annotate Tools/scripts/ndiff.py @ 8831:00a6bb65ff29 legacy-trunk

Getting rid of this ancient stuff.
author Guido van Rossum <guido@python.org>
date Wed, 05 Aug 1998 21:10:09 +0000
parents 32bc17f7c77b
children 30c94cd49a68
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
8228
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
1 #! /usr/bin/env python
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
2
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
3 # Released to the public domain $JustDate: 3/16/98 $,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
4 # by Tim Peters (email tim_one@email.msn.com).
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
5
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
6 # ndiff file1 file2 -- a human-friendly file differencer.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
7
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
8 # $Revision$
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
9
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
10 # SequenceMatcher tries to compute a "human-friendly diff" between
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
11 # two sequences (chiefly picturing a file as a sequence of lines,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
12 # and a line as a sequence of characters, here). Unlike UNIX(tm) diff,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
13 # e.g., the fundamental notion is the longest *contiguous* & junk-free
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
14 # matching subsequence. That's what catches peoples' eyes. The
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
15 # Windows(tm) windiff has another interesting notion, pairing up elements
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
16 # that appear uniquely in each sequence. That, and the method here,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
17 # appear to yield more intuitive difference reports than does diff. This
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
18 # method appears to be the least vulnerable to synching up on blocks
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
19 # of "junk lines", though (like blank lines in ordinary text files,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
20 # or maybe "<P>" lines in HTML files). That may be because this is
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
21 # the only method of the 3 that has a *concept* of "junk" <wink>.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
22 #
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
23 # Note that ndiff makes no claim to produce a *minimal* diff. To the
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
24 # contrary, minimal diffs are often counter-intuitive, because they
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
25 # synch up anywhere possible, sometimes accidental matches 100 pages
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
26 # apart. Restricting synch points to contiguous matches preserves some
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
27 # notion of locality, at the occasional cost of producing a longer diff.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
28 #
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
29 # With respect to junk, an earlier verion of ndiff simply refused to
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
30 # *start* a match with a junk element. The result was cases like this:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
31 # before: private Thread currentThread;
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
32 # after: private volatile Thread currentThread;
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
33 # If you consider whitespace to be junk, the longest continguous match
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
34 # not starting with junk is "e Thread currentThread". So ndiff reported
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
35 # that "e volatil" was inserted between the 't' and the 'e' in "private".
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
36 # While an accurate view, to people that's absurd. The current version
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
37 # looks for matching blocks that are entirely junk-free, then extends the
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
38 # longest one of those as far as possible but only with matching junk.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
39 # So now "currentThread" is matched, then extended to suck up the
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
40 # preceding blank; then "private" is matched, and extended to suck up the
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
41 # following blank; then "Thread" is matched; and finally ndiff reports
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
42 # that "volatile " was inserted before "Thread". The only quibble
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
43 # remaining is that perhaps it was really the case that " volative"
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
44 # was inserted after "private". I can live with that <wink>.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
45 #
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
46 # NOTE on the output: From an ndiff report,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
47 # 1) The first file can be recovered by retaining only lines that begin
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
48 # with " " or "- ", and deleting those 2-character prefixes.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
49 # 2) The second file can be recovered similarly, but by retaining only
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
50 # " " and "+ " lines.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
51 # 3) Lines beginning with "? " attempt to guide the eye to intraline
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
52 # differences, and were not present in either input file.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
53 #
8285
32bc17f7c77b Add the specific sed statement to recover the second output file to
Guido van Rossum <guido@python.org>
parents: 8228
diff changeset
54 # COROLLARY:
32bc17f7c77b Add the specific sed statement to recover the second output file to
Guido van Rossum <guido@python.org>
parents: 8228
diff changeset
55 # On Unix, the second file can be recovered by piping the output through
32bc17f7c77b Add the specific sed statement to recover the second output file to
Guido van Rossum <guido@python.org>
parents: 8228
diff changeset
56 # sed -n '/^[+ ] /s/^..//p'
32bc17f7c77b Add the specific sed statement to recover the second output file to
Guido van Rossum <guido@python.org>
parents: 8228
diff changeset
57 # Modifications to recover the first file are left as an exercise for
32bc17f7c77b Add the specific sed statement to recover the second output file to
Guido van Rossum <guido@python.org>
parents: 8228
diff changeset
58 # the reader.
