voku /
portable-utf8
| @@ 1850-1868 (lines=19) @@ | ||
| 1847 | * |
|
| 1848 | * @return string |
|
| 1849 | */ |
|
| 1850 | public static function fix_simple_utf8($str) |
|
| 1851 | { |
|
| 1852 | // init |
|
| 1853 | $str = (string)$str; |
|
| 1854 | ||
| 1855 | if (!isset($str[0])) { |
|
| 1856 | return ''; |
|
| 1857 | } |
|
| 1858 | ||
| 1859 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1860 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1861 | ||
| 1862 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1863 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1864 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = array_values(self::$BROKEN_UTF8_FIX); |
|
| 1865 | } |
|
| 1866 | ||
| 1867 | return str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1868 | } |
|
| 1869 | ||
| 1870 | /** |
|
| 1871 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3692-3710 (lines=19) @@ | ||
| 3689 | * |
|
| 3690 | * @return string |
|
| 3691 | */ |
|
| 3692 | public static function normalize_msword($str) |
|
| 3693 | { |
|
| 3694 | $str = (string)$str; |
|
| 3695 | ||
| 3696 | if (!isset($str[0])) { |
|
| 3697 | return ''; |
|
| 3698 | } |
|
| 3699 | ||
| 3700 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3701 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3702 | ||
| 3703 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3704 | $UTF8_MSWORD_KEYS_CACHE = array_keys(self::$UTF8_MSWORD); |
|
| 3705 | $UTF8_MSWORD_VALUES_CACHE = array_values(self::$UTF8_MSWORD); |
|
| 3706 | } |
|
| 3707 | ||
| 3708 | return \str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3709 | } |
|
| 3710 | ||
| 3711 | /** |
|
| 3712 | * Normalize the whitespace. |
|
| 3713 | * |
|
