1
# Copyright (C) 2005, 2006 by 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."""
23
from copy import deepcopy
24
from unittest import TestSuite
27
import bzrlib.errors as errors
28
from bzrlib.inter import InterObject
29
from bzrlib.textmerge import TextMerge
30
from bzrlib.transport.memory import MemoryTransport
31
from bzrlib.tsort import topo_sort
33
from bzrlib.symbol_versioning import (deprecated_function,
39
class VersionedFile(object):
40
"""Versioned text file storage.
42
A versioned file manages versions of line-based text files,
43
keeping track of the originating version for each line.
45
To clients the "lines" of the file are represented as a list of
46
strings. These strings will typically have terminal newline
47
characters, but this is not required. In particular files commonly
48
do not have a newline at the end of the file.
50
Texts are identified by a version-id string.
53
def __init__(self, access_mode):
55
self._access_mode = access_mode
57
def copy_to(self, name, transport):
58
"""Copy this versioned file to name on transport."""
59
raise NotImplementedError(self.copy_to)
61
@deprecated_method(zero_eight)
63
"""Return a list of all the versions in this versioned file.
65
Please use versionedfile.versions() now.
67
return self.versions()
70
"""Return a unsorted list of versions."""
71
raise NotImplementedError(self.versions)
73
def has_ghost(self, version_id):
74
"""Returns whether version is present as a ghost."""
75
raise NotImplementedError(self.has_ghost)
77
def has_version(self, version_id):
78
"""Returns whether version is present."""
79
raise NotImplementedError(self.has_version)
81
def add_delta(self, version_id, parents, delta_parent, sha1, noeol, delta):
82
"""Add a text to the versioned file via a pregenerated delta.
84
:param version_id: The version id being added.
85
:param parents: The parents of the version_id.
86
:param delta_parent: The parent this delta was created against.
87
:param sha1: The sha1 of the full text.
88
:param delta: The delta instructions. See get_delta for details.
90
self._check_write_ok()
91
if self.has_version(version_id):
92
raise errors.RevisionAlreadyPresent(version_id, self)
93
return self._add_delta(version_id, parents, delta_parent, sha1, noeol, delta)
95
def _add_delta(self, version_id, parents, delta_parent, sha1, noeol, delta):
96
"""Class specific routine to add a delta.
98
This generic version simply applies the delta to the delta_parent and
101
# strip annotation from delta
103
for start, stop, delta_len, delta_lines in delta:
104
new_delta.append((start, stop, delta_len, [text for origin, text in delta_lines]))
105
if delta_parent is not None:
106
parent_full = self.get_lines(delta_parent)
109
new_full = self._apply_delta(parent_full, new_delta)
110
# its impossible to have noeol on an empty file
111
if noeol and new_full[-1][-1] == '\n':
112
new_full[-1] = new_full[-1][:-1]
113
self.add_lines(version_id, parents, new_full)
115
def add_lines(self, version_id, parents, lines, parent_texts=None):
116
"""Add a single text on top of the versioned file.
118
Must raise RevisionAlreadyPresent if the new version is
119
already present in file history.
121
Must raise RevisionNotPresent if any of the given parents are
122
not present in file history.
123
:param parent_texts: An optional dictionary containing the opaque
124
representations of some or all of the parents of
125
version_id to allow delta optimisations.
126
VERY IMPORTANT: the texts must be those returned
127
by add_lines or data corruption can be caused.
128
:return: An opaque representation of the inserted version which can be
129
provided back to future add_lines calls in the parent_texts
132
self._check_write_ok()
133
return self._add_lines(version_id, parents, lines, parent_texts)
135
def _add_lines(self, version_id, parents, lines, parent_texts):
136
"""Helper to do the class specific add_lines."""
137
raise NotImplementedError(self.add_lines)
139
def add_lines_with_ghosts(self, version_id, parents, lines,
141
"""Add lines to the versioned file, allowing ghosts to be present.
143
This takes the same parameters as add_lines.
