|
1
|
|
|
<?php |
|
2
|
|
|
namespace SimilarText; |
|
3
|
|
|
|
|
4
|
|
|
/** |
|
5
|
|
|
* Class Finder |
|
6
|
|
|
* @package SimilarText |
|
7
|
|
|
* |
|
8
|
|
|
* Thanks to: http://php.net/manual/fr/function.levenshtein.php#113702 |
|
9
|
|
|
*/ |
|
10
|
|
|
class Finder |
|
11
|
|
|
{ |
|
12
|
|
|
/** |
|
13
|
|
|
* @var string Needle |
|
14
|
|
|
*/ |
|
15
|
|
|
protected $needle; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* @var array Haystack |
|
19
|
|
|
*/ |
|
20
|
|
|
protected $haystack; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* Hold the sorted comparison stack. |
|
24
|
|
|
* |
|
25
|
|
|
* @var array Haystack |
|
26
|
|
|
*/ |
|
27
|
|
|
protected $sorted_haystack; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @var null|int Threshold |
|
31
|
|
|
*/ |
|
32
|
|
|
protected $threshold; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* Finder constructor. |
|
36
|
|
|
* |
|
37
|
|
|
* @param string $needle |
|
38
|
|
|
* @param array $haystack |
|
39
|
|
|
* @return void |
|
|
|
|
|
|
40
|
|
|
*/ |
|
41
|
15 |
|
public function __construct($needle, $haystack) |
|
42
|
|
|
{ |
|
43
|
15 |
|
$this->needle = $needle; |
|
44
|
15 |
|
$this->haystack = $haystack; |
|
45
|
15 |
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Sort Haystack. |
|
49
|
|
|
* |
|
50
|
|
|
* @return void |
|
51
|
|
|
*/ |
|
52
|
9 |
|
protected function sortHaystack() |
|
53
|
|
|
{ |
|
54
|
9 |
|
$sorted_haystack = []; |
|
55
|
9 |
|
foreach ($this->haystack as $string) { |
|
56
|
9 |
|
$sorted_haystack[$string] = $this->levenshteinUtf8($this->needle, $string); |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
// Apply threshold when set. |
|
60
|
9 |
|
if(!is_null($this->threshold)){ |
|
61
|
3 |
|
$sorted_haystack = array_filter($sorted_haystack, function ($score){ |
|
62
|
3 |
|
return $score <= $this->threshold; |
|
63
|
3 |
|
}); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
9 |
|
asort($sorted_haystack); |
|
67
|
|
|
|
|
68
|
9 |
|
$this->sorted_haystack = $sorted_haystack; |
|
69
|
9 |
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Apply threshold to filter only relevant results. The higher |
|
73
|
|
|
* the threshold the more results there will be returned. |
|
74
|
|
|
* |
|
75
|
|
|
* @param int|null $threshold |
|
76
|
|
|
* @return Finder |
|
77
|
|
|
*/ |
|
78
|
3 |
|
public function threshold($threshold = null) |
|
79
|
|
|
{ |
|
80
|
3 |
|
$this->threshold = $threshold; |
|
81
|
|
|
|
|
82
|
3 |
|
return $this; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
/** |
|
86
|
|
|
* Return the highest match. |
|
87
|
|
|
* |
|
88
|
|
|
* @return mixed |
|
89
|
|
|
*/ |
|
90
|
6 |
|
public function first() |
|
91
|
|
|
{ |
|
92
|
6 |
|
$this->sortHaystack(); |
|
93
|
6 |
|
reset($this->sorted_haystack); |
|
94
|
6 |
|
return key($this->sorted_haystack); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
/** |
|
98
|
|
|
* Return all strings in sorted match order. |
|
99
|
|
|
* |
|
100
|
|
|
* @return array |
|
101
|
|
|
*/ |
|
102
|
6 |
|
public function all() |
|
103
|
|
|
{ |
|
104
|
6 |
|
$this->sortHaystack(); |
|
105
|
6 |
|
return array_keys($this->sorted_haystack); |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
/** |
|
109
|
|
|
* Return whether there is an exact match. |
|
110
|
|
|
* |
|
111
|
|
|
* @return bool |
|
112
|
|
|
*/ |
|
113
|
6 |
|
public function hasExactMatch() |
|
114
|
|
|
{ |
|
115
|
6 |
|
return in_array($this->needle, $this->haystack); |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* Ensure a string only uses ascii characters. |
|
120
|
|
|
* |
|
121
|
|
|
* @param string $str |
|
122
|
|
|
* @param array $map |
|
123
|
|
|
* @return string |
|
124
|
|
|
*/ |
|
125
|
9 |
|
protected function utf8ToExtendedAscii($str, &$map) |
|
126
|
|
|
{ |
|
127
|
|
|
// Find all multi-byte characters (cf. utf-8 encoding specs). |
|
128
|
9 |
|
$matches = array(); |
|
129
|
9 |
|
if (!preg_match_all('/[\xC0-\xF7][\x80-\xBF]+/', $str, $matches)) { |
|
130
|
9 |
|
return $str; // plain ascii string |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
// Update the encoding map with the characters not already met. |
|
134
|
3 |
|
foreach ($matches[0] as $mbc) { |
|
135
|
3 |
|
if (!isset($map[$mbc])) { |
|
136
|
3 |
|
$map[$mbc] = chr(128 + count($map)); |
|
137
|
|
|
} |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
// Finally remap non-ascii characters. |
|
141
|
3 |
|
return strtr($str, $map); |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
/** |
|
145
|
|
|
* Calculate the levenshtein distance between two strings. |
|
146
|
|
|
* |
|
147
|
|
|
* @param string $string1 |
|
148
|
|
|
* @param string $string2 |
|
149
|
|
|
* @return int |
|
150
|
|
|
*/ |
|
151
|
9 |
|
protected function levenshteinUtf8($string1, $string2) |
|
152
|
|
|
{ |
|
153
|
9 |
|
$charMap = array(); |
|
154
|
9 |
|
$string1 = $this->utf8ToExtendedAscii($string1, $charMap); |
|
155
|
9 |
|
$string2 = $this->utf8ToExtendedAscii($string2, $charMap); |
|
156
|
|
|
|
|
157
|
9 |
|
return levenshtein($string1, $string2); |
|
158
|
|
|
} |
|
159
|
|
|
} |
|
160
|
|
|
|
rap2hpoutre /
similar-text-finder
Adding a
@returnannotation to a constructor is not recommended, since a constructor does not have a meaningful return value.Please refer to the PHP core documentation on constructors.