2
# Copyright (C) 2005 Bram Cohen, Copyright (C) 2005, 2006 Canonical Ltd
4
# This program is free software; you can redistribute it and/or modify
5
# it under the terms of the GNU General Public License as published by
6
# the Free Software Foundation; either version 2 of the License, or
7
# (at your option) any later version.
9
# This program is distributed in the hope that it will be useful,
10
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
# GNU General Public License for more details.
14
# You should have received a copy of the GNU General Public License
15
# along with this program; if not, write to the Free Software
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19
from bisect import bisect
26
from bzrlib.trace import mutter
29
__all__ = ['PatienceSequenceMatcher', 'unified_diff', 'unified_diff_files']
33
"""Find the longest common subset for unique lines.
35
:param a: An indexable object (such as string or list of strings)
36
:param b: Another indexable object (such as string or list of strings)
37
:return: A list of tuples, one for each line which is matched.
38
[(line_in_a, line_in_b), ...]
40
This only matches lines which are unique on both sides.
41
This helps prevent common lines from over influencing match
43
The longest common subset uses the Patience Sorting algorithm:
44
http://en.wikipedia.org/wiki/Patience_sorting
46
# set index[line in a] = position of line in a unless
47
# unless a is a duplicate, in which case it's set to None
49
for i in xrange(len(a)):
55
# make btoa[i] = position of line i in a, unless
56
# that line doesn't occur exactly once in both,
57
# in which case it's set to None
58
btoa = [None] * len(b)
60
for pos, line in enumerate(b):
61
next = index.get(line)
64
# unset the previous mapping, which we now know to
65
# be invalid because the line isn't unique
66
btoa[index2[line]] = None
71
# this is the Patience sorting algorithm
72
# see http://en.wikipedia.org/wiki/Patience_sorting
73
backpointers = [None] * len(b)
77
for bpos, apos in enumerate(btoa):
80
# as an optimization, check if the next line comes at the end,
81
# because it usually does
82
if stacks and stacks[-1] < apos:
84
# as an optimization, check if the next line comes right after
85
# the previous line, because usually it does
86
elif stacks and stacks[k] < apos and (k == len(stacks) - 1 or
90
k = bisect(stacks, apos)
92
backpointers[bpos] = lasts[k-1]
104
result.append((btoa[k], k))
110
def recurse_matches(a, b, ahi, bhi, answer, maxrecursion):
111
"""Find all of the matching text in the lines of a and b.
114
:param b: Another sequence
115
:param ahi: The maximum length of a to check, typically len(a)
116
:param bhi: The maximum length of b to check, typically len(b)
117
:param answer: The return array. Will be filled with tuples
118
indicating [(line_in_a), (line_in_b)]
119
:param maxrecursion: The maximum depth to recurse.
120
Must be a positive integer.
121
:return: None, the return value is in the parameter answer, which
127
mutter('max recursion depth reached')
128
# this will never happen normally, this check is to prevent DOS attacks
130
oldlength = len(answer)
134
alo, blo = answer[-1]
137
if alo == ahi or blo == bhi:
139
for apos, bpos in unique_lcs(a[alo:ahi], b[blo:bhi]):
140
# recurse between lines which are unique in each file and match
143
recurse_matches(a, b, apos, bpos, answer, maxrecursion - 1)
144
answer.append((apos, bpos))
145
if len(answer) > oldlength:
146
# find matches between the last match and the end
147
recurse_matches(a, b, ahi, bhi, answer, maxrecursion - 1)
148
elif a[alo] == b[blo]:
149
# find matching lines at the very beginning
150
while alo < ahi and blo < bhi and a[alo] == b[blo]:
151
answer.append((alo, blo))
154
recurse_matches(a, b, ahi, bhi, answer, maxrecursion - 1)
155
elif a[ahi - 1] == b[bhi - 1]:
156
# find matching lines at the very end
159
while nahi > alo and nbhi > blo and a[nahi - 1] == b[nbhi - 1]:
162
recurse_matches(a, b, nahi, nbhi, answer, maxrecursion - 1)
163
for i in xrange(ahi - nahi):
164
answer.append((nahi + i, nbhi + i))
167
class PatienceSequenceMatcher(difflib.SequenceMatcher):
168
"""Compare a pair of sequences using longest common subset."""
