33
from bzrlib.graph import Graph
34
33
from bzrlib.transport.memory import MemoryTransport
37
36
from cStringIO import StringIO
39
38
from bzrlib.inter import InterObject
40
from bzrlib.registry import Registry
41
from bzrlib.symbol_versioning import *
42
39
from bzrlib.textmerge import TextMerge
45
adapter_registry = Registry()
46
adapter_registry.register_lazy(('knit-delta-gz', 'fulltext'), 'bzrlib.knit',
47
'DeltaPlainToFullText')
48
adapter_registry.register_lazy(('knit-ft-gz', 'fulltext'), 'bzrlib.knit',
50
adapter_registry.register_lazy(('knit-annotated-delta-gz', 'knit-delta-gz'),
51
'bzrlib.knit', 'DeltaAnnotatedToUnannotated')
52
adapter_registry.register_lazy(('knit-annotated-delta-gz', 'fulltext'),
53
'bzrlib.knit', 'DeltaAnnotatedToFullText')
54
adapter_registry.register_lazy(('knit-annotated-ft-gz', 'knit-ft-gz'),
55
'bzrlib.knit', 'FTAnnotatedToUnannotated')
56
adapter_registry.register_lazy(('knit-annotated-ft-gz', 'fulltext'),
57
'bzrlib.knit', 'FTAnnotatedToFullText')
60
class ContentFactory(object):
61
"""Abstract interface for insertion and retrieval from a VersionedFile.
63
:ivar sha1: None, or the sha1 of the content fulltext.
64
:ivar storage_kind: The native storage kind of this factory. One of
65
'mpdiff', 'knit-annotated-ft', 'knit-annotated-delta', 'knit-ft',
66
'knit-delta', 'fulltext', 'knit-annotated-ft-gz',
67
'knit-annotated-delta-gz', 'knit-ft-gz', 'knit-delta-gz'.
68
:ivar key: The key of this content. Each key is a tuple with a single
70
:ivar parents: A tuple of parent keys for self.key. If the object has
71
no parent information, None (as opposed to () for an empty list of
76
"""Create a ContentFactory."""
78
self.storage_kind = None
83
class AbsentContentFactory(object):
84
"""A placeholder content factory for unavailable texts.
87
:ivar storage_kind: 'absent'.
88
:ivar key: The key of this content. Each key is a tuple with a single
93
def __init__(self, key):
94
"""Create a ContentFactory."""
96
self.storage_kind = 'absent'
101
def filter_absent(record_stream):
102
"""Adapt a record stream to remove absent records."""
103
for record in record_stream:
104
if record.storage_kind != 'absent':
108
42
class VersionedFile(object):
109
43
"""Versioned text file storage.
127
65
"""Copy this versioned file to name on transport."""
128
66
raise NotImplementedError(self.copy_to)
130
def get_record_stream(self, versions, ordering, include_delta_closure):
131
"""Get a stream of records for versions.
69
"""Return a unsorted list of versions."""
70
raise NotImplementedError(self.versions)
133
:param versions: The versions to include. Each version is a tuple
135
:param ordering: Either 'unordered' or 'topological'. A topologically
136
sorted stream has compression parents strictly before their
138
:param include_delta_closure: If True then the closure across any
139
compression parents will be included (in the data content of the
140
stream, not in the emitted records). This guarantees that
141
'fulltext' can be used successfully on every record.
142
:return: An iterator of ContentFactory objects, each of which is only
143
valid until the iterator is advanced.
145
raise NotImplementedError(self.get_record_stream)
72
def has_ghost(self, version_id):
73
"""Returns whether version is present as a ghost."""
74
raise NotImplementedError(self.has_ghost)
147
76
def has_version(self, version_id):
148
77
"""Returns whether version is present."""
149
78
raise NotImplementedError(self.has_version)
151
def insert_record_stream(self, stream):
152
"""Insert a record stream into this versioned file.
