voku /
portable-utf8
| @@ 1787-1805 (lines=19) @@ | ||
| 1784 | * |
|
| 1785 | * @return string |
|
| 1786 | */ |
|
| 1787 | public static function fix_simple_utf8($str) |
|
| 1788 | { |
|
| 1789 | // init |
|
| 1790 | $str = (string)$str; |
|
| 1791 | ||
| 1792 | if (!isset($str[0])) { |
|
| 1793 | return ''; |
|
| 1794 | } |
|
| 1795 | ||
| 1796 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1797 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1798 | ||
| 1799 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1800 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1801 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = array_values(self::$BROKEN_UTF8_FIX); |
|
| 1802 | } |
|
| 1803 | ||
| 1804 | return str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1805 | } |
|
| 1806 | ||
| 1807 | /** |
|
| 1808 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3428-3446 (lines=19) @@ | ||
| 3425 | * |
|
| 3426 | * @return string |
|
| 3427 | */ |
|
| 3428 | public static function normalize_msword($str) |
|
| 3429 | { |
|
| 3430 | // init |
|
| 3431 | $str = (string)$str; |
|
| 3432 | ||
| 3433 | if (!isset($str[0])) { |
|
| 3434 | return ''; |
|
| 3435 | } |
|
| 3436 | ||
| 3437 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3438 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3439 | ||
| 3440 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3441 | $UTF8_MSWORD_KEYS_CACHE = array_keys(self::$UTF8_MSWORD); |
|
| 3442 | $UTF8_MSWORD_VALUES_CACHE = array_values(self::$UTF8_MSWORD); |
|
| 3443 | } |
|
| 3444 | ||
| 3445 | return str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3446 | } |
|
| 3447 | ||
| 3448 | /** |
|
| 3449 | * Normalize the whitespace. |
|