170
def __init__(self, isjunk=None, a='', b=''):
171
if isjunk is not None:
172
raise NotImplementedError('Currently we do not support'
173
' isjunk for sequence matching')
174
difflib.SequenceMatcher.__init__(self, isjunk, a, b)
176
def _check_with_diff(self, alo, ahi, blo, bhi, answer):
177
"""Use the original diff algorithm on an unmatched section.
179
This will check to make sure the range is worth checking,
180
before doing any work.
182
:param alo: The last line that actually matched
183
:param ahi: The next line that actually matches
184
:param blo: Same as alo, only for the 'b' set
185
:param bhi: Same as ahi
186
:param answer: An array which will have the new ranges appended to it
190
# recurse_matches has an implementation design
191
# which does not match non-unique lines in the
192
# if they do not touch matching unique lines
193
# so we rerun the regular diff algorithm
194
# if find a large enough chunk.
196
# recurse_matches already looked at the direct
197
# neighbors, so we only need to run if there is
198
# enough space to do so
199
if ahi - alo > 2 and bhi - blo > 2:
200
a = self.a[alo+1:ahi-1]
201
b = self.b[blo+1:bhi-1]
202
m = difflib.SequenceMatcher(None, a, b)
203
new_blocks = m.get_matching_blocks()
204
# difflib always adds a final match
206
for blk in new_blocks:
207
answer.append((blk[0]+alo+1,
211
def get_matching_blocks(self):
212
"""Return list of triples describing matching subsequences.
214
Each triple is of the form (i, j, n), and means that
215
a[i:i+n] == b[j:j+n]. The triples are monotonically increasing in
218
The last triple is a dummy, (len(a), len(b), 0), and is the only
221
>>> s = PatienceSequenceMatcher(None, "abxcd", "abcd")
222
>>> s.get_matching_blocks()
223
[(0, 0, 2), (3, 2, 2), (5, 4, 0)]
225
# jam 20060525 This is the python 2.4.1 difflib get_matching_blocks
226
# implementation which uses __helper. 2.4.3 got rid of helper for
227
# doing it inline with a queue.
228
# We should consider doing the same for recurse_matches
230
if self.matching_blocks is not None:
231
return self.matching_blocks
232
self.matching_blocks = []
233
la, lb = len(self.a), len(self.b)
234
self.__helper(0, la, 0, lb, self.matching_blocks)
235
self.matching_blocks.append( (la, lb, 0) )
236
return self.matching_blocks
238
def __helper(self, alo, ahi, blo, bhi, answer):
242
recurse_matches(a, b, len(a), len(b), matches, 10)
243
# Matches now has individual line pairs of
244
# line A matches line B, at the given offsets
246
start_a = start_b = None
248
for i_a, i_b in matches:
249
if (start_a is not None
250
and (i_a == start_a + length)
251
and (i_b == start_b + length)):
256
# We need to check from 0,0 until the current match
257
self._check_with_diff(alo-1, i_a+alo, blo-1, i_b+blo,
260
answer.append((start_a+alo, start_b+blo, length))
261
self._check_with_diff(start_a+alo+length, i_a+alo,
262
start_b+blo+length, i_b+blo,
270
answer.append((start_a+alo, start_b+blo, length))
271
self._check_with_diff(start_a+alo+length, ahi+1,
272
start_b+blo+length, bhi+1,
275
# Nothing matched, so we need to send the complete text
276
self._check_with_diff(alo-1, ahi+1, blo-1, bhi+1, answer)
278
# For consistency sake, make sure all matches are only increasing
282
for a,b,match_len in answer:
283
assert a >= next_a, 'Non increasing matches for a'
284
assert b >= next_b, 'Not increasing matches for b'
285
next_a = a + match_len
286
next_b = b + match_len
289
# This is a version of unified_diff which only adds a factory parameter
290
# so that you can override the default SequenceMatcher
291
# this has been submitted as a patch to python
292
def unified_diff(a, b, fromfile='', tofile='', fromfiledate='',
293
tofiledate='', n=3, lineterm='\n',
294
sequencematcher=None):
296
Compare two sequences of lines; generate the delta as a unified diff.
