解码UTF-8之JS

JS的任何字符串操作之前都将会转义。so会看到如下：

JS无法捕获反斜杠，所以输入限制为输入\xe6

  1 function utf8Decode(str){
  2     var b=[],
  3         arr=str.replace(/\\x/g," ").split(" ");
  4     arr.shift();
  5     
  6     var i=0,
  7         j=0,
  8         length=arr.length;
  9     while(i<length){
 10        arr[i]=parseInt(arr[i],16).toString("2");
 11        if(arr[i].length<=7){
 12            b.push(parseInt(arr[i],2));
 14        }else{
 15            j=0;
 16            while(j<8){
 17                if(arr[i].charAt(j)==='1'){
 18                    j++;
 19                }else{
 20                    break;
 21                }   
 22            }   
 23            
 24            //已经判断出首个字节前连续几个1
 25            //判断出几个字节表示的字符
 26            if(j===2){
 27                b.push(parseInt(arr[i].slice(3)+parseInt(arr[++i],16).toString("2").slice(2),2));
 28            }else if(j===3){
 29                b.push(parseInt(arr[i].slice(4)+parseInt(arr[++i],16).toString("2").slice(2)+parseInt(arr[++i],16).toString("2").slice(2),2));
 30            }else{
 31                b.push(parseInt(arr[i].slice(5)+parseInt(arr[++i],16).toString("2").slice(2)+parseInt(arr[++i],16).toString("2").slice(2)+parseInt(arr[++i],16).toString("2").slice(2),2));
 33            }
               i++;   
 34        }   
 35     }  
 36     
 37     return b;
 38     
 39 }