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, random_id=False,
83
"""Add a single text on top of the versioned file.
85
Must raise RevisionAlreadyPresent if the new version is
86
already present in file history.
88
Must raise RevisionNotPresent if any of the given parents are
89
not present in file history.
91
:param lines: A list of lines. Each line must be a bytestring. And all
92
of them except the last must be terminated with \n and contain no
93
other \n's. The last line may either contain no \n's or a single
94
terminated \n. If the lines list does meet this constraint the add
95
routine may error or may succeed - but you will be unable to read
96
the data back accurately. (Checking the lines have been split
97
correctly is expensive and extremely unlikely to catch bugs so it
98
is not done at runtime unless check_content is True.)
99
:param parent_texts: An optional dictionary containing the opaque
100
representations of some or all of the parents of version_id to
101
allow delta optimisations. VERY IMPORTANT: the texts must be those
102
returned by add_lines or data corruption can be caused.
103
:param left_matching_blocks: a hint about which areas are common
104
between the text and its left-hand-parent. The format is
105
the SequenceMatcher.get_matching_blocks format.
106
:param nostore_sha: Raise ExistingContent and do not add the lines to
107
the versioned file if the digest of the lines matches this.
108
:param random_id: If True a random id has been selected rather than
109
an id determined by some deterministic process such as a converter
110
from a foreign VCS. When True the backend may choose not to check
111
for uniqueness of the resulting key within the versioned file, so
112
this should only be done when the result is expected to be unique
114
:param check_content: If True, the lines supplied are verified to be
115
bytestrings that are correctly formed lines.
116
:return: The text sha1, the number of bytes in the text, and an opaque
117
representation of the inserted version which can be provided
118
back to future add_lines calls in the parent_texts dictionary.
120
self._check_write_ok()
121
return self._add_lines(version_id, parents, lines, parent_texts,
122
left_matching_blocks, nostore_sha, random_id, check_content)
124
def _add_lines(self, version_id, parents, lines, parent_texts,
125
left_matching_blocks, nostore_sha, random_id, check_content):
126
"""Helper to do the class specific add_lines."""
127
raise NotImplementedError(self.add_lines)
129
def add_lines_with_ghosts(self, version_id, parents, lines,
130
parent_texts=None, nostore_sha=None, random_id=False,
132
"""Add lines to the versioned file, allowing ghosts to be present.
134
This takes the same parameters as add_lines and returns the same.
136
self._check_write_ok()
137
return self._add_lines_with_ghosts(version_id, parents, lines,
138
parent_texts, nostore_sha, random_id, check_content)
140
def _add_lines_with_ghosts(self, version_id, parents, lines, parent_texts,
141
nostore_sha, random_id, check_content):
142
"""Helper to do class specific add_lines_with_ghosts."""
143
raise NotImplementedError(self.add_lines_with_ghosts)
145
def check(self, progress_bar=None):
146
"""Check the versioned file for integrity."""
147
raise NotImplementedError(self.check)
149
def _check_lines_not_unicode(self, lines):
150
"""Check that lines being added to a versioned file are not unicode."""
152
if line.__class__ is not str:
153
raise errors.BzrBadParameterUnicode("lines")
155
def _check_lines_are_lines(self, lines):
156
"""Check that the lines really are full lines without inline EOL."""
158
if '\n' in line[:-1]:
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)
206
def get_format_signature(self):
207
"""Get a text description of the data encoding in this file.
211
raise NotImplementedError(self.get_format_signature)
213
def make_mpdiffs(self, version_ids):
214
"""Create multiparent diffs for specified versions."""
215
knit_versions = set()
216
for version_id in version_ids:
217
knit_versions.add(version_id)
218
knit_versions.update(self.get_parents(version_id))
219
lines = dict(zip(knit_versions,
220
self._get_lf_split_line_list(knit_versions)))
222
for version_id in version_ids:
223
target = lines[version_id]
224
parents = [lines[p] for p in self.get_parents(version_id)]
226
left_parent_blocks = self._extract_blocks(version_id,
229
left_parent_blocks = None
230
diffs.append(multiparent.MultiParent.from_lines(target, parents,
234
def _extract_blocks(self, version_id, source, target):
237
def add_mpdiffs(self, records):
238
"""Add mpdiffs to this VersionedFile.
240
Records should be iterables of version, parents, expected_sha1,
241
mpdiff. mpdiff should be a MultiParent instance.
