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(), """
33
from bzrlib.transport.memory import MemoryTransport
36
from cStringIO import StringIO
38
from bzrlib.inter import InterObject
39
from bzrlib.textmerge import TextMerge
42
class VersionedFile(object):
43
"""Versioned text file storage.
45
A versioned file manages versions of line-based text files,
46
keeping track of the originating version for each line.
48
To clients the "lines" of the file are represented as a list of
49
strings. These strings will typically have terminal newline
50
characters, but this is not required. In particular files commonly
51
do not have a newline at the end of the file.
53
Texts are identified by a version-id string.
56
def __init__(self, access_mode):
58
self._access_mode = access_mode
61
def check_not_reserved_id(version_id):
62
revision.check_not_reserved_id(version_id)
64
def copy_to(self, name, transport):
65
"""Copy this versioned file to name on transport."""
66
raise NotImplementedError(self.copy_to)
69
"""Return a unsorted list of versions."""
70
raise NotImplementedError(self.versions)
72
def has_ghost(self, version_id):
73
"""Returns whether version is present as a ghost."""
74
raise NotImplementedError(self.has_ghost)
76
def has_version(self, version_id):
77
"""Returns whether version is present."""
78
raise NotImplementedError(self.has_version)
80
def add_lines(self, version_id, parents, lines, parent_texts=None,
81
left_matching_blocks=None, nostore_sha=None):
82
"""Add a single text on top of the versioned file.
84
Must raise RevisionAlreadyPresent if the new version is
85
already present in file history.
87
Must raise RevisionNotPresent if any of the given parents are
88
not present in file history.
89
:param parent_texts: An optional dictionary containing the opaque
90
representations of some or all of the parents of
91
version_id to allow delta optimisations.
92
VERY IMPORTANT: the texts must be those returned
93
by add_lines or data corruption can be caused.
94
:param left_matching_blocks: a hint about which areas are common
95
between the text and its left-hand-parent. The format is
96
the SequenceMatcher.get_matching_blocks format.
97
:param nostore_sha: Raise ExistingContent and do not add the lines to
98
the versioned file if the digest of the lines matches this.
99
:return: The text sha1, the number of bytes in the text, and an opaque
100
representation of the inserted version which can be provided
101
back to future add_lines calls in the parent_texts dictionary.
103
version_id = osutils.safe_revision_id(version_id)
104
parents = [osutils.safe_revision_id(v) for v in parents]
105
self._check_write_ok()
106
return self._add_lines(version_id, parents, lines, parent_texts,
107
left_matching_blocks, nostore_sha)
109
def _add_lines(self, version_id, parents, lines, parent_texts,
110
left_matching_blocks, nostore_sha):
111
"""Helper to do the class specific add_lines."""
112
raise NotImplementedError(self.add_lines)
114
def add_lines_with_ghosts(self, version_id, parents, lines,
115
parent_texts=None, nostore_sha=None):
116
"""Add lines to the versioned file, allowing ghosts to be present.
118
This takes the same parameters as add_lines and returns the same.
120
version_id = osutils.safe_revision_id(version_id)
121
parents = [osutils.safe_revision_id(v) for v in parents]
122
self._check_write_ok()
123
return self._add_lines_with_ghosts(version_id, parents, lines,
124
parent_texts, nostore_sha)
126
def _add_lines_with_ghosts(self, version_id, parents, lines, parent_texts,
128
"""Helper to do class specific add_lines_with_ghosts."""
129
raise NotImplementedError(self.add_lines_with_ghosts)
131
def check(self, progress_bar=None):
132
"""Check the versioned file for integrity."""
133
raise NotImplementedError(self.check)
135
def _check_lines_not_unicode(self, lines):
136
"""Check that lines being added to a versioned file are not unicode."""
138
if line.__class__ is not str:
139
raise errors.BzrBadParameterUnicode("lines")
141
def _check_lines_are_lines(self, lines):
142
"""Check that the lines really are full lines without inline EOL."""
