')
2067
self.pb.note('repository converted')
2069
def step(self, message):
2070
"""Update the pb by a step."""
2072
self.pb.update(message, self.count, self.total)
2075
class CommitBuilder(object):
2076
"""Provides an interface to build up a commit.
2078
This allows describing a tree to be committed without needing to
2079
know the internals of the format of the repository.
2082
# all clients should supply tree roots.
2083
record_root_entry = True
2085
def __init__(self, repository, parents, config, timestamp=None,
2086
timezone=None, committer=None, revprops=None,
2088
"""Initiate a CommitBuilder.
2090
:param repository: Repository to commit to.
2091
:param parents: Revision ids of the parents of the new revision.
2092
:param config: Configuration to use.
2093
:param timestamp: Optional timestamp recorded for commit.
2094
:param timezone: Optional timezone for timestamp.
2095
:param committer: Optional committer to set for commit.
2096
:param revprops: Optional dictionary of revision properties.
2097
:param revision_id: Optional revision id.
2099
self._config = config
2101
if committer is None:
2102
self._committer = self._config.username()
2104
assert isinstance(committer, basestring), type(committer)
2105
self._committer = committer
2107
self.new_inventory = Inventory(None)
2108
self._new_revision_id = osutils.safe_revision_id(revision_id)
2109
self.parents = parents
2110
self.repository = repository
2113
if revprops is not None:
2114
self._revprops.update(revprops)
2116
if timestamp is None:
2117
timestamp = time.time()
2118
# Restrict resolution to 1ms
2119
self._timestamp = round(timestamp, 3)
2121
if timezone is None:
2122
self._timezone = osutils.local_time_offset()
2124
self._timezone = int(timezone)
2126
self._generate_revision_if_needed()
2128
def commit(self, message):
2129
"""Make the actual commit.
2131
:return: The revision id of the recorded revision.
2133
rev = _mod_revision.Revision(
2134
timestamp=self._timestamp,
2135
timezone=self._timezone,
2136
committer=self._committer,
2138
inventory_sha1=self.inv_sha1,
2139
revision_id=self._new_revision_id,
2140
properties=self._revprops)
2141
rev.parent_ids = self.parents
2142
self.repository.add_revision(self._new_revision_id, rev,
2143
self.new_inventory, self._config)
2144
self.repository.commit_write_group()
2145
return self._new_revision_id
2148
"""Abort the commit that is being built.
2150
self.repository.abort_write_group()
2152
def revision_tree(self):
2153
"""Return the tree that was just committed.
2155
After calling commit() this can be called to get a RevisionTree
2156
representing the newly committed tree. This is preferred to
2157
calling Repository.revision_tree() because that may require
2158
deserializing the inventory, while we already have a copy in
2161
return RevisionTree(self.repository, self.new_inventory,
2162
self._new_revision_id)
2164
def finish_inventory(self):
2165
"""Tell the builder that the inventory is finished."""
2166
if self.new_inventory.root is None:
2167
symbol_versioning.warn('Root entry should be supplied to'
2168
' record_entry_contents, as of bzr 0.10.',
2169
DeprecationWarning, stacklevel=2)
2170
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2171
self.new_inventory.revision_id = self._new_revision_id
2172
self.inv_sha1 = self.repository.add_inventory(
2173
self._new_revision_id,
2178
def _gen_revision_id(self):
2179
"""Return new revision-id."""
2180
return generate_ids.gen_revision_id(self._config.username(),
2183
def _generate_revision_if_needed(self):
2184
"""Create a revision id if None was supplied.
2186
If the repository can not support user-specified revision ids
2187
they should override this function and raise CannotSetRevisionId
2188
if _new_revision_id is not None.
2190
:raises: CannotSetRevisionId
2192
if self._new_revision_id is None:
2193
self._new_revision_id = self._gen_revision_id()
2194
self.random_revid = True
2196
self.random_revid = False
2198
def _check_root(self, ie, parent_invs, tree):
2199
"""Helper for record_entry_contents.
2201
:param ie: An entry being added.
2202
:param parent_invs: The inventories of the parent revisions of the
2204
:param tree: The tree that is being committed.
2206
if ie.parent_id is not None:
2207
# if ie is not root, add a root automatically.
2208
symbol_versioning.warn('Root entry should be supplied to'
2209
' record_entry_contents, as of bzr 0.10.',
2210
DeprecationWarning, stacklevel=2)
2211
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2214
# In this revision format, root entries have no knit or weave When
2215
# serializing out to disk and back in root.revision is always
2217
ie.revision = self._new_revision_id
2219
def record_entry_contents(self, ie, parent_invs, path, tree):
2220
"""Record the content of ie from tree into the commit if needed.
2222
Side effect: sets ie.revision when unchanged
2224
:param ie: An inventory entry present in the commit.