243
# Does this need to call self._check_write_ok()? (IanC 20070919)
245
mpvf = multiparent.MultiMemoryVersionedFile()
247
for version, parent_ids, expected_sha1, mpdiff in records:
248
versions.append(version)
249
mpvf.add_diff(mpdiff, version, parent_ids)
250
needed_parents = set()
251
for version, parent_ids, expected_sha1, mpdiff in records:
252
needed_parents.update(p for p in parent_ids
253
if not mpvf.has_version(p))
254
for parent_id, lines in zip(needed_parents,
255
self._get_lf_split_line_list(needed_parents)):
256
mpvf.add_version(lines, parent_id, [])
257
for (version, parent_ids, expected_sha1, mpdiff), lines in\
258
zip(records, mpvf.get_line_list(versions)):
259
if len(parent_ids) == 1:
260
left_matching_blocks = list(mpdiff.get_matching_blocks(0,
261
mpvf.get_diff(parent_ids[0]).num_lines()))
263
left_matching_blocks = None
264
_, _, version_text = self.add_lines(version, parent_ids, lines,
265
vf_parents, left_matching_blocks=left_matching_blocks)
266
vf_parents[version] = version_text
267
for (version, parent_ids, expected_sha1, mpdiff), sha1 in\
268
zip(records, self.get_sha1s(versions)):
269
if expected_sha1 != sha1:
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)
279
def get_sha1s(self, version_ids):
280
"""Get the stored sha1 sums for the given revisions.
282
:param version_ids: The names of the versions to lookup
283
:return: a list of sha1s in order according to the version_ids
285
raise NotImplementedError(self.get_sha1s)
287
def get_suffixes(self):
288
"""Return the file suffixes associated with this versioned file."""
289
raise NotImplementedError(self.get_suffixes)
291
def get_text(self, version_id):
292
"""Return version contents as a text string.
294
Raises RevisionNotPresent if version is not present in
297
return ''.join(self.get_lines(version_id))
298
get_string = get_text
300
def get_texts(self, version_ids):
301
"""Return the texts of listed versions as a list of strings.
303
Raises RevisionNotPresent if version is not present in
306
return [''.join(self.get_lines(v)) for v in version_ids]
308
def get_lines(self, version_id):
309
"""Return version contents as a sequence of lines.
311
Raises RevisionNotPresent if version is not present in
314
raise NotImplementedError(self.get_lines)
316
def _get_lf_split_line_list(self, version_ids):
317
return [StringIO(t).readlines() for t in self.get_texts(version_ids)]
319
def get_ancestry(self, version_ids, topo_sorted=True):
320
"""Return a list of all ancestors of given version(s). This
321
will not include the null revision.
323
This list will not be topologically sorted if topo_sorted=False is
326
Must raise RevisionNotPresent if any of the given versions are
327
not present in file history."""
328
if isinstance(version_ids, basestring):
329
version_ids = [version_ids]
330
raise NotImplementedError(self.get_ancestry)
332
def get_ancestry_with_ghosts(self, version_ids):
333
"""Return a list of all ancestors of given version(s). This
334
will not include the null revision.
336
Must raise RevisionNotPresent if any of the given versions are
337
not present in file history.
339
Ghosts that are known about will be included in ancestry list,
340
but are not explicitly marked.
342
raise NotImplementedError(self.get_ancestry_with_ghosts)
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)
382
def get_parents_with_ghosts(self, version_id):
383
"""Return version names for parents of version_id.
385
Will raise RevisionNotPresent if version_id is not present
388
Ghosts that are known about will be included in the parent list,
389
but are not explicitly marked.
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)
402
def annotate(self, version_id):
403
return list(self.annotate_iter(version_id))
405
def join(self, other, pb=None, msg=None, version_ids=None,
406
ignore_missing=False):
407
"""Integrate versions from other into this versioned file.
409
If version_ids is None all versions from other should be
410
incorporated into this versioned file.
412
Must raise RevisionNotPresent if any of the specified versions
413
are not present in the other file's history unless ignore_missing
414
is supplied in which case they are silently skipped.
416
self._check_write_ok()
417
return InterVersionedFile.get(other, self).join(
423
def iter_lines_added_or_present_in_versions(self, version_ids=None,
425
"""Iterate over the lines in the versioned file from version_ids.
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.
433
If a progress bar is supplied, it may be used to indicate progress.
434
The caller is responsible for cleaning up progress bars (because this
437
NOTES: Lines are normalised: they will all have \n terminators.
438
Lines are returned in arbitrary order.
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
465
def plan_merge(self, ver_a, ver_b):
466
"""Return pseudo-annotation indicating how the two versions merge.
468
This is computed between versions a and b and their common
471
Weave lines present in none of them are skipped entirely.
474
killed-base Dead in base revision
475
killed-both Killed in each revision
478
unchanged Alive in both a and b (possibly created in both)
481
ghost-a Killed in a, unborn in b
482
ghost-b Killed in b, unborn in a
483
irrelevant Not in either revision
485
raise NotImplementedError(VersionedFile.plan_merge)
487
def weave_merge(self, plan, a_marker=TextMerge.A_MARKER,
488
b_marker=TextMerge.B_MARKER):
489
return PlanWeaveMerge(plan, a_marker, b_marker).merge_lines()[0]
492
class PlanWeaveMerge(TextMerge):
493
"""Weave merge that takes a plan as its input.
