voku /
portable-utf8
| @@ 1531-1551 (lines=21) @@ | ||
| 1528 | * |
|
| 1529 | * @return string |
|
| 1530 | */ |
|
| 1531 | public static function fix_simple_utf8(string $str): string |
|
| 1532 | { |
|
| 1533 | if (!isset($str[0])) { |
|
| 1534 | return ''; |
|
| 1535 | } |
|
| 1536 | ||
| 1537 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1538 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1539 | ||
| 1540 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1541 | ||
| 1542 | if (self::$BROKEN_UTF8_FIX === null) { |
|
| 1543 | self::$BROKEN_UTF8_FIX = self::getData('utf8_fix'); |
|
| 1544 | } |
|
| 1545 | ||
| 1546 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = \array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1547 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = \array_values(self::$BROKEN_UTF8_FIX); |
|
| 1548 | } |
|
| 1549 | ||
| 1550 | return \str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1551 | } |
|
| 1552 | ||
| 1553 | /** |
|
| 1554 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3673-3694 (lines=22) @@ | ||
| 3670 | * |
|
| 3671 | * @return string |
|
| 3672 | */ |
|
| 3673 | public static function normalize_msword(string $str): string |
|
| 3674 | { |
|
| 3675 | if (!isset($str[0])) { |
|
| 3676 | return ''; |
|
| 3677 | } |
|
| 3678 | ||
| 3679 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3680 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3681 | ||
| 3682 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3683 | ||
| 3684 | if (self::$UTF8_MSWORD === null) { |
|
| 3685 | self::$UTF8_MSWORD = self::getData('utf8_msword'); |
|
| 3686 | } |
|
| 3687 | ||
| 3688 | $UTF8_MSWORD_KEYS_CACHE = \array_keys(self::$UTF8_MSWORD); |
|
| 3689 | $UTF8_MSWORD_VALUES_CACHE = \array_values(self::$UTF8_MSWORD); |
|
| 3690 | } |
|
| 3691 | ||
| 3692 | return \str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3693 | } |
|
| 3694 | ||
| 3695 | /** |
|
| 3696 | * Normalize the whitespace. |
|
| 3697 | * |
|