145
self._check_write_ok()
146
return self._add_lines_with_ghosts(version_id, parents, lines,
149
def _add_lines_with_ghosts(self, version_id, parents, lines, parent_texts):
150
"""Helper to do class specific add_lines_with_ghosts."""
151
raise NotImplementedError(self.add_lines_with_ghosts)
153
def check(self, progress_bar=None):
154
"""Check the versioned file for integrity."""
155
raise NotImplementedError(self.check)
157
def _check_lines_not_unicode(self, lines):
158
"""Check that lines being added to a versioned file are not unicode."""
160
if line.__class__ is not str:
161
raise errors.BzrBadParameterUnicode("lines")
163
def _check_lines_are_lines(self, lines):
164
"""Check that the lines really are full lines without inline EOL."""
166
if '\n' in line[:-1]:
167
raise errors.BzrBadParameterContainsNewline("lines")
169
def _check_write_ok(self):
170
"""Is the versioned file marked as 'finished' ? Raise if it is."""
172
raise errors.OutSideTransaction()
173
if self._access_mode != 'w':
174
raise errors.ReadOnlyObjectDirtiedError(self)
176
def clear_cache(self):
177
"""Remove any data cached in the versioned file object."""
179
def clone_text(self, new_version_id, old_version_id, parents):
180
"""Add an identical text to old_version_id as new_version_id.
182
Must raise RevisionNotPresent if the old version or any of the
183
parents are not present in file history.
185
Must raise RevisionAlreadyPresent if the new version is
186
already present in file history."""
187
self._check_write_ok()
188
return self._clone_text(new_version_id, old_version_id, parents)
190
def _clone_text(self, new_version_id, old_version_id, parents):
191
"""Helper function to do the _clone_text work."""
192
raise NotImplementedError(self.clone_text)
194
def create_empty(self, name, transport, mode=None):
195
"""Create a new versioned file of this exact type.
197
:param name: the file name
198
:param transport: the transport
199
:param mode: optional file mode.
201
raise NotImplementedError(self.create_empty)
203
def fix_parents(self, version, new_parents):
204
"""Fix the parents list for version.
206
This is done by appending a new version to the index
207
with identical data except for the parents list.
208
the parents list must be a superset of the current
211
self._check_write_ok()
212
return self._fix_parents(version, new_parents)
214
def _fix_parents(self, version, new_parents):
215
"""Helper for fix_parents."""
216
raise NotImplementedError(self.fix_parents)
218
def get_delta(self, version):
219
"""Get a delta for constructing version from some other version.
221
:return: (delta_parent, sha1, noeol, delta)
222
Where delta_parent is a version id or None to indicate no parent.
224
raise NotImplementedError(self.get_delta)
226
def get_deltas(self, versions):
227
"""Get multiple deltas at once for constructing versions.
229
:return: dict(version_id:(delta_parent, sha1, noeol, delta))
230
Where delta_parent is a version id or None to indicate no parent, and
231
version_id is the version_id created by that delta.
234
for version in versions:
235
result[version] = self.get_delta(version)
238
def get_sha1(self, version_id):
239
"""Get the stored sha1 sum for the given revision.
241
:param name: The name of the version to lookup
243
raise NotImplementedError(self.get_sha1)
245
def get_suffixes(self):
246
"""Return the file suffixes associated with this versioned file."""
247
raise NotImplementedError(self.get_suffixes)
249
def get_text(self, version_id):
250
"""Return version contents as a text string.
252
Raises RevisionNotPresent if version is not present in
255
return ''.join(self.get_lines(version_id))
256
get_string = get_text
258
def get_texts(self, version_ids):
259
"""Return the texts of listed versions as a list of strings.
261
Raises RevisionNotPresent if version is not present in
264
return [''.join(self.get_lines(v)) for v in version_ids]
266
def get_lines(self, version_id):
267
"""Return version contents as a sequence of lines.
269
Raises RevisionNotPresent if version is not present in
272
raise NotImplementedError(self.get_lines)
274
def get_ancestry(self, version_ids):
275
"""Return a list of all ancestors of given version(s). This
276
will not include the null revision.
