[原创代码] Perl版汉字批量转UTF-8编码[纯编码]

PakTC

保存脚本的时候须为utf-8格式。

#!/usr/bin/perl
#Code by Paktc
#2010-09

$lv_[1]=00000000;
$lv_[2]=11000000;
$lv_[3]=11100000;
$lv_[4]=11110000;
$lv_[5]=11111000;
$lv_[6]=11111100;

use utf8;

$a="千与千寻";
$L=length($a)-1;
for $i (0..$L) {
    $b=$i;
    printf "%s %s\n",&code2utf(ord(substr($a,$i,2)));
}
print "END!";
<STDIN>;


sub code2utf {
    my ($code,$bin,$level,$hex,$tmp,$result);
    $code=$_[0];
    $bin=sprintf "%b",$code;

    $level=1;
    if (length($bin)>7) { 
        while (length($bin)>=6) {
            $level+=1;
            $bin=~s/.{6}$//;
            $hex=sprintf "%X",&bin2int(10 . $&);
            $result=$hex ."-". $result;
        }
    }

    $tmp=substr($lv_[$level],0,8-length($bin));
    $hex=sprintf "%X",&bin2int($tmp . $bin);
    $result=$hex ."-". $result;
    $result=~s/-$//;  
    $result;   #result - string
}

sub bin2int {
    my ($a,$str,$all,$act);
    $a=$_[0];
    $str="";
    $all=0;
    $act=0;
    do { 
        $b=($a % 10) * (2**$act);  #** as ^ of the calc
        $all+=$b;
        $act+=1;
        $a=int ($a/10);
    } until ($a==0);
    int ($all);
}
复制代码

输出
E5-8D-83
E4-B8-8E
E5-8D-83
E5-AF-BB
END!