« back to all changes in this revision
Viewing changes to bzrlib/weave.py
- browse files at revision 1417.1.10
- compare with another revision
- download diff
- download tarball
- view history from revision 1417.1.10
- Committer: Robert Collins
- Date: 2005-10-11 01:05:24 UTC
- mto: This revision was merged to the branch mainline in revision 1438.
- Revision ID: robertc@robertcollins.net-20051011010524-e258bc8d051cc9d2
add a cache bound to Transactions, and a precious facility, so that we keep inventory.weave in memory, but can discard weaves for other such files.
- files added:
- HACKING
- bzrlib/plugins
- bzrlib/selftest
- bzrlib/store
- bzrlib/transport
- bzrlib/util
- bzrlib/util/effbot
- bzrlib/util/effbot/org
- contrib
- contrib/bash
- contrib/emacs
- contrib/zsh
- patches
- tools
- files removed:
- bzrlib/remotebranch.py
- files renamed:
- bzrlib/store.py => bzrlib/store/__init__.py
- elementtree/ => bzrlib/util/elementtree/
- urlgrabber/ => bzrlib/util/urlgrabber/
- bzrlib/newinventory.py => contrib/newinventory.py
- files modified:
- .bzrignore
- bzr *
- setup.py *
added
removed
bzrlib/weave.py
1
#! /usr/bin/python
2
3
# Copyright (C) 2005 Canonical Ltd
4
5
# This program is free software; you can redistribute it and/or modify
6
# it under the terms of the GNU General Public License as published by
7
# the Free Software Foundation; either version 2 of the License, or
8
# (at your option) any later version.
9
10
# This program is distributed in the hope that it will be useful,
11
# but WITHOUT ANY WARRANTY; without even the implied warranty of
12
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13
# GNU General Public License for more details.
14
15
# You should have received a copy of the GNU General Public License
16
# along with this program; if not, write to the Free Software
17
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
19
# Author: Martin Pool <mbp@canonical.com>
20
21
22
"""Weave - storage of related text file versions"""
23
24
25
# XXX: If we do weaves this way, will a merge still behave the same
26
# way if it's done in a different order? That's a pretty desirable
27
# property.
28
29
# TODO: Nothing here so far assumes the lines are really \n newlines,
30
# rather than being split up in some other way. We could accomodate
31
# binaries, perhaps by naively splitting on \n or perhaps using
32
# something like a rolling checksum.
33
34
# TODO: End marker for each version so we can stop reading?
35
36
# TODO: Check that no insertion occurs inside a deletion that was
37
# active in the version of the insertion.
38
39
# TODO: In addition to the SHA-1 check, perhaps have some code that
40
# checks structural constraints of the weave: ie that insertions are
41
# properly nested, that there is no text outside of an insertion, that
42
# insertions or deletions are not repeated, etc.
43
44
# TODO: Parallel-extract that passes back each line along with a
45
# description of which revisions include it. Nice for checking all
46
# shas or calculating stats in parallel.
47
48
# TODO: Using a single _extract routine and then processing the output
49
# is probably inefficient. It's simple enough that we can afford to
50
# have slight specializations for different ways its used: annotate,
51
# basis for add, get, etc.
52
53
# TODO: Perhaps the API should work only in names to hide the integer
54
# indexes from the user?
55
56
# TODO: Is there any potential performance win by having an add()
57
# variant that is passed a pre-cooked version of the single basis
58
# version?
59
60
# TODO: Have a way to go back and insert a revision V1 that is a parent
61
# of an already-stored revision V2. This means some lines previously
62
# counted as new in V2 will be discovered to have actually come from V1.
63
# It is probably necessary to insert V1, then compute a whole new diff
64
# from the mashed ancestors to V2. This must be repeated for every
65
# direct child of V1. The deltas from V2 to its descendents won't change,
66
# although their location within the weave may change. It may be possible
67
# to just adjust the location of those instructions rather than
68
# re-weaving the whole thing. This is expected to be a fairly rare
69
# operation, only used when inserting data that was previously a ghost.
70
71
72
73
74
import sha
75
from difflib import SequenceMatcher
76
77
from bzrlib.trace import mutter
78
79
80
class WeaveError(Exception):
81
"""Exception in processing weave"""
82
83
84
class WeaveFormatError(WeaveError):
85
"""Weave invariant violated"""
86
87
88
class WeaveParentMismatch(WeaveError):
89
"""Parents are mismatched between two revisions."""
90
91
92
class Weave(object):
93
"""weave - versioned text file storage.
94
95
A Weave manages versions of line-based text files, keeping track
96
of the originating version for each line.
97
98
To clients the "lines" of the file are represented as a list of strings.
99
These strings will typically have terminal newline characters, but
100
this is not required. In particular files commonly do not have a newline
101
at the end of the file.
