|
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 levenshteinDistance extends distance |
|
16
|
|
|
{ |
|
17
|
|
|
public static function levenshtein($a, $b) |
|
18
|
|
|
{ |
|
19
|
|
|
if (!is_string($a) || !is_string($b)) { |
|
20
|
|
|
return false; |
|
21
|
|
|
} |
|
22
|
|
|
|
|
23
|
|
|
|
|
24
|
|
|
static $distance = 0; |
|
25
|
|
|
static $previous = array(); |
|
26
|
|
|
if (array($a, $b) === $previous) { |
|
27
|
|
|
return $distance; |
|
28
|
|
|
} |
|
29
|
|
|
$previous = array($a, $b); |
|
30
|
|
|
$a = self::split($a); |
|
31
|
|
|
$b = self::split($b); |
|
32
|
|
|
$ca = count((array) $a); |
|
33
|
|
|
$cb = count((array) $b); |
|
34
|
|
|
$dis = range(0, $cb); |
|
35
|
|
|
self::BuildLevenshteinCostMatrix($a, $b, $ca, $cb, $dis); |
|
36
|
|
|
|
|
37
|
|
|
return $distance = $dis[$cb]; |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
|
|
41
|
|
|
public static function levenshteinDamerau($a, $b) |
|
42
|
|
|
{ |
|
43
|
|
|
if (!is_string($a) || !is_string($b)) { |
|
44
|
|
|
return false; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
static $distance = 0; |
|
48
|
|
|
static $previous = array(); |
|
49
|
|
|
if (array($a, $b) === $previous) { |
|
50
|
|
|
return $distance; |
|
51
|
|
|
} |
|
52
|
|
|
$previous = array($a, $b); |
|
53
|
|
|
$a = self::split($a); |
|
54
|
|
|
$b = self::split($b); |
|
55
|
|
|
$ca = count((array) $a); |
|
56
|
|
|
$cb = count((array) $b); |
|
57
|
|
|
$dis = range(0, $cb); |
|
58
|
|
|
self::BuildLevenshteinCostMatrix($a, $b, $ca, $cb, $dis, true); |
|
59
|
|
|
|
|
60
|
|
|
return $distance = $dis[$cb]; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
private static function BuildLevenshteinCostMatrix($a, $b, $ca, $cb, &$dis, $damerau = false) |
|
64
|
|
|
{ |
|
65
|
|
|
$dis_new = array(); |
|
66
|
|
|
for ($x = 1; $x <= $ca; $x++) { |
|
67
|
|
|
$dis_new[0] = $x; |
|
68
|
|
|
for ($y = 1; $y <= $cb; $y++) { |
|
69
|
|
|
self::costMatrix($a, $b, $dis_new, $dis, $damerau, $x, $y); |
|
70
|
|
|
} |
|
71
|
|
|
$dis = $dis_new; |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
private static function costMatrix(&$a, &$b, &$dis_new, &$dis, $damerau, $x, $y) |
|
76
|
|
|
{ |
|
77
|
|
|
$c = ($a[$x - 1] == $b[$y - 1]) ? 0 : 1; |
|
78
|
|
|
$dis_new[$y] = min($dis[$y] + 1, $dis_new[$y - 1] + 1, $dis[$y - 1] + $c); |
|
79
|
|
|
if ($damerau) { |
|
80
|
|
|
self::handleDamerau($a, $b, $dis_new, $dis, $x, $y, $c); |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
private static function handleDamerau(&$a, &$b, &$dis_new, &$dis, $x, $y, $c) |
|
85
|
|
|
{ |
|
86
|
|
|
if ($x > 1 && $y > 1 && $a[$x - 1] == $b[$y - 2] && $a[$x - 2] == $b[$y - 1]) { |
|
87
|
|
|
$dis_new[$y] = min($dis_new[$y - 1], $dis[$y - 3] + $c); |
|
88
|
|
|
} |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
} |
|
92
|
|
|
|