8
|
1 |
/*
|
|
2 |
* analysisutil.cpp
|
|
3 |
*
|
|
4 |
* Created on: Mar 25, 2010
|
|
5 |
* Author: admin
|
|
6 |
*/
|
|
7 |
|
|
8 |
#include "localetestinfos.h"
|
|
9 |
#include "config.h"
|
|
10 |
|
|
11 |
const char* EnglishLocale::LOCALE = "en";
|
|
12 |
const char* EnglishLocale::FILES[] = {
|
|
13 |
STEM_TEST_CORPUS_PATH "\\en\\1.txt",
|
|
14 |
STEM_TEST_CORPUS_PATH "\\en\\2.txt",
|
|
15 |
STEM_TEST_CORPUS_PATH "\\en\\3.txt",
|
|
16 |
STEM_TEST_CORPUS_PATH "\\en\\4.txt",
|
|
17 |
0
|
|
18 |
};
|
|
19 |
const wchar_t* EnglishLocale::QUERIES[] = {
|
|
20 |
L"happy",
|
|
21 |
L"happiness",
|
|
22 |
L"happening",
|
|
23 |
0
|
|
24 |
};
|
|
25 |
|
|
26 |
|
|
27 |
const char* FrenchLocale::LOCALE = "fr";
|
|
28 |
|
|
29 |
const char* FrenchLocale::FILES[] = {
|
|
30 |
LOC_TEST_CORPUS_PATH "\\fr\\1.txt",
|
|
31 |
LOC_TEST_CORPUS_PATH "\\fr\\2.txt",
|
|
32 |
LOC_TEST_CORPUS_PATH "\\fr\\3.txt",
|
|
33 |
LOC_TEST_CORPUS_PATH "\\fr\\4.txt",
|
|
34 |
0
|
|
35 |
};
|
|
36 |
|
|
37 |
const wchar_t* FrenchLocale::QUERIES[] = {
|
|
38 |
L"d'\xe9nergie",
|
|
39 |
L"\xe9nergie",
|
|
40 |
L"elle",
|
|
41 |
0
|
|
42 |
};
|
|
43 |
|
|
44 |
const char* HebrewLocale::LOCALE = "he";
|
|
45 |
|
|
46 |
const char* HebrewLocale::FILES[] = {
|
|
47 |
LOC_TEST_CORPUS_PATH "\\he\\1.txt",
|
|
48 |
LOC_TEST_CORPUS_PATH "\\he\\2.txt",
|
|
49 |
LOC_TEST_CORPUS_PATH "\\he\\3.txt",
|
|
50 |
LOC_TEST_CORPUS_PATH "\\he\\4.txt",
|
|
51 |
0
|
|
52 |
};
|
|
53 |
|
|
54 |
const wchar_t* HebrewLocale::QUERIES[] = {
|
|
55 |
L"\x05e9\x05e4\x05e2\x05ea", // L"שפעת",
|
|
56 |
L"\x05e4\x05e2\x05ea", // L"פעת",
|
|
57 |
L"\x05e9\x05e4\x05e2*", // L"שפע*",
|
|
58 |
L"\x05e4\x05e8*", //L"פר*",
|
|
59 |
0
|
|
60 |
};
|
|
61 |
|
|
62 |
const char* ThaiLocale::LOCALE = "th";
|
|
63 |
const char* ThaiLocale::FILES[] = {
|
|
64 |
LOC_TEST_CORPUS_PATH "\\th\\1.txt",
|
|
65 |
LOC_TEST_CORPUS_PATH "\\th\\2.txt",
|
|
66 |
LOC_TEST_CORPUS_PATH "\\th\\3.txt",
|
|
67 |
LOC_TEST_CORPUS_PATH "\\th\\4.txt",
|
|
68 |
0
|
|
69 |
};
|
|
70 |
const wchar_t* ThaiLocale::QUERIES[] = {
|
|
71 |
L"\x0E14\x0E32\x0E27\x0E15\x0E01", // a thai word
|
|
72 |
L"\x0E21\x0E35", // another thai word
|
|
73 |
0
|
|
74 |
};
|
|
75 |
|
|
76 |
|
|
77 |
const char* ChineseLocale::LOCALE = "ch";
|
|
78 |