278
Must raise RevisionNotPresent if any of the given versions are
279
not present in file history."""
280
if isinstance(version_ids, basestring):
281
version_ids = [version_ids]
282
raise NotImplementedError(self.get_ancestry)
284
def get_ancestry_with_ghosts(self, version_ids):
285
"""Return a list of all ancestors of given version(s). This
286
will not include the null revision.
288
Must raise RevisionNotPresent if any of the given versions are
289
not present in file history.
291
Ghosts that are known about will be included in ancestry list,
292
but are not explicitly marked.
294
raise NotImplementedError(self.get_ancestry_with_ghosts)
296
def get_graph(self, version_ids=None):
297
"""Return a graph from the versioned file.
299
Ghosts are not listed or referenced in the graph.
300
:param version_ids: Versions to select.
301
None means retrieve all versions.
304
if version_ids is None:
305
for version in self.versions():
306
result[version] = self.get_parents(version)
308
pending = set(version_ids)
310
version = pending.pop()
311
if version in result:
313
parents = self.get_parents(version)
314
for parent in parents:
318
result[version] = parents
321
def get_graph_with_ghosts(self):
322
"""Return a graph for the entire versioned file.
324
Ghosts are referenced in parents list but are not
327
raise NotImplementedError(self.get_graph_with_ghosts)
329
@deprecated_method(zero_eight)
330
def parent_names(self, version):
331
"""Return version names for parents of a version.
333
See get_parents for the current api.
335
return self.get_parents(version)
337
def get_parents(self, version_id):
338
"""Return version names for parents of a version.
340
Must raise RevisionNotPresent if version is not present in
343
raise NotImplementedError(self.get_parents)
345
def get_parents_with_ghosts(self, version_id):
346
"""Return version names for parents of version_id.
348
Will raise RevisionNotPresent if version_id is not present
351
Ghosts that are known about will be included in the parent list,
352
but are not explicitly marked.
354
raise NotImplementedError(self.get_parents_with_ghosts)
356
def annotate_iter(self, version_id):
357
"""Yield list of (version-id, line) pairs for the specified
360
Must raise RevisionNotPresent if any of the given versions are
361
not present in file history.
363
raise NotImplementedError(self.annotate_iter)
365
def annotate(self, version_id):
366
return list(self.annotate_iter(version_id))
368
def _apply_delta(self, lines, delta):
369
"""Apply delta to lines."""
372
for start, end, count, delta_lines in delta:
373
lines[offset+start:offset+end] = delta_lines
374
offset = offset + (start - end) + count
377
def join(self, other, pb=None, msg=None, version_ids=None,
378
ignore_missing=False):
379
"""Integrate versions from other into this versioned file.
381
If version_ids is None all versions from other should be
382
incorporated into this versioned file.
384
Must raise RevisionNotPresent if any of the specified versions
385
are not present in the other files history unless ignore_missing
386
is supplied when they are silently skipped.
388
self._check_write_ok()
389
return InterVersionedFile.get(other, self).join(
395
def iter_lines_added_or_present_in_versions(self, version_ids=None):
396
"""Iterate over the lines in the versioned file from version_ids.
398
This may return lines from other versions, and does not return the
399
specific version marker at this point. The api may be changed
400
during development to include the version that the versioned file
401
thinks is relevant, but given that such hints are just guesses,
402
its better not to have it if we don't need it.
404
NOTES: Lines are normalised: they will all have \n terminators.
405
Lines are returned in arbitrary order.
407
raise NotImplementedError(self.iter_lines_added_or_present_in_versions)
409
def transaction_finished(self):
410
"""The transaction that this file was opened in has finished.
412
This records self.finished = True and should cause all mutating
417
@deprecated_method(zero_eight)
418
def walk(self, version_ids=None):
419
"""Walk the versioned file as a weave-like structure, for
420
versions relative to version_ids. Yields sequence of (lineno,
421
insert, deletes, text) for each relevant line.
423
Must raise RevisionNotPresent if any of the specified versions
424
are not present in the file history.
