1
# Copyright (C) 2005, 2006 Canonical Ltd
4
# Johan Rydberg <jrydberg@gnu.org>
6
# This program is free software; you can redistribute it and/or modify
7
# it under the terms of the GNU General Public License as published by
8
# the Free Software Foundation; either version 2 of the License, or
9
# (at your option) any later version.
11
# This program is distributed in the hope that it will be useful,
12
# but WITHOUT ANY WARRANTY; without even the implied warranty of
13
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14
# GNU General Public License for more details.
16
# You should have received a copy of the GNU General Public License
17
# along with this program; if not, write to the Free Software
18
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20
"""Versioned text file storage api."""
22
from bzrlib.lazy_import import lazy_import
23
lazy_import(globals(), """
24
from copy import deepcopy
32
from bzrlib.transport.memory import MemoryTransport
35
from bzrlib.inter import InterObject
36
from bzrlib.textmerge import TextMerge
37
from bzrlib.symbol_versioning import (deprecated_function,
43
class VersionedFile(object):
44
"""Versioned text file storage.
46
A versioned file manages versions of line-based text files,
47
keeping track of the originating version for each line.
49
To clients the "lines" of the file are represented as a list of
50
strings. These strings will typically have terminal newline
51
characters, but this is not required. In particular files commonly
52
do not have a newline at the end of the file.
54
Texts are identified by a version-id string.
57
def __init__(self, access_mode):
59
self._access_mode = access_mode
61
def copy_to(self, name, transport):
62
"""Copy this versioned file to name on transport."""
63
raise NotImplementedError(self.copy_to)
65
@deprecated_method(zero_eight)
67
"""Return a list of all the versions in this versioned file.
69
Please use versionedfile.versions() now.
71
return self.versions()
74
"""Return a unsorted list of versions."""
75
raise NotImplementedError(self.versions)
77
def has_ghost(self, version_id):
78
"""Returns whether version is present as a ghost."""
79
raise NotImplementedError(self.has_ghost)
81
def has_version(self, version_id):
82
"""Returns whether version is present."""
83
raise NotImplementedError(self.has_version)
85
def add_delta(self, version_id, parents, delta_parent, sha1, noeol, delta):
86
"""Add a text to the versioned file via a pregenerated delta.
88
:param version_id: The version id being added.
89
:param parents: The parents of the version_id.
90
:param delta_parent: The parent this delta was created against.
91
:param sha1: The sha1 of the full text.
92
:param delta: The delta instructions. See get_delta for details.
94
self._check_write_ok()
95
if self.has_version(version_id):
96
raise errors.RevisionAlreadyPresent(version_id, self)
97
return self._add_delta(version_id, parents, delta_parent, sha1, noeol, delta)
99
def _add_delta(self, version_id, parents, delta_parent, sha1, noeol, delta):
100
"""Class specific routine to add a delta.
102
This generic version simply applies the delta to the delta_parent and
105
# strip annotation from delta
107
for start, stop, delta_len, delta_lines in delta:
108
new_delta.append((start, stop, delta_len, [text for origin, text in delta_lines]))
109
if delta_parent is not None:
110
parent_full = self.get_lines(delta_parent)
113
new_full = self._apply_delta(parent_full, new_delta)
114
# its impossible to have noeol on an empty file
115
if noeol and new_full[-1][-1] == '\n':
116
new_full[-1] = new_full[-1][:-1]
117
self.add_lines(version_id, parents, new_full)
119
def add_lines(self, version_id, parents, lines, parent_texts=None):
120
"""Add a single text on top of the versioned file.
122
Must raise RevisionAlreadyPresent if the new version is
123
already present in file history.
125
Must raise RevisionNotPresent if any of the given parents are
126
not present in file history.
127
:param parent_texts: An optional dictionary containing the opaque
128
representations of some or all of the parents of
129
version_id to allow delta optimisations.
130
VERY IMPORTANT: the texts must be those returned
131
by add_lines or data corruption can be caused.
132
:return: An opaque representation of the inserted version which can be
133
provided back to future add_lines calls in the parent_texts
136
self._check_write_ok()
137
return self._add_lines(version_id, parents, lines, parent_texts)
139
def _add_lines(self, version_id, parents, lines, parent_texts):
140
"""Helper to do the class specific add_lines."""