144
if '\n' in line[:-1]:
145
raise errors.BzrBadParameterContainsNewline("lines")
147
def _check_write_ok(self):
148
"""Is the versioned file marked as 'finished' ? Raise if it is."""
150
raise errors.OutSideTransaction()
151
if self._access_mode != 'w':
152
raise errors.ReadOnlyObjectDirtiedError(self)
154
def enable_cache(self):
155
"""Tell this versioned file that it should cache any data it reads.
157
This is advisory, implementations do not have to support caching.
161
def clear_cache(self):
162
"""Remove any data cached in the versioned file object.
164
This only needs to be supported if caches are supported
168
def clone_text(self, new_version_id, old_version_id, parents):
169
"""Add an identical text to old_version_id as new_version_id.
171
Must raise RevisionNotPresent if the old version or any of the
172
parents are not present in file history.
174
Must raise RevisionAlreadyPresent if the new version is
175
already present in file history."""
176
new_version_id = osutils.safe_revision_id(new_version_id)
177
old_version_id = osutils.safe_revision_id(old_version_id)
178
self._check_write_ok()
179
return self._clone_text(new_version_id, old_version_id, parents)
181
def _clone_text(self, new_version_id, old_version_id, parents):
182
"""Helper function to do the _clone_text work."""
183
raise NotImplementedError(self.clone_text)
185
def create_empty(self, name, transport, mode=None):
186
"""Create a new versioned file of this exact type.
188
:param name: the file name
189
:param transport: the transport
190
:param mode: optional file mode.
192
raise NotImplementedError(self.create_empty)
194
def fix_parents(self, version_id, new_parents):
195
"""Fix the parents list for version.
197
This is done by appending a new version to the index
198
with identical data except for the parents list.
199
the parents list must be a superset of the current
202
version_id = osutils.safe_revision_id(version_id)
203
new_parents = [osutils.safe_revision_id(p) for p in new_parents]
204
self._check_write_ok()
205
return self._fix_parents(version_id, new_parents)
207
def _fix_parents(self, version_id, new_parents):
208
"""Helper for fix_parents."""
209
raise NotImplementedError(self.fix_parents)
211
def get_format_signature(self):
212
"""Get a text description of the data encoding in this file.
216
raise NotImplementedError(self.get_format_signature)
218
def make_mpdiffs(self, version_ids):
219
"""Create multiparent diffs for specified versions"""
220
knit_versions = set()
221
for version_id in version_ids:
222
knit_versions.add(version_id)
223
knit_versions.update(self.get_parents(version_id))
224
lines = dict(zip(knit_versions,
225
self._get_lf_split_line_list(knit_versions)))
227
for version_id in version_ids:
228
target = lines[version_id]
229
parents = [lines[p] for p in self.get_parents(version_id)]
231
left_parent_blocks = self._extract_blocks(version_id,
234
left_parent_blocks = None
235
diffs.append(multiparent.MultiParent.from_lines(target, parents,
239
def _extract_blocks(self, version_id, source, target):
242
def add_mpdiffs(self, records):
243
"""Add mpdiffs to this versionedfile
245
Records should be iterables of version, parents, expected_sha1,
246
mpdiff. mpdiff should be a MultiParent instance.
249
mpvf = multiparent.MultiMemoryVersionedFile()
251
for version, parent_ids, expected_sha1, mpdiff in records:
252
versions.append(version)
253
mpvf.add_diff(mpdiff, version, parent_ids)
254
needed_parents = set()
255
for version, parent_ids, expected_sha1, mpdiff in records:
256
needed_parents.update(p for p in parent_ids
257
if not mpvf.has_version(p))
258
for parent_id, lines in zip(needed_parents,
259
self._get_lf_split_line_list(needed_parents)):
260
mpvf.add_version(lines, parent_id, [])
261
for (version, parent_ids, expected_sha1, mpdiff), lines in\
262
zip(records, mpvf.get_line_list(versions)):
263
if len(parent_ids) == 1:
264
left_matching_blocks = list(mpdiff.get_matching_blocks(0,
265
mpvf.get_diff(parent_ids[0]).num_lines()))
267
left_matching_blocks = None
268
_, _, version_text = self.add_lines(version, parent_ids, lines,
269
vf_parents, left_matching_blocks=left_matching_blocks)
270
vf_parents[version] = version_text
271
for (version, parent_ids, expected_sha1, mpdiff), sha1 in\
272
zip(records, self.get_sha1s(versions)):
273
if expected_sha1 != sha1:
274
raise errors.VersionedFileInvalidChecksum(version)
276
def get_sha1(self, version_id):
277
"""Get the stored sha1 sum for the given revision.