426
:param version_ids: the version_ids to walk with respect to. If not
427
supplied the entire weave-like structure is walked.
429
walk is deprecated in favour of iter_lines_added_or_present_in_versions
431
raise NotImplementedError(self.walk)
433
@deprecated_method(zero_eight)
434
def iter_names(self):
435
"""Walk the names list."""
436
return iter(self.versions())
438
def plan_merge(self, ver_a, ver_b):
439
"""Return pseudo-annotation indicating how the two versions merge.
441
This is computed between versions a and b and their common
444
Weave lines present in none of them are skipped entirely.
447
killed-base Dead in base revision
448
killed-both Killed in each revision
451
unchanged Alive in both a and b (possibly created in both)
454
ghost-a Killed in a, unborn in b
455
ghost-b Killed in b, unborn in a
456
irrelevant Not in either revision
458
raise NotImplementedError(VersionedFile.plan_merge)
460
def weave_merge(self, plan, a_marker=TextMerge.A_MARKER,
461
b_marker=TextMerge.B_MARKER):
462
return PlanWeaveMerge(plan, a_marker, b_marker).merge_lines()[0]
465
class PlanWeaveMerge(TextMerge):
466
"""Weave merge that takes a plan as its input.
468
This exists so that VersionedFile.plan_merge is implementable.
469
Most callers will want to use WeaveMerge instead.
472
def __init__(self, plan, a_marker=TextMerge.A_MARKER,
473
b_marker=TextMerge.B_MARKER):
474
TextMerge.__init__(self, a_marker, b_marker)
477
def _merge_struct(self):
482
def outstanding_struct():
483
if not lines_a and not lines_b:
485
elif ch_a and not ch_b:
488
elif ch_b and not ch_a:
490
elif lines_a == lines_b:
493
yield (lines_a, lines_b)
495
# We previously considered either 'unchanged' or 'killed-both' lines
496
# to be possible places to resynchronize. However, assuming agreement
497
# on killed-both lines may be too aggressive. -- mbp 20060324
498
for state, line in self.plan:
499
if state == 'unchanged':
500
# resync and flush queued conflicts changes if any
501
for struct in outstanding_struct():
507
if state == 'unchanged':
510
elif state == 'killed-a':
513
elif state == 'killed-b':
516
elif state == 'new-a':
519
elif state == 'new-b':
523
assert state in ('irrelevant', 'ghost-a', 'ghost-b',
524
'killed-base', 'killed-both'), state
525
for struct in outstanding_struct():
529
class WeaveMerge(PlanWeaveMerge):
530
"""Weave merge that takes a VersionedFile and two versions as its input"""
532
def __init__(self, versionedfile, ver_a, ver_b,
533
a_marker=PlanWeaveMerge.A_MARKER, b_marker=PlanWeaveMerge.B_MARKER):
534
plan = versionedfile.plan_merge(ver_a, ver_b)
535
PlanWeaveMerge.__init__(self, plan, a_marker, b_marker)
538
class InterVersionedFile(InterObject):
539
"""This class represents operations taking place between two versionedfiles..
541
Its instances have methods like join, and contain
542
references to the source and target versionedfiles these operations can be
545
Often we will provide convenience methods on 'versionedfile' which carry out
546
operations with another versionedfile - they will always forward to
547
InterVersionedFile.get(other).method_name(parameters).
551
"""The available optimised InterVersionedFile types."""
553
def join(self, pb=None, msg=None, version_ids=None, ignore_missing=False):
554
"""Integrate versions from self.source into self.target.
556
If version_ids is None all versions from source should be
557
incorporated into this versioned file.
559
Must raise RevisionNotPresent if any of the specified versions
560
are not present in the other files history unless ignore_missing is
561
supplied when they are silently skipped.
