1
# Copyright (C) 2005 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.symbol_versioning import *
30
from bzrlib.textmerge import TextMerge
31
from bzrlib.transport.memory import MemoryTransport
32
from bzrlib.tsort import topo_sort
36
class VersionedFile(object):
37
"""Versioned text file storage.
39
A versioned file manages versions of line-based text files,
40
keeping track of the originating version for each line.
42
To clients the "lines" of the file are represented as a list of
43
strings. These strings will typically have terminal newline
44
characters, but this is not required. In particular files commonly
45
do not have a newline at the end of the file.
47
Texts are identified by a version-id string.
50
def __init__(self, access_mode):
52
self._access_mode = access_mode
54
def copy_to(self, name, transport):
55
"""Copy this versioned file to name on transport."""
56
raise NotImplementedError(self.copy_to)
58
@deprecated_method(zero_eight)
60
"""Return a list of all the versions in this versioned file.
62
Please use versionedfile.versions() now.
64
return self.versions()
67
"""Return a unsorted list of versions."""
68
raise NotImplementedError(self.versions)
70
def has_ghost(self, version_id):
71
"""Returns whether version is present as a ghost."""
72
raise NotImplementedError(self.has_ghost)
74
def has_version(self, version_id):
75
"""Returns whether version is present."""
76
raise NotImplementedError(self.has_version)
78
def add_delta(self, version_id, parents, delta_parent, sha1, noeol, delta):
79
"""Add a text to the versioned file via a pregenerated delta.
81
:param version_id: The version id being added.
82
:param parents: The parents of the version_id.
83
:param delta_parent: The parent this delta was created against.
84
:param sha1: The sha1 of the full text.
85
:param delta: The delta instructions. See get_delta for details.
87
self._check_write_ok()
88
if self.has_version(version_id):
89
raise errors.RevisionAlreadyPresent(version_id, self)
90
return self._add_delta(version_id, parents, delta_parent, sha1, noeol, delta)
92
def _add_delta(self, version_id, parents, delta_parent, sha1, noeol, delta):
93
"""Class specific routine to add a delta.
95
This generic version simply applies the delta to the delta_parent and
98
# strip annotation from delta
100
for start, stop, delta_len, delta_lines in delta:
101
new_delta.append((start, stop, delta_len, [text for origin, text in delta_lines]))
102
if delta_parent is not None:
103
parent_full = self.get_lines(delta_parent)
106
new_full = self._apply_delta(parent_full, new_delta)
107
# its impossible to have noeol on an empty file
108
if noeol and new_full[-1][-1] == '\n':
109
new_full[-1] = new_full[-1][:-1]
110
self.add_lines(version_id, parents, new_full)
112
def add_lines(self, version_id, parents, lines, parent_texts=None):
113
"""Add a single text on top of the versioned file.
115
Must raise RevisionAlreadyPresent if the new version is
116
already present in file history.
118
Must raise RevisionNotPresent if any of the given parents are
119
not present in file history.
120
:param parent_texts: An optional dictionary containing the opaque
121
representations of some or all of the parents of
122
version_id to allow delta optimisations.
123
VERY IMPORTANT: the texts must be those returned
124
by add_lines or data corruption can be caused.
125
:return: An opaque representation of the inserted version which can be
126
provided back to future add_lines calls in the parent_texts
129
self._check_write_ok()
130
return self._add_lines(version_id, parents, lines, parent_texts)
132
def _add_lines(self, version_id, parents, lines, parent_texts):
133
"""Helper to do the class specific add_lines."""
134
raise NotImplementedError(self.add_lines)
136
def add_lines_with_ghosts(self, version_id, parents, lines,
138
"""Add lines to the versioned file, allowing ghosts to be present.
140
This takes the same parameters as add_lines.
142
self._check_write_ok()
143
return self._add_lines_with_ghosts(version_id, parents, lines,
146
def _add_lines_with_ghosts(self, version_id, parents, lines, parent_texts):
147
"""Helper to do class specific add_lines_with_ghosts."""
