1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Onoi\Tesa\Tests; |
4
|
|
|
|
5
|
|
|
use Onoi\Tesa\SanitizerFactory; |
6
|
|
|
|
7
|
|
|
/** |
8
|
|
|
* @covers \Onoi\Tesa\SanitizerFactory |
9
|
|
|
* @group onoi-tesa |
10
|
|
|
* |
11
|
|
|
* @license GNU GPL v2+ |
12
|
|
|
* @since 0.1 |
13
|
|
|
* |
14
|
|
|
* @author mwjames |
15
|
|
|
*/ |
16
|
|
|
class SanitizerFactoryTest extends \PHPUnit_Framework_TestCase { |
17
|
|
|
|
18
|
|
|
private $sanitizerFactory; |
19
|
|
|
|
20
|
|
|
protected function setUp() { |
21
|
|
|
$this->sanitizerFactory = new SanitizerFactory(); |
22
|
|
|
} |
23
|
|
|
|
24
|
|
|
public function testCanConstructSanitizer() { |
25
|
|
|
|
26
|
|
|
$this->assertInstanceOf( |
27
|
|
|
'\Onoi\Tesa\Sanitizer', |
28
|
|
|
$this->sanitizerFactory->newSanitizer() |
29
|
|
|
); |
30
|
|
|
} |
31
|
|
|
|
32
|
|
|
/* StopwordAnalyzer */ |
33
|
|
|
|
34
|
|
|
public function testCanConstructStopwordAnalyzerByNullLanguage() { |
35
|
|
|
|
36
|
|
|
$this->assertInstanceOf( |
37
|
|
|
'\Onoi\Tesa\StopwordAnalyzer\StopwordAnalyzer', |
38
|
|
|
$this->sanitizerFactory->newStopwordAnalyzerByLanguage( null ) |
39
|
|
|
); |
40
|
|
|
} |
41
|
|
|
|
42
|
|
|
public function testCanConstructStopwordAnalyzerByEnLanguage() { |
43
|
|
|
|
44
|
|
|
$this->assertInstanceOf( |
45
|
|
|
'\Onoi\Tesa\StopwordAnalyzer\CdbStopwordAnalyzer', |
46
|
|
|
$this->sanitizerFactory->newStopwordAnalyzerByLanguage( 'EN' ) |
47
|
|
|
); |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
public function testCanConstructCdbStopwordAnalyzer() { |
51
|
|
|
|
52
|
|
|
$this->assertInstanceOf( |
53
|
|
|
'\Onoi\Tesa\StopwordAnalyzer\CdbStopwordAnalyzer', |
54
|
|
|
$this->sanitizerFactory->newCdbStopwordAnalyzer() |
55
|
|
|
); |
56
|
|
|
} |
57
|
|
|
|
58
|
|
|
public function testCanConstructArrayStopwordAnalyzer() { |
59
|
|
|
|
60
|
|
|
$this->assertInstanceOf( |
61
|
|
|
'\Onoi\Tesa\StopwordAnalyzer\ArrayStopwordAnalyzer', |
62
|
|
|
$this->sanitizerFactory->newArrayStopwordAnalyzer() |
63
|
|
|
); |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
public function testCanConstructNullStopwordAnalyzer() { |
67
|
|
|
|
68
|
|
|
$this->assertInstanceOf( |
69
|
|
|
'\Onoi\Tesa\StopwordAnalyzer\NullStopwordAnalyzer', |
70
|
|
|
$this->sanitizerFactory->newNullStopwordAnalyzer() |
71
|
|
|
); |
72
|
|
|
} |
73
|
|
|
|
74
|
|
|
/* Synonymizer */ |
75
|
|
|
|
76
|
|
|
public function testCanConstructSynonymizerByLanguage() { |
77
|
|
|
|
78
|
|
|
$this->assertInstanceOf( |
79
|
|
|
'\Onoi\Tesa\Synonymizer\Synonymizer', |
80
|
|
|
$this->sanitizerFactory->newSynonymizerByLanguage() |
81
|
|
|
); |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
public function testCanConstructNullSynonymizer() { |
85
|
|
|
|
86
|
|
|
$this->assertInstanceOf( |
87
|
|
|
'\Onoi\Tesa\Synonymizer\NullSynonymizer', |
88
|
|
|
$this->sanitizerFactory->newNullSynonymizer() |
89
|
|
|
); |
90
|
|
|
} |
91
|
|
|
|
92
|
|
|
/* LanguageDetector */ |
93
|
|
|
|
94
|
|
|
public function testCanConstructNullLanguageDetector() { |
95
|
|
|
|
96
|
|
|
$this->assertInstanceOf( |
97
|
|
|
'\Onoi\Tesa\LanguageDetector\NullLanguageDetector', |
98
|
|
|
$this->sanitizerFactory->newNullLanguageDetector() |
99
|
|
|
); |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
public function testCanConstructTextCatLanguageDetector() { |
