|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace GeoFixer\helpers; |
|
4
|
|
|
|
|
5
|
|
|
/** |
|
6
|
|
|
* Class FuzzySearchHelper |
|
7
|
|
|
* |
|
8
|
|
|
* @package GeoFixer\helpers |
|
9
|
|
|
*/ |
|
10
|
|
|
class FuzzySearchHelper |
|
11
|
|
|
{ |
|
12
|
|
|
public $max_similarity; |
|
13
|
|
|
public $meta_similarity; |
|
14
|
|
|
public $min_levenshtein; |
|
15
|
|
|
public $meta_min_levenshtein; |
|
16
|
|
|
|
|
17
|
|
|
public $result_array = []; |
|
18
|
|
|
public $words_array = []; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Строгий поиск |
|
22
|
|
|
* |
|
23
|
|
|
* @param $word |
|
24
|
|
|
* @param $translited_words |
|
25
|
|
|
* |
|
26
|
|
|
* @return bool|mixed |
|
27
|
|
|
*/ |
|
28
|
9 |
|
public function findBestMatch($word, $translited_words) |
|
29
|
|
|
{ |
|
30
|
9 |
|
$this->refreshVariables(); |
|
31
|
|
|
|
|
32
|
9 |
|
array_walk($translited_words, array($this, 'strictWordsLevenshteinHandlerCallback'), $word); |
|
33
|
9 |
|
$most_similar = $this->findMostSimilarWords($word, $this->words_array); |
|
34
|
9 |
|
array_walk($most_similar, array($this, 'findFinalSimilirityAndLevenshtein'), $word); |
|
35
|
9 |
|
array_walk($most_similar, array($this, 'strictSimilarityCallback'), $word); |
|
36
|
|
|
|
|
37
|
9 |
|
return is_null(key($this->result_array)) ? false : key($this->result_array); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* Нестрогий поиск |
|
42
|
|
|
* |
|
43
|
|
|
* @param $word |
|
44
|
|
|
* @param $words_array |
|
45
|
|
|
* @return array |
|
46
|
|
|
*/ |
|
47
|
19 |
|
public function findMostSimilarWords($word, $words_array) |
|
48
|
|
|
{ |
|
49
|
19 |
|
$this->refreshVariables(); |
|
50
|
|
|
|
|
51
|
19 |
|
array_walk($words_array, array($this, 'similarityCallback'), $word); |
|
52
|
|
|
|
|
53
|
19 |
|
return $this->result_array; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* Вспомогательный метод для findBestMatch |
|
58
|
|
|
* |
|
59
|
|
|
* @param $item |
|
60
|
|
|
* @param $key |
|
61
|
|
|
* @param $word |
|
62
|
|
|
*/ |
|
63
|
2 |
|
private function findFinalSimilirityAndLevenshtein($item, $key, $word) |
|
64
|
|
|
{ |
|
65
|
2 |
|
$this->meta_min_levenshtein = min($this->meta_min_levenshtein, levenshtein(metaphone($item), metaphone($word))); |
|
66
|
2 |
|
if (levenshtein($item, $word) == $this->meta_min_levenshtein) { |
|
67
|
2 |
|
$this->meta_similarity = max($this->meta_similarity, similar_text(metaphone($item), metaphone($word))); |
|
68
|
2 |
|
} |
|
69
|
2 |
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Вспомогательный метод для findBestMatch |
|
73
|
|
|
* |
|
74
|
|
|
* @param $latin |
|
75
|
|
|
* @param $russian |
|
76
|
|
|
* @param $word |
|
77
|
|
|
*/ |
|
78
|
2 |
|
private function strictSimilarityCallback($latin, $russian, $word) |
|
79
|
|
|
{ |
|
80
|
2 |
|
if (levenshtein(metaphone($latin), metaphone($word)) <= $this->meta_min_levenshtein) { |
|
81
|
2 |
|
if (similar_text(metaphone($latin), metaphone($word)) >= $this->meta_similarity) { |
|
82
|
2 |
|
$this->result_array[$russian] = $latin; |
|
83
|
2 |
|
} |
|
84
|
2 |
|
} |
|
85
|
2 |
|
} |
|
86
|
|
|
|
|
87
|
|
|
/** |
|
88
|
|
|
* Вспомогательный метод для findBestMatch |
|
89
|
|
|
* |
|
90
|
|
|
* @param $translit |
|
91
|
|
|
* @param $russian |
|
92
|
|
|
* @param $word |
|
93
|
|
|
*/ |
|
94
|
9 |
|
private function strictWordsLevenshteinHandlerCallback($translit, $russian, $word) |
|
95
|
|
|
{ |
|
96
|
9 |
|
if (levenshtein(metaphone($word), metaphone($translit)) < mb_strlen(metaphone($word)) / 2) { |
|
97
|
2 |
|
if (levenshtein($word, $translit) < mb_strlen($word) / 2) { |
|
98
|
2 |
|
$this->words_array[$russian] = $translit; |
|
99
|
2 |
|
} |
|
100
|
2 |
|
} |
|
101
|
9 |
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* Вспомогательный метод для findMostSimilarWords |
|
105
|
|
|
* |
|
106
|
|
|
* @param $translit |
|
107
|
|
|
* @param $russian |
|
108
|
|
|
* @param $word |
|
109
|
|
|
*/ |
|
110
|
12 |
|
private function similarityCallback($translit, $russian, $word) |
|
111
|
|
|
{ |
|
112
|
12 |
|
if (levenshtein($translit, $word) <= $this->min_levenshtein) { |
|
113
|
12 |
|
if (similar_text($translit, $word) >= $this->max_similarity) { |
|
114
|
12 |
|
$this->result_array = []; |
|
115
|
12 |
|
$this->min_levenshtein = levenshtein($translit, $word); |
|
116
|
12 |
|
$this->max_similarity = similar_text($translit, $word); |
|
117
|
12 |
|
$this->result_array[$russian] = $translit; |
|
118
|
12 |
|
} |
|
119
|
12 |
|
} |
|
120
|
12 |
|
} |
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* Refresh variables |
|
124
|
|
|
*/ |
|
125
|
19 |
|
private function refreshVariables() |
|
126
|
|
|
{ |
|
127
|
19 |
|
$this->max_similarity = 0; |
|
128
|
19 |
|
$this->meta_similarity = 0; |
|
129
|
19 |
|
$this->min_levenshtein = 1000; |
|
130
|
19 |
|
$this->meta_min_levenshtein = 1000; |
|
131
|
19 |
|
} |
|
132
|
|
|
|
|
133
|
|
|
} |
|
134
|
|
|
|
Phrlog /
GeoFixer
