voku /
portable-utf8
| @@ 1828-1846 (lines=19) @@ | ||
| 1825 | * |
|
| 1826 | * @return string |
|
| 1827 | */ |
|
| 1828 | public static function fix_simple_utf8($str) |
|
| 1829 | { |
|
| 1830 | // init |
|
| 1831 | $str = (string)$str; |
|
| 1832 | ||
| 1833 | if (!isset($str[0])) { |
|
| 1834 | return ''; |
|
| 1835 | } |
|
| 1836 | ||
| 1837 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1838 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1839 | ||
| 1840 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1841 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1842 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = array_values(self::$BROKEN_UTF8_FIX); |
|
| 1843 | } |
|
| 1844 | ||
| 1845 | return str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1846 | } |
|
| 1847 | ||
| 1848 | /** |
|
| 1849 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3594-3612 (lines=19) @@ | ||
| 3591 | * |
|
| 3592 | * @return string |
|
| 3593 | */ |
|
| 3594 | public static function normalize_msword($str) |
|
| 3595 | { |
|
| 3596 | $str = (string)$str; |
|
| 3597 | ||
| 3598 | if (!isset($str[0])) { |
|
| 3599 | return ''; |
|
| 3600 | } |
|
| 3601 | ||
| 3602 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3603 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3604 | ||
| 3605 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3606 | $UTF8_MSWORD_KEYS_CACHE = array_keys(self::$UTF8_MSWORD); |
|
| 3607 | $UTF8_MSWORD_VALUES_CACHE = array_values(self::$UTF8_MSWORD); |
|
| 3608 | } |
|
| 3609 | ||
| 3610 | return str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3611 | } |
|
| 3612 | ||
| 3613 | /** |
|
| 3614 | * Normalize the whitespace. |
|
| 3615 | * |
|