102
103
Texts can be identified in either of two ways:
104
105
* a nonnegative index number.
106
107
* a version-id string.
108
109
Typically the index number will be valid only inside this weave and
110
the version-id is used to reference it in the larger world.
111
112
The weave is represented as a list mixing edit instructions and
113
literal text. Each entry in _weave can be either a string (or
114
unicode), or a tuple. If a string, it means that the given line
115
should be output in the currently active revisions.
116
117
If a tuple, it gives a processing instruction saying in which
118
revisions the enclosed lines are active. The tuple has the form
119
(instruction, version).
120
121
The instruction can be '{' or '}' for an insertion block, and '['
122
and ']' for a deletion block respectively. The version is the
123
integer version index. There is no replace operator, only deletes
124
and inserts. For '}', the end of an insertion, there is no
125
version parameter because it always closes the most recently
126
opened insertion.
127
128
Constraints/notes:
129
130
* A later version can delete lines that were introduced by any
131
number of ancestor versions; this implies that deletion
132
instructions can span insertion blocks without regard to the
133
insertion block's nesting.
134
135
* Similarly, deletions need not be properly nested with regard to
136
each other, because they might have been generated by
137
independent revisions.
138
139
* Insertions are always made by inserting a new bracketed block
140
into a single point in the previous weave. This implies they
141
can nest but not overlap, and the nesting must always have later
142
insertions on the inside.
143
144
* It doesn't seem very useful to have an active insertion
145
inside an inactive insertion, but it might happen.
146
147
* Therefore, all instructions are always"considered"; that
148
is passed onto and off the stack. An outer inactive block
149
doesn't disable an inner block.
150
151
* Lines are enabled if the most recent enclosing insertion is
152
active and none of the enclosing deletions are active.
153
154
* There is no point having a deletion directly inside its own
155
insertion; you might as well just not write it. And there
156
should be no way to get an earlier version deleting a later
157
version.
158
159
_weave
160
Text of the weave; list of control instruction tuples and strings.
161
162
_parents
163
List of parents, indexed by version number.
164
It is only necessary to store the minimal set of parents for
165
each version; the parent's parents are implied.
166
167
_sha1s
168
List of hex SHA-1 of each version.
169
170
_names
171
List of symbolic names for each version. Each should be unique.
172
173
_name_map
174
For each name, the version number.
175
176
_weave_name
177
Descriptive name of this weave; typically the filename if known.
178
Set by read_weave.
179
"""
180
181
__slots__ = ['_weave', '_parents', '_sha1s', '_names', '_name_map',
182
'_weave_name']
183
184
def __init__(self, weave_name=None):
185
self._weave = []
186
self._parents = []
187
self._sha1s = []
188
self._names = []
189
self._name_map = {}
190
self._weave_name = weave_name
191
192
193
def copy(self):
194
"""Return a deep copy of self.
195
196
The copy can be modified without affecting the original weave."""
197
other = Weave()
198
other._weave = self._weave[:]
199
other._parents = self._parents[:]
200
other._sha1s = self._sha1s[:]
201
other._names = self._names[:]
202
other._name_map = self._name_map.copy()
203
other._weave_name = self._weave_name
204
return other
205
206
def __eq__(self, other):
207
if not isinstance(other, Weave):
208
return False
209
return self._parents == other._parents \
210
and self._weave == other._weave \
211
and self._sha1s == other._sha1s
212
213
214
def __ne__(self, other):
215
return not self.__eq__(other)
216
217
218
def maybe_lookup(self, name_or_index):
219
"""Convert possible symbolic name to index, or pass through indexes."""
220
if isinstance(name_or_index, (int, long)):
221
return name_or_index
222
else:
223
return self.lookup(name_or_index)
224
225
226
def lookup(self, name):
227
"""Convert symbolic version name to index."""
228
try:
229
return self._name_map[name]
230
except KeyError:
231
raise WeaveError("name %r not present in weave %r" %
232
(name, self._weave_name))
233
234
235
def iter_names(self):
236
"""Yield a list of all names in this weave."""
237
return iter(self._names)
238
239
def idx_to_name(self, version):
240
return self._names[version]
241
242
243
def _check_repeated_add(self, name, parents, text, sha1):
244
"""Check that a duplicated add is OK.
245
246
If it is, return the (old) index; otherwise raise an exception.
247
"""
248
idx = self.lookup(name)
249
if sorted(self._parents[idx]) != sorted(parents):
250
raise WeaveError("name \"%s\" already present in weave "
251
"with different parents" % name)
252
if sha1 != self._sha1s[idx]:
253
raise WeaveError("name \"%s\" already present in weave "
254
"with different text" % name)
255
return idx
256
257
258
259
def add(self, name, parents, text, sha1=None):
260
"""Add a single text on top of the weave.