154
:param stream: A stream of records to insert.
156
:seealso VersionedFile.get_record_stream:
158
raise NotImplementedError
160
80
def add_lines(self, version_id, parents, lines, parent_texts=None,
161
81
left_matching_blocks=None, nostore_sha=None, random_id=False,
162
82
check_content=True):
209
129
def add_lines_with_ghosts(self, version_id, parents, lines,
210
130
parent_texts=None, nostore_sha=None, random_id=False,
211
check_content=True, left_matching_blocks=None):
212
132
"""Add lines to the versioned file, allowing ghosts to be present.
214
134
This takes the same parameters as add_lines and returns the same.
216
136
self._check_write_ok()
217
137
return self._add_lines_with_ghosts(version_id, parents, lines,
218
parent_texts, nostore_sha, random_id, check_content, left_matching_blocks)
138
parent_texts, nostore_sha, random_id, check_content)
220
140
def _add_lines_with_ghosts(self, version_id, parents, lines, parent_texts,
221
nostore_sha, random_id, check_content, left_matching_blocks):
141
nostore_sha, random_id, check_content):
222
142
"""Helper to do class specific add_lines_with_ghosts."""
223
143
raise NotImplementedError(self.add_lines_with_ghosts)
238
158
if '\n' in line[:-1]:
239
159
raise errors.BzrBadParameterContainsNewline("lines")
161
def _check_write_ok(self):
162
"""Is the versioned file marked as 'finished' ? Raise if it is."""
164
raise errors.OutSideTransaction()
165
if self._access_mode != 'w':
166
raise errors.ReadOnlyObjectDirtiedError(self)
168
def enable_cache(self):
169
"""Tell this versioned file that it should cache any data it reads.
171
This is advisory, implementations do not have to support caching.
175
def clear_cache(self):
176
"""Remove any data cached in the versioned file object.
178
This only needs to be supported if caches are supported
182
def clone_text(self, new_version_id, old_version_id, parents):
183
"""Add an identical text to old_version_id as new_version_id.
185
Must raise RevisionNotPresent if the old version or any of the
186
parents are not present in file history.
188
Must raise RevisionAlreadyPresent if the new version is
189
already present in file history."""
190
self._check_write_ok()
191
return self._clone_text(new_version_id, old_version_id, parents)
193
def _clone_text(self, new_version_id, old_version_id, parents):
194
"""Helper function to do the _clone_text work."""
195
raise NotImplementedError(self.clone_text)
197
def create_empty(self, name, transport, mode=None):
198
"""Create a new versioned file of this exact type.
200
:param name: the file name
201
:param transport: the transport
202
:param mode: optional file mode.
204
raise NotImplementedError(self.create_empty)
241
206
def get_format_signature(self):
242
207
"""Get a text description of the data encoding in this file.
248
213
def make_mpdiffs(self, version_ids):
249
214
"""Create multiparent diffs for specified versions."""
250
215
knit_versions = set()
251
knit_versions.update(version_ids)
252
parent_map = self.get_parent_map(version_ids)
253
216
for version_id in version_ids:
255
knit_versions.update(parent_map[version_id])
257
raise RevisionNotPresent(version_id, self)
258
# We need to filter out ghosts, because we can't diff against them.
259
knit_versions = set(self.get_parent_map(knit_versions).keys())
217
knit_versions.add(version_id)
218
knit_versions.update(self.get_parents(version_id))
260
219
lines = dict(zip(knit_versions,
261
220
self._get_lf_split_line_list(knit_versions)))
263
222
for version_id in version_ids:
264
223
target = lines[version_id]
266
parents = [lines[p] for p in parent_map[version_id] if p in
269
raise RevisionNotPresent(version_id, self)
224
parents = [lines[p] for p in self.get_parents(version_id)]
270
225
if len(parents) > 0:
271
226
left_parent_blocks = self._extract_blocks(version_id,
272
227
parents[0], target)
307
261
mpvf.get_diff(parent_ids[0]).num_lines()))
309
263
left_matching_blocks = None
311
_, _, version_text = self.add_lines_with_ghosts(version,
312
parent_ids, lines, vf_parents,
313
left_matching_blocks=left_matching_blocks)
314
except NotImplementedError:
