Mercurial > dedupe

annotate CachedEditDistance.hpp @ 83:c17475848bf4

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Allow for paths to be specified on commandline.
author Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
date Thu, 10 Oct 2013 14:13:54 +0200
parents 9744ec195be3
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
78
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
1 #ifndef CACHEDEDITDISTANCE_HPP
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
2 #define CACHEDEDITDISTANCE_HPP
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
3
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
4 #include "DBCache.hpp"
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
5
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
6 #include "EditDistance.hpp"
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
7
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
8 template<typename Value>
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
9 struct InsertRegulator<OrderedPair<UniqueString>, Value >
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
10 {
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
11 uint n;
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
12 void start()
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
13 {
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
14 n = 0;
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
15 HuffmanString::getSet().setAutoRebuild(false);
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
16 }
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
17
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
18 static void finish()
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
19 {
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
20 HuffmanString::getSet().rebuild();
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
21 HuffmanString::getSet().setAutoRebuild(true);
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
22 }
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
23
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
24 void next()
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
25 {
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
26 if (++n == 2048)
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
27 HuffmanString::getSet().rebuild();
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
28 }
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
29 };
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
30
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
31 class CachedEditDistance {
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
32 protected:
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
33 typedef DBCache<OrderedPair<UniqueString>, int, true> cacheType;
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
34 public:
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
35 static int Compute(QString a, QString b, bool removeDiacritics = false);
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
36 static void removeDiacriticsNoCopy(QString& in)
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
37 {
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
38 EditDistance::removeDiacriticsNoCopy(in);
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
39 }
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
40 static QString removeDiacritics(const QString& in)
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
41 {
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
42 return EditDistance::removeDiacritics(in);
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
43 }
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
44
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
45 static cacheType* cache;
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
46 };
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
47
9744ec195be3 Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
diff changeset
48 #endif //CACHEDEDITDISTANCE_HPP