|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* |
|
4
|
|
|
* @Name : similar-text |
|
5
|
|
|
* @Programmer : Akpé Aurelle Emmanuel Moïse Zinsou |
|
6
|
|
|
* @Date : 2019-04-01 |
|
7
|
|
|
* @Released under : https://github.com/manuwhat/similar-text/blob/master/LICENSE |
|
8
|
|
|
* @Repository : https://github.com/manuwhat/similar |
|
9
|
|
|
* |
|
10
|
|
|
**/ |
|
11
|
|
|
|
|
12
|
|
|
|
|
13
|
|
|
namespace EZAMA{ |
|
14
|
|
|
|
|
15
|
|
|
class similar_text |
|
16
|
|
|
{ |
|
17
|
|
|
private function __construct() |
|
18
|
|
|
{ |
|
19
|
|
|
} |
|
20
|
|
|
public static function similarText($a, $b, $round = 2, $insensitive = true, &$stats = false, $getParts = false, $checkposition = false) |
|
21
|
|
|
{ |
|
22
|
|
|
if (!is_string($a) || !is_string($b)) { |
|
23
|
|
|
return false; |
|
24
|
|
|
} |
|
25
|
|
|
if ($insensitive) { |
|
26
|
|
|
$a = self::strtolower($a); |
|
27
|
|
|
$b = self::strtolower($b); |
|
28
|
|
|
} else { |
|
29
|
|
|
$a = self::split($a); |
|
30
|
|
|
$b = self::split($b); |
|
31
|
|
|
} |
|
32
|
|
|
/* prevent bad types and useless memory usage due to for example array instead of simple boolean */ |
|
33
|
|
|
unset($insensitive); |
|
34
|
|
|
$getParts = (bool) $getParts; |
|
35
|
|
|
/* ******************************************************************************************** */ |
|
36
|
|
|
$ca = count($a); |
|
37
|
|
|
$cb = count($b); |
|
38
|
|
|
if ($ca < $cb) { |
|
39
|
|
|
$stats = self::getStats($cb, $a, self::_check($a, $b, $getParts, $round, $checkposition), $getParts, $round); |
|
40
|
|
|
} else { |
|
41
|
|
|
$stats = self::getStats($ca, $b, self::_check($b, $a, $getParts, $round, $checkposition), $getParts, $round); |
|
42
|
|
|
} |
|
43
|
|
|
return $stats['similar']; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
protected static function _check($a, $b, $getParts, $round, $checkposition = false) |
|
47
|
|
|
{ |
|
48
|
|
|
$diff = array(); |
|
49
|
|
|
if ($getParts) { |
|
50
|
|
|
$diff[] = array_diff($a, $b); |
|
51
|
|
|
$diff[] = array_diff($b, $a); |
|
52
|
|
|
} |
|
53
|
|
|
$diff[] = $checkposition ?array_intersect_assoc($a, $b) : array_intersect($a, $b); |
|
54
|
|
|
$diff[] = round(count(array_intersect(self::getParts($a, $c), self::getParts($b))) / $c * 100, $round); |
|
55
|
|
|
$diff[] = $a === $b; |
|
56
|
|
|
return $diff; |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
protected static function getStats($ca, $b, $diff, $getParts, $round) |
|
60
|
|
|
{ |
|
61
|
|
|
$stats = array(); |
|
62
|
|
|
if ($getParts) { |
|
63
|
|
|
$stats['similar'] = round(count($diff[2]) * 100 / $ca, $round); |
|
64
|
|
|
$stats['substr'] = $diff[3]; |
|
65
|
|
|
$stats['contain'] = ($diff[2] === $b) ?true:false; |
|
66
|
|
|
$stats['equal'] = $diff[4]; |
|
67
|
|
|
$stats['a-b'] = $diff[0]; |
|
68
|
|
|
$stats['b-a'] = $diff[1]; |
|
69
|
|
|
$stats['a&b'] = $diff[2]; |
|
70
|
|
|
} else { |
|
71
|
|
|
$stats['similar'] = round(count($diff[0]) * 100 / $ca, $round); |
|
72
|
|
|
$stats['substr'] = $diff[1]; |
|
73
|
|
|
$stats['contain'] = ($diff[0] === $b) ?true:false; |
|
74
|
|
|
$stats['equal'] = $diff[2]; |
|
75
|
|
|
} |
|
76
|
|
|
return $stats; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
protected static function getParts($b, &$c = 0, $lengthCapture=false) |
|
