|
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 diceDistance extends distance |
|
16
|
|
|
{ |
|
17
|
|
|
public static function dice($a, $b, $round = 2) |
|
18
|
|
|
{ |
|
19
|
|
|
if ($distance = in_array(self::handleVeryCommonDiceCases($a, $b), array(false, 0.0, 1.0), true)) { |
|
20
|
|
|
return $distance; |
|
21
|
|
|
} |
|
22
|
|
|
static $distance = 0; |
|
23
|
|
|
static $previous = array(); |
|
24
|
|
|
if (array($a, $b) === $previous) { |
|
25
|
|
|
return $distance; |
|
26
|
|
|
} |
|
27
|
|
|
$previous = array($a, $b); |
|
28
|
|
|
$a = self::split($a, 2); |
|
29
|
|
|
$b = self::split($b, 2); |
|
30
|
|
|
self::diceDistance($distance, $a, $b, $round); |
|
31
|
|
|
return $distance; |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
private static function handleVeryCommonDiceCases(&$a, &$b) |
|
35
|
|
|
{ |
|
36
|
|
|
if (!is_string($a) || !is_string($b)) { |
|
37
|
|
|
return false; |
|
38
|
|
|
} |
|
39
|
|
|
if (empty($a) || empty($b)) { |
|
40
|
|
|
return 0.0; |
|
41
|
|
|
} |
|
42
|
|
|
if ($a === $b) { |
|
43
|
|
|
return 1.0; |
|
44
|
|
|
} |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
private static function diceDistance(&$distance, &$a, &$b, $round) |
|
48
|
|
|
{ |
|
49
|
|
|
$ca = ($caGrams = count($a)) * 2 - self::getEndStrLen($a); |
|
50
|
|
|
$cb = ($cbGrams = count($b)) * 2 - self::getEndStrLen($b); |
|
51
|
|
|
$distance = round(2 * count($caGrams > $cbGrams ?array_intersect($a, $b) : array_intersect($b, $a)) / ($ca + $cb), $round); |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
private static function getEndStrLen($a) |
|
55
|
|
|
{ |
|
56
|
|
|
if (function_exists('array_key_last')) { |
|
57
|
|
|
$end = array_key_last($a); |
|
58
|
|
|
$end = (isset($end[1])) ? 0 : 1; |
|
59
|
|
|
} else { |
|
60
|
|
|
$end = end($a); |
|
61
|
|
|
$end = (isset($end[1])) ? 0 : 1; |
|
62
|
|
|
reset($a); |
|
63
|
|
|
} |
|
64
|
|
|
return $end; |
|
65
|
|
|
} |
|
66
|
|
|
} |
|
67
|
|
|
} |
|
68
|
|
|
|