261
262
Returns the index number of the newly added version.
263
264
name
265
Symbolic name for this version.
266
(Typically the revision-id of the revision that added it.)
267
268
parents
269
List or set of direct parent version numbers.
270
271
text
272
Sequence of lines to be added in the new version.
273
274
sha -- SHA-1 of the file, if known. This is trusted to be
275
correct if supplied.
276
"""
277
from bzrlib.osutils import sha_strings
278
279
assert isinstance(name, basestring)
280
if sha1 is None:
281
sha1 = sha_strings(text)
282
if name in self._name_map:
283
return self._check_repeated_add(name, parents, text, sha1)
284
285
parents = map(self.maybe_lookup, parents)
286
self._check_versions(parents)
287
## self._check_lines(text)
288
new_version = len(self._parents)
289
290
291
# if we abort after here the (in-memory) weave will be corrupt because only
292
# some fields are updated
293
self._parents.append(parents[:])
294
self._sha1s.append(sha1)
295
self._names.append(name)
296
self._name_map[name] = new_version
297
298
299
if not parents:
300
# special case; adding with no parents revision; can do
301
# this more quickly by just appending unconditionally.
302
# even more specially, if we're adding an empty text we
303
# need do nothing at all.
304
if text:
305
self._weave.append(('{', new_version))
306
self._weave.extend(text)
307
self._weave.append(('}', None))
308
309
return new_version
310
311
if len(parents) == 1:
312
pv = list(parents)[0]
313
if sha1 == self._sha1s[pv]:
314
# special case: same as the single parent
315
return new_version
316
317
318
ancestors = self.inclusions(parents)
319
320
l = self._weave
321
322
# basis a list of (origin, lineno, line)
323
basis_lineno = []
324
basis_lines = []
325
for origin, lineno, line in self._extract(ancestors):
326
basis_lineno.append(lineno)
327
basis_lines.append(line)
328
329
# another small special case: a merge, producing the same text
330
# as auto-merge
331
if text == basis_lines:
332
return new_version
333
334
# add a sentinal, because we can also match against the final line
335
basis_lineno.append(len(self._weave))
336
337
# XXX: which line of the weave should we really consider
338
# matches the end of the file? the current code says it's the
339
# last line of the weave?
340
341
#print 'basis_lines:', basis_lines
342
#print 'new_lines: ', lines
343
344
s = SequenceMatcher(None, basis_lines, text)
345
346
# offset gives the number of lines that have been inserted
347
# into the weave up to the current point; if the original edit instruction
348
# says to change line A then we actually change (A+offset)
349
offset = 0
350
351
for tag, i1, i2, j1, j2 in s.get_opcodes():
352
# i1,i2 are given in offsets within basis_lines; we need to map them
353
# back to offsets within the entire weave
354
#print 'raw match', tag, i1, i2, j1, j2
355
if tag == 'equal':
356
continue
357
358
i1 = basis_lineno[i1]
359
i2 = basis_lineno[i2]
360
361
assert 0 <= j1 <= j2 <= len(text)
362
363
#print tag, i1, i2, j1, j2
364
365
# the deletion and insertion are handled separately.
366
# first delete the region.
367
if i1 != i2:
368
self._weave.insert(i1+offset, ('[', new_version))
369
self._weave.insert(i2+offset+1, (']', new_version))
370
offset += 2
371
372
if j1 != j2:
373
# there may have been a deletion spanning up to
374
# i2; we want to insert after this region to make sure
375
# we don't destroy ourselves
376
i = i2 + offset
377
self._weave[i:i] = ([('{', new_version)]
378
+ text[j1:j2]
379
+ [('}', None)])
380
offset += 2 + (j2 - j1)
381
382
return new_version
383
384
def add_identical(self, old_rev_id, new_rev_id, parents):
385
"""Add an identical text to old_rev_id as new_rev_id."""
386
old_lines = self.get(self.lookup(old_rev_id))
387
self.add(new_rev_id, parents, old_lines)
388
389
def inclusions(self, versions):
390
"""Return set of all ancestors of given version(s)."""
391
i = set(versions)
392
for v in xrange(max(versions), 0, -1):
393
if v in i:
394
# include all its parents
395
i.update(self._parents[v])
396
return i
397
## except IndexError:
398
## raise ValueError("version %d not present in weave" % v)
399
400
401
def parents(self, version):
402
return self._parents[version]
403
404
405
def parent_names(self, version):
406
"""Return version names for parents of a version."""
407
return map(self.idx_to_name, self._parents[self.lookup(version)])
408
409
410
def minimal_parents(self, version):
411
"""Find the minimal set of parents for the version."""
