| 844,7 → 844,7 |
|---|
| } |
| |
| function init_byte_map(){ |
| global $byte_map; |
| $byte_map = array(); |
| for($x=128;$x<256;++$x){ |
| $byte_map[chr($x)]=utf8_encode(chr($x)); |
| } |
| 880,12 → 880,14 |
|---|
| foreach($cp1252_map as $k=>$v){ |
| $byte_map[$k]=$v; |
| } |
| |
| return $byte_map; |
| } |
| |
| function fix_latin($instr){ |
| |
| $byte_map=array(); |
| init_byte_map(); |
| $byte_map = init_byte_map(); |
| |
| $ascii_char='[\x00-\x7F]'; |
| $cont_byte='[\x80-\xBF]'; |
| $utf8_2='[\xC0-\xDF]'.$cont_byte; |
| 892,6 → 894,7 |
|---|
| $utf8_3='[\xE0-\xEF]'.$cont_byte.'{2}'; |
| $utf8_4='[\xF0-\xF7]'.$cont_byte.'{3}'; |
| $utf8_5='[\xF8-\xFB]'.$cont_byte.'{4}'; |
| |
| $nibble_good_chars = "@^($ascii_char+|$utf8_2|$utf8_3|$utf8_4|$utf8_5)(.*)$@s"; |
| |
| if(mb_check_encoding($instr,'UTF-8'))return $instr; // no need for the rest if it's all valid UTF-8 already |