32bc17f7c77b Add the specific sed statement to recover the second output file to
Guido van Rossum <guido@python.org>
parents: 8228
diff changeset
59 #
8228
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
60 # NOTE on junk: the module-level names
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
61 # IS_LINE_JUNK
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
62 # IS_CHARACTER_JUNK
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
63 # can be set to any functions you like. The first one should accept
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
64 # a single string argument, and return true iff the string is junk.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
65 # The default is whether the regexp r"\s*#?\s*$" matches (i.e., a
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
66 # line without visible characters, except for at most one splat).
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
67 # The second should accept a string of length 1 etc. The default is
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
68 # whether the character is a blank or tab (note: bad idea to include
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
69 # newline in this!).
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
70 #
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
71 # After setting those, you can call fcompare(f1name, f2name) with the
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
72 # names of the files you want to compare. The difference report
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
73 # is sent to stdout. Or you can call main(), which expects to find
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
74 # (exactly) the two file names in sys.argv.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
75
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
76 import string
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
77 TRACE = 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
78
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
79 # define what "junk" means
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
80 import re
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
81
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
82 def IS_LINE_JUNK(line, pat=re.compile(r"\s*#?\s*$").match):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
83 return pat(line) is not None
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
84
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
85 def IS_CHARACTER_JUNK(ch, ws=" \t"):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
86 return ch in ws
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
87
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
88 del re
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
89
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
90 class SequenceMatcher:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
91 def __init__(self, isjunk=None, a='', b=''):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
92 # Members:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
93 # a
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
94 # first sequence
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
95 # b
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
96 # second sequence; differences are computed as "what do
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
97 # we need to do to 'a' to change it into 'b'?"
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
98 # b2j
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
99 # for x in b, b2j[x] is a list of the indices (into b)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
100 # at which x appears; junk elements do not appear
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
101 # b2jhas
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
102 # b2j.has_key
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
103 # fullbcount
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
104 # for x in b, fullbcount[x] == the number of times x
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
105 # appears in b; only materialized if really needed (used
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
106 # only for computing quick_ratio())
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
107 # matching_blocks
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
108 # a list of (i, j, k) triples, where a[i:i+k] == b[j:j+k];
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
109 # ascending & non-overlapping in i and in j; terminated by
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
110 # a dummy (len(a), len(b), 0) sentinel
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
111 # opcodes
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
112 # a list of (tag, i1, i2, j1, j2) tuples, where tag is
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
113 # one of
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
114 # 'replace' a[i1:i2] should be replaced by b[j1:j2]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
115 # 'delete' a[i1:i2] should be deleted
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
116 # 'insert' b[j1:j2] should be inserted
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
117 # 'equal' a[i1:i2] == b[j1:j2]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
118 # isjunk
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
119 # a user-supplied function taking a sequence element and
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
120 # returning true iff the element is "junk" -- this has
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
121 # subtle but helpful effects on the algorithm, which I'll
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
122 # get around to writing up someday <0.9 wink>.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
123 # DON'T USE! Only __chain_b uses this. Use isbjunk.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
124 # isbjunk
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
125 # for x in b, isbjunk(x) == isjunk(x) but much faster;
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
126 # it's really the has_key method of a hidden dict.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
127 # DOES NOT WORK for x in a!