141
raise NotImplementedError(self.add_lines)
143
def add_lines_with_ghosts(self, version_id, parents, lines,
145
"""Add lines to the versioned file, allowing ghosts to be present.
147
This takes the same parameters as add_lines.
149
self._check_write_ok()
150
return self._add_lines_with_ghosts(version_id, parents, lines,
153
def _add_lines_with_ghosts(self, version_id, parents, lines, parent_texts):
154
"""Helper to do class specific add_lines_with_ghosts."""
155
raise NotImplementedError(self.add_lines_with_ghosts)
157
def check(self, progress_bar=None):
158
"""Check the versioned file for integrity."""
159
raise NotImplementedError(self.check)
161
def _check_lines_not_unicode(self, lines):
162
"""Check that lines being added to a versioned file are not unicode."""
164
if line.__class__ is not str:
165
raise errors.BzrBadParameterUnicode("lines")
167
def _check_lines_are_lines(self, lines):
168
"""Check that the lines really are full lines without inline EOL."""
170
if '\n' in line[:-1]:
171
raise errors.BzrBadParameterContainsNewline("lines")
173
def _check_write_ok(self):
174
"""Is the versioned file marked as 'finished' ? Raise if it is."""
176
raise errors.OutSideTransaction()
177
if self._access_mode != 'w':
178
raise errors.ReadOnlyObjectDirtiedError(self)
180
def enable_cache(self):
181
"""Tell this versioned file that it should cache any data it reads.
183
This is advisory, implementations do not have to support caching.
187
def clear_cache(self):
188
"""Remove any data cached in the versioned file object.
190
This only needs to be supported if caches are supported
194
def clone_text(self, new_version_id, old_version_id, parents):
195
"""Add an identical text to old_version_id as new_version_id.
197
Must raise RevisionNotPresent if the old version or any of the
198
parents are not present in file history.
200
Must raise RevisionAlreadyPresent if the new version is
201
already present in file history."""
202
self._check_write_ok()
203
return self._clone_text(new_version_id, old_version_id, parents)
205
def _clone_text(self, new_version_id, old_version_id, parents):
206
"""Helper function to do the _clone_text work."""
207
raise NotImplementedError(self.clone_text)
209
def create_empty(self, name, transport, mode=None):
210
"""Create a new versioned file of this exact type.
212
:param name: the file name
213
:param transport: the transport
214
:param mode: optional file mode.
216
raise NotImplementedError(self.create_empty)
218
def fix_parents(self, version, new_parents):
219
"""Fix the parents list for version.
221
This is done by appending a new version to the index
222
with identical data except for the parents list.
223
the parents list must be a superset of the current
226
self._check_write_ok()
227
return self._fix_parents(version, new_parents)
229
def _fix_parents(self, version, new_parents):
230
"""Helper for fix_parents."""
231
raise NotImplementedError(self.fix_parents)
233
def get_delta(self, version):
234
"""Get a delta for constructing version from some other version.
236
:return: (delta_parent, sha1, noeol, delta)
237
Where delta_parent is a version id or None to indicate no parent.
239
raise NotImplementedError(self.get_delta)
241
def get_deltas(self, versions):
242
"""Get multiple deltas at once for constructing versions.
244
:return: dict(version_id:(delta_parent, sha1, noeol, delta))
245
Where delta_parent is a version id or None to indicate no parent, and
246
version_id is the version_id created by that delta.
249
for version in versions:
250
result[version] = self.get_delta(version)
253
def get_sha1(self, version_id):
254
"""Get the stored sha1 sum for the given revision.
256
:param name: The name of the version to lookup
258
raise NotImplementedError(self.get_sha1)
260
def get_suffixes(self):
261
"""Return the file suffixes associated with this versioned file."""
262
raise NotImplementedError(self.get_suffixes)
264
def get_text(self, version_id):
265
"""Return version contents as a text string.
267
Raises RevisionNotPresent if version is not present in
270
return ''.join(self.get_lines(version_id))
271
get_string = get_text
273
def get_texts(self, version_ids):
274
"""Return the texts of listed versions as a list of strings.