315
# The vf can't handle ghosts, so add lines normally, which will
316
# (reasonably) fail if there are ghosts in the data.
317
_, _, version_text = self.add_lines(version,
318
parent_ids, lines, vf_parents,
319
left_matching_blocks=left_matching_blocks)
264
_, _, version_text = self.add_lines(version, parent_ids, lines,
265
vf_parents, left_matching_blocks=left_matching_blocks)
320
266
vf_parents[version] = version_text
321
267
for (version, parent_ids, expected_sha1, mpdiff), sha1 in\
322
268
zip(records, self.get_sha1s(versions)):
323
269
if expected_sha1 != sha1:
324
270
raise errors.VersionedFileInvalidChecksum(version)
272
def get_sha1(self, version_id):
273
"""Get the stored sha1 sum for the given revision.
275
:param version_id: The name of the version to lookup
277
raise NotImplementedError(self.get_sha1)
326
279
def get_sha1s(self, version_ids):
327
280
"""Get the stored sha1 sums for the given revisions.
383
340
but are not explicitly marked.
385
342
raise NotImplementedError(self.get_ancestry_with_ghosts)
387
def get_parent_map(self, version_ids):
388
"""Get a map of the parents of version_ids.
390
:param version_ids: The version ids to look up parents for.
391
:return: A mapping from version id to parents.
393
raise NotImplementedError(self.get_parent_map)
344
def get_graph(self, version_ids=None):
345
"""Return a graph from the versioned file.
347
Ghosts are not listed or referenced in the graph.
348
:param version_ids: Versions to select.
349
None means retrieve all versions.
351
if version_ids is None:
352
return dict(self.iter_parents(self.versions()))
354
pending = set(version_ids)
356
this_iteration = pending
358
for version, parents in self.iter_parents(this_iteration):
359
result[version] = parents
360
for parent in parents:
366
def get_graph_with_ghosts(self):
367
"""Return a graph for the entire versioned file.
369
Ghosts are referenced in parents list but are not
372
raise NotImplementedError(self.get_graph_with_ghosts)
374
def get_parents(self, version_id):
375
"""Return version names for parents of a version.
377
Must raise RevisionNotPresent if version is not present in
380
raise NotImplementedError(self.get_parents)
395
382
def get_parents_with_ghosts(self, version_id):
396
383
"""Return version names for parents of version_id.
401
388
Ghosts that are known about will be included in the parent list,
402
389
but are not explicitly marked.
405
return list(self.get_parent_map([version_id])[version_id])
407
raise errors.RevisionNotPresent(version_id, self)
391
raise NotImplementedError(self.get_parents_with_ghosts)
393
def annotate_iter(self, version_id):
394
"""Yield list of (version-id, line) pairs for the specified
397
Must raise RevisionNotPresent if the given version is
398
not present in file history.
400
raise NotImplementedError(self.annotate_iter)
409
402
def annotate(self, version_id):
410
"""Return a list of (version-id, line) tuples for version_id.
412
:raise RevisionNotPresent: If the given version is
413
not present in file history.
415
raise NotImplementedError(self.annotate)
417
@deprecated_method(one_five)
403
return list(self.annotate_iter(version_id))
418
405
def join(self, other, pb=None, msg=None, version_ids=None,
419
406
ignore_missing=False):
420
407
"""Integrate versions from other into this versioned file.
436
def iter_lines_added_or_present_in_versions(self, version_ids=None,
423
def iter_lines_added_or_present_in_versions(self, version_ids=None,
438
425
"""Iterate over the lines in the versioned file from version_ids.
