voku /
portable-utf8
| @@ 1823-1841 (lines=19) @@ | ||
| 1820 | * |
|
| 1821 | * @return string |
|
| 1822 | */ |
|
| 1823 | public static function fix_simple_utf8($str) |
|
| 1824 | { |
|
| 1825 | // init |
|
| 1826 | $str = (string)$str; |
|
| 1827 | ||
| 1828 | if (!isset($str[0])) { |
|
| 1829 | return ''; |
|
| 1830 | } |
|
| 1831 | ||
| 1832 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1833 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1834 | ||
| 1835 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1836 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1837 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = array_values(self::$BROKEN_UTF8_FIX); |
|
| 1838 | } |
|
| 1839 | ||
| 1840 | return str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1841 | } |
|
| 1842 | ||
| 1843 | /** |
|
| 1844 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3499-3517 (lines=19) @@ | ||
| 3496 | * |
|
| 3497 | * @return string |
|
| 3498 | */ |
|
| 3499 | public static function normalize_msword($str) |
|
| 3500 | { |
|
| 3501 | // init |
|
| 3502 | $str = (string)$str; |
|
| 3503 | ||
| 3504 | if (!isset($str[0])) { |
|
| 3505 | return ''; |
|
| 3506 | } |
|
| 3507 | ||
| 3508 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3509 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3510 | ||
| 3511 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3512 | $UTF8_MSWORD_KEYS_CACHE = array_keys(self::$UTF8_MSWORD); |
|
| 3513 | $UTF8_MSWORD_VALUES_CACHE = array_values(self::$UTF8_MSWORD); |
|
| 3514 | } |
|
| 3515 | ||
| 3516 | return str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3517 | } |
|
| 3518 | ||
| 3519 | /** |
|
| 3520 | * Normalize the whitespace. |
|