2225
:param parent_invs: The inventories of the parent revisions of the
2227
:param path: The path the entry is at in the tree.
2228
:param tree: The tree which contains this entry and should be used to
2231
if self.new_inventory.root is None:
2232
self._check_root(ie, parent_invs, tree)
2233
self.new_inventory.add(ie)
2235
# ie.revision is always None if the InventoryEntry is considered
2236
# for committing. ie.snapshot will record the correct revision
2237
# which may be the sole parent if it is untouched.
2238
if ie.revision is not None:
2241
parent_candiate_entries = ie.parent_candidates(parent_invs)
2242
heads = self.repository.get_graph().heads(parent_candiate_entries.keys())
2243
# XXX: Note that this is unordered - and this is tolerable because
2244
# the previous code was also unordered.
2245
previous_entries = dict((head, parent_candiate_entries[head]) for head
2247
# we are creating a new revision for ie in the history store and
2249
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2251
def modified_directory(self, file_id, file_parents):
2252
"""Record the presence of a symbolic link.
2254
:param file_id: The file_id of the link to record.
2255
:param file_parents: The per-file parent revision ids.
2257
self._add_text_to_weave(file_id, [], file_parents.keys())
2259
def modified_reference(self, file_id, file_parents):
2260
"""Record the modification of a reference.
2262
:param file_id: The file_id of the link to record.
2263
:param file_parents: The per-file parent revision ids.
2265
self._add_text_to_weave(file_id, [], file_parents.keys())
2267
def modified_file_text(self, file_id, file_parents,
2268
get_content_byte_lines, text_sha1=None,
2270
"""Record the text of file file_id
2272
:param file_id: The file_id of the file to record the text of.
2273
:param file_parents: The per-file parent revision ids.
2274
:param get_content_byte_lines: A callable which will return the byte
2276
:param text_sha1: Optional SHA1 of the file contents.
2277
:param text_size: Optional size of the file contents.
2279
# mutter('storing text of file {%s} in revision {%s} into %r',
2280
# file_id, self._new_revision_id, self.repository.weave_store)
2281
# special case to avoid diffing on renames or
2283
if (len(file_parents) == 1
2284
and text_sha1 == file_parents.values()[0].text_sha1
2285
and text_size == file_parents.values()[0].text_size):
2286
previous_ie = file_parents.values()[0]
2287
versionedfile = self.repository.weave_store.get_weave(file_id,
2288
self.repository.get_transaction())
2289
versionedfile.clone_text(self._new_revision_id,
2290
previous_ie.revision, file_parents.keys())
2291
return text_sha1, text_size
2293
new_lines = get_content_byte_lines()
2294
return self._add_text_to_weave(file_id, new_lines,
2295
file_parents.keys())
2297
def modified_link(self, file_id, file_parents, link_target):
2298
"""Record the presence of a symbolic link.
2300
:param file_id: The file_id of the link to record.
2301
:param file_parents: The per-file parent revision ids.
2302
:param link_target: Target location of this link.
2304
self._add_text_to_weave(file_id, [], file_parents.keys())
2306
def _add_text_to_weave(self, file_id, new_lines, parents):
2307
versionedfile = self.repository.weave_store.get_weave_or_empty(
2308
file_id, self.repository.get_transaction())
2309
# Don't change this to add_lines - add_lines_with_ghosts is cheaper
2310
# than add_lines, and allows committing when a parent is ghosted for
2312
# Note: as we read the content directly from the tree, we know its not
2313
# been turned into unicode or badly split - but a broken tree
2314
# implementation could give us bad output from readlines() so this is
2315
# not a guarantee of safety. What would be better is always checking
2316
# the content during test suite execution. RBC 20070912
2317
result = versionedfile.add_lines_with_ghosts(
2318
self._new_revision_id, parents, new_lines,
2319
random_id=self.random_revid, check_content=False)[0:2]
2320
versionedfile.clear_cache()
2324
class RootCommitBuilder(CommitBuilder):
2325
"""This commitbuilder actually records the root id"""
2327
def _check_root(self, ie, parent_invs, tree):
2328
"""Helper for record_entry_contents.
2330
:param ie: An entry being added.
2331
:param parent_invs: The inventories of the parent revisions of the
2333
:param tree: The tree that is being committed.
2335
# ie must be root for this builder
2336
assert ie.parent_id is None
2348
def _unescaper(match, _map=_unescape_map):
2349
code = match.group(1)
2353
if not code.startswith('#'):
2355
return unichr(int(code[1:])).encode('utf8')
2361
def _unescape_xml(data):
2362
"""Unescape predefined XML entities in a string of data."""
2364
if _unescape_re is None:
2365
_unescape_re = re.compile('\&([^;]*);')
2366
return _unescape_re.sub(_unescaper, data)