495
This exists so that VersionedFile.plan_merge is implementable.
496
Most callers will want to use WeaveMerge instead.
499
def __init__(self, plan, a_marker=TextMerge.A_MARKER,
500
b_marker=TextMerge.B_MARKER):
501
TextMerge.__init__(self, a_marker, b_marker)
504
def _merge_struct(self):
509
def outstanding_struct():
510
if not lines_a and not lines_b:
512
elif ch_a and not ch_b:
515
elif ch_b and not ch_a:
517
elif lines_a == lines_b:
520
yield (lines_a, lines_b)
522
# We previously considered either 'unchanged' or 'killed-both' lines
523
# to be possible places to resynchronize. However, assuming agreement
524
# on killed-both lines may be too aggressive. -- mbp 20060324
525
for state, line in self.plan:
526
if state == 'unchanged':
527
# resync and flush queued conflicts changes if any
528
for struct in outstanding_struct():
534
if state == 'unchanged':
537
elif state == 'killed-a':
540
elif state == 'killed-b':
543
elif state == 'new-a':
546
elif state == 'new-b':
550
assert state in ('irrelevant', 'ghost-a', 'ghost-b',
551
'killed-base', 'killed-both'), state
552
for struct in outstanding_struct():
556
class WeaveMerge(PlanWeaveMerge):
557
"""Weave merge that takes a VersionedFile and two versions as its input."""
559
def __init__(self, versionedfile, ver_a, ver_b,
560
a_marker=PlanWeaveMerge.A_MARKER, b_marker=PlanWeaveMerge.B_MARKER):
561
plan = versionedfile.plan_merge(ver_a, ver_b)
562
PlanWeaveMerge.__init__(self, plan, a_marker, b_marker)
565
class InterVersionedFile(InterObject):
566
"""This class represents operations taking place between two VersionedFiles.
568
Its instances have methods like join, and contain
569
references to the source and target versionedfiles these operations can be
572
Often we will provide convenience methods on 'versionedfile' which carry out
573
operations with another versionedfile - they will always forward to
574
InterVersionedFile.get(other).method_name(parameters).
578
"""The available optimised InterVersionedFile types."""
580
def join(self, pb=None, msg=None, version_ids=None, ignore_missing=False):
581
"""Integrate versions from self.source into self.target.
583
If version_ids is None all versions from source should be
584
incorporated into this versioned file.
586
Must raise RevisionNotPresent if any of the specified versions
587
are not present in the other file's history unless ignore_missing is
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())
602
version_ids = self._get_source_version_ids(version_ids, ignore_missing)
603
graph = self.source.get_graph(version_ids)
604
order = tsort.topo_sort(graph.items())
605
pb = ui.ui_factory.nested_progress_bar()
608
# TODO for incremental cross-format work:
609
# make a versioned file with the following content:
610
# all revisions we have been asked to join
611
# all their ancestors that are *not* in target already.
612
# the immediate parents of the above two sets, with
613
# empty parent lists - these versions are in target already
614
# and the incorrect version data will be ignored.
615
# TODO: for all ancestors that are present in target already,
616
# check them for consistent data, this requires moving sha1 from
618
# TODO: remove parent texts when they are not relevant any more for
619
# memory pressure reduction. RBC 20060313
620
# pb.update('Converting versioned data', 0, len(order))
622
for index, version in enumerate(order):
623
pb.update('Converting versioned data', index, total)
624
_, _, parent_text = target.add_lines(version,
625
self.source.get_parents(version),
626
self.source.get_lines(version),
627
parent_texts=parent_texts)
628
parent_texts[version] = parent_text
630
# this should hit the native code path for target
631
if target is not self.target:
632
return self.target.join(temp_source,
642
def _get_source_version_ids(self, version_ids, ignore_missing):
643
"""Determine the version ids to be used from self.source.
645
:param version_ids: The caller-supplied version ids to check. (None
646
for all). If None is in version_ids, it is stripped.
647
:param ignore_missing: if True, remove missing ids from the version
648
list. If False, raise RevisionNotPresent on
649
a missing version id.
650
:return: A set of version ids.
652
if version_ids is None:
653
# None cannot be in source.versions
654
return set(self.source.versions())
657
return set(self.source.versions()).intersection(set(version_ids))
659
new_version_ids = set()
660
for version in version_ids:
663
if not self.source.has_version(version):
664
raise errors.RevisionNotPresent(version, str(self.source))
666
new_version_ids.add(version)
667
return new_version_ids
added
removed
bzrlib/versionedfile.py