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
128
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
129 self.isjunk = isjunk
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
130 self.a = self.b = None
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
131 self.set_seqs(a, b)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
132
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
133 def set_seqs(self, a, b):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
134 self.set_seq1(a)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
135 self.set_seq2(b)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
136
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
137 def set_seq1(self, a):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
138 if a is self.a:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
139 return
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
140 self.a = a
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
141 self.matching_blocks = self.opcodes = None
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
142
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
143 def set_seq2(self, b):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
144 if b is self.b:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
145 return
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
146 self.b = b
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
147 self.matching_blocks = self.opcodes = None
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
148 self.fullbcount = None
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
149 self.__chain_b()
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
150
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
151 # for each element x in b, set b2j[x] to a list of the indices in
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
152 # b where x appears; the indices are in increasing order; note that
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
153 # the number of times x appears in b is len(b2j[x]) ...
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
154 # when self.isjunk is defined, junk elements don't show up in this
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
155 # map at all, which stops the central find_longest_match method
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
156 # from starting any matching block at a junk element ...
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
157 # also creates the fast isbjunk function ...
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
158 # note that this is only called when b changes; so for cross-product
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
159 # kinds of matches, it's best to call set_seq2 once, then set_seq1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
160 # repeatedly
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
161
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
162 def __chain_b(self):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
163 # Because isjunk is a user-defined (not C) function, and we test
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
164 # for junk a LOT, it's important to minimize the number of calls.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
165 # Before the tricks described here, __chain_b was by far the most
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
166 # time-consuming routine in the whole module! If anyone sees
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
167 # Jim Roskind, thank him again for profile.py -- I never would
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
168 # have guessed that.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
169 # The first trick is to build b2j ignoring the possibility
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
170 # of junk. I.e., we don't call isjunk at all yet. Throwing
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
171 # out the junk later is much cheaper than building b2j "right"
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
172 # from the start.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
173 b = self.b
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
174 self.b2j = b2j = {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
175 self.b2jhas = b2jhas = b2j.has_key
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
176 for i in xrange(0, len(b)):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
177 elt = b[i]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
178 if b2jhas(elt):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
179 b2j[elt].append(i)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
180 else:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
181 b2j[elt] = [i]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
182
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
183 # Now b2j.keys() contains elements uniquely, and especially when
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
184 # the sequence is a string, that's usually a good deal smaller
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
185 # than len(string). The difference is the number of isjunk calls
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
186 # saved.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
187 isjunk, junkdict = self.isjunk, {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
188 if isjunk:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
189 for elt in b2j.keys():
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
190 if isjunk(elt):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
191 junkdict[elt] = 1 # value irrelevant; it's a set
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
192 del b2j[elt]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
193
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
194 # Now for x in b, isjunk(x) == junkdict.has_key(x), but the
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
195 # latter is much faster. Note too that while there may be a
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
196 # lot of junk in the sequence, the number of *unique* junk
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
197 # elements is probably small. So the memory burden of keeping
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
198 # this dict alive is likely trivial compared to the size of b2j.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
199 self.isbjunk = junkdict.has_key
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
200
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
201 def find_longest_match(self, alo, ahi, blo, bhi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
202 """Find longest matching block in a[alo:ahi] and b[blo:bhi].
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
203
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
204 If isjunk is not defined:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
205
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
206 Return (i,j,k) such that a[i:i+k] is equal to b[j:j+k], where
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
207 alo <= i <= i+k <= ahi
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
208 blo <= j <= j+k <= bhi
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
209 and for all (i',j',k') meeting those conditions,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
210 k >= k'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
211 i <= i'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
212 and if i == i', j <= j'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
213 In other words, of all maximal matching blocks, returns one
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
214 that starts earliest in a, and of all those maximal matching
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
215 blocks that start earliest in a, returns the one that starts
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
216 earliest in b.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
217
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
218 If isjunk is defined, first the longest matching block is
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
219 determined as above, but with the additional restriction that
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
220 no junk element appears in the block. Then that block is
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
221 extended as far as possible by matching (only) junk elements on
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
222 both sides. So the resulting block never matches on junk except
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
223 as identical junk happens to be adjacent to an "interesting"