412
included = self._parents[version]
413
if not included:
414
return []
415
416
li = list(included)
417
li.sort(reverse=True)
418
419
mininc = []
420
gotit = set()
421
422
for pv in li:
423
if pv not in gotit:
424
mininc.append(pv)
425
gotit.update(self.inclusions(pv))
426
427
assert mininc[0] >= 0
428
assert mininc[-1] < version
429
return mininc
430
431
432
433
def _check_lines(self, text):
434
if not isinstance(text, list):
435
raise ValueError("text should be a list, not %s" % type(text))
436
437
for l in text:
438
if not isinstance(l, basestring):
439
raise ValueError("text line should be a string or unicode, not %s"
440
% type(l))
441
442
443
444
def _check_versions(self, indexes):
445
"""Check everything in the sequence of indexes is valid"""
446
for i in indexes:
447
try:
448
self._parents[i]
449
except IndexError:
450
raise IndexError("invalid version number %r" % i)
451
452
453
def annotate(self, name_or_index):
454
return list(self.annotate_iter(name_or_index))
455
456
457
def annotate_iter(self, name_or_index):
458
"""Yield list of (index-id, line) pairs for the specified version.
459
460
The index indicates when the line originated in the weave."""
461
incls = [self.maybe_lookup(name_or_index)]
462
for origin, lineno, text in self._extract(incls):
463
yield origin, text
464
465
466
def _walk(self):
467
"""Walk the weave.
468
469
Yields sequence of
470
(lineno, insert, deletes, text)
471
for each literal line.
472
"""
473
474
istack = []
475
dset = set()
476
477
lineno = 0 # line of weave, 0-based
478
479
for l in self._weave:
480
if isinstance(l, tuple):
481
c, v = l
482
isactive = None
483
if c == '{':
484
istack.append(v)
485
elif c == '}':
486
istack.pop()
487
elif c == '[':
488
assert v not in dset
489
dset.add(v)
490
elif c == ']':
491
dset.remove(v)
492
else:
493
raise WeaveFormatError('unexpected instruction %r'
494
% v)
495
else:
496
assert isinstance(l, basestring)
497
assert istack
498
yield lineno, istack[-1], dset, l
499
lineno += 1
500
501
502
503
def _extract(self, versions):
504
"""Yield annotation of lines in included set.
505
506
Yields a sequence of tuples (origin, lineno, text), where
507
origin is the origin version, lineno the index in the weave,
508
and text the text of the line.
509
510
The set typically but not necessarily corresponds to a version.
511
"""
512
for i in versions:
513
if not isinstance(i, int):
514
raise ValueError(i)
515
516
included = self.inclusions(versions)
517
518
istack = []
519
dset = set()
520
521
lineno = 0 # line of weave, 0-based
522
523
isactive = None
524
525
result = []
526
527
WFE = WeaveFormatError
528
529
for l in self._weave:
530
if isinstance(l, tuple):
531
c, v = l
532
isactive = None
533
if c == '{':
534
assert v not in istack
535
istack.append(v)
536
elif c == '}':
537
istack.pop()
538
elif c == '[':
539
if v in included:
540
assert v not in dset
541
dset.add(v)
542
else:
543
assert c == ']'
544
if v in included:
545
assert v in dset
546
dset.remove(v)
547
else:
548
assert isinstance(l, basestring)
549
if isactive is None:
550
isactive = (not dset) and istack and (istack[-1] in included)
551
if isactive:
552
result.append((istack[-1], lineno, l))
553
lineno += 1
554
555
if istack:
556
raise WFE("unclosed insertion blocks at end of weave",
557
istack)
558
if dset:
559
raise WFE("unclosed deletion blocks at end of weave",
560
dset)
561
562
return result
563
564
565
566
def get_iter(self, name_or_index):
567
"""Yield lines for the specified version."""
568
incls = [self.maybe_lookup(name_or_index)]
569
for origin, lineno, line in self._extract(incls):
570
yield line
571
572
573
def get_text(self, name_or_index):
574
return ''.join(self.get_iter(name_or_index))
575
assert isinstance(version, int)
576
577
578
def get_lines(self, name_or_index):
579
return list(self.get_iter(name_or_index))
580
581
582
get = get_lines
583
584
585
def mash_iter(self, included):
586
"""Return composed version of multiple included versions."""