279
:param name: The name of the version to lookup
281
raise NotImplementedError(self.get_sha1)
283
def get_sha1s(self, version_ids):
284
"""Get the stored sha1 sums for the given revisions.
286
:param version_ids: The names of the versions to lookup
287
:return: a list of sha1s in order according to the version_ids
289
raise NotImplementedError(self.get_sha1)
291
def get_suffixes(self):
292
"""Return the file suffixes associated with this versioned file."""
293
raise NotImplementedError(self.get_suffixes)
295
def get_text(self, version_id):
296
"""Return version contents as a text string.
298
Raises RevisionNotPresent if version is not present in
301
return ''.join(self.get_lines(version_id))
302
get_string = get_text
304
def get_texts(self, version_ids):
305
"""Return the texts of listed versions as a list of strings.
307
Raises RevisionNotPresent if version is not present in
310
return [''.join(self.get_lines(v)) for v in version_ids]
312
def get_lines(self, version_id):
313
"""Return version contents as a sequence of lines.
315
Raises RevisionNotPresent if version is not present in
318
raise NotImplementedError(self.get_lines)
320
def _get_lf_split_line_list(self, version_ids):
321
return [StringIO(t).readlines() for t in self.get_texts(version_ids)]
323
def get_ancestry(self, version_ids, topo_sorted=True):
324
"""Return a list of all ancestors of given version(s). This
325
will not include the null revision.
327
This list will not be topologically sorted if topo_sorted=False is
330
Must raise RevisionNotPresent if any of the given versions are
331
not present in file history."""
332
if isinstance(version_ids, basestring):
333
version_ids = [version_ids]
334
raise NotImplementedError(self.get_ancestry)
336
def get_ancestry_with_ghosts(self, version_ids):
337
"""Return a list of all ancestors of given version(s). This
338
will not include the null revision.
340
Must raise RevisionNotPresent if any of the given versions are
341
not present in file history.
343
Ghosts that are known about will be included in ancestry list,
344
but are not explicitly marked.
346
raise NotImplementedError(self.get_ancestry_with_ghosts)
348
def get_graph(self, version_ids=None):
349
"""Return a graph from the versioned file.
351
Ghosts are not listed or referenced in the graph.
352
:param version_ids: Versions to select.
353
None means retrieve all versions.
355
if version_ids is None:
356
return dict(self.iter_parents(self.versions()))
358
pending = set(osutils.safe_revision_id(v) for v in version_ids)
360
this_iteration = pending
362
for version, parents in self.iter_parents(this_iteration):
363
result[version] = parents
364
for parent in parents:
370
def get_graph_with_ghosts(self):
371
"""Return a graph for the entire versioned file.
373
Ghosts are referenced in parents list but are not
376
raise NotImplementedError(self.get_graph_with_ghosts)
378
def get_parents(self, version_id):
379
"""Return version names for parents of a version.
381
Must raise RevisionNotPresent if version is not present in
384
raise NotImplementedError(self.get_parents)
386
def get_parents_with_ghosts(self, version_id):
387
"""Return version names for parents of version_id.
389
Will raise RevisionNotPresent if version_id is not present
392
Ghosts that are known about will be included in the parent list,
393
but are not explicitly marked.
395
raise NotImplementedError(self.get_parents_with_ghosts)
397
def annotate_iter(self, version_id):
398
"""Yield list of (version-id, line) pairs for the specified
401
Must raise RevisionNotPresent if any of the given versions are
402
not present in file history.
