|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace TextUtils; |
|
4
|
|
|
|
|
5
|
|
|
use TextUtils\Exceptions\InvalidArgumentException; |
|
6
|
|
|
|
|
7
|
|
|
class NGram |
|
8
|
|
|
{ |
|
9
|
|
|
/** |
|
10
|
|
|
* The length of the n-gram. |
|
11
|
|
|
* |
|
12
|
|
|
* @var int |
|
13
|
|
|
*/ |
|
14
|
|
|
protected $n; |
|
15
|
|
|
|
|
16
|
|
|
protected $string; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* NGram constructor. |
|
20
|
|
|
* |
|
21
|
|
|
* @param int $n |
|
22
|
|
|
* @param string $string |
|
23
|
|
|
* |
|
24
|
|
|
* @throws \TextUtils\Exceptions\InvalidArgumentException |
|
25
|
|
|
*/ |
|
26
|
|
|
public function __construct(int $n, string $string) |
|
27
|
|
|
{ |
|
28
|
|
|
$this->setN($n); |
|
29
|
|
|
$this->setString($string); |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* Static wrapper for n-gram generator. |
|
34
|
|
|
* |
|
35
|
|
|
* @param string $text |
|
36
|
|
|
* @param int $n |
|
37
|
|
|
* |
|
38
|
|
|
* @throws \TextUtils\Exceptions\InvalidArgumentException |
|
39
|
|
|
* |
|
40
|
|
|
* @return array |
|
41
|
|
|
*/ |
|
42
|
|
|
public static function for(string $text, int $n = 3) |
|
43
|
|
|
{ |
|
44
|
|
|
return (new static($n, $text))->get(); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Static wrapper to generate a bigram. |
|
49
|
|
|
* |
|
50
|
|
|
* @param string $text |
|
51
|
|
|
* |
|
52
|
|
|
* @throws \TextUtils\Exceptions\InvalidArgumentException |
|
53
|
|
|
* |
|
54
|
|
|
* @return array |
|
55
|
|
|
*/ |
|
56
|
|
|
public static function bigram(string $text) : array |
|
57
|
|
|
{ |
|
58
|
|
|
return self::for($text, 2); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* Static wrapper to generate a trigram. |
|
63
|
|
|
* |
|
64
|
|
|
* @param string $text |
|
65
|
|
|
* |
|
66
|
|
|
* @throws \TextUtils\Exceptions\InvalidArgumentException |
|
67
|
|
|
* |
|
68
|
|
|
* @return array |
|
69
|
|
|
*/ |
|
70
|
|
|
public static function trigram(string $text) : array |
|
71
|
|
|
{ |
|
72
|
|
|
return self::for($text, 3); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
/** |
|
76
|
|
|
* Generate the N-gram for the provided string. |
|
77
|
|
|
* |
|
78
|
|
|
* @return array |
|
79
|
|
|
*/ |
|
80
|
|
|
public function get() : array |
|
81
|
|
|
{ |
|
82
|
|
|
$nGrams = []; |
|
83
|
|
|
|
|
84
|
|
|
$text = $this->getString(); |
|
85
|
|
|
$max = \mb_strlen($text); |
|
86
|
|
|
$n = $this->getN(); |
|
87
|
|
|
for ($i = 0; $i + $n <= $max; $i++) { |
|
88
|
|
|
$partial = ''; |
|
89
|
|
|
for ($j = 0; $j < $n; $j++) { |
|
90
|
|
|
$partial .= $text[$j + $i]; |
|
91
|
|
|
} |
|
92
|
|
|
$nGrams[] = $partial; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
return $nGrams; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* @return int |
|
100
|
|
|
*/ |
|
101
|
|
|
public function getN() : int |
|
102
|
|
|
{ |
|
103
|
|
|
return $this->n; |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
/** |
|
107
|
|
|
* Set the length of the n-gram. |
|
108
|
|
|
* |
|
109
|
|
|
* @param int $n |
|
110
|
|
|
* |
|
111
|
|
|
* @throws \TextUtils\Exceptions\InvalidArgumentException |
|
112
|
|
|
* |
|
113
|
|
|
* @return \TextUtils\NGram |
|
114
|
|
|
*/ |
|
115
|
|
|
public function setN(int $n) : NGram |
|
116
|
|
|
{ |
|
117
|
|
|
if ($n < 1) { |
|
118
|
|
|
throw new InvalidArgumentException('Provided number cannot be smaller than 1'); |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
$this->n = $n; |
|
122
|
|
|
|
|
123
|
|
|
return $this; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
/** |
|
127
|
|
|
* Set the string to create the n-gram for. |
|
128
|
|
|
* |
|
129
|
|
|
* @param string $string |
|
130
|
|
|
* |
|
131
|
|
|
* @return \TextUtils\NGram |
|
132
|
|
|
*/ |
|
133
|
|
|
public function setString(string $string) : NGram |
|
134
|
|
|
{ |
|
135
|
|
|
$this->string = $string; |
|
136
|
|
|
|
|
137
|
|
|
return $this; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
/** |
|
141
|
|
|
* Get the string used for the n-gram. |
|
142
|
|
|
* |
|
143
|
|
|
* @return string |
|
144
|
|
|
*/ |
|
145
|
|
|
public function getString() : string |
|
146
|
|
|
{ |
|
147
|
|
|
return $this->string; |
|
148
|
|
|
} |
|
149
|
|
|
} |
|
150
|
|
|
|
devinbeeuwkes /
ngram