587
included = map(self.maybe_lookup, included)
588
for origin, lineno, text in self._extract(included):
589
yield text
590
591
592
def dump(self, to_file):
593
from pprint import pprint
594
print >>to_file, "Weave._weave = ",
595
pprint(self._weave, to_file)
596
print >>to_file, "Weave._parents = ",
597
pprint(self._parents, to_file)
598
599
600
601
def numversions(self):
602
l = len(self._parents)
603
assert l == len(self._sha1s)
604
return l
605
606
607
def __len__(self):
608
return self.numversions()
609
610
611
def check(self, progress_bar=None):
612
# check no circular inclusions
613
for version in range(self.numversions()):
614
inclusions = list(self._parents[version])
615
if inclusions:
616
inclusions.sort()
617
if inclusions[-1] >= version:
618
raise WeaveFormatError("invalid included version %d for index %d"
619
% (inclusions[-1], version))
620
621
# try extracting all versions; this is a bit slow and parallel
622
# extraction could be used
623
nv = self.numversions()
624
for version in range(nv):
625
if progress_bar:
626
progress_bar.update('checking text', version, nv)
627
s = sha.new()
628
for l in self.get_iter(version):
629
s.update(l)
630
hd = s.hexdigest()
631
expected = self._sha1s[version]
632
if hd != expected:
633
raise WeaveError("mismatched sha1 for version %d; "
634
"got %s, expected %s"
635
% (version, hd, expected))
636
637
# TODO: check insertions are properly nested, that there are
638
# no lines outside of insertion blocks, that deletions are
639
# properly paired, etc.
640
641
642
643
def merge(self, merge_versions):
644
"""Automerge and mark conflicts between versions.
645
646
This returns a sequence, each entry describing alternatives
647
for a chunk of the file. Each of the alternatives is given as
648
a list of lines.
649
650
If there is a chunk of the file where there's no diagreement,
651
only one alternative is given.
652
"""
653
# approach: find the included versions common to all the
654
# merged versions
655
raise NotImplementedError()
656
657
658
659
def _delta(self, included, lines):
660
"""Return changes from basis to new revision.
661
662
The old text for comparison is the union of included revisions.
663
664
This is used in inserting a new text.
665
666
Delta is returned as a sequence of
667
(weave1, weave2, newlines).
668
669
This indicates that weave1:weave2 of the old weave should be
670
replaced by the sequence of lines in newlines. Note that
671
these line numbers are positions in the total weave and don't
672
correspond to the lines in any extracted version, or even the
673
extracted union of included versions.
674
675
If line1=line2, this is a pure insert; if newlines=[] this is a
676
pure delete. (Similar to difflib.)
677
"""
678
raise NotImplementedError()
679
680
681
def plan_merge(self, ver_a, ver_b):
682
"""Return pseudo-annotation indicating how the two versions merge.
683
684
This is computed between versions a and b and their common
685
base.
686
687
Weave lines present in none of them are skipped entirely.
688
"""
689
inc_a = self.inclusions([ver_a])
690
inc_b = self.inclusions([ver_b])
691
inc_c = inc_a & inc_b
692
693
for lineno, insert, deleteset, line in self._walk():
694
if deleteset & inc_c:
695
# killed in parent; can't be in either a or b
696
# not relevant to our work
697
yield 'killed-base', line
698
elif insert in inc_c:
699
# was inserted in base
700
killed_a = bool(deleteset & inc_a)
701
killed_b = bool(deleteset & inc_b)
702
if killed_a and killed_b:
703
yield 'killed-both', line
704
elif killed_a:
705
yield 'killed-a', line
706
elif killed_b:
707
yield 'killed-b', line
708
else:
709
yield 'unchanged', line
710
elif insert in inc_a:
711
if deleteset & inc_a:
712
yield 'ghost-a', line
713
else:
714
# new in A; not in B
715
yield 'new-a', line
716
elif insert in inc_b:
717
if deleteset & inc_b:
718
yield 'ghost-b', line
719
else:
720
yield 'new-b', line
721
else:
722
# not in either revision
723
yield 'irrelevant', line
724
725
yield 'unchanged', '' # terminator
726
727
728
729
def weave_merge(self, plan):
730
lines_a = []
731
lines_b = []
732
ch_a = ch_b = False
733
734
for state, line in plan:
735
if state == 'unchanged' or state == 'killed-both':
736
# resync and flush queued conflicts changes if any
737
if not lines_a and not lines_b:
738
pass
739
elif ch_a and not ch_b:
740
# one-sided change:
741
for l in lines_a: yield l
742
elif ch_b and not ch_a:
743
for l in lines_b: yield l
744
elif lines_a == lines_b:
745
for l in lines_a: yield l
746
else:
747
yield '<<<<\n'
748
for l in lines_a: yield l
749
yield '====\n'
750
for l in lines_b: yield l
751
yield '>>>>\n'
752
753
del lines_a[:]
754
del lines_b[:]
755
ch_a = ch_b = False
756
757
if state == 'unchanged':
758
if line:
759
yield line
760
elif state == 'killed-a':
761
ch_a = True
762
lines_b.append(line)
763
elif state == 'killed-b':
764
ch_b = True
765
lines_a.append(line)
766
elif state == 'new-a':
767
ch_a = True
768
lines_a.append(line)
769
elif state == 'new-b':
770
ch_b = True
771
lines_b.append(line)
772
else:
773
assert state in ('irrelevant', 'ghost-a', 'ghost-b', 'killed-base',
774
'killed-both'), \
775
state
776
777
778
def join(self, other):
779
"""Integrate versions from other into this weave.