276
Raises RevisionNotPresent if version is not present in
279
return [''.join(self.get_lines(v)) for v in version_ids]
281
def get_lines(self, version_id):
282
"""Return version contents as a sequence of lines.
284
Raises RevisionNotPresent if version is not present in
287
raise NotImplementedError(self.get_lines)
289
def get_ancestry(self, version_ids):
290
"""Return a list of all ancestors of given version(s). This
291
will not include the null revision.
293
Must raise RevisionNotPresent if any of the given versions are
294
not present in file history."""
295
if isinstance(version_ids, basestring):
296
version_ids = [version_ids]
297
raise NotImplementedError(self.get_ancestry)
299
def get_ancestry_with_ghosts(self, version_ids):
300
"""Return a list of all ancestors of given version(s). This
301
will not include the null revision.
303
Must raise RevisionNotPresent if any of the given versions are
304
not present in file history.
306
Ghosts that are known about will be included in ancestry list,
307
but are not explicitly marked.
309
raise NotImplementedError(self.get_ancestry_with_ghosts)
311
def get_graph(self, version_ids=None):
312
"""Return a graph from the versioned file.
314
Ghosts are not listed or referenced in the graph.
315
:param version_ids: Versions to select.
316
None means retrieve all versions.
319
if version_ids is None:
320
for version in self.versions():
321
result[version] = self.get_parents(version)
323
pending = set(version_ids)
325
version = pending.pop()
326
if version in result:
328
parents = self.get_parents(version)
329
for parent in parents:
333
result[version] = parents
336
def get_graph_with_ghosts(self):
337
"""Return a graph for the entire versioned file.
339
Ghosts are referenced in parents list but are not
342
raise NotImplementedError(self.get_graph_with_ghosts)
344
@deprecated_method(zero_eight)
345
def parent_names(self, version):
346
"""Return version names for parents of a version.
348
See get_parents for the current api.
350
return self.get_parents(version)
352
def get_parents(self, version_id):
353
"""Return version names for parents of a version.
355
Must raise RevisionNotPresent if version is not present in
358
raise NotImplementedError(self.get_parents)
360
def get_parents_with_ghosts(self, version_id):
361
"""Return version names for parents of version_id.
363
Will raise RevisionNotPresent if version_id is not present
366
Ghosts that are known about will be included in the parent list,
367
but are not explicitly marked.
369
raise NotImplementedError(self.get_parents_with_ghosts)
371
def annotate_iter(self, version_id):
372
"""Yield list of (version-id, line) pairs for the specified
375
Must raise RevisionNotPresent if any of the given versions are
376
not present in file history.
378
raise NotImplementedError(self.annotate_iter)
380
def annotate(self, version_id):
381
return list(self.annotate_iter(version_id))
383
def _apply_delta(self, lines, delta):
384
"""Apply delta to lines."""
387
for start, end, count, delta_lines in delta:
388
lines[offset+start:offset+end] = delta_lines
389
offset = offset + (start - end) + count
392
def join(self, other, pb=None, msg=None, version_ids=None,
393
ignore_missing=False):
394
"""Integrate versions from other into this versioned file.
396
If version_ids is None all versions from other should be
397
incorporated into this versioned file.
399
Must raise RevisionNotPresent if any of the specified versions
400
are not present in the other files history unless ignore_missing
401
is supplied when they are silently skipped.
403
self._check_write_ok()
404
return InterVersionedFile.get(other, self).join(
410
def iter_lines_added_or_present_in_versions(self, version_ids=None,
412
"""Iterate over the lines in the versioned file from version_ids.
414
This may return lines from other versions, and does not return the
415
specific version marker at this point. The api may be changed
416
during development to include the version that the versioned file
417
thinks is relevant, but given that such hints are just guesses,
418
its better not to have it if we don't need it.
420
If a progress bar is supplied, it may be used to indicate progress.
421
The caller is responsible for cleaning up progress bars (because this
424
NOTES: Lines are normalised: they will all have \n terminators.
425
Lines are returned in arbitrary order.
427
raise NotImplementedError(self.iter_lines_added_or_present_in_versions)
429
def transaction_finished(self):
430
"""The transaction that this file was opened in has finished.