404
raise NotImplementedError(self.annotate_iter)
406
def annotate(self, version_id):
407
return list(self.annotate_iter(version_id))
409
def _apply_delta(self, lines, delta):
410
"""Apply delta to lines."""
413
for start, end, count, delta_lines in delta:
414
lines[offset+start:offset+end] = delta_lines
415
offset = offset + (start - end) + count
418
def join(self, other, pb=None, msg=None, version_ids=None,
419
ignore_missing=False):
420
"""Integrate versions from other into this versioned file.
422
If version_ids is None all versions from other should be
423
incorporated into this versioned file.
425
Must raise RevisionNotPresent if any of the specified versions
426
are not present in the other files history unless ignore_missing
427
is supplied when they are silently skipped.
429
self._check_write_ok()
430
return InterVersionedFile.get(other, self).join(
436
def iter_lines_added_or_present_in_versions(self, version_ids=None,
438
"""Iterate over the lines in the versioned file from version_ids.
440
This may return lines from other versions, and does not return the
441
specific version marker at this point. The api may be changed
442
during development to include the version that the versioned file
443
thinks is relevant, but given that such hints are just guesses,
444
its better not to have it if we don't need it.
446
If a progress bar is supplied, it may be used to indicate progress.
447
The caller is responsible for cleaning up progress bars (because this
450
NOTES: Lines are normalised: they will all have \n terminators.
451
Lines are returned in arbitrary order.
453
raise NotImplementedError(self.iter_lines_added_or_present_in_versions)
455
def iter_parents(self, version_ids):
456
"""Iterate through the parents for many version ids.
458
:param version_ids: An iterable yielding version_ids.
459
:return: An iterator that yields (version_id, parents). Requested
460
version_ids not present in the versioned file are simply skipped.
461
The order is undefined, allowing for different optimisations in
462
the underlying implementation.
464
for version_id in version_ids:
466
yield version_id, tuple(self.get_parents(version_id))
467
except errors.RevisionNotPresent:
470
def transaction_finished(self):
471
"""The transaction that this file was opened in has finished.
473
This records self.finished = True and should cause all mutating
478
def plan_merge(self, ver_a, ver_b):
479
"""Return pseudo-annotation indicating how the two versions merge.
481
This is computed between versions a and b and their common
484
Weave lines present in none of them are skipped entirely.
487
killed-base Dead in base revision
488
killed-both Killed in each revision
491
unchanged Alive in both a and b (possibly created in both)
494
ghost-a Killed in a, unborn in b
495
ghost-b Killed in b, unborn in a
496
irrelevant Not in either revision
498
raise NotImplementedError(VersionedFile.plan_merge)
500
def weave_merge(self, plan, a_marker=TextMerge.A_MARKER,
501
b_marker=TextMerge.B_MARKER):
502
return PlanWeaveMerge(plan, a_marker, b_marker).merge_lines()[0]
505
class PlanWeaveMerge(TextMerge):
506
"""Weave merge that takes a plan as its input.
508
This exists so that VersionedFile.plan_merge is implementable.
509
Most callers will want to use WeaveMerge instead.