780
781
The resulting weave contains all the history of both weaves;
782
any version you could retrieve from either self or other can be
783
retrieved from self after this call.
784
785
It is illegal for the two weaves to contain different values
786
or different parents for any version. See also reweave().
787
"""
788
if other.numversions() == 0:
789
return # nothing to update, easy
790
# two loops so that we do not change ourselves before verifying it
791
# will be ok
792
# work through in index order to make sure we get all dependencies
793
for other_idx, name in enumerate(other._names):
794
if self._check_version_consistent(other, other_idx, name):
795
continue
796
for other_idx, name in enumerate(other._names):
797
# TODO: If all the parents of the other version are already
798
# present then we can avoid some work by just taking the delta
799
# and adjusting the offsets.
800
new_parents = self._imported_parents(other, other_idx)
801
lines = other.get_lines(other_idx)
802
sha1 = other._sha1s[other_idx]
803
self.add(name, new_parents, lines, sha1)
804
805
806
def _imported_parents(self, other, other_idx):
807
"""Return list of parents in self corresponding to indexes in other."""
808
new_parents = []
809
for parent_idx in other._parents[other_idx]:
810
parent_name = other._names[parent_idx]
811
if parent_name not in self._names:
812
# should not be possible
813
raise WeaveError("missing parent {%s} of {%s} in %r"
814
% (parent_name, other._name_map[other_idx], self))
815
new_parents.append(self._name_map[parent_name])
816
return new_parents
817
818
def _check_version_consistent(self, other, other_idx, name):
819
"""Check if a version in consistent in this and other.
820
821
To be consistent it must have:
822
823
* the same text
824
* the same direct parents (by name, not index, and disregarding
825
order)
826
827
If present & correct return True;
828
if not present in self return False;
829
if inconsistent raise error."""
830
this_idx = self._name_map.get(name, -1)
831
if this_idx != -1:
832
if self._sha1s[this_idx] != other._sha1s[other_idx]:
833
raise WeaveError("inconsistent texts for version {%s} "
834
"when joining weaves"
835
% (name))
836
self_parents = self._parents[this_idx]
837
other_parents = other._parents[other_idx]
838
n1 = [self._names[i] for i in self_parents]
839
n2 = [other._names[i] for i in other_parents]
840
n1.sort()
841
n2.sort()
842
if n1 != n2:
843
raise WeaveParentMismatch("inconsistent parents "
844
"for version {%s}: %s vs %s" % (name, n1, n2))
845
else:
846
return True # ok!
847
else:
848
return False
849
850
def reweave(self, other):
851
"""Reweave self with other."""
852
new_weave = reweave(self, other)
853
for attr in self.__slots__:
854
setattr(self, attr, getattr(new_weave, attr))
855
856
857
def reweave(wa, wb):
858
"""Combine two weaves and return the result.
859
860
This works even if a revision R has different parents in
861
wa and wb. In the resulting weave all the parents are given.
862
863
This is done by just building up a new weave, maintaining ordering
864
of the versions in the two inputs. More efficient approaches
865
might be possible but it should only be necessary to do
866
this operation rarely, when a new previously ghost version is
867
inserted.
868
"""
869
wr = Weave()
870
ia = ib = 0
871
queue_a = range(wa.numversions())
872
queue_b = range(wb.numversions())
873
# first determine combined parents of all versions
874
# map from version name -> all parent names
875
combined_parents = _reweave_parent_graphs(wa, wb)
876
mutter("combined parents: %r", combined_parents)
877
order = _make_reweave_order(wa._names, wb._names, combined_parents)
878
mutter("order to reweave: %r", order)
879
for name in order:
880
if name in wa._name_map:
881
lines = wa.get_lines(name)
882
if name in wb._name_map:
883
assert lines == wb.get_lines(name)
884
else:
885
lines = wb.get_lines(name)
886
wr.add(name, combined_parents[name], lines)
887
return wr
888
889
890
def _reweave_parent_graphs(wa, wb):
891
"""Return combined parent ancestry for two weaves.
892
893
Returned as a list of (version_name, set(parent_names))"""
894
combined = {}
895
for weave in [wa, wb]:
896
for idx, name in enumerate(weave._names):
897
p = combined.setdefault(name, set())
898
p.update(map(weave.idx_to_name, weave._parents[idx]))
899
return combined
900
901
902
def _make_reweave_order(wa_order, wb_order, combined_parents):
903
"""Return an order to reweave versions respecting parents."""