148
raise NotImplementedError(self.add_lines_with_ghosts)
150
def check(self, progress_bar=None):
151
"""Check the versioned file for integrity."""
152
raise NotImplementedError(self.check)
154
def _check_lines_not_unicode(self, lines):
155
"""Check that lines being added to a versioned file are not unicode."""
157
if line.__class__ is not str:
158
raise errors.BzrBadParameterUnicode("lines")
160
def _check_lines_are_lines(self, lines):
161
"""Check that the lines really are full lines without inline EOL."""
163
if '\n' in line[:-1]:
164
raise errors.BzrBadParameterContainsNewline("lines")
166
def _check_write_ok(self):
167
"""Is the versioned file marked as 'finished' ? Raise if it is."""
169
raise errors.OutSideTransaction()
170
if self._access_mode != 'w':
171
raise errors.ReadOnlyObjectDirtiedError(self)
173
def clear_cache(self):
174
"""Remove any data cached in the versioned file object."""
176
def clone_text(self, new_version_id, old_version_id, parents):
177
"""Add an identical text to old_version_id as new_version_id.
179
Must raise RevisionNotPresent if the old version or any of the
180
parents are not present in file history.
182
Must raise RevisionAlreadyPresent if the new version is
183
already present in file history."""
184
self._check_write_ok()
185
return self._clone_text(new_version_id, old_version_id, parents)
187
def _clone_text(self, new_version_id, old_version_id, parents):
188
"""Helper function to do the _clone_text work."""
189
raise NotImplementedError(self.clone_text)
191
def create_empty(self, name, transport, mode=None):
192
"""Create a new versioned file of this exact type.
194
:param name: the file name
195
:param transport: the transport
196
:param mode: optional file mode.
198
raise NotImplementedError(self.create_empty)
200
def fix_parents(self, version, new_parents):
201
"""Fix the parents list for version.
203
This is done by appending a new version to the index
204
with identical data except for the parents list.
205
the parents list must be a superset of the current
208
self._check_write_ok()
209
return self._fix_parents(version, new_parents)
211
def _fix_parents(self, version, new_parents):
212
"""Helper for fix_parents."""
213
raise NotImplementedError(self.fix_parents)
215
def get_delta(self, version):
216
"""Get a delta for constructing version from some other version.
218
:return: (delta_parent, sha1, noeol, delta)
219
Where delta_parent is a version id or None to indicate no parent.
221
raise NotImplementedError(self.get_delta)
223
def get_deltas(self, versions):
224
"""Get multiple deltas at once for constructing versions.
226
:return: dict(version_id:(delta_parent, sha1, noeol, delta))
227
Where delta_parent is a version id or None to indicate no parent, and
228
version_id is the version_id created by that delta.
231
for version in versions:
232
result[version] = self.get_delta(version)
235
def get_sha1(self, version_id):
236
"""Get the stored sha1 sum for the given revision.
238
:param name: The name of the version to lookup
240
raise NotImplementedError(self.get_sha1)
242
def get_suffixes(self):
243
"""Return the file suffixes associated with this versioned file."""
244
raise NotImplementedError(self.get_suffixes)
246
def get_text(self, version_id):
247
"""Return version contents as a text string.
249
Raises RevisionNotPresent if version is not present in
252
return ''.join(self.get_lines(version_id))
253
get_string = get_text
255
def get_lines(self, version_id):
256
"""Return version contents as a sequence of lines.
258
Raises RevisionNotPresent if version is not present in
261
raise NotImplementedError(self.get_lines)
263
def get_ancestry(self, version_ids):
264
"""Return a list of all ancestors of given version(s). This
265
will not include the null revision.
267
Must raise RevisionNotPresent if any of the given versions are
268
not present in file history."""
269
if isinstance(version_ids, basestring):
270
version_ids = [version_ids]
271
raise NotImplementedError(self.get_ancestry)
273
def get_ancestry_with_ghosts(self, version_ids):
274
"""Return a list of all ancestors of given version(s). This
275
will not include the null revision.
