|
@@ 1165-1185 (lines=21) @@
|
| 1162 |
|
* |
| 1163 |
|
* @return string |
| 1164 |
|
*/ |
| 1165 |
|
public static function fix_simple_utf8(string $str): string |
| 1166 |
|
{ |
| 1167 |
|
if (!isset($str[0])) { |
| 1168 |
|
return ''; |
| 1169 |
|
} |
| 1170 |
|
|
| 1171 |
|
static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null; |
| 1172 |
|
static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null; |
| 1173 |
|
|
| 1174 |
|
if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) { |
| 1175 |
|
|
| 1176 |
|
if (self::$BROKEN_UTF8_FIX === null) { |
| 1177 |
|
self::$BROKEN_UTF8_FIX = self::getData('utf8_fix'); |
| 1178 |
|
} |
| 1179 |
|
|
| 1180 |
|
$BROKEN_UTF8_TO_UTF8_KEYS_CACHE = \array_keys(self::$BROKEN_UTF8_FIX); |
| 1181 |
|
$BROKEN_UTF8_TO_UTF8_VALUES_CACHE = \array_values(self::$BROKEN_UTF8_FIX); |
| 1182 |
|
} |
| 1183 |
|
|
| 1184 |
|
return \str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str); |
| 1185 |
|
} |
| 1186 |
|
|
| 1187 |
|
/** |
| 1188 |
|
* Fix a double (or multiple) encoded UTF8 string. |
|
@@ 2950-2971 (lines=22) @@
|
| 2947 |
|
* |
| 2948 |
|
* @return string |
| 2949 |
|
*/ |
| 2950 |
|
public static function normalize_msword(string $str): string |
| 2951 |
|
{ |
| 2952 |
|
if (!isset($str[0])) { |
| 2953 |
|
return ''; |
| 2954 |
|
} |
| 2955 |
|
|
| 2956 |
|
static $UTF8_MSWORD_KEYS_CACHE = null; |
| 2957 |
|
static $UTF8_MSWORD_VALUES_CACHE = null; |
| 2958 |
|
|
| 2959 |
|
if ($UTF8_MSWORD_KEYS_CACHE === null) { |
| 2960 |
|
|
| 2961 |
|
|
| 2962 |
|
if (self::$UTF8_MSWORD === null) { |
| 2963 |
|
self::$UTF8_MSWORD = self::getData('utf8_msword'); |
| 2964 |
|
} |
| 2965 |
|
|
| 2966 |
|
$UTF8_MSWORD_KEYS_CACHE = \array_keys(self::$UTF8_MSWORD); |
| 2967 |
|
$UTF8_MSWORD_VALUES_CACHE = \array_values(self::$UTF8_MSWORD); |
| 2968 |
|
} |
| 2969 |
|
|
| 2970 |
|
return \str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str); |
| 2971 |
|
} |
| 2972 |
|
|
| 2973 |
|
/** |
| 2974 |
|
* Normalize the whitespace. |
voku /
portable-utf8