const char* ChineseLocale::FILES[] = {
|
|
79 |
LOC_TEST_CORPUS_PATH "\\ch_hk\\1.txt",
|
|
80 |
LOC_TEST_CORPUS_PATH "\\ch_hk\\2.txt",
|
|
81 |
LOC_TEST_CORPUS_PATH "\\ch_hk\\3.txt",
|
|
82 |
LOC_TEST_CORPUS_PATH "\\ch_hk\\4.txt",
|
|
83 |
|
|
84 |
LOC_TEST_CORPUS_PATH "\\ch_prc\\1.txt",
|
|
85 |
LOC_TEST_CORPUS_PATH "\\ch_prc\\2.txt",
|
|
86 |
LOC_TEST_CORPUS_PATH "\\ch_prc\\3.txt",
|
|
87 |
LOC_TEST_CORPUS_PATH "\\ch_prc\\4.txt",
|
|
88 |
|
|
89 |
LOC_TEST_CORPUS_PATH "\\ch_simple\\1.txt",
|
|
90 |
LOC_TEST_CORPUS_PATH "\\ch_simple\\2.txt",
|
|
91 |
LOC_TEST_CORPUS_PATH "\\ch_simple\\3.txt",
|
|
92 |
LOC_TEST_CORPUS_PATH "\\ch_simple\\4.txt",
|
|
93 |
|
|
94 |
LOC_TEST_CORPUS_PATH "\\ch_tw\\1.txt",
|
|
95 |
LOC_TEST_CORPUS_PATH "\\ch_tw\\2.txt",
|
|
96 |
LOC_TEST_CORPUS_PATH "\\ch_tw\\3.txt",
|
|
97 |
LOC_TEST_CORPUS_PATH "\\ch_tw\\4.txt",
|
|
98 |
|
|
99 |
0
|
|
100 |
};
|
|
101 |
const wchar_t* ChineseLocale::QUERIES[] = {
|
|
102 |
L"\x53f0\x6e7e", // a chinese word
|
|
103 |
L"\x4e2d\x56fd", // another chinese word
|
|
104 |
0
|
|
105 |
};
|
|
106 |
|
|
107 |
const char* KoreanLocale::LOCALE = "ko";
|
|
108 |
const char* KoreanLocale::FILES[] = {
|
|
109 |
LOC_TEST_CORPUS_PATH "\\ko\\1.txt",
|
|
110 |
LOC_TEST_CORPUS_PATH "\\ko\\2.txt",
|
|
111 |
LOC_TEST_CORPUS_PATH "\\ko\\3.txt",
|
|
112 |
LOC_TEST_CORPUS_PATH "\\ko\\4.txt",
|
|
113 |
|
|
114 |
0
|
|
115 |
};
|
|
116 |
const wchar_t* KoreanLocale::QUERIES[] = {
|
|
117 |
L"\xc2a4\xd1a0\xb9ac", // a korean word
|
|
118 |
L"\xc778\xbb3c", // another korean word
|
|
119 |
0
|
|
120 |
};
|
|
121 |
|
|
122 |
const char* JapaneseLocale::LOCALE = "jp";
|
|
123 |
const char* JapaneseLocale::FILES[] = {
|
|
124 |
LOC_TEST_CORPUS_PATH "\\jp\\1.txt",
|
|
125 |
LOC_TEST_CORPUS_PATH "\\jp\\2.txt",
|
|
126 |
LOC_TEST_CORPUS_PATH "\\jp\\3.txt",
|
|
127 |
LOC_TEST_CORPUS_PATH "\\jp\\4.txt",
|
|
128 |
|
|
129 |
LOC_TEST_CORPUS_PATH "\\jp_old\\1.txt",
|
|
130 |
LOC_TEST_CORPUS_PATH "\\jp_old\\2.txt",
|
|
131 |
LOC_TEST_CORPUS_PATH "\\jp_old\\3.txt",
|
|
132 |
LOC_TEST_CORPUS_PATH "\\jp_old\\4.txt",
|
|
133 |
|
|
134 |
0
|
|
135 |
};
|
|
136 |
const wchar_t* JapaneseLocale::QUERIES[] = {
|
|
137 |
L"\x56fd\x969b", // a japanese word
|
|
138 |
L"\x65e5\x672c", // another japanese word
|
|
139 |
0
|
|
140 |
};
|