277
Must raise RevisionNotPresent if any of the given versions are
278
not present in file history.
280
Ghosts that are known about will be included in ancestry list,
281
but are not explicitly marked.
283
raise NotImplementedError(self.get_ancestry_with_ghosts)
285
def get_graph(self, version_ids=None):
286
"""Return a graph from the versioned file.
288
Ghosts are not listed or referenced in the graph.
289
:param version_ids: Versions to select.
290
None means retreive all versions.
293
if version_ids is None:
294
for version in self.versions():
295
result[version] = self.get_parents(version)
297
pending = set(version_ids)
299
version = pending.pop()
300
if version in result:
302
parents = self.get_parents(version)
303
for parent in parents:
307
result[version] = parents
310
def get_graph_with_ghosts(self):
311
"""Return a graph for the entire versioned file.
313
Ghosts are referenced in parents list but are not
316
raise NotImplementedError(self.get_graph_with_ghosts)
318
@deprecated_method(zero_eight)
319
def parent_names(self, version):
320
"""Return version names for parents of a version.
322
See get_parents for the current api.
324
return self.get_parents(version)
326
def get_parents(self, version_id):
327
"""Return version names for parents of a version.
329
Must raise RevisionNotPresent if version is not present in
332
raise NotImplementedError(self.get_parents)
334
def get_parents_with_ghosts(self, version_id):
335
"""Return version names for parents of version_id.
337
Will raise RevisionNotPresent if version_id is not present
340
Ghosts that are known about will be included in the parent list,
341
but are not explicitly marked.
343
raise NotImplementedError(self.get_parents_with_ghosts)
345
def annotate_iter(self, version_id):
346
"""Yield list of (version-id, line) pairs for the specified
349
Must raise RevisionNotPresent if any of the given versions are
350
not present in file history.
352
raise NotImplementedError(self.annotate_iter)
354
def annotate(self, version_id):
355
return list(self.annotate_iter(version_id))
357
def _apply_delta(self, lines, delta):
358
"""Apply delta to lines."""
361
for start, end, count, delta_lines in delta:
362
lines[offset+start:offset+end] = delta_lines
363
offset = offset + (start - end) + count
366
def join(self, other, pb=None, msg=None, version_ids=None,
367
ignore_missing=False):
368
"""Integrate versions from other into this versioned file.
370
If version_ids is None all versions from other should be
371
incorporated into this versioned file.
373
Must raise RevisionNotPresent if any of the specified versions
374
are not present in the other files history unless ignore_missing
375
is supplied when they are silently skipped.
377
self._check_write_ok()
378
return InterVersionedFile.get(other, self).join(
384
def iter_lines_added_or_present_in_versions(self, version_ids=None):
385
"""Iterate over the lines in the versioned file from version_ids.
387
This may return lines from other versions, and does not return the
388
specific version marker at this point. The api may be changed
389
during development to include the version that the versioned file
390
thinks is relevant, but given that such hints are just guesses,
391
its better not to have it if we dont need it.
393
NOTES: Lines are normalised: they will all have \n terminators.
394
Lines are returned in arbitrary order.
396
raise NotImplementedError(self.iter_lines_added_or_present_in_versions)
398
def transaction_finished(self):
399
"""The transaction that this file was opened in has finished.
401
This records self.finished = True and should cause all mutating
406
@deprecated_method(zero_eight)
407
def walk(self, version_ids=None):
408
"""Walk the versioned file as a weave-like structure, for
409
versions relative to version_ids. Yields sequence of (lineno,
410
insert, deletes, text) for each relevant line.
412
Must raise RevisionNotPresent if any of the specified versions
413
are not present in the file history.
415
:param version_ids: the version_ids to walk with respect to. If not
416
supplied the entire weave-like structure is walked.