564
# - if the target is empty, just add all the versions from
565
# source to target, otherwise:
566
# - make a temporary versioned file of type target
567
# - insert the source content into it one at a time
569
if not self.target.versions():
572
# Make a new target-format versioned file.
573
temp_source = self.target.create_empty("temp", MemoryTransport())
575
version_ids = self._get_source_version_ids(version_ids, ignore_missing)
576
graph = self.source.get_graph(version_ids)
577
order = topo_sort(graph.items())
578
pb = ui.ui_factory.nested_progress_bar()
581
# TODO for incremental cross-format work:
582
# make a versioned file with the following content:
583
# all revisions we have been asked to join
584
# all their ancestors that are *not* in target already.
585
# the immediate parents of the above two sets, with
586
# empty parent lists - these versions are in target already
587
# and the incorrect version data will be ignored.
588
# TODO: for all ancestors that are present in target already,
589
# check them for consistent data, this requires moving sha1 from
591
# TODO: remove parent texts when they are not relevant any more for
592
# memory pressure reduction. RBC 20060313
593
# pb.update('Converting versioned data', 0, len(order))
594
# deltas = self.source.get_deltas(order)
595
for index, version in enumerate(order):
596
pb.update('Converting versioned data', index, len(order))
597
parent_text = target.add_lines(version,
598
self.source.get_parents(version),
599
self.source.get_lines(version),
600
parent_texts=parent_texts)
601
parent_texts[version] = parent_text
602
#delta_parent, sha1, noeol, delta = deltas[version]
603
#target.add_delta(version,
604
# self.source.get_parents(version),
609
#target.get_lines(version)
611
# this should hit the native code path for target
612
if target is not self.target:
613
return self.target.join(temp_source,
621
def _get_source_version_ids(self, version_ids, ignore_missing):
622
"""Determine the version ids to be used from self.source.
624
:param version_ids: The caller-supplied version ids to check. (None
625
for all). If None is in version_ids, it is stripped.
626
:param ignore_missing: if True, remove missing ids from the version
627
list. If False, raise RevisionNotPresent on
628
a missing version id.
629
:return: A set of version ids.
631
if version_ids is None:
632
# None cannot be in source.versions
633
return set(self.source.versions())
636
return set(self.source.versions()).intersection(set(version_ids))
638
new_version_ids = set()
639
for version in version_ids:
642
if not self.source.has_version(version):
643
raise errors.RevisionNotPresent(version, str(self.source))
645
new_version_ids.add(version)
646
return new_version_ids
649
class InterVersionedFileTestProviderAdapter(object):
650
"""A tool to generate a suite testing multiple inter versioned-file classes.
652
This is done by copying the test once for each InterVersionedFile provider
653
and injecting the transport_server, transport_readonly_server,
654
versionedfile_factory and versionedfile_factory_to classes into each copy.
655
Each copy is also given a new id() to make it easy to identify.
658
def __init__(self, transport_server, transport_readonly_server, formats):
659
self._transport_server = transport_server
660
self._transport_readonly_server = transport_readonly_server
661
self._formats = formats
663
def adapt(self, test):
665
for (interversionedfile_class,
666
versionedfile_factory,
667
versionedfile_factory_to) in self._formats:
668
new_test = deepcopy(test)
669
new_test.transport_server = self._transport_server
670
new_test.transport_readonly_server = self._transport_readonly_server
671
new_test.interversionedfile_class = interversionedfile_class
672
new_test.versionedfile_factory = versionedfile_factory
673
new_test.versionedfile_factory_to = versionedfile_factory_to
674
def make_new_test_id():
675
new_id = "%s(%s)" % (new_test.id(), interversionedfile_class.__name__)
676
return lambda: new_id
677
new_test.id = make_new_test_id()
678
result.addTest(new_test)
682
def default_test_list():
683
"""Generate the default list of interversionedfile permutations to test."""
684
from bzrlib.weave import WeaveFile
685
from bzrlib.knit import KnitVersionedFile
687
# test the fallback InterVersionedFile from annotated knits to weave
688
result.append((InterVersionedFile,
691
for optimiser in InterVersionedFile._optimisers:
692
result.append((optimiser,
693
optimiser._matching_file_from_factory,
694
optimiser._matching_file_to_factory
696
# if there are specific combinations we want to use, we can add them
added
removed
bzrlib/versionedfile.py