904
done = set()
905
result = []
906
ia = ib = 0
907
next_a = next_b = None
908
len_a = len(wa_order)
909
len_b = len(wb_order)
910
while ia < len(wa_order) or ib < len(wb_order):
911
if ia < len_a:
912
next_a = wa_order[ia]
913
if next_a in done:
914
ia += 1
915
continue
916
if combined_parents[next_a].issubset(done):
917
ia += 1
918
result.append(next_a)
919
done.add(next_a)
920
continue
921
if ib < len_b:
922
next_b = wb_order[ib]
923
if next_b in done:
924
ib += 1
925
continue
926
elif combined_parents[next_b].issubset(done):
927
ib += 1
928
result.append(next_b)
929
done.add(next_b)
930
continue
931
raise WeaveError("don't know how to reweave at {%s} and {%s}"
932
% (next_a, next_b))
933
return result
934
935
936
def weave_toc(w):
937
"""Show the weave's table-of-contents"""
938
print '%6s %50s %10s %10s' % ('ver', 'name', 'sha1', 'parents')
939
for i in (6, 50, 10, 10):
940
print '-' * i,
941
print
942
for i in range(w.numversions()):
943
sha1 = w._sha1s[i]
944
name = w._names[i]
945
parent_str = ' '.join(map(str, w._parents[i]))
946
print '%6d %-50.50s %10.10s %s' % (i, name, sha1, parent_str)
947
948
949
950
def weave_stats(weave_file, pb):
951
from bzrlib.weavefile import read_weave
952
953
wf = file(weave_file, 'rb')
954
w = read_weave(wf)
955
# FIXME: doesn't work on pipes
956
weave_size = wf.tell()
957
958
total = 0
959
vers = len(w)
960
for i in range(vers):
961
pb.update('checking sizes', i, vers)
962
for origin, lineno, line in w._extract([i]):
963
total += len(line)
964
965
pb.clear()
966
967
print 'versions %9d' % vers
968
print 'weave file %9d bytes' % weave_size
969
print 'total contents %9d bytes' % total
970
print 'compression ratio %9.2fx' % (float(total) / float(weave_size))
971
if vers:
972
avg = total/vers
973
print 'average size %9d bytes' % avg
974
print 'relative size %9.2fx' % (float(weave_size) / float(avg))
975
976
977
def usage():
978
print """bzr weave tool
979
980
Experimental tool for weave algorithm.
981
982
usage:
983
weave init WEAVEFILE
984
Create an empty weave file
985
weave get WEAVEFILE VERSION
986
Write out specified version.
987
weave check WEAVEFILE
988
Check consistency of all versions.
989
weave toc WEAVEFILE
990
Display table of contents.
991
weave add WEAVEFILE NAME [BASE...] < NEWTEXT
992
Add NEWTEXT, with specified parent versions.
993
weave annotate WEAVEFILE VERSION
994
Display origin of each line.
995
weave mash WEAVEFILE VERSION...
996
Display composite of all selected versions.
997
weave merge WEAVEFILE VERSION1 VERSION2 > OUT
998
Auto-merge two versions and display conflicts.
999
weave diff WEAVEFILE VERSION1 VERSION2
1000
Show differences between two versions.
1001
1002
example:
1003
1004
% weave init foo.weave
1005
% vi foo.txt
1006
% weave add foo.weave ver0 < foo.txt
1007
added version 0
1008
1009
(create updated version)
1010
% vi foo.txt
1011
% weave get foo.weave 0 | diff -u - foo.txt
1012
% weave add foo.weave ver1 0 < foo.txt
1013
added version 1
1014
1015
% weave get foo.weave 0 > foo.txt (create forked version)
1016
% vi foo.txt
1017
% weave add foo.weave ver2 0 < foo.txt
1018
added version 2
1019
1020
% weave merge foo.weave 1 2 > foo.txt (merge them)
1021
% vi foo.txt (resolve conflicts)
1022
% weave add foo.weave merged 1 2 < foo.txt (commit merged version)
1023
1024
"""
1025
1026
1027
1028
def main(argv):
1029
import sys
1030
import os
1031
try:
1032
import bzrlib
1033
except ImportError:
1034
# in case we're run directly from the subdirectory
1035
sys.path.append('..')