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
224 match.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
225
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
226 If no blocks match, returns (alo, blo, 0).
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
227 """
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
228
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
229 # CAUTION: stripping common prefix or suffix would be incorrect.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
230 # E.g.,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
231 # ab
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
232 # acab
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
233 # Longest matching block is "ab", but if common prefix is
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
234 # stripped, it's "a" (tied with "b"). UNIX(tm) diff does so
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
235 # strip, so ends up claiming that ab is changed to acab by
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
236 # inserting "ca" in the middle. That's minimal but unintuitive:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
237 # "it's obvious" that someone inserted "ac" at the front.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
238 # Windiff ends up at the same place as diff, but by pairing up
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
239 # the unique 'b's and then matching the first two 'a's.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
240
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
241 # find longest junk-free match
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
242 a, b, b2j, isbjunk = self.a, self.b, self.b2j, self.isbjunk
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
243 besti, bestj, bestsize = alo, blo, 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
244 for i in xrange(alo, ahi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
245 # check for longest match starting at a[i]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
246 if i + bestsize >= ahi:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
247 # we're too far right to get a new best
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
248 break
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
249 # look at all instances of a[i] in b; note that because
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
250 # b2j has no junk keys, the loop is skipped if a[i] is junk
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
251 for j in b2j.get(a[i], []):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
252 # a[i] matches b[j]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
253 if j < blo:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
254 continue
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
255 if j + bestsize >= bhi:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
256 # we're too far right to get a new best, here or
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
257 # anywhere to the right
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
258 break
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
259 if a[i + bestsize] != b[j + bestsize]:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
260 # can't be longer match; this test is not necessary
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
261 # for correctness, but is a huge win for efficiency
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
262 continue
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
263 # set k to length of match
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
264 k = 1 # a[i] == b[j] already known
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
265 while i + k < ahi and j + k < bhi and \
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
266 a[i+k] == b[j+k] and not isbjunk(b[j+k]):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
267 k = k + 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
268 if k > bestsize:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
269 besti, bestj, bestsize = i, j, k
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
270 if i + bestsize >= ahi:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
271 # only time in my life I really wanted a
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
272 # labelled break <wink> -- we're done with
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
273 # both loops now
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
274 break
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
275
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
276 # Now that we have a wholly interesting match (albeit possibly
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
277 # empty!), we may as well suck up the matching junk on each
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
278 # side of it too. Can't think of a good reason not to, and it
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
279 # saves post-processing the (possibly considerable) expense of
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
280 # figuring out what to do with it. In the case of an empty
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
281 # interesting match, this is clearly the right thing to do,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
282 # because no other kind of match is possible in the regions.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
283 while besti > alo and bestj > blo and \
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
284 isbjunk(b[bestj-1]) and \
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
285 a[besti-1] == b[bestj-1]:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
286 besti, bestj, bestsize = besti-1, bestj-1, bestsize+1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
287 while besti+bestsize < ahi and bestj+bestsize < bhi and \
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
288 isbjunk(b[bestj+bestsize]) and \
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
289 a[besti+bestsize] == b[bestj+bestsize]:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
290 bestsize = bestsize + 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
291
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
292 if TRACE:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
293 print "get_matching_blocks", alo, ahi, blo, bhi
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
294 print " returns", besti, bestj, bestsize
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
295 return besti, bestj, bestsize
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
296
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
297 # A different implementation, using a binary doubling technique that
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
298 # does far fewer element compares (trades 'em for integer compares),
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
299 # and has n*lg n worst-case behavior. Alas, the code is much harder
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
300 # to follow (the details are tricky!), and in most cases I've seen,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
301 # it takes at least 50% longer than the "clever dumb" method above;
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
302 # probably due to creating layers of small dicts.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
303 # NOTE: this no longer matches the version above wrt junk; remains
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
304 # too unpromising to update it; someday, though ...