418
walk is deprecated in favour of iter_lines_added_or_present_in_versions
420
raise NotImplementedError(self.walk)
422
@deprecated_method(zero_eight)
423
def iter_names(self):
424
"""Walk the names list."""
425
return iter(self.versions())
427
def plan_merge(self, ver_a, ver_b):
428
"""Return pseudo-annotation indicating how the two versions merge.
430
This is computed between versions a and b and their common
433
Weave lines present in none of them are skipped entirely.
436
killed-base Dead in base revision
437
killed-both Killed in each revision
440
unchanged Alive in both a and b (possibly created in both)
443
ghost-a Killed in a, unborn in b
444
ghost-b Killed in b, unborn in a
445
irrelevant Not in either revision
447
raise NotImplementedError(VersionedFile.plan_merge)
449
def weave_merge(self, plan, a_marker=TextMerge.A_MARKER,
450
b_marker=TextMerge.B_MARKER):
451
return PlanWeaveMerge(plan, a_marker, b_marker).merge_lines()[0]
454
class PlanWeaveMerge(TextMerge):
455
"""Weave merge that takes a plan as its input.
457
This exists so that VersionedFile.plan_merge is implementable.
458
Most callers will want to use WeaveMerge instead.
461
def __init__(self, plan, a_marker=TextMerge.A_MARKER,
462
b_marker=TextMerge.B_MARKER):
463
TextMerge.__init__(self, a_marker, b_marker)
466
def _merge_struct(self):
471
def outstanding_struct():
472
if not lines_a and not lines_b:
474
elif ch_a and not ch_b:
477
elif ch_b and not ch_a:
479
elif lines_a == lines_b:
482
yield (lines_a, lines_b)
484
# We previously considered either 'unchanged' or 'killed-both' lines
485
# to be possible places to resynchronize. However, assuming agreement
486
# on killed-both lines may be too agressive. -- mbp 20060324
487
for state, line in self.plan:
488
if state == 'unchanged':
489
# resync and flush queued conflicts changes if any
490
for struct in outstanding_struct():
496
if state == 'unchanged':
499
elif state == 'killed-a':
502
elif state == 'killed-b':
505
elif state == 'new-a':
508
elif state == 'new-b':
512
assert state in ('irrelevant', 'ghost-a', 'ghost-b',
513
'killed-base', 'killed-both'), state
514
for struct in outstanding_struct():
518
class WeaveMerge(PlanWeaveMerge):
519
"""Weave merge that takes a VersionedFile and two versions as its input"""
521
def __init__(self, versionedfile, ver_a, ver_b,
522
a_marker=PlanWeaveMerge.A_MARKER, b_marker=PlanWeaveMerge.B_MARKER):
523
plan = versionedfile.plan_merge(ver_a, ver_b)
524
PlanWeaveMerge.__init__(self, plan, a_marker, b_marker)
527
class InterVersionedFile(InterObject):
528
"""This class represents operations taking place between two versionedfiles..
530
Its instances have methods like join, and contain
531
references to the source and target versionedfiles these operations can be
534
Often we will provide convenience methods on 'versionedfile' which carry out
535
operations with another versionedfile - they will always forward to
536
InterVersionedFile.get(other).method_name(parameters).
540
"""The available optimised InterVersionedFile types."""
542
def join(self, pb=None, msg=None, version_ids=None, ignore_missing=False):
543
"""Integrate versions from self.source into self.target.
545
If version_ids is None all versions from source should be
546
incorporated into this versioned file.
548
Must raise RevisionNotPresent if any of the specified versions
549
are not present in the other files history unless ignore_missing is
550
supplied when they are silently skipped.