432
This records self.finished = True and should cause all mutating
437
@deprecated_method(zero_eight)
438
def walk(self, version_ids=None):
439
"""Walk the versioned file as a weave-like structure, for
440
versions relative to version_ids. Yields sequence of (lineno,
441
insert, deletes, text) for each relevant line.
443
Must raise RevisionNotPresent if any of the specified versions
444
are not present in the file history.
446
:param version_ids: the version_ids to walk with respect to. If not
447
supplied the entire weave-like structure is walked.
449
walk is deprecated in favour of iter_lines_added_or_present_in_versions
451
raise NotImplementedError(self.walk)
453
@deprecated_method(zero_eight)
454
def iter_names(self):
455
"""Walk the names list."""
456
return iter(self.versions())
458
def plan_merge(self, ver_a, ver_b):
459
"""Return pseudo-annotation indicating how the two versions merge.
461
This is computed between versions a and b and their common
464
Weave lines present in none of them are skipped entirely.
467
killed-base Dead in base revision
468
killed-both Killed in each revision
471
unchanged Alive in both a and b (possibly created in both)
474
ghost-a Killed in a, unborn in b
475
ghost-b Killed in b, unborn in a
476
irrelevant Not in either revision
478
raise NotImplementedError(VersionedFile.plan_merge)
480
def weave_merge(self, plan, a_marker=TextMerge.A_MARKER,
481
b_marker=TextMerge.B_MARKER):
482
return PlanWeaveMerge(plan, a_marker, b_marker).merge_lines()[0]
485
class PlanWeaveMerge(TextMerge):
486
"""Weave merge that takes a plan as its input.
488
This exists so that VersionedFile.plan_merge is implementable.
489
Most callers will want to use WeaveMerge instead.
492
def __init__(self, plan, a_marker=TextMerge.A_MARKER,
493
b_marker=TextMerge.B_MARKER):
494
TextMerge.__init__(self, a_marker, b_marker)
497
def _merge_struct(self):
502
def outstanding_struct():
503
if not lines_a and not lines_b:
505
elif ch_a and not ch_b:
508
elif ch_b and not ch_a:
510
elif lines_a == lines_b:
513
yield (lines_a, lines_b)
515
# We previously considered either 'unchanged' or 'killed-both' lines
516
# to be possible places to resynchronize. However, assuming agreement
517
# on killed-both lines may be too aggressive. -- mbp 20060324
518
for state, line in self.plan:
519
if state == 'unchanged':
520
# resync and flush queued conflicts changes if any
521
for struct in outstanding_struct():
527
if state == 'unchanged':
530
elif state == 'killed-a':
533
elif state == 'killed-b':
536
elif state == 'new-a':
539
elif state == 'new-b':
543
assert state in ('irrelevant', 'ghost-a', 'ghost-b',
544
'killed-base', 'killed-both'), state
545
for struct in outstanding_struct():
549
class WeaveMerge(PlanWeaveMerge):
550
"""Weave merge that takes a VersionedFile and two versions as its input"""
552
def __init__(self, versionedfile, ver_a, ver_b,
553
a_marker=PlanWeaveMerge.A_MARKER, b_marker=PlanWeaveMerge.B_MARKER):
554
plan = versionedfile.plan_merge(ver_a, ver_b)
555
PlanWeaveMerge.__init__(self, plan, a_marker, b_marker)
558
class InterVersionedFile(InterObject):
559
"""This class represents operations taking place between two versionedfiles..
561
Its instances have methods like join, and contain
562
references to the source and target versionedfiles these operations can be
565
Often we will provide convenience methods on 'versionedfile' which carry out
566
operations with another versionedfile - they will always forward to
567
InterVersionedFile.get(other).method_name(parameters).
571
"""The available optimised InterVersionedFile types."""
573
def join(self, pb=None, msg=None, version_ids=None, ignore_missing=False):
574
"""Integrate versions from self.source into self.target.
576
If version_ids is None all versions from source should be
577
incorporated into this versioned file.
579
Must raise RevisionNotPresent if any of the specified versions
580
are not present in the other files history unless ignore_missing is
581
supplied when they are silently skipped.
