voku /
portable-utf8
| @@ 1845-1863 (lines=19) @@ | ||
| 1842 | * |
|
| 1843 | * @return string |
|
| 1844 | */ |
|
| 1845 | public static function fix_simple_utf8($str) |
|
| 1846 | { |
|
| 1847 | // init |
|
| 1848 | $str = (string)$str; |
|
| 1849 | ||
| 1850 | if (!isset($str[0])) { |
|
| 1851 | return ''; |
|
| 1852 | } |
|
| 1853 | ||
| 1854 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1855 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1856 | ||
| 1857 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1858 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1859 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = array_values(self::$BROKEN_UTF8_FIX); |
|
| 1860 | } |
|
| 1861 | ||
| 1862 | return str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1863 | } |
|
| 1864 | ||
| 1865 | /** |
|
| 1866 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3677-3695 (lines=19) @@ | ||
| 3674 | * |
|
| 3675 | * @return string |
|
| 3676 | */ |
|
| 3677 | public static function normalize_msword($str) |
|
| 3678 | { |
|
| 3679 | $str = (string)$str; |
|
| 3680 | ||
| 3681 | if (!isset($str[0])) { |
|
| 3682 | return ''; |
|
| 3683 | } |
|
| 3684 | ||
| 3685 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3686 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3687 | ||
| 3688 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3689 | $UTF8_MSWORD_KEYS_CACHE = array_keys(self::$UTF8_MSWORD); |
|
| 3690 | $UTF8_MSWORD_VALUES_CACHE = array_values(self::$UTF8_MSWORD); |
|
| 3691 | } |
|
| 3692 | ||
| 3693 | return str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3694 | } |
|
| 3695 | ||
| 3696 | /** |
|
| 3697 | * Normalize the whitespace. |
|
| 3698 | * |
|
