voku /
portable-utf8
| @@ 1848-1866 (lines=19) @@ | ||
| 1845 | * |
|
| 1846 | * @return string |
|
| 1847 | */ |
|
| 1848 | public static function fix_simple_utf8($str) |
|
| 1849 | { |
|
| 1850 | // init |
|
| 1851 | $str = (string)$str; |
|
| 1852 | ||
| 1853 | if (!isset($str[0])) { |
|
| 1854 | return ''; |
|
| 1855 | } |
|
| 1856 | ||
| 1857 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1858 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1859 | ||
| 1860 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1861 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1862 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = array_values(self::$BROKEN_UTF8_FIX); |
|
| 1863 | } |
|
| 1864 | ||
| 1865 | return str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1866 | } |
|
| 1867 | ||
| 1868 | /** |
|
| 1869 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3624-3642 (lines=19) @@ | ||
| 3621 | * |
|
| 3622 | * @return string |
|
| 3623 | */ |
|
| 3624 | public static function normalize_msword($str) |
|
| 3625 | { |
|
| 3626 | $str = (string)$str; |
|
| 3627 | ||
| 3628 | if (!isset($str[0])) { |
|
| 3629 | return ''; |
|
| 3630 | } |
|
| 3631 | ||
| 3632 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3633 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3634 | ||
| 3635 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3636 | $UTF8_MSWORD_KEYS_CACHE = array_keys(self::$UTF8_MSWORD); |
|
| 3637 | $UTF8_MSWORD_VALUES_CACHE = array_values(self::$UTF8_MSWORD); |
|
| 3638 | } |
|
| 3639 | ||
| 3640 | return str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3641 | } |
|
| 3642 | ||
| 3643 | /** |
|
| 3644 | * Normalize the whitespace. |
|
| 3645 | * |
|
