voku /
portable-utf8
| @@ 1816-1834 (lines=19) @@ | ||
| 1813 | * |
|
| 1814 | * @return string |
|
| 1815 | */ |
|
| 1816 | public static function fix_simple_utf8($str) |
|
| 1817 | { |
|
| 1818 | // init |
|
| 1819 | $str = (string)$str; |
|
| 1820 | ||
| 1821 | if (!isset($str[0])) { |
|
| 1822 | return ''; |
|
| 1823 | } |
|
| 1824 | ||
| 1825 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1826 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1827 | ||
| 1828 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1829 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1830 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = array_values(self::$BROKEN_UTF8_FIX); |
|
| 1831 | } |
|
| 1832 | ||
| 1833 | return str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1834 | } |
|
| 1835 | ||
| 1836 | /** |
|
| 1837 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3492-3510 (lines=19) @@ | ||
| 3489 | * |
|
| 3490 | * @return string |
|
| 3491 | */ |
|
| 3492 | public static function normalize_msword($str) |
|
| 3493 | { |
|
| 3494 | // init |
|
| 3495 | $str = (string)$str; |
|
| 3496 | ||
| 3497 | if (!isset($str[0])) { |
|
| 3498 | return ''; |
|
| 3499 | } |
|
| 3500 | ||
| 3501 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3502 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3503 | ||
| 3504 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3505 | $UTF8_MSWORD_KEYS_CACHE = array_keys(self::$UTF8_MSWORD); |
|
| 3506 | $UTF8_MSWORD_VALUES_CACHE = array_values(self::$UTF8_MSWORD); |
|
| 3507 | } |
|
| 3508 | ||
| 3509 | return str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3510 | } |
|
| 3511 | ||
| 3512 | /** |
|
| 3513 | * Normalize the whitespace. |
|