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
305
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
306 # def find_longest_match(self, alo, ahi, blo, bhi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
307 # """Find longest matching block in a[alo:ahi] and b[blo:bhi].
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
308 #
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
309 # Return (i,j,k) such that a[i:i+k] is equal to b[j:j+k], where
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
310 # alo <= i <= i+k <= ahi
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
311 # blo <= j <= j+k <= bhi
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
312 # and for all (i',j',k') meeting those conditions,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
313 # k >= k'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
314 # i <= i'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
315 # and if i == i', j <= j'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
316 # In other words, of all maximal matching blocks, returns one
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
317 # that starts earliest in a, and of all those maximal matching
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
318 # blocks that start earliest in a, returns the one that starts
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
319 # earliest in b.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
320 #
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
321 # If no blocks match, returns (alo, blo, 0).
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
322 # """
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
323 #
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
324 # a, b2j = self.a, self.b2j
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
325 # # alljs[size][i] is a set of all j's s.t. a[i:i+len] matches
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
326 # # b[j:j+len]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
327 # alljs = {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
328 # alljs[1] = js = {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
329 # ahits = {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
330 # for i in xrange(alo, ahi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
331 # elt = a[i]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
332 # if ahits.has_key(elt):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
333 # js[i] = ahits[elt]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
334 # continue
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
335 # if b2j.has_key(elt):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
336 # in_range = {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
337 # for j in b2j[elt]:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
338 # if j >= blo:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
339 # if j >= bhi:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
340 # break
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
341 # in_range[j] = 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
342 # if in_range:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
343 # ahits[elt] = js[i] = in_range
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
344 # del ahits
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
345 # size = 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
346 # while js:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
347 # oldsize = size
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
348 # size = size + size
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
349 # oldjs = js
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
350 # alljs[size] = js = {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
351 # for i in oldjs.keys():
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
352 # # i has matches of size oldsize
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
353 # if not oldjs.has_key(i + oldsize):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
354 # # can't double it
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
355 # continue
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
356 # second_js = oldjs[i + oldsize]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
357 # answer = {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
358 # for j in oldjs[i].keys():
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
359 # if second_js.has_key(j + oldsize):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
360 # answer[j] = 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
361 # if answer:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
362 # js[i] = answer
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
363 # del alljs[size]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
364 # size = size >> 1 # max power of 2 with a match
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
365 # if not size:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
366 # return alo, blo, 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
367 # besti, bestj, bestsize = alo, blo, 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
368 # fatis = alljs[size].keys()
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
369 # fatis.sort()
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
370 # for i in fatis:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
371 # # figure out longest match starting at a[i]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
372 # totalsize = halfsize = size
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
373 # # i has matches of len totalsize at the indices in js
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
374 # js = alljs[size][i].keys()
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
375 # while halfsize > 1:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
376 # halfsize = halfsize >> 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
377 # # is there a match of len halfsize starting at
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
378 # # i + totalsize?
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
379 # newjs = []
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
380 # if alljs[halfsize].has_key(i + totalsize):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
381 # second_js = alljs[halfsize][i + totalsize]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
382 # for j in js:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
383 # if second_js.has_key(j + totalsize):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
384 # newjs.append(j)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
385 # if newjs:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
386 # totalsize = totalsize + halfsize
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
387 # js = newjs
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
388 # if totalsize > bestsize:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
389 # besti, bestj, bestsize = i, min(js), totalsize
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
390 # return besti, bestj, bestsize
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
391
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
392 def get_matching_blocks(self):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
393 if self.matching_blocks is not None:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
394 return self.matching_blocks
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
395 self.matching_blocks = []
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
396 la, lb = len(self.a), len(self.b)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
397 self.__helper(0, la, 0, lb, self.matching_blocks)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
398 self.matching_blocks.append( (la, lb, 0) )
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
399 if TRACE:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
400 print '*** matching blocks', self.matching_blocks
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
401 return self.matching_blocks
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
402
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
403 # builds list of matching blocks covering a[alo:ahi] and
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
404 # b[blo:bhi], appending them in increasing order to answer
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
405
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
406 def __helper(self, alo, ahi, blo, bhi, answer):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
407 i, j, k = x = self.find_longest_match(alo, ahi, blo, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
408 # a[alo:i] vs b[blo:j] unknown
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
409 # a[i:i+k] same as b[j:j+k]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
410 # a[i+k:ahi] vs b[j+k:bhi] unknown
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
411 if k:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
412 if alo < i and blo < j:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
413 self.__helper(alo, i, blo, j, answer)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
414 answer.append( x )
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
415 if i+k < ahi and j+k < bhi:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
416 self.__helper(i+k, ahi, j+k, bhi, answer)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
417
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
418 def ratio(self):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
419 """Return a measure of the sequences' similarity (float in [0,1]).