298
Unified diffs are a compact way of showing line changes and a few
299
lines of context. The number of context lines is set by 'n' which
302
By default, the diff control lines (those with ---, +++, or @@) are
303
created with a trailing newline. This is helpful so that inputs
304
created from file.readlines() result in diffs that are suitable for
305
file.writelines() since both the inputs and outputs have trailing
308
For inputs that do not have trailing newlines, set the lineterm
309
argument to "" so that the output will be uniformly newline free.
311
The unidiff format normally has a header for filenames and modification
312
times. Any or all of these may be specified using strings for
313
'fromfile', 'tofile', 'fromfiledate', and 'tofiledate'. The modification
314
times are normally expressed in the format returned by time.ctime().
318
>>> for line in unified_diff('one two three four'.split(),
319
... 'zero one tree four'.split(), 'Original', 'Current',
320
... 'Sat Jan 26 23:30:50 1991', 'Fri Jun 06 10:20:52 2003',
323
--- Original Sat Jan 26 23:30:50 1991
324
+++ Current Fri Jun 06 10:20:52 2003
333
if sequencematcher is None:
334
sequencematcher = difflib.SequenceMatcher
337
for group in sequencematcher(None,a,b).get_grouped_opcodes(n):
339
yield '--- %s %s%s' % (fromfile, fromfiledate, lineterm)
340
yield '+++ %s %s%s' % (tofile, tofiledate, lineterm)
342
i1, i2, j1, j2 = group[0][1], group[-1][2], group[0][3], group[-1][4]
343
yield "@@ -%d,%d +%d,%d @@%s" % (i1+1, i2-i1, j1+1, j2-j1, lineterm)
344
for tag, i1, i2, j1, j2 in group:
346
for line in a[i1:i2]:
349
if tag == 'replace' or tag == 'delete':
350
for line in a[i1:i2]:
352
if tag == 'replace' or tag == 'insert':
353
for line in b[j1:j2]:
357
def unified_diff_files(a, b, sequencematcher=None):
358
"""Generate the diff for two files.
360
# Should this actually be an error?
367
file_a = open(a, 'rb')
368
time_a = os.stat(a).st_mtime
374
file_b = open(b, 'rb')
375
time_b = os.stat(b).st_mtime
377
# TODO: Include fromfiledate and tofiledate
378
return unified_diff(file_a.readlines(), file_b.readlines(),
379
fromfile=a, tofile=b,
380
sequencematcher=sequencematcher)
385
p = optparse.OptionParser(usage='%prog [options] file_a file_b'
386
'\nFiles can be "-" to read from stdin')
387
p.add_option('--patience', dest='matcher', action='store_const', const='patience',
388
default='patience', help='Use the patience difference algorithm')
389
p.add_option('--difflib', dest='matcher', action='store_const', const='difflib',
390
default='patience', help='Use python\'s difflib algorithm')
392
algorithms = {'patience':PatienceSequenceMatcher, 'difflib':difflib.SequenceMatcher}
394
(opts, args) = p.parse_args(args)
395
matcher = algorithms[opts.matcher]
398
print 'You must supply 2 filenames to diff'
401
for line in unified_diff_files(args[0], args[1], sequencematcher=matcher):
402
sys.stdout.write(line)
404
if __name__ == '__main__':
405
sys.exit(main(sys.argv[1:]))