|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Onoi\Tesa\Tokenizer; |
|
4
|
|
|
|
|
5
|
|
|
use Onoi\Tesa\CharacterExaminer; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* @license GNU GPL v2+ |
|
9
|
|
|
* @since 0.1 |
|
10
|
|
|
* |
|
11
|
|
|
* @author mwjames |
|
12
|
|
|
*/ |
|
13
|
|
|
class JaCompoundGroupTokenizer implements Tokenizer { |
|
14
|
|
|
|
|
15
|
|
|
private $compound = [ |
|
16
|
|
|
"あっ", |
|
17
|
|
|
"あり", |
|
18
|
|
|
"ある", |
|
19
|
|
|
"い", |
|
20
|
|
|
"いう", |
|
21
|
|
|
"いる", |
|
22
|
|
|
"う", |
|
23
|
|
|
"うち", |
|
24
|
|
|
"お", |
|
25
|
|
|
"および", |
|
26
|
|
|
"おり", |
|
27
|
|
|
"か", |
|
28
|
|
|
"かつて", |
|
29
|
|
|
"から", |
|
30
|
|
|
"が", |
|
31
|
|
|
"き", |
|
32
|
|
|
"ここ", |
|
33
|
|
|
"こと", |
|
34
|
|
|
"この", |
|
35
|
|
|
"これ", |
|
36
|
|
|
"これら", |
|
37
|
|
|
"さ", |
|
38
|
|
|
"さらに", |
|
39
|
|
|
"し", |
|
40
|
|
|
"しかし", |
|
41
|
|
|
"する", |
|
42
|
|
|
"ず", |
|
43
|
|
|
"せ", |
|
44
|
|
|
"せる", |
|
45
|
|
|
"そして", |
|
46
|
|
|
"その", |
|
47
|
|
|
"その他", |
|
48
|
|
|
"その後", |
|
49
|
|
|
"それ", |
|
50
|
|
|
"それぞれ", |
|
51
|
|
|
"た", |
|
52
|
|
|
"ただし", |
|
53
|
|
|
"たち", |
|
54
|
|
|
"ため", |
|
55
|
|
|
"たり", |
|
56
|
|
|
"だ", |
|
57
|
|
|
"だっ", |
|
58
|
|
|
"つ", |
|
59
|
|
|
"て", |
|
60
|
|
|
"で", |
|
61
|
|
|
"でき", |
|
62
|
|
|
"できる", |
|
63
|
|
|
"です", |
|
64
|
|
|
"では", |
|
65
|
|
|
"でも", |
|
66
|
|
|
"と", |
|
67
|
|
|
"という", |
|
68
|
|
|
"といった", |
|
69
|
|
|
"とき", |
|
70
|
|
|
"ところ", |
|
71
|
|
|
"として", |
|
72
|
|
|
"とともに", |
|
73
|
|
|
"とも", |
|
74
|
|
|
"と共に", |
|
75
|
|
|
"な", |
|
76
|
|
|
"ない", |
|
77
|
|
|
"なお", |
|
78
|
|
|
"なかっ", |
|
79
|
|
|
"ながら", |
|
80
|
|
|
"なく", |
|
81
|
|
|
"なっ", |
|
82
|
|
|
"など", |
|
83
|
|
|
"なら", |
|
84
|
|
|
"なり", |
|
85
|
|
|
"なる", |
|
86
|
|
|
"に", |
|
87
|
|
|
"において", |
|
88
|
|
|
"における", |
|
89
|
|
|
"について", |
|
90
|
|
|
"にて", |
|
91
|
|
|
"によって", |
|
92
|
|
|
"により", |
|
93
|
|
|
"による", |
|
94
|
|
|
"に対して", |
|
95
|
|
|
"に対する", |
|
96
|
|
|
"に関する", |
|
97
|
|
|
"の", |
|
98
|
|
|
"ので", |
|
99
|
|
|
"のみ", |
|
100
|
|
|
"は", |
|
101
|
|
|
"ば", |
|
102
|
|
|
"へ", |
|
103
|
|
|
"ほか", |
|
104
|
|
|
"ほとんど", |
|
105
|
|
|
"ほど", |
|
106
|
|
|
"ます", |
|
107
|
|
|
"また", |
|
108
|
|
|
"または", |
|
109
|
|
|
"まで", |
|
110
|
|
|
"も", |
|
111
|
|
|
"もの", |
|
112
|
|
|
"ものの", |
|
113
|
|
|
"や", |
|
114
|
|
|
"よう", |
|
115
|
|
|
"より", |
|
116
|
|
|
"ら", |
|
117
|
|
|
"られ", |
|
118
|
|
|
"られる", |
|
119
|
|
|
"れ", |
|
120
|
|
|
"れる", |
|
121
|
|
|
"を", |
|
122
|
|
|
"ん", |
|
123
|
|
|
"及び", |
|
124
|
|
|
"特に", |
|
125
|
|
|
"、", |
|
126
|
|
|
"。", |
|
127
|
|
|
"「", |
|
128
|
|
|
"」" |
|
129
|
|
|
]; |
|
130
|
|
|
|
|
131
|
|
|
/** |
|
132
|
|
|
* @var Tokenizer |
|
133
|
|
|
*/ |
|
134
|
|
|
private $tokenizer; |
|
135
|
|
|
|
|
136
|
|
|
/** |
|
137
|
|
|
* @since 0.1 |
|
138
|
|
|
* |
|
139
|
|
|
* @param Tokenizer $tokenizer |
|
140
|
|
|
*/ |
|
141
|
|
|
public function __construct( Tokenizer $tokenizer = null ) { |
