| 1 |
<?php |
| 2 |
|
| 3 |
|
| 4 |
if (!defined('ABSPATH')) { |
| 5 |
exit; |
| 6 |
} |
| 7 |
|
| 8 |
/** |
| 9 |
* Native string-backed implementation of ABJ_404_Solution_MbStringAdapter. |
| 10 |
* |
| 11 |
* Use when the PHP mbstring extension is unavailable. Regex primitives |
| 12 |
* live on ABJ_404_Solution_RegexHelperPreg (sibling extraction, task |
| 13 |
* i826). |
| 14 |
*/ |
| 15 |
class ABJ_404_Solution_MbStringAdapterPreg extends ABJ_404_Solution_MbStringAdapter { |
| 16 |
|
| 17 |
/** @var self|null */ |
| 18 |
private static $instance = null; |
| 19 |
/** |
| 20 |
* Test seam: install or clear the cached singleton instance without |
| 21 |
* private-field reflection. Pass null to reset between tests; pass a |
| 22 |
* configured instance (or double) to install it (M105 singleton-reset seam). |
| 23 |
* |
| 24 |
* @param self|null $instance |
| 25 |
* @return void |
| 26 |
*/ |
| 27 |
public static function setInstance($instance) { |
| 28 |
self::$instance = $instance; |
| 29 |
} |
| 30 |
|
| 31 |
|
| 32 |
public static function getInstance(): self { |
| 33 |
if (self::$instance === null) { |
| 34 |
self::$instance = new self(); |
| 35 |
} |
| 36 |
return self::$instance; |
| 37 |
} |
| 38 |
|
| 39 |
public function ord(string $char): int { |
| 40 |
return ord($char); |
| 41 |
} |
| 42 |
|
| 43 |
public function strtolower(?string $string): string { |
| 44 |
if ($string === null) { |
| 45 |
return ''; |
| 46 |
} |
| 47 |
return strtolower($string); |
| 48 |
} |
| 49 |
|
| 50 |
public function strlen(string $string): int { |
| 51 |
return strlen($string); |
| 52 |
} |
| 53 |
|
| 54 |
/** @return int|false */ |
| 55 |
public function strpos(string $haystack, string $needle, int $offset = 0) { |
| 56 |
if ($offset === 0) { |
| 57 |
return strpos($haystack, $needle); |
| 58 |
} |
| 59 |
return strpos($haystack, $needle, $offset); |
| 60 |
} |
| 61 |
|
| 62 |
public function substr(?string $str, int $start, ?int $length = null): string { |
| 63 |
if ($str === null) { |
| 64 |
return ''; |
| 65 |
} |
| 66 |
if ($length === null) { |
| 67 |
return substr($str, $start); |
| 68 |
} |
| 69 |
return substr($str, $start, $length); |
| 70 |
} |
| 71 |
|
| 72 |
public function sanitizeInvalidUTF8(?string $string): string { |
| 73 |
if ($string === null || $string === '') { |
| 74 |
return ''; |
| 75 |
} |
| 76 |
|
| 77 |
if (function_exists('iconv')) { |
| 78 |
// iconv may emit warnings on malformed input; suppress and fall |
| 79 |
// back if it returns false. |
| 80 |
// allow-silent-catch: iconv //IGNORE intentionally swallows |
| 81 |
// malformed-byte warnings; we fall through to preg below. |
| 82 |
$sanitized = @iconv('UTF-8', 'UTF-8//IGNORE', $string); |
| 83 |
if ($sanitized !== false) { |
| 84 |
$sanitized = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F]/u', '', $sanitized) ?? $sanitized; |
| 85 |
return $sanitized; |
| 86 |
} |
| 87 |
} |
| 88 |
|
| 89 |
// Validate UTF-8 with the //u modifier; preg_replace returns null on |
| 90 |
// invalid input, which signals fall-through. |
| 91 |
$sanitized = @preg_replace('//u', '', $string); |
| 92 |
|
| 93 |
if ($sanitized === null) { |
| 94 |
// Strip invalid UTF-8 lead bytes and stranded continuation bytes. |
| 95 |
$sanitized = preg_replace('/[\xC0\xC1\xF5-\xFF][\x80-\xBF]*/', '', $string) ?? $string; |
| 96 |
$sanitized = preg_replace('/[\x80-\xBF]+/', '', $sanitized) ?? $sanitized; |
| 97 |
|
| 98 |
if (@preg_match('//u', $sanitized) === false) { |
| 99 |
// Last-resort ASCII-only filter. |
| 100 |
$sanitized = preg_replace('/[^\x09\x0A\x0D\x20-\x7E]/', '', $string) ?? ''; |
| 101 |
} |
| 102 |
} |
| 103 |
|
| 104 |
$sanitized = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F]/', '', $sanitized) ?? $sanitized; |
| 105 |
return $sanitized; |
| 106 |
} |
| 107 |
} |
| 108 |
|
