|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* User: jensk |
|
4
|
|
|
* Date: 21-2-2017 |
|
5
|
|
|
* Time: 10:44 |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
namespace CloudControl\Cms\search; |
|
9
|
|
|
|
|
10
|
|
|
class CharacterFilter |
|
11
|
|
|
{ |
|
12
|
|
|
protected $originalString; |
|
13
|
|
|
/** |
|
14
|
|
|
* @var string |
|
15
|
|
|
*/ |
|
16
|
|
|
protected $filteredString = ''; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* CharacterFilter constructor. |
|
20
|
|
|
* |
|
21
|
|
|
* @param string $string |
|
22
|
|
|
*/ |
|
23
|
|
|
public function __construct($string) |
|
24
|
|
|
{ |
|
25
|
|
|
$this->originalString = $string; |
|
26
|
|
|
$string = $this->convertToUTF8($string); |
|
27
|
|
|
$string = mb_strtolower($string); |
|
28
|
|
|
$string = $this->filterSpecialCharacters($string); |
|
29
|
|
|
$this->filteredString = $string; |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* Returns the filtered string |
|
34
|
|
|
* @return string |
|
35
|
|
|
*/ |
|
36
|
|
|
public function __toString() |
|
37
|
|
|
{ |
|
38
|
|
|
return (string)$this->filteredString; |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Filter out all special characters, like punctuation and characters with accents |
|
43
|
|
|
* |
|
44
|
|
|
* @param $string |
|
45
|
|
|
* |
|
46
|
|
|
* @return mixed|string |
|
47
|
|
|
*/ |
|
48
|
|
|
private function filterSpecialCharacters($string) |
|
49
|
|
|
{ |
|
50
|
|
|
$string = str_replace('<', ' <', |
|
51
|
|
|
$string); // This is need, otherwise this: <h1>something</h1><h2>something</h2> will result in somethingsomething |
|
52
|
|
|
$string = strip_tags($string); |
|
53
|
|
|
$string = trim($string); |
|
54
|
|
|
$string = iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $string); // Remove special alphanumeric characters |
|
55
|
|
|
$string = str_replace(array('+', '=', '!', ',', '.', ';', ':', '?'), ' ', |
|
56
|
|
|
$string); // Replace sentence breaking charaters with spaces |
|
57
|
|
|
$string = preg_replace("/[\r\n]+/", " ", $string); // Replace multiple newlines with a single space. |
|
58
|
|
|
$string = preg_replace("/[\t]+/", " ", $string); // Replace multiple tabs with a single space. |
|
59
|
|
|
$string = preg_replace("/[^a-zA-Z0-9 ]/", '', |
|
60
|
|
|
$string); // Filter out everything that is not alphanumeric or a space |
|
61
|
|
|
$string = preg_replace('!\s+!', ' ', $string); // Replace multiple spaces with a single space |
|
62
|
|
|
return $string; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* Convert the string to UTF-8 encoding |
|
67
|
|
|
* @param $string |
|
68
|
|
|
* |
|
69
|
|
|
* @return string |
|
70
|
|
|
*/ |
|
71
|
|
|
private function convertToUTF8($string) |
|
72
|
|
|
{ |
|
73
|
|
|
$encoding = mb_detect_encoding($string, mb_detect_order(), false); |
|
74
|
|
|
|
|
75
|
|
|
if ($encoding == "UTF-8") { |
|
76
|
|
|
$string = mb_convert_encoding($string, 'UTF-8', 'UTF-8'); |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
$out = iconv(mb_detect_encoding($string, mb_detect_order(), false), "UTF-8//IGNORE", $string); |
|
80
|
|
|
return $out; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* @return mixed|string |
|
85
|
|
|
*/ |
|
86
|
|
|
public function getFilteredString() |
|
87
|
|
|
{ |
|
88
|
|
|
return $this->filteredString; |
|
89
|
|
|
} |
|
90
|
|
|
} |
jenskooij /
cloudcontrol