1036
import bzrlib
1037
from bzrlib.weavefile import write_weave, read_weave
1038
from bzrlib.progress import ProgressBar
1039
1040
try:
1041
import psyco
1042
psyco.full()
1043
except ImportError:
1044
pass
1045
1046
if len(argv) < 2:
1047
usage()
1048
return 0
1049
1050
cmd = argv[1]
1051
1052
def readit():
1053
return read_weave(file(argv[2], 'rb'))
1054
1055
if cmd == 'help':
1056
usage()
1057
elif cmd == 'add':
1058
w = readit()
1059
# at the moment, based on everything in the file
1060
name = argv[3]
1061
parents = map(int, argv[4:])
1062
lines = sys.stdin.readlines()
1063
ver = w.add(name, parents, lines)
1064
write_weave(w, file(argv[2], 'wb'))
1065
print 'added version %r %d' % (name, ver)
1066
elif cmd == 'init':
1067
fn = argv[2]
1068
if os.path.exists(fn):
1069
raise IOError("file exists")
1070
w = Weave()
1071
write_weave(w, file(fn, 'wb'))
1072
elif cmd == 'get': # get one version
1073
w = readit()
1074
sys.stdout.writelines(w.get_iter(int(argv[3])))
1075
1076
elif cmd == 'mash': # get composite
1077
w = readit()
1078
sys.stdout.writelines(w.mash_iter(map(int, argv[3:])))
1079
1080
elif cmd == 'diff':
1081
from difflib import unified_diff
1082
w = readit()
1083
fn = argv[2]
1084
v1, v2 = map(int, argv[3:5])
1085
lines1 = w.get(v1)
1086
lines2 = w.get(v2)
1087
diff_gen = unified_diff(lines1, lines2,
1088
'%s version %d' % (fn, v1),
1089
'%s version %d' % (fn, v2))
1090
sys.stdout.writelines(diff_gen)
1091
1092
elif cmd == 'annotate':
1093
w = readit()
1094
# newline is added to all lines regardless; too hard to get
1095
# reasonable formatting otherwise
1096
lasto = None
1097
for origin, text in w.annotate(int(argv[3])):
1098
text = text.rstrip('\r\n')
1099
if origin == lasto:
1100
print ' | %s' % (text)
1101
else:
1102
print '%5d | %s' % (origin, text)
1103
lasto = origin
1104
1105
elif cmd == 'toc':
1106
weave_toc(readit())
1107
1108
elif cmd == 'stats':
1109
weave_stats(argv[2], ProgressBar())
1110
1111
elif cmd == 'check':
1112
w = readit()
1113
pb = ProgressBar()
1114
w.check(pb)
1115
pb.clear()
1116
print '%d versions ok' % w.numversions()
1117
1118
elif cmd == 'inclusions':
1119
w = readit()
1120
print ' '.join(map(str, w.inclusions([int(argv[3])])))
1121
1122
elif cmd == 'parents':
1123
w = readit()
1124
print ' '.join(map(str, w._parents[int(argv[3])]))
1125
1126
elif cmd == 'plan-merge':
1127
w = readit()
1128
for state, line in w.plan_merge(int(argv[3]), int(argv[4])):
1129
if line:
1130
print '%14s | %s' % (state, line),
1131
1132
elif cmd == 'merge':
1133
w = readit()
1134
p = w.plan_merge(int(argv[3]), int(argv[4]))
1135
sys.stdout.writelines(w.weave_merge(p))
1136
1137
elif cmd == 'mash-merge':
1138
if len(argv) != 5:
1139
usage()
1140
return 1
1141
1142
w = readit()
1143
v1, v2 = map(int, argv[3:5])
1144
1145
basis = w.inclusions([v1]).intersection(w.inclusions([v2]))
1146
1147
base_lines = list(w.mash_iter(basis))
1148
a_lines = list(w.get(v1))
1149
b_lines = list(w.get(v2))
1150
1151
from bzrlib.merge3 import Merge3
1152
m3 = Merge3(base_lines, a_lines, b_lines)
1153
1154
name_a = 'version %d' % v1
1155
name_b = 'version %d' % v2
1156
sys.stdout.writelines(m3.merge_lines(name_a=name_a, name_b=name_b))
1157
else:
1158
raise ValueError('unknown command %r' % cmd)
1159
1160
1161
1162
def profile_main(argv):
1163
import tempfile, hotshot, hotshot.stats
1164
1165
prof_f = tempfile.NamedTemporaryFile()
1166
1167
prof = hotshot.Profile(prof_f.name)
1168
1169
ret = prof.runcall(main, argv)
1170
prof.close()
1171
1172
stats = hotshot.stats.load(prof_f.name)
1173
#stats.strip_dirs()
1174
stats.sort_stats('cumulative')
1175
## XXX: Might like to write to stderr or the trace file instead but
1176
## print_stats seems hardcoded to stdout
1177
stats.print_stats(20)
1178
1179
return ret
1180
1181
1182
if __name__ == '__main__':
1183
import sys
1184
if '--profile' in sys.argv:
1185
args = sys.argv[:]
1186
args.remove('--profile')
1187
sys.exit(profile_main(args))
1188
else:
1189
sys.exit(main(sys.argv))
1190
Loggerhead is a web-based interface for Breezy
Version: 2.0.1