584
# - if the target is empty, just add all the versions from
585
# source to target, otherwise:
586
# - make a temporary versioned file of type target
587
# - insert the source content into it one at a time
589
if not self.target.versions():
592
# Make a new target-format versioned file.
593
temp_source = self.target.create_empty("temp", MemoryTransport())
595
version_ids = self._get_source_version_ids(version_ids, ignore_missing)
596
graph = self.source.get_graph(version_ids)
597
order = tsort.topo_sort(graph.items())
598
pb = ui.ui_factory.nested_progress_bar()
601
# TODO for incremental cross-format work:
602
# make a versioned file with the following content:
603
# all revisions we have been asked to join
604
# all their ancestors that are *not* in target already.
605
# the immediate parents of the above two sets, with
606
# empty parent lists - these versions are in target already
607
# and the incorrect version data will be ignored.
608
# TODO: for all ancestors that are present in target already,
609
# check them for consistent data, this requires moving sha1 from
611
# TODO: remove parent texts when they are not relevant any more for
612
# memory pressure reduction. RBC 20060313
613
# pb.update('Converting versioned data', 0, len(order))
614
# deltas = self.source.get_deltas(order)
615
for index, version in enumerate(order):
616
pb.update('Converting versioned data', index, len(order))
617
parent_text = target.add_lines(version,
618
self.source.get_parents(version),
619
self.source.get_lines(version),
620
parent_texts=parent_texts)
621
parent_texts[version] = parent_text
622
#delta_parent, sha1, noeol, delta = deltas[version]
623
#target.add_delta(version,
624
# self.source.get_parents(version),
629
#target.get_lines(version)
631
# this should hit the native code path for target
632
if target is not self.target:
633
return self.target.join(temp_source,
641
def _get_source_version_ids(self, version_ids, ignore_missing):
642
"""Determine the version ids to be used from self.source.
644
:param version_ids: The caller-supplied version ids to check. (None
645
for all). If None is in version_ids, it is stripped.
646
:param ignore_missing: if True, remove missing ids from the version
647
list. If False, raise RevisionNotPresent on
648
a missing version id.
649
:return: A set of version ids.
651
if version_ids is None:
652
# None cannot be in source.versions
653
return set(self.source.versions())
656
return set(self.source.versions()).intersection(set(version_ids))
658
new_version_ids = set()
659
for version in version_ids:
662
if not self.source.has_version(version):
663
raise errors.RevisionNotPresent(version, str(self.source))
665
new_version_ids.add(version)
666
return new_version_ids
669
class InterVersionedFileTestProviderAdapter(object):
670
"""A tool to generate a suite testing multiple inter versioned-file classes.
672
This is done by copying the test once for each InterVersionedFile provider
673
and injecting the transport_server, transport_readonly_server,
674
versionedfile_factory and versionedfile_factory_to classes into each copy.
675
Each copy is also given a new id() to make it easy to identify.
678
def __init__(self, transport_server, transport_readonly_server, formats):
679
self._transport_server = transport_server
680
self._transport_readonly_server = transport_readonly_server
681
self._formats = formats
683
def adapt(self, test):
684
result = unittest.TestSuite()
685
for (interversionedfile_class,
686
versionedfile_factory,
687
versionedfile_factory_to) in self._formats:
688
new_test = deepcopy(test)
689
new_test.transport_server = self._transport_server
690
new_test.transport_readonly_server = self._transport_readonly_server
691
new_test.interversionedfile_class = interversionedfile_class
692
new_test.versionedfile_factory = versionedfile_factory
693
new_test.versionedfile_factory_to = versionedfile_factory_to
694
def make_new_test_id():
695
new_id = "%s(%s)" % (new_test.id(), interversionedfile_class.__name__)
696
return lambda: new_id
697
new_test.id = make_new_test_id()
698
result.addTest(new_test)
702
def default_test_list():
703
"""Generate the default list of interversionedfile permutations to test."""
704
from bzrlib.weave import WeaveFile
705
from bzrlib.knit import KnitVersionedFile
707
# test the fallback InterVersionedFile from annotated knits to weave
708
result.append((InterVersionedFile,
711
for optimiser in InterVersionedFile._optimisers:
712
result.append((optimiser,
713
optimiser._matching_file_from_factory,
714
optimiser._matching_file_to_factory
716
# if there are specific combinations we want to use, we can add them
added
removed
bzrlib/versionedfile.py