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
420
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
421 Where T is the total number of elements in both sequences, and
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
422 M is the number of matches, this is 2*M / T.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
423 Note that this is 1 if the sequences are identical, and 0 if
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
424 they have nothing in common.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
425 """
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
426
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
427 matches = reduce(lambda sum, triple: sum + triple[-1],
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
428 self.get_matching_blocks(), 0)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
429 return 2.0 * matches / (len(self.a) + len(self.b))
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
430
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
431 def quick_ratio(self):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
432 """Return an upper bound on ratio() relatively quickly."""
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
433 # viewing a and b as multisets, set matches to the cardinality
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
434 # of their intersection; this counts the number of matches
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
435 # without regard to order, so is clearly an upper bound
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
436 if self.fullbcount is None:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
437 self.fullbcount = fullbcount = {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
438 for elt in self.b:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
439 fullbcount[elt] = fullbcount.get(elt, 0) + 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
440 fullbcount = self.fullbcount
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
441 # avail[x] is the number of times x appears in 'b' less the
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
442 # number of times we've seen it in 'a' so far ... kinda
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
443 avail = {}
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
444 availhas, matches = avail.has_key, 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
445 for elt in self.a:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
446 if availhas(elt):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
447 numb = avail[elt]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
448 else:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
449 numb = fullbcount.get(elt, 0)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
450 avail[elt] = numb - 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
451 if numb > 0:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
452 matches = matches + 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
453 return 2.0 * matches / (len(self.a) + len(self.b))
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
454
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
455 def real_quick_ratio(self):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
456 """Return an upper bound on ratio() very quickly"""
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
457 la, lb = len(self.a), len(self.b)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
458 # can't have more matches than the number of elements in the
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
459 # shorter sequence
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
460 return 2.0 * min(la, lb) / (la + lb)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
461
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
462 def get_opcodes(self):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
463 if self.opcodes is not None:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
464 return self.opcodes
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
465 i = j = 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
466 self.opcodes = answer = []
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
467 for ai, bj, size in self.get_matching_blocks():
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
468 # invariant: we've pumped out correct diffs to change
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
469 # a[:i] into b[:j], and the next matching block is
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
470 # a[ai:ai+size] == b[bj:bj+size]. So we need to pump
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
471 # out a diff to change a[i:ai] into b[j:bj], pump out
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
472 # the matching block, and move (i,j) beyond the match
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
473 tag = ''
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
474 if i < ai and j < bj:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
475 tag = 'replace'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
476 elif i < ai:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
477 tag = 'delete'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
478 elif j < bj:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
479 tag = 'insert'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
480 if tag:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
481 answer.append( (tag, i, ai, j, bj) )
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
482 i, j = ai+size, bj+size
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
483 # the list of matching blocks is terminated by a
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
484 # sentinel with size 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
485 if size:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
486 answer.append( ('equal', ai, i, bj, j) )
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
487 return answer
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
488
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
489 # meant for dumping lines
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
490 def dump(tag, x, lo, hi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
491 for i in xrange(lo, hi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
492 print tag, x[i],
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
493
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
494 # figure out which mark to stick under characters in lines that
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
495 # have changed (blank = same, - = deleted, + = inserted, ^ = replaced)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
496 _combine = { ' ': ' ',
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
497 '. ': '-',
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
498 ' .': '+',
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
499 '..': '^' }
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
500
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
501 def plain_replace(a, alo, ahi, b, blo, bhi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
502 assert alo < ahi and blo < bhi
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