80
|
|
|
{ |
|
81
|
|
|
$parts = array(); |
|
82
|
|
|
$tmp = ''; |
|
83
|
|
|
$c = 0; |
|
84
|
|
|
$length=0; |
|
85
|
|
|
$lengthCapture=(bool)$lengthCapture; |
|
86
|
|
|
if ($lengthCapture) { |
|
87
|
|
|
self::capturePartsWithLength($b, $length, $tmp, $c, $parts); |
|
88
|
|
|
} else { |
|
89
|
|
|
self::capturePartsWithoutLength($b, $tmp, $c, $parts); |
|
90
|
|
|
} |
|
91
|
|
|
return $parts; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
private static function capturePartsWithoutLength(&$b, $tmp, &$c, &$parts) |
|
95
|
|
|
{ |
|
96
|
|
|
foreach ($b as $k=>$v) { |
|
97
|
|
|
if (ctype_space($v) || ctype_punct($v)) { |
|
98
|
|
|
$parts[] = $tmp; |
|
99
|
|
|
$parts[] = $v; |
|
100
|
|
|
$c += 2; |
|
101
|
|
|
$tmp = ''; |
|
102
|
|
|
continue; |
|
103
|
|
|
} |
|
104
|
|
|
$tmp .= $v; |
|
105
|
|
|
} |
|
106
|
|
|
if (!empty($tmp)) { |
|
107
|
|
|
$parts[] = $tmp; |
|
108
|
|
|
$c++; |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
private static function capturePartsWithLength(&$b, $length, $tmp, &$c, &$parts) |
|
113
|
|
|
{ |
|
114
|
|
|
foreach ($b as $k=>$v) { |
|
115
|
|
|
$length++; |
|
116
|
|
|
if (ctype_space($v) || ctype_punct($v)) { |
|
117
|
|
|
$parts[] =array($tmp,$length-1); |
|
118
|
|
|
$parts[] = array($v,1); |
|
119
|
|
|
$c += 2; |
|
120
|
|
|
$tmp = ''; |
|
121
|
|
|
$length=0; |
|
122
|
|
|
continue; |
|
123
|
|
|
} |
|
124
|
|
|
$tmp .= $v; |
|
125
|
|
|
} |
|
126
|
|
|
if (!empty($tmp)) { |
|
127
|
|
|
$parts[] = array($tmp,$length); |
|
128
|
|
|
$c++; |
|
129
|
|
|
} |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
|
|
133
|
|
|
|
|
134
|
|
|
protected static function is_ascii($str) |
|
135
|
|
|
{ |
|
136
|
|
|
if ('' === $str) { |
|
137
|
|
|
return true; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
return !preg_match('/[^\x09\x10\x13\x0A\x0D\x20-\x7E]/', $str); |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
protected static function strtolower($str) |
|
144
|
|
|
{ |
|
145
|
|
|
$split = self::split($str); |
|
146
|
|
|
if (is_array($split)) { |
|
147
|
|
|
return |
|
148
|
|
|
array_map( |
|
149
|
|
|
function ($val) { |
|
150
|
|
|
if (self::is_ascii($val)) { |
|
151
|
|
|
return strtolower($val); |
|
152
|
|
|
} |
|
153
|
|
|
return $val; |
|
154
|
|
|
}, |
|
155
|
|
|
$split |
|
156
|
|
|
) |
|
157
|
|
|
|
|
158
|
|
|
; |
|
159
|
|
|
} else { |
|
160
|
|
|
return array(); |
|
161
|
|
|
} |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
protected static function split($str, $grams=false) |
|
165
|
|
|
{ |
|
166
|
|
|
if (!is_string($str)) { |
|
167
|
|
|
return array(); |
|
168
|
|
|
} |
|
169
|
|
|
static $split = []; |
|
170
|
|
|
static $old = ''; |
|
171
|
|
|
static $oldGrams=1; |
|
172
|
|
|
$grams=is_int($grams) && $grams >=1 && $grams <= strlen($str) ? $grams : false; |
|
173
|
|
|
if ($old === $str && $oldGrams===$grams) { |
|
174
|
|
|
return $split; |
|
175
|
|
|
} else { |
|
176
|
|
|
$old = $str; |
|
177
|
|
|
$oldGrams=$grams; |
|
178
|
|
|
$split = !$grams ? preg_split('//u', $str, -1, PREG_SPLIT_NO_EMPTY):preg_split('/(.{'.$grams.'})/su', $str, -1, PREG_SPLIT_NO_EMPTY|PREG_SPLIT_DELIM_CAPTURE); |
|
179
|
|
|
return $split; |
|
180
|
|
|
} |
|
181
|
|
|
} |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
} |
|
185
|
|
|
|
manuwhat /
similar-text