|
142
|
|
|
$this->tokenizer = $tokenizer; |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
/** |
|
146
|
|
|
* @since 0.1 |
|
147
|
|
|
* |
|
148
|
|
|
* {@inheritDoc} |
|
149
|
|
|
*/ |
|
150
|
|
|
public function setOption( $name, $value ) { |
|
151
|
|
|
if ( $this->tokenizer !== null ) { |
|
152
|
|
|
$this->tokenizer->setOption( $name, $value ); |
|
153
|
|
|
} |
|
154
|
|
|
} |
|
155
|
|
|
|
|
156
|
|
|
/** |
|
157
|
|
|
* @since 0.1 |
|
158
|
|
|
* |
|
159
|
|
|
* {@inheritDoc} |
|
160
|
|
|
*/ |
|
161
|
|
|
public function isWordTokenizer() { |
|
162
|
|
|
return false; |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
/** |
|
166
|
|
|
* @since 0.1 |
|
167
|
|
|
* |
|
168
|
|
|
* {@inheritDoc} |
|
169
|
|
|
*/ |
|
170
|
|
|
public function tokenize( $string ) { |
|
171
|
|
|
|
|
172
|
|
|
if ( $this->tokenizer !== null ) { |
|
173
|
|
|
$string = implode( " ", $this->tokenizer->tokenize( $string ) ); |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
$result = explode( " " , $this->splitByCharacterGroup( |
|
177
|
|
|
str_replace( $this->compound, ' ', $string ) ) |
|
178
|
|
|
); |
|
179
|
|
|
|
|
180
|
|
|
foreach ( $result as $key => $value ) { |
|
181
|
|
|
if ( $value === '' ) { |
|
182
|
|
|
unset( $result[$key] ); |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
|
|
// Single katakana/hiragana are exempted |
|
186
|
|
|
if ( mb_strlen( $value ) === 1 && CharacterExaminer::contains( CharacterExaminer::HIRAGANA_KATAKANA, $value ) ) { |
|
187
|
|
|
unset( $result[$key] ); |
|
188
|
|
|
} |
|
189
|
|
|
} |
|
190
|
|
|
|
|
191
|
|
|
if ( $result !== false ) { |
|
192
|
|
|
return array_values( $result ); |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
return array(); |
|
196
|
|
|
} |
|
197
|
|
|
|
|
198
|
|
|
/** |
|
199
|
|
|
* @see MediaWiki LanguageJa::segmentByWord |
|
200
|
|
|
* |
|
201
|
|
|
* @since 0.1 |
|
202
|
|
|
* |
|
203
|
|
|
* {@inheritDoc} |
|
204
|
|
|
*/ |
|
205
|
|
|
public function splitByCharacterGroup( $string ) { |
|
206
|
|
|
|
|
207
|
|
|
// Space strings of like hiragana/katakana/kanji |
|
208
|
|
|
$hiragana = '(?:\xe3(?:\x81[\x80-\xbf]|\x82[\x80-\x9f]))'; # U3040-309f |
|
209
|
|
|
$katakana = '(?:\xe3(?:\x82[\xa0-\xbf]|\x83[\x80-\xbf]))'; # U30a0-30ff |
|
210
|
|
|
$kanji = '(?:\xe3[\x88-\xbf][\x80-\xbf]' |
|
211
|
|
|
. '|[\xe4-\xe8][\x80-\xbf]{2}' |
|
212
|
|
|
. '|\xe9[\x80-\xa5][\x80-\xbf]' |
|
213
|
|
|
. '|\xe9\xa6[\x80-\x99])'; |
|
214
|
|
|
# U3200-9999 = \xe3\x88\x80-\xe9\xa6\x99 |
|
215
|
|
|
|
|
216
|
|
|
$reg = "/({$hiragana}+|{$katakana}+|{$kanji}+)/"; |
|
217
|
|
|
|
|
218
|
|
|
return $this->insertSpace( $string, $reg ); |
|
219
|
|
|
} |
|
220
|
|
|
|
|
221
|
|
|
private function insertSpace( $string, $pattern ) { |
|
222
|
|
|
return preg_replace( '/ +/', ' ', preg_replace( $pattern, " $1 ", $string ) ); |
|
223
|
|
|
} |
|
224
|
|
|
|
|
225
|
|
|
} |
|
226
|
|
|
|
onoi /
tesa