103
|
|
|
|
104
|
|
|
$this->assertInstanceOf( |
105
|
|
|
'\Onoi\Tesa\LanguageDetector\TextCatLanguageDetector', |
106
|
|
|
$this->sanitizerFactory->newTextCatLanguageDetector() |
107
|
|
|
); |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
/* Tokenizer */ |
111
|
|
|
|
112
|
|
|
public function testCanConstructPreferredTokenizerByLanguage() { |
113
|
|
|
|
114
|
|
|
$this->assertInstanceOf( |
115
|
|
|
'\Onoi\Tesa\Tokenizer\Tokenizer', |
116
|
|
|
$this->sanitizerFactory->newPreferredTokenizerByLanguage( 'テスト' ) |
117
|
|
|
); |
118
|
|
|
|
119
|
|
|
$this->assertInstanceOf( |
120
|
|
|
'\Onoi\Tesa\Tokenizer\Tokenizer', |
121
|
|
|
$this->sanitizerFactory->newPreferredTokenizerByLanguage( '在延安更名为新' ) |
122
|
|
|
); |
123
|
|
|
} |
124
|
|
|
|
125
|
|
|
public function testCanConstructnewCJKMatchableTokenizer() { |
126
|
|
|
|
127
|
|
|
$this->assertInstanceOf( |
128
|
|
|
'\Onoi\Tesa\Tokenizer\Tokenizer', |
129
|
|
|
$this->sanitizerFactory->newCJKMatchableTokenizer( 'テスト' ) |
130
|
|
|
); |
131
|
|
|
|
132
|
|
|
$this->assertInstanceOf( |
133
|
|
|
'\Onoi\Tesa\Tokenizer\Tokenizer', |
134
|
|
|
$this->sanitizerFactory->newCJKMatchableTokenizer( '在延安更名为新' ) |
135
|
|
|
); |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
public function testCanConstructIcuWordBoundaryTokenizer() { |
139
|
|
|
|
140
|
|
|
$this->assertInstanceOf( |
141
|
|
|
'\Onoi\Tesa\Tokenizer\IcuWordBoundaryTokenizer', |
142
|
|
|
$this->sanitizerFactory->newIcuWordBoundaryTokenizer() |
143
|
|
|
); |
144
|
|
|
} |
145
|
|
|
|
146
|
|
|
public function testCanConstructGenericRegExTokenizer() { |
147
|
|
|
|
148
|
|
|
$this->assertInstanceOf( |
149
|
|
|
'\Onoi\Tesa\Tokenizer\GenericRegExTokenizer', |
150
|
|
|
$this->sanitizerFactory->newGenericRegExTokenizer() |
151
|
|
|
); |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
public function testCanConstructPunctuationRegExTokenizer() { |
155
|
|
|
|
156
|
|
|
$this->assertInstanceOf( |
157
|
|
|
'\Onoi\Tesa\Tokenizer\PunctuationRegExTokenizer', |
158
|
|
|
$this->sanitizerFactory->newPunctuationRegExTokenizer() |
159
|
|
|
); |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
public function testCanConstructJaCompoundGroupTokenizer() { |
163
|
|
|
|
164
|
|
|
$this->assertInstanceOf( |
165
|
|
|
'\Onoi\Tesa\Tokenizer\JaCompoundGroupTokenizer', |
166
|
|
|
$this->sanitizerFactory->newJaCompoundGroupTokenizer() |
167
|
|
|
); |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
public function testCanConstructJaTinySegmenterTokenizer() { |
171
|
|
|
|
172
|
|
|
$this->assertInstanceOf( |
173
|
|
|
'\Onoi\Tesa\Tokenizer\JaTinySegmenterTokenizer', |
174
|
|
|
$this->sanitizerFactory->newJaTinySegmenterTokenizer() |
175
|
|
|
); |
176
|
|
|
} |
177
|
|
|
|
178
|
|
|
public function testCanConstructCJKSimpleCharacterRegExTokenizer() { |
179
|
|
|
|
180
|
|
|
$this->assertInstanceOf( |
181
|
|
|
'\Onoi\Tesa\Tokenizer\CJKSimpleCharacterRegExTokenizer', |
182
|
|
|
$this->sanitizerFactory->newCJKSimpleCharacterRegExTokenizer() |
183
|
|
|
); |
184
|
|
|
} |
185
|
|
|
|
186
|
|
|
public function testCanConstructNGramTokenizer() { |
187
|
|
|
|
188
|
|
|
$this->assertInstanceOf( |
189
|
|
|
'\Onoi\Tesa\Tokenizer\NGramTokenizer', |
190
|
|
|
$this->sanitizerFactory->newNGramTokenizer() |
191
|
|
|
); |
192
|
|
|
} |
193
|
|
|
|
194
|
|
|
} |
195
|
|
|
|