503 # dump the shorter block first -- reduces the burden on short-term
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
504 # memory if the blocks are of very different sizes
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
505 if bhi - blo < ahi - alo:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
506 dump('+', b, blo, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
507 dump('-', a, alo, ahi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
508 else:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
509 dump('-', a, alo, ahi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
510 dump('+', b, blo, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
511
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
512 # When replacing one block of lines with another, this guy searches
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
513 # the blocks for *similar* lines; the best-matching pair (if any) is
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
514 # used as a synch point, and intraline difference marking is done on
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
515 # the similar pair. Lots of work, but often worth it.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
516
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
517 def fancy_replace(a, alo, ahi, b, blo, bhi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
518 if TRACE:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
519 print '*** fancy_replace', alo, ahi, blo, bhi
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
520 dump('>', a, alo, ahi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
521 dump('<', b, blo, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
522
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
523 # don't synch up unless the lines have a similarity score of at
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
524 # least cutoff; best_ratio tracks the best score seen so far
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
525 best_ratio, cutoff = 0.74, 0.75
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
526 cruncher = SequenceMatcher(IS_CHARACTER_JUNK)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
527 eqi, eqj = None, None # 1st indices of equal lines (if any)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
528
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
529 # search for the pair that matches best without being identical
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
530 # (identical lines must be junk lines, & we don't want to synch up
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
531 # on junk -- unless we have to)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
532 for j in xrange(blo, bhi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
533 bj = b[j]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
534 cruncher.set_seq2(bj)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
535 for i in xrange(alo, ahi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
536 ai = a[i]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
537 if ai == bj:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
538 if eqi is None:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
539 eqi, eqj = i, j
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
540 continue
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
541 cruncher.set_seq1(ai)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
542 # computing similarity is expensive, so use the quick
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
543 # upper bounds first -- have seen this speed up messy
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
544 # compares by a factor of 3.
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
545 # note that ratio() is only expensive to compute the first
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
546 # time it's called on a sequence pair; the expensive part
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
547 # of the computation is cached by cruncher
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
548 if cruncher.real_quick_ratio() > best_ratio and \
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
549 cruncher.quick_ratio() > best_ratio and \
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
550 cruncher.ratio() > best_ratio:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
551 best_ratio, best_i, best_j = cruncher.ratio(), i, j
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
552 if best_ratio < cutoff:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
553 # no non-identical "pretty close" pair
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
554 if eqi is None:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
555 # no identical pair either -- treat it as a straight replace
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
556 plain_replace(a, alo, ahi, b, blo, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
557 return
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
558 # no close pair, but an identical pair -- synch up on that
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
559 best_i, best_j, best_ratio = eqi, eqj, 1.0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
560 else:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
561 # there's a close pair, so forget the identical pair (if any)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
562 eqi = None
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
563
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
564 # a[best_i] very similar to b[best_j]; eqi is None iff they're not
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
565 # identical
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
566 if TRACE:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
567 print '*** best_ratio', best_ratio, best_i, best_j
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
568 dump('>', a, best_i, best_i+1)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
569 dump('<', b, best_j, best_j+1)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
570
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
571 # pump out diffs from before the synch point
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
572 fancy_helper(a, alo, best_i, b, blo, best_j)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
573
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
574 # do intraline marking on the synch pair
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
575 aelt, belt = a[best_i], b[best_j]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
576 if eqi is None:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
577 # pump out a '-', '+', '?' triple for the synched lines;
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
578 atags = btags = ""
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
579 cruncher.set_seqs(aelt, belt)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
580 for tag, ai1, ai2, bj1, bj2 in cruncher.get_opcodes():
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
581 la, lb = ai2 - ai1, bj2 - bj1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
582 if tag == 'replace':
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
