| 1 |
<?php |
| 2 |
|
| 3 |
class Red_Url_Encode { |
| 4 |
/** |
| 5 |
* URL |
| 6 |
* |
| 7 |
* @var string |
| 8 |
*/ |
| 9 |
private $url; |
| 10 |
|
| 11 |
/** |
| 12 |
* Is regex? |
| 13 |
* |
| 14 |
* @var boolean |
| 15 |
*/ |
| 16 |
private $is_regex; |
| 17 |
|
| 18 |
/** |
| 19 |
* Constructor |
| 20 |
* |
| 21 |
* @param string $url URL. |
| 22 |
* @param boolean $is_regex Is Regex. |
| 23 |
*/ |
| 24 |
public function __construct( $url, $is_regex = false ) { |
| 25 |
// Remove any newlines |
| 26 |
$url = preg_replace( "/[\r\n\t].*?$/s", '', $url ); |
| 27 |
|
| 28 |
// Remove invalid characters |
| 29 |
$url = preg_replace( '/[^\PC\s]/u', '', $url ); |
| 30 |
|
| 31 |
// Make sure spaces are quoted |
| 32 |
$url = str_replace( ' ', '%20', $url ); |
| 33 |
$url = str_replace( '%24', '$', $url ); |
| 34 |
|
| 35 |
$this->url = $url; |
| 36 |
$this->is_regex = $is_regex; |
| 37 |
} |
| 38 |
|
| 39 |
/** |
| 40 |
* URL encode some things, but other things can be passed through |
| 41 |
* |
| 42 |
* @return string |
| 43 |
*/ |
| 44 |
public function get_as_target() { |
| 45 |
$allowed = [ |
| 46 |
'%2F' => '/', |
| 47 |
'%3F' => '?', |
| 48 |
'%3A' => ':', |
| 49 |
'%3D' => '=', |
| 50 |
'%26' => '&', |
| 51 |
'%25' => '%', |
| 52 |
'+' => '%20', |
| 53 |
'%24' => '$', |
| 54 |
'%23' => '#', |
| 55 |
]; |
| 56 |
|
| 57 |
$url = rawurlencode( $this->url ); |
| 58 |
$url = $this->replace_encoding( $url, $allowed ); |
| 59 |
|
| 60 |
return $this->encode_regex( $url ); |
| 61 |
} |
| 62 |
|
| 63 |
/** |
| 64 |
* Encode a URL |
| 65 |
* |
| 66 |
* @return string |
| 67 |
*/ |
| 68 |
public function get_as_source() { |
| 69 |
$allowed = [ |
| 70 |
'%2F' => '/', |
| 71 |
'%3F' => '?', |
| 72 |
'+' => '%20', |
| 73 |
'.' => '\\.', |
| 74 |
]; |
| 75 |
|
| 76 |
$url = $this->replace_encoding( rawurlencode( $this->url ), $allowed ); |
| 77 |
return $this->encode_regex( $url ); |
| 78 |
} |
| 79 |
|
| 80 |
|
| 81 |
/** |
| 82 |
* Replace encoded characters in a URL |
| 83 |
* |
| 84 |
* @param string $str Source string. |
| 85 |
* @param array $allowed Allowed encodings. |
| 86 |
* @return string |
| 87 |
*/ |
| 88 |
private function replace_encoding( $str, $allowed ) { |
| 89 |
foreach ( $allowed as $before => $after ) { |
| 90 |
$str = str_replace( $before, $after, $str ); |
| 91 |
} |
| 92 |
|
| 93 |
return $str; |
| 94 |
} |
| 95 |
|
| 96 |
/** |
| 97 |
* Encode a regex URL |
| 98 |
* |
| 99 |
* @param string $url URL. |
| 100 |
* @return string |
| 101 |
*/ |
| 102 |
private function encode_regex( $url ) { |
| 103 |
if ( $this->is_regex ) { |
| 104 |
// No leading slash |
| 105 |
$url = ltrim( $url, '/' ); |
| 106 |
|
| 107 |
// If pattern has a ^ at the start then ensure we don't have a slash immediatley after |
| 108 |
$url = preg_replace( '@^\^/@', '^', $url ); |
| 109 |
|
| 110 |
$url = $this->replace_encoding( $url, [ |
| 111 |
'%2A' => '*', |
| 112 |
'%3F' => '?', |
| 113 |
'%28' => '(', |
| 114 |
'%29' => ')', |
| 115 |
'%5B' => '[', |
| 116 |
'%5C' => ']', |
| 117 |
'%24' => '$', |
| 118 |
'%2B' => '+', |
| 119 |
'%7C' => '|', |
| 120 |
'\\.' => '.', |
| 121 |
] ); |
| 122 |
} |
| 123 |
|
| 124 |
return $url; |
| 125 |
} |
| 126 |
} |