440
This may return lines from other versions. Each item the returned
441
iterator yields is a tuple of a line and a text version that that line
442
is present in (not introduced in).
444
Ordering of results is in whatever order is most suitable for the
445
underlying storage format.
427
This may return lines from other versions, and does not return the
428
specific version marker at this point. The api may be changed
429
during development to include the version that the versioned file
430
thinks is relevant, but given that such hints are just guesses,
431
its better not to have it if we don't need it.
447
433
If a progress bar is supplied, it may be used to indicate progress.
448
434
The caller is responsible for cleaning up progress bars (because this
451
437
NOTES: Lines are normalised: they will all have \n terminators.
452
438
Lines are returned in arbitrary order.
454
:return: An iterator over (line, version_id).
456
440
raise NotImplementedError(self.iter_lines_added_or_present_in_versions)
442
def iter_parents(self, version_ids):
443
"""Iterate through the parents for many version ids.
445
:param version_ids: An iterable yielding version_ids.
446
:return: An iterator that yields (version_id, parents). Requested
447
version_ids not present in the versioned file are simply skipped.
448
The order is undefined, allowing for different optimisations in
449
the underlying implementation.
451
for version_id in version_ids:
453
yield version_id, tuple(self.get_parents(version_id))
454
except errors.RevisionNotPresent:
457
def transaction_finished(self):
458
"""The transaction that this file was opened in has finished.
460
This records self.finished = True and should cause all mutating
458
465
def plan_merge(self, ver_a, ver_b):
459
466
"""Return pseudo-annotation indicating how the two versions merge.
482
489
return PlanWeaveMerge(plan, a_marker, b_marker).merge_lines()[0]
485
class RecordingVersionedFileDecorator(object):
486
"""A minimal versioned file that records calls made on it.
488
Only enough methods have been added to support tests using it to date.
490
:ivar calls: A list of the calls made; can be reset at any time by
494
def __init__(self, backing_vf):
495
"""Create a RecordingVersionedFileDecorator decorating backing_vf.
497
:param backing_vf: The versioned file to answer all methods.
499
self._backing_vf = backing_vf
502
def get_lines(self, version_ids):
503
self.calls.append(("get_lines", version_ids))
504
return self._backing_vf.get_lines(version_ids)
507
class _PlanMergeVersionedFile(object):
508
"""A VersionedFile for uncommitted and committed texts.
510
It is intended to allow merges to be planned with working tree texts.
511
It implements only the small part of the VersionedFile interface used by
512
PlanMerge. It falls back to multiple versionedfiles for data not stored in
513
_PlanMergeVersionedFile itself.
516
def __init__(self, file_id, fallback_versionedfiles=None):
519
:param file_id: Used when raising exceptions.
520
:param fallback_versionedfiles: If supplied, the set of fallbacks to
521
use. Otherwise, _PlanMergeVersionedFile.fallback_versionedfiles
522
can be appended to later.
524
self._file_id = file_id
525
if fallback_versionedfiles is None:
526
self.fallback_versionedfiles = []
528
self.fallback_versionedfiles = fallback_versionedfiles
532
def plan_merge(self, ver_a, ver_b, base=None):
533
"""See VersionedFile.plan_merge"""
534
from bzrlib.merge import _PlanMerge
536
return _PlanMerge(ver_a, ver_b, self).plan_merge()
537
old_plan = list(_PlanMerge(ver_a, base, self).plan_merge())
538
new_plan = list(_PlanMerge(ver_a, ver_b, self).plan_merge())
539
return _PlanMerge._subtract_plans(old_plan, new_plan)
541
def plan_lca_merge(self, ver_a, ver_b, base=None):
542
from bzrlib.merge import _PlanLCAMerge
543
graph = self._get_graph()
544
new_plan = _PlanLCAMerge(ver_a, ver_b, self, graph).plan_merge()
547
old_plan = _PlanLCAMerge(ver_a, base, self, graph).plan_merge()
548
return _PlanLCAMerge._subtract_plans(list(old_plan), list(new_plan))
550
def add_lines(self, version_id, parents, lines):
551
"""See VersionedFile.add_lines
553
Lines are added locally, not fallback versionedfiles. Also, ghosts are
554
permitted. Only reserved ids are permitted.
