php中gb2312跟unicode码的转换

wss1051

php中gb2312跟unicode码的转换
2008年01月06日 星期日 下午 09:55

<?php
// utf8 ---> unicode
function utf8_unicode($c) {
switch(strlen($c)) {
    case 1:
      return ord($c);
    case 2:
      $n = (ord($c[0]) & 0x3f) << 6;
      $n += ord($c[1]) & 0x3f;
      return $n;
    case 3:
      $n = (ord($c[0]) & 0x1f) << 12;
      $n += (ord($c[1]) & 0x3f) << 6;
      $n += ord($c[2]) & 0x3f;
      return $n;
    case 4:
      $n = (ord($c[0]) & 0x0f) << 18;
      $n += (ord($c[1]) & 0x3f) << 12;
      $n += (ord($c[2]) & 0x3f) << 6;
      $n += ord($c[3]) & 0x3f;
      return $n;
}
}

//gb2312转换为unicode
function gb2un($g)//传入gb2312字符串返回unicode码
{
preg_match_all("/[\x80-\xff]?./",$g,$ar);
$str = "";
foreach($ar[0] as $v)
{
   $str = $str."&#".utf8_unicode(iconv("gb2312","utf-8",$v)).";";
}
return $str;
}


//unicode ---> gb2312
function unescape($str) {
$str = rawurldecode($str);
preg_match_all("/(?:%u.{4})|&#x.{4};|&#\d+;|.+/u",$str,$r);
$ar = $r[0];
//print_r($ar);
foreach($ar as $k=>$v) {
    if(substr($v,0,2) == "%u")
      $ar[$k] = iconv("ucs-2","gb2312",pack("h4",substr($v,-4)));
    elseif(substr($v,0,3) == "&#x")
      $ar[$k] = iconv("ucs-2","gb2312",pack("h4",substr($v,3,-1)));
    elseif(substr($v,0,2) == "&#") {
//echo substr($v,2,-1)."<br>";
      $ar[$k] = iconv("ucs-2","gb2312",pack("n",substr($v,2,-1)));
    }
}
return join("",$ar);
}

//unicode转换为gb2312
function un2gb($y)//传入unicode码返回gb2312
{
return unescape($y);
}
?>