553
# - if the target is empty, just add all the versions from
554
# source to target, otherwise:
555
# - make a temporary versioned file of type target
556
# - insert the source content into it one at a time
558
if not self.target.versions():
561
# Make a new target-format versioned file.
562
temp_source = self.target.create_empty("temp", MemoryTransport())
564
version_ids = self._get_source_version_ids(version_ids, ignore_missing)
565
graph = self.source.get_graph(version_ids)
566
order = topo_sort(graph.items())
567
pb = ui.ui_factory.nested_progress_bar()
570
# TODO for incremental cross-format work:
571
# make a versioned file with the following content:
572
# all revisions we have been asked to join
573
# all their ancestors that are *not* in target already.
574
# the immediate parents of the above two sets, with
575
# empty parent lists - these versions are in target already
576
# and the incorrect version data will be ignored.
577
# TODO: for all ancestors that are present in target already,
578
# check them for consistent data, this requires moving sha1 from
580
# TODO: remove parent texts when they are not relevant any more for
581
# memory pressure reduction. RBC 20060313
582
# pb.update('Converting versioned data', 0, len(order))
583
# deltas = self.source.get_deltas(order)
584
for index, version in enumerate(order):
585
pb.update('Converting versioned data', index, len(order))
586
parent_text = target.add_lines(version,
587
self.source.get_parents(version),
588
self.source.get_lines(version),
589
parent_texts=parent_texts)
590
parent_texts[version] = parent_text
591
#delta_parent, sha1, noeol, delta = deltas[version]
592
#target.add_delta(version,
593
# self.source.get_parents(version),
598
#target.get_lines(version)
600
# this should hit the native code path for target
601
if target is not self.target:
602
return self.target.join(temp_source,
610
def _get_source_version_ids(self, version_ids, ignore_missing):
611
"""Determine the version ids to be used from self.source.
613
:param version_ids: The caller-supplied version ids to check. (None
614
for all). If None is in version_ids, it is stripped.
615
:param ignore_missing: if True, remove missing ids from the version
616
list. If False, raise RevisionNotPresent on
617
a missing version id.
618
:return: A set of version ids.
620
if version_ids is None:
621
# None cannot be in source.versions
622
return set(self.source.versions())
625
return set(self.source.versions()).intersection(set(version_ids))
627
new_version_ids = set()
628
for version in version_ids:
631
if not self.source.has_version(version):
632
raise errors.RevisionNotPresent(version, str(self.source))
634
new_version_ids.add(version)
635
return new_version_ids
638
class InterVersionedFileTestProviderAdapter(object):
639
"""A tool to generate a suite testing multiple inter versioned-file classes.
641
This is done by copying the test once for each interversionedfile provider
642
and injecting the transport_server, transport_readonly_server,
643
versionedfile_factory and versionedfile_factory_to classes into each copy.
644
Each copy is also given a new id() to make it easy to identify.
647
def __init__(self, transport_server, transport_readonly_server, formats):
648
self._transport_server = transport_server
649
self._transport_readonly_server = transport_readonly_server
650
self._formats = formats
652
def adapt(self, test):
654
for (interversionedfile_class,
655
versionedfile_factory,
656
versionedfile_factory_to) in self._formats:
657
new_test = deepcopy(test)
658
new_test.transport_server = self._transport_server
659
new_test.transport_readonly_server = self._transport_readonly_server
660
new_test.interversionedfile_class = interversionedfile_class
661
new_test.versionedfile_factory = versionedfile_factory
662
new_test.versionedfile_factory_to = versionedfile_factory_to
663
def make_new_test_id():
664
new_id = "%s(%s)" % (new_test.id(), interversionedfile_class.__name__)
665
return lambda: new_id
666
new_test.id = make_new_test_id()
667
result.addTest(new_test)
671
def default_test_list():
672
"""Generate the default list of interversionedfile permutations to test."""
673
from bzrlib.weave import WeaveFile
674
from bzrlib.knit import KnitVersionedFile
676
# test the fallback InterVersionedFile from annotated knits to weave
677
result.append((InterVersionedFile,
680
for optimiser in InterVersionedFile._optimisers:
681
result.append((optimiser,
682
optimiser._matching_file_from_factory,
683
optimiser._matching_file_to_factory
685
# if there are specific combinations we want to use, we can add them
added
removed
bzrlib/versionedfile.py
