functionis_utf8($string){

returnpreg_match('%^(?:

[x09x0Ax0Dx20-x7E]#ASCII

|[xC2-xDF][x80-xBF]#non-overlong2-byte

|xE0[xA0-xBF][x80-xBF]#excludingoverlongs

|[xE1-xECxEExEF][x80-xBF]{2}#straight3-byte

|xED[x80-x9F][x80-xBF]#excludingsurrogates

|xF0[x90-xBF][x80-xBF]{2}#planes1-3

|[xF1-xF3][x80-xBF]{3}#planes4-15

|xF4[x80-x8F][x80-xBF]{2}#plane16

)*$%xs',$string);

}

准确率基本和mb_detect_encoding一样，要对一起对，要错一起错。

编码检测不可能100%准确，这个东西已经可以基本满足要求了。