|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Matecat\Finder\Helper; |
|
4
|
|
|
|
|
5
|
|
|
class Strings |
|
6
|
|
|
{ |
|
7
|
|
|
/** |
|
8
|
|
|
* @param string $string |
|
9
|
|
|
* |
|
10
|
|
|
* @return string |
|
11
|
|
|
*/ |
|
12
|
|
|
public static function cutNbsps($string) |
|
13
|
|
|
{ |
|
14
|
|
|
return str_replace([' ', ' '], ' ', $string); |
|
15
|
|
|
} |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* @param string $string |
|
19
|
|
|
* |
|
20
|
|
|
* @return string |
|
21
|
|
|
*/ |
|
22
|
|
|
public static function htmlEntityDecode($string) |
|
23
|
|
|
{ |
|
24
|
|
|
return html_entity_decode($string, ENT_QUOTES|ENT_XHTML, 'UTF-8'); |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @param string $string |
|
29
|
|
|
* |
|
30
|
|
|
* @return array|bool|string[]|null |
|
31
|
|
|
*/ |
|
32
|
|
|
public static function split($string) |
|
33
|
|
|
{ |
|
34
|
|
|
return mb_str_split($string); |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @param int $length |
|
39
|
|
|
* |
|
40
|
|
|
* @return string |
|
41
|
|
|
*/ |
|
42
|
|
|
public static function token($length = 8) |
|
43
|
|
|
{ |
|
44
|
|
|
$key = ''; |
|
45
|
|
|
$keys = array_merge(range(0, 9), range('a', 'z')); |
|
46
|
|
|
|
|
47
|
|
|
for ($i = 0; $i < $length; $i++) { |
|
48
|
|
|
$key .= $keys[array_rand($keys)]; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
return $key; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* @param string $string |
|
56
|
|
|
* |
|
57
|
|
|
* @return bool |
|
58
|
|
|
*/ |
|
59
|
|
|
public static function isMultibyte($string) |
|
60
|
|
|
{ |
|
61
|
|
|
return ((strlen($string) - mb_strlen($string)) > 0); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* @param $needle |
|
66
|
|
|
* @param $haystack |
|
67
|
|
|
* |
|
68
|
|
|
* @return bool |
|
69
|
|
|
*/ |
|
70
|
|
|
public static function contains($needle, $haystack) |
|
71
|
|
|
{ |
|
72
|
|
|
if(empty($haystack)){ |
|
73
|
|
|
return false; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
return strpos($haystack, $needle) !== false; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
public static function firstChar($string) |
|
80
|
|
|
{ |
|
81
|
|
|
return $string[0]; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
public static function lastChar($string) |
|
85
|
|
|
{ |
|
86
|
|
|
return $string[strlen($string) - 1]; |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* @param string $string |
|
91
|
|
|
* @return string |
|
92
|
|
|
*/ |
|
93
|
|
|
public static function protectHTMLTags($string) |
|
94
|
|
|
{ |
|
95
|
|
|
preg_match_all('/<(.*?)>|<(.*?)>/sm', $string, $matches); |
|
96
|
|
|
|
|
97
|
|
|
if(!empty($matches[0])){ |
|
98
|
|
|
foreach ($matches[0] as $index => $element){ |
|
99
|
|
|
|
|
100
|
|
|
$tag = explode(" ", $element); |
|
101
|
|
|
$tag = str_replace(["<", ">", "<", ">", "/"], "", $tag[0]); |
|
102
|
|
|
|
|
103
|
|
|
$nextElement = $matches[0][$index+1] ?? null; |
|
104
|
|
|
|
|
105
|
|
|
if(!self::contains("/", $element)){ |
|
106
|
|
|
|
|
107
|
|
|
if($nextElement === null){ |
|
108
|
|
|
continue; |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
$nextTag = explode(" ", $nextElement); |
|
112
|
|
|
$nextTag = str_replace(["<", ">", "<", ">", "/"], "", $nextTag[0]); |
|
113
|
|
|
|
|
114
|
|
|
if($nextTag !== $tag){ |
|
115
|
|
|
continue; |
|
116
|
|
|
} |
|
117
|
|
|
} else { |
|
118
|
|
|
// self closing tag |
|
119
|
|
|
$closingTag = explode(" ", $element); |
|
120
|
|
|
$closingTag = str_replace(["<", ">", "<", ">"], "", $closingTag[0]); |
|
121
|
|
|
|
|
122
|
|
|
if(empty($closingTag)){ |
|
123
|
|
|
continue; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
if(!(self::firstChar($closingTag) === "/" or self::lastChar($closingTag) === "/")){ |
|
127
|
|
|
continue; |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
$charMap = self::charMap(); |
|
132
|
|
|
$protectedTag = str_replace(["<", ">", "<", ">"], [$charMap["<"], $charMap[">"], $charMap["<"], $charMap[">"]], $element); |
|
133
|
|
|
$string = str_replace($element, $protectedTag, $string); |
|
134
|
|
|
} |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
return $string; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
/** |
|
141
|
|
|
* @param string $string |
|
142
|
|
|
* @return string |
|
143
|
|
|
*/ |
|
144
|
|
|
public static function unprotectHTMLTags($string) |
|
145
|
|
|
{ |
|
146
|
|
|
$charMap = self::charMap(); |
|
147
|
|
|
$string = str_replace([$charMap["<"], $charMap[">"], $charMap["<"], $charMap[">"]], ["<", ">", "<", ">"], $string); |
|
148
|
|
|
|
|
149
|
|
|
return $string; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
/** |
|
153
|
|
|
* @return array |
|
154
|
|
|
*/ |
|
155
|
|
|
private static function charMap() |
|
156
|
|
|
{ |
|
157
|
|
|
return [ |
|
158
|
|
|
"<" => "ʃʃʃʃ", |
|
159
|
|
|
">" => "¶¶¶¶", |
|
160
|
|
|
"<" => "ɑɑɑɑ", |
|
161
|
|
|
">" => "ʒʒʒʒ", |
|
162
|
|
|
]; |
|
163
|
|
|
} |
|
164
|
|
|
} |
matecat /
whole-text-finder
