voku /
portable-utf8
| @@ 1226-1246 (lines=21) @@ | ||
| 1223 | * |
|
| 1224 | * @return string |
|
| 1225 | */ |
|
| 1226 | public static function fix_simple_utf8(string $str): string |
|
| 1227 | { |
|
| 1228 | if (!isset($str[0])) { |
|
| 1229 | return ''; |
|
| 1230 | } |
|
| 1231 | ||
| 1232 | static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
|
| 1233 | static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
|
| 1234 | ||
| 1235 | if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
|
| 1236 | ||
| 1237 | if (self::$BROKEN_UTF8_FIX === null) { |
|
| 1238 | self::$BROKEN_UTF8_FIX = self::getData('utf8_fix'); |
|
| 1239 | } |
|
| 1240 | ||
| 1241 | $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = \array_keys(self::$BROKEN_UTF8_FIX); |
|
| 1242 | $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = \array_values(self::$BROKEN_UTF8_FIX); |
|
| 1243 | } |
|
| 1244 | ||
| 1245 | return \str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
|
| 1246 | } |
|
| 1247 | ||
| 1248 | /** |
|
| 1249 | * Fix a double (or multiple) encoded UTF8 string. |
|
| @@ 3044-3065 (lines=22) @@ | ||
| 3041 | * |
|
| 3042 | * @return string |
|
| 3043 | */ |
|
| 3044 | public static function normalize_msword(string $str): string |
|
| 3045 | { |
|
| 3046 | if (!isset($str[0])) { |
|
| 3047 | return ''; |
|
| 3048 | } |
|
| 3049 | ||
| 3050 | static $UTF8_MSWORD_KEYS_CACHE = null; |
|
| 3051 | static $UTF8_MSWORD_VALUES_CACHE = null; |
|
| 3052 | ||
| 3053 | if ($UTF8_MSWORD_KEYS_CACHE === null) { |
|
| 3054 | ||
| 3055 | if (self::$UTF8_MSWORD === null) { |
|
| 3056 | self::$UTF8_MSWORD = self::getData('utf8_msword'); |
|
| 3057 | } |
|
| 3058 | ||
| 3059 | $UTF8_MSWORD_KEYS_CACHE = \array_keys(self::$UTF8_MSWORD); |
|
| 3060 | $UTF8_MSWORD_VALUES_CACHE = \array_values(self::$UTF8_MSWORD); |
|
| 3061 | } |
|
| 3062 | ||
| 3063 | return \str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
|
| 3064 | } |
|
| 3065 | ||
| 3066 | /** |
|
| 3067 | * Normalize the whitespace. |
|
| 3068 | * |
|