583 atags = atags + '.' * la
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
584 btags = btags + '.' * lb
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
585 elif tag == 'delete':
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
586 atags = atags + '.' * la
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
587 elif tag == 'insert':
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
588 btags = btags + '.' * lb
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
589 elif tag == 'equal':
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
590 atags = atags + ' ' * la
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
591 btags = btags + ' ' * lb
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
592 else:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
593 raise ValueError, 'unknown tag ' + `tag`
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
594 la, lb = len(atags), len(btags)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
595 if la < lb:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
596 atags = atags + ' ' * (lb - la)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
597 elif lb < la:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
598 btags = btags + ' ' * (la - lb)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
599 combined = map(lambda x,y: _combine[x+y], atags, btags)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
600 print '-', aelt, '+', belt, '?', \
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
601 string.rstrip(string.join(combined, ''))
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
602 else:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
603 # the synch pair is identical
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
604 print ' ', aelt,
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
605
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
606 # pump out diffs from after the synch point
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
607 fancy_helper(a, best_i+1, ahi, b, best_j+1, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
608
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
609 def fancy_helper(a, alo, ahi, b, blo, bhi):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
610 if alo < ahi:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
611 if blo < bhi:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
612 fancy_replace(a, alo, ahi, b, blo, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
613 else:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
614 dump('-', a, alo, ahi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
615 elif blo < bhi:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
616 dump('+', b, blo, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
617
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
618 # open a file & return the file object; gripe and return 0 if it
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
619 # couldn't be opened
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
620 def fopen(fname):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
621 try:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
622 return open(fname, 'r')
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
623 except IOError, detail:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
624 print "couldn't open " + fname + ": " + `detail`
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
625 return 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
626
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
627 # open two files & spray the diff to stdout; return false iff a problem
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
628 def fcompare(f1name, f2name):
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
629 f1 = fopen(f1name)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
630 f2 = fopen(f2name)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
631 if not f1 or not f2:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
632 return 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
633
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
634 a = f1.readlines(); f1.close()
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
635 b = f2.readlines(); f2.close()
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
636
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
637 cruncher = SequenceMatcher(IS_LINE_JUNK, a, b)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
638 for tag, alo, ahi, blo, bhi in cruncher.get_opcodes():
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
639 if tag == 'replace':
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
640 fancy_replace(a, alo, ahi, b, blo, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
641 elif tag == 'delete':
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
642 dump('-', a, alo, ahi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
643 elif tag == 'insert':
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
644 dump('+', b, blo, bhi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
645 elif tag == 'equal':
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
646 dump(' ', a, alo, ahi)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
647 else:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
648 raise ValueError, 'unknown tag ' + `tag`
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
649
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
650 return 1
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
651
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
652 # get file names from argv & compare; return false iff a problem
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
653 def main():
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
654 from sys import argv
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
655 if len(argv) != 3:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
656 print 'need 2 args'
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
657 return 0
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
658 [f1name, f2name] = argv[1:3]
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
659 print '-:', f1name
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
660 print '+:', f2name
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
661 return fcompare(f1name, f2name)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
662
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
663 if __name__ == '__main__':
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
664 if 1:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
665 main()
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
666 else:
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
667 import profile, pstats
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
668 statf = "ndiff.pro"
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
669 profile.run("main()", statf)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
670 stats = pstats.Stats(statf)
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
671 stats.strip_dirs().sort_stats('time').print_stats()
9d642a735a13 Adding Tim Peters' ndiff utility.
Guido van Rossum <guido@python.org>
parents:
diff changeset
672