voku /
portable-utf8
| @@ 1803-1821 (lines=19) @@ | ||
| 1800 | * |
|
| 1801 | * @return string |
|
| 1802 | */ |
|
| 1803 | public static function fix_simple_utf8($str) |
|
| 1804 | { |
|
| 1805 | // init |
|
| 1806 | $str = (string)$str; |
|
| 1807 | ||
| 1808 | if (!isset($str[0])) { |
|
| 1809 | return ''; |
|
| 1810 | } |
|
| 1811 | ||
| 1812 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1813 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1814 | ||
| 1815 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1816 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1817 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = array_values(self::$BROKEN_UTF8_FIX); |
|
| 1818 | } |
|
| 1819 | ||
| 1820 | return str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1821 | } |
|
| 1822 | ||
| 1823 | /** |
|
| 1824 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3444-3462 (lines=19) @@ | ||
| 3441 | * |
|
| 3442 | * @return string |
|
| 3443 | */ |
|
| 3444 | public static function normalize_msword($str) |
|
| 3445 | { |
|
| 3446 | // init |
|
| 3447 | $str = (string)$str; |
|
| 3448 | ||
| 3449 | if (!isset($str[0])) { |
|
| 3450 | return ''; |
|
| 3451 | } |
|
| 3452 | ||
| 3453 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3454 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3455 | ||
| 3456 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3457 | $UTF8_MSWORD_KEYS_CACHE = array_keys(self::$UTF8_MSWORD); |
|
| 3458 | $UTF8_MSWORD_VALUES_CACHE = array_values(self::$UTF8_MSWORD); |
|
| 3459 | } |
|
| 3460 | ||
| 3461 | return str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3462 | } |
|
| 3463 | ||
| 3464 | /** |
|
| 3465 | * Normalize the whitespace. |
|