556
if not revision.is_reserved_id(version_id):
557
raise ValueError('Only reserved ids may be used')
559
raise ValueError('Parents may not be None')
561
raise ValueError('Lines may not be None')
562
self._parents[version_id] = tuple(parents)
563
self._lines[version_id] = lines
565
def get_lines(self, version_id):
566
"""See VersionedFile.get_ancestry"""
567
lines = self._lines.get(version_id)
568
if lines is not None:
570
for versionedfile in self.fallback_versionedfiles:
572
return versionedfile.get_lines(version_id)
573
except errors.RevisionNotPresent:
576
raise errors.RevisionNotPresent(version_id, self._file_id)
578
def get_ancestry(self, version_id, topo_sorted=False):
579
"""See VersionedFile.get_ancestry.
581
Note that this implementation assumes that if a VersionedFile can
582
answer get_ancestry at all, it can give an authoritative answer. In
583
fact, ghosts can invalidate this assumption. But it's good enough
584
99% of the time, and far cheaper/simpler.
586
Also note that the results of this version are never topologically
587
sorted, and are a set.
590
raise ValueError('This implementation does not provide sorting')
591
parents = self._parents.get(version_id)
593
for vf in self.fallback_versionedfiles:
595
return vf.get_ancestry(version_id, topo_sorted=False)
596
except errors.RevisionNotPresent:
599
raise errors.RevisionNotPresent(version_id, self._file_id)
600
ancestry = set([version_id])
601
for parent in parents:
602
ancestry.update(self.get_ancestry(parent, topo_sorted=False))
605
def get_parent_map(self, version_ids):
606
"""See VersionedFile.get_parent_map"""
608
pending = set(version_ids)
609
for key in version_ids:
611
result[key] = self._parents[key]
614
pending = pending - set(result.keys())
615
for versionedfile in self.fallback_versionedfiles:
616
parents = versionedfile.get_parent_map(pending)
617
result.update(parents)
618
pending = pending - set(parents.keys())
623
def _get_graph(self):
624
from bzrlib.graph import (
627
_StackedParentsProvider,
629
from bzrlib.repofmt.knitrepo import _KnitParentsProvider
630
parent_providers = [DictParentsProvider(self._parents)]
631
for vf in self.fallback_versionedfiles:
632
parent_providers.append(_KnitParentsProvider(vf))
633
return Graph(_StackedParentsProvider(parent_providers))
636
492
class PlanWeaveMerge(TextMerge):
637
493
"""Weave merge that takes a plan as its input.
738
587
are not present in the other file's history unless ignore_missing is
739
588
supplied in which case they are silently skipped.
591
# - if the target is empty, just add all the versions from
592
# source to target, otherwise:
593
# - make a temporary versioned file of type target
594
# - insert the source content into it one at a time
596
if not self.target.versions():
599
# Make a new target-format versioned file.
600
temp_source = self.target.create_empty("temp", MemoryTransport())
742
602
version_ids = self._get_source_version_ids(version_ids, ignore_missing)
743
graph = Graph(self.source)
744
search = graph._make_breadth_first_searcher(version_ids)
745
transitive_ids = set()
746
map(transitive_ids.update, list(search))
747
parent_map = self.source.get_parent_map(transitive_ids)
748
order = tsort.topo_sort(parent_map.items())
603
graph = self.source.get_graph(version_ids)
604
order = tsort.topo_sort(graph.items())
749
605
pb = ui.ui_factory.nested_progress_bar()
750
606
parent_texts = {}
added
removed
bzrlib/versionedfile.py
