|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* |
|
5
|
|
|
* @Name : similar-text |
|
6
|
|
|
* @Programmer : Akpé Aurelle Emmanuel Moïse Zinsou |
|
7
|
|
|
* @Date : 2019-04-01 |
|
8
|
|
|
* @Released under : https://github.com/manuwhat/similar-text/blob/master/LICENSE |
|
9
|
|
|
* @Repository : https://github.com/manuwhat/similar |
|
10
|
|
|
* |
|
11
|
|
|
**/ |
|
12
|
|
|
|
|
13
|
|
|
|
|
14
|
|
|
namespace EZAMA{ |
|
15
|
|
|
class jaroWinklerDistance extends distance |
|
16
|
|
|
{ |
|
17
|
|
|
public static function jaroWinkler($a, $b, $round = 2) |
|
18
|
|
|
{ |
|
19
|
|
|
if (!is_string($a) || !is_string($b)) { |
|
20
|
|
|
return false; |
|
21
|
|
|
} |
|
22
|
|
|
static $distance = array(); |
|
23
|
|
|
static $previous = array(); |
|
24
|
|
|
if (array($a, $b) === $previous) { |
|
25
|
|
|
return $distance; |
|
26
|
|
|
} |
|
27
|
|
|
$previous = array($a, $b); |
|
28
|
|
|
return self::getJWDistance($a, $b, $distance, $round); |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
|
|
32
|
|
|
|
|
33
|
|
|
private static function getJWDistance(&$a, &$b, &$distance, $round) |
|
34
|
|
|
{ |
|
35
|
|
|
extract(self::prepareJaroWinkler($a, $b)); |
|
36
|
|
|
for ($i = 0, $min = min(count((array) $a), count((array) $b)), $t = 0; $i < $min; $i++) { |
|
37
|
|
|
if ($a[$i] !== $b[$i]) { |
|
38
|
|
|
$t++; |
|
39
|
|
|
} |
|
40
|
|
|
} |
|
41
|
|
|
$t /= 2; |
|
42
|
|
|
$distance['jaro'] = 1 / 3 * ($corresponding / $ca + $corresponding / $cb + ($corresponding - $t) / $corresponding); |
|
43
|
|
|
$distance['jaro-winkler'] = $distance['jaro'] + (min($longCommonSubstr, 4) * 0.1 * (1 - $distance['jaro'])); |
|
44
|
|
|
$distance = array_map(function($v) use ($round) { |
|
45
|
|
|
return round($v, $round); |
|
46
|
|
|
}, $distance); |
|
47
|
|
|
|
|
48
|
|
|
return $distance; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
private static function prepareJaroWinkler(&$a, &$b) |
|
52
|
|
|
{ |
|
53
|
|
|
$a = self::split($a); |
|
54
|
|
|
$b = self::split($b); |
|
55
|
|
|
$transpositions = array('a'=>array(), 'b'=>array(), 'corresponding'=>0, 'longCommonSubstr'=>0, 'ca'=>count((array) $a), 'cb'=>count((array) $b)); |
|
56
|
|
|
$Δ = max($transpositions['ca'], $transpositions['cb']) / 2 - 1; |
|
57
|
|
|
self::jwMatches($a, $b, $transpositions, $Δ); |
|
58
|
|
|
ksort($transpositions['a']); |
|
59
|
|
|
ksort($transpositions['b']); |
|
60
|
|
|
$transpositions['a'] = array_values($transpositions['a']); |
|
61
|
|
|
$transpositions['b'] = array_values($transpositions['b']); |
|
62
|
|
|
return $transpositions; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
private static function jwMatches(&$a, &$b, &$transpositions, $Δ) |
|
66
|
|
|
{ |
|
67
|
|
|
foreach ($a as $ind=>$chr) { |
|
68
|
|
|
foreach ($b as $index=>$char) { |
|
69
|
|
|
self::_jwMatches($chr, $char, $index, $ind, $transpositions, $Δ); |
|
70
|
|
|
} |
|
71
|
|
|
} |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
private static function _jwMatches($chr, $char, $index, $ind, &$transpositions, $Δ) |
|
75
|
|
|
{ |
|
76
|
|
|
if ($chr === $char && (abs($index - $ind) <= $Δ)) { |
|
77
|
|
|
if ($ind !== $index) { |
|
78
|
|
|
$transpositions['a'][$ind] = $chr; |
|
79
|
|
|
$transpositions['b'][$index] = $char; |
|
80
|
|
|
} else { |
|
81
|
|
|
if ($ind - 1 <= $transpositions['longCommonSubstr']) { |
|
82
|
|
|
$transpositions['longCommonSubstr']++; |
|
83
|
|
|
} |
|
84
|
|
|
} |
|
85
|
|
|
$transpositions['corresponding']++; |
|
86
|
|
|
} |
|
87
|
|
|
} |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
} |
|
91
|
|
|
|
manuwhat /
similar-text