512
def __init__(self, plan, a_marker=TextMerge.A_MARKER,
513
b_marker=TextMerge.B_MARKER):
514
TextMerge.__init__(self, a_marker, b_marker)
517
def _merge_struct(self):
522
def outstanding_struct():
523
if not lines_a and not lines_b:
525
elif ch_a and not ch_b:
528
elif ch_b and not ch_a:
530
elif lines_a == lines_b:
533
yield (lines_a, lines_b)
535
# We previously considered either 'unchanged' or 'killed-both' lines
536
# to be possible places to resynchronize. However, assuming agreement
537
# on killed-both lines may be too aggressive. -- mbp 20060324
538
for state, line in self.plan:
539
if state == 'unchanged':
540
# resync and flush queued conflicts changes if any
541
for struct in outstanding_struct():
547
if state == 'unchanged':
550
elif state == 'killed-a':
553
elif state == 'killed-b':
556
elif state == 'new-a':
559
elif state == 'new-b':
563
assert state in ('irrelevant', 'ghost-a', 'ghost-b',
564
'killed-base', 'killed-both'), state
565
for struct in outstanding_struct():
569
class WeaveMerge(PlanWeaveMerge):
570
"""Weave merge that takes a VersionedFile and two versions as its input"""
572
def __init__(self, versionedfile, ver_a, ver_b,
573
a_marker=PlanWeaveMerge.A_MARKER, b_marker=PlanWeaveMerge.B_MARKER):
574
plan = versionedfile.plan_merge(ver_a, ver_b)
575
PlanWeaveMerge.__init__(self, plan, a_marker, b_marker)
578
class InterVersionedFile(InterObject):
579
"""This class represents operations taking place between two versionedfiles..
581
Its instances have methods like join, and contain
582
references to the source and target versionedfiles these operations can be
585
Often we will provide convenience methods on 'versionedfile' which carry out
586
operations with another versionedfile - they will always forward to
587
InterVersionedFile.get(other).method_name(parameters).
591
"""The available optimised InterVersionedFile types."""
593
def join(self, pb=None, msg=None, version_ids=None, ignore_missing=False):
594
"""Integrate versions from self.source into self.target.
596
If version_ids is None all versions from source should be
597
incorporated into this versioned file.
599
Must raise RevisionNotPresent if any of the specified versions
600
are not present in the other files history unless ignore_missing is
601
supplied when they are silently skipped.
604
# - if the target is empty, just add all the versions from
605
# source to target, otherwise:
606
# - make a temporary versioned file of type target
607
# - insert the source content into it one at a time
609
if not self.target.versions():
612
# Make a new target-format versioned file.
613
temp_source = self.target.create_empty("temp", MemoryTransport())
615
version_ids = self._get_source_version_ids(version_ids, ignore_missing)
616
graph = self.source.get_graph(version_ids)
617
order = tsort.topo_sort(graph.items())
618
pb = ui.ui_factory.nested_progress_bar()
621
# TODO for incremental cross-format work:
622
# make a versioned file with the following content:
623
# all revisions we have been asked to join
624
# all their ancestors that are *not* in target already.
625
# the immediate parents of the above two sets, with
626
# empty parent lists - these versions are in target already
627
# and the incorrect version data will be ignored.
628
# TODO: for all ancestors that are present in target already,
629
# check them for consistent data, this requires moving sha1 from
631
# TODO: remove parent texts when they are not relevant any more for
632
# memory pressure reduction. RBC 20060313
633
# pb.update('Converting versioned data', 0, len(order))
634
for index, version in enumerate(order):
635
pb.update('Converting versioned data', index, len(order))
636
_, _, parent_text = target.add_lines(version,
637
self.source.get_parents(version),
638
self.source.get_lines(version),
639
parent_texts=parent_texts)
640
parent_texts[version] = parent_text
642
# this should hit the native code path for target
643
if target is not self.target:
644
return self.target.join(temp_source,
652
def _get_source_version_ids(self, version_ids, ignore_missing):
653
"""Determine the version ids to be used from self.source.
655
:param version_ids: The caller-supplied version ids to check. (None
656
for all). If None is in version_ids, it is stripped.
657
:param ignore_missing: if True, remove missing ids from the version
658
list. If False, raise RevisionNotPresent on
659
a missing version id.
660
:return: A set of version ids.
662
if version_ids is None:
663
# None cannot be in source.versions
664
return set(self.source.versions())
666
version_ids = [osutils.safe_revision_id(v) for v in version_ids]
668
return set(self.source.versions()).intersection(set(version_ids))
670
new_version_ids = set()
671
for version in version_ids:
674
if not self.source.has_version(version):
675
raise errors.RevisionNotPresent(version, str(self.source))
677
new_version_ids.add(version)
678
return new_version_ids
added
removed
bzrlib/versionedfile.py
