最近研究了jsp中作HTTP认证的问题，它的工作方式如下：
1、server发送一个要求认证代码401和一个头信息WWW-authenticate，激发browser弹出一个认证窗口
2、server取得browser送来的认证头"Authorization",它是加密的了，要用Base64方法解密，取得明文的用户名和密码
3、检查用户名和密码，根据结果传送不同的页面
以下是jsp的片断，你也可以把它做成include文件。和Base64的加解密的class源码。
如有兴趣可与我联系：unixboy@yeah.net
<jsp:useBean id="base64"scope="page"class="Base64"/>
<%
if(request.getHeader("Authorization")==null){
response.setStatus(401);
response.setHeader("WWW-authenticate","Basic realm=\"unixboy.com\"");
}else{
String encoded=(request.getHeader("Authorization"));
String tmp=encoded.substring(6);
String up=Base64.decode(tmp);
String user="";
String password="";
if(up!=null){
user=up.substring(0,up.indexOf(":"));
password=up.substring(up.indexOf(":")+1);
}
if(user.equals("unixboy")&&password.equals("123456")){
//认证成功
}else{
//认证失败
}
}
%>
//消息加解密class
public class Base64
{
/** decode a Base 64 encoded String.
*<p><h4>String to byte conversion</h4>
* This method uses a naive String to byte interpretation, it simply gets each
* char of the String and calls it a byte.</p>
*<p>Since we should be dealing with Base64 encoded Strings that is a reasonable
* assumption.</p>
*<p><h4>End of data</h4>
* We don''t try to stop the converion when we find the"="end of data padding char.
* We simply add zero bytes to the unencode buffer.</p>
*/
public static String decode(String encoded)
{
StringBuffer sb=new StringBuffer();
int maxturns;
//work out how long to loop for.
if(encoded.length()%3==0)
maxturns=encoded.length();
else
maxturns=encoded.length()+(3-(encoded.length()%3));
//tells us whether to include the char in the unencode
boolean skip;
//the unencode buffer
byte[] unenc=new byte[4];
byte b;
for(int i=0,j=0;i<maxturns;i++)
{
skip=false;
//get the byte to convert or 0
if(i<encoded.length())
b=(byte)encoded.charAt(i);
else
b=0;
//test and convert first capital letters, lowercase, digits then ''+'' and ''/''
if(b>=65&&b<91)
unenc[j]=(byte)(b-65);
else if(b>=97&&b<123)
unenc[j]=(byte)(b-71);
else if(b>=48&&b<58)
unenc[j]=(byte)(b+4);
else if(b==''+'')
unenc[j]=62;
else if(b==''/'')
unenc[j]=63;
//if we find"="then data has finished, we''re not really dealing with this now
else if(b==''='')
unenc[j]=0;
else
{
char c=(char)b;
if(c==''\n'' c==''\r'' c=='' '' c==''\t'')
skip=true;
else
//could throw an exception here? it''s input we don''t understand.
;
}
//once the array has boiled convert the bytes back into chars
if(!skip&&++j==4)
{
//shift the 6 bit bytes into a single 4 octet word
int res=(unenc[0]<<18)+(unenc[1]<<12)+(unenc[2]<<6)+unenc[3];
byte c;
int k=16;
//shift each octet down to read it as char and add to StringBuffer
while(k>=0)
{
c=(byte)(res>>k);
if ( c>0 )
sb.append((char)c);
k-=8;
}
//reset j and the unencode buffer
j=0;
unenc[0]=0;unenc[1]=0;unenc[2]=0;unenc[3]=0;
}
}
return sb.toString();
}
/** encode plaintext data to a base 64 string
* @param plain the text to convert. If plain is longer than 76 characters this method
* returns null (see RFC2045).
* @return the encoded text (or null if string was longer than 76 chars).
*/
public static String encode(String plain)
{
if(plain.length()>76)
return null;
int maxturns;
StringBuffer sb=new StringBuffer();
//the encode buffer
byte[] enc=new byte[3];
boolean end=false;
for(int i=0,j=0;!end;i++)
{
char _ch=plain.charAt(i);
if(i==plain.length()-1)
end=true;
enc[j++]=(byte)plain.charAt(i);
if(j==3 end)
{
int res;
//this is a bit inefficient at the end point
//worth it for the small decrease in code size?
res=(enc[0]<<16)+(enc[1]<<8)+enc[2];
int b;
int lowestbit=18-(j*6);
for(int toshift=18;toshift>=lowestbit;toshift-=6)
{
b=res>>>toshift;
b&=63;
if(b>=0&&b<26)
sb.append((char)(b+65));
if(b>=26&&b<52)
sb.append((char)(b+71));
if(b>=52&&b<62)
sb.append((char)(b-4));
if(b==62)
sb.append(''+'');
if(b==63)
sb.append(''/'');
if(sb.length()%76==0)
sb.append(''\n'');
}
//now set the end chars to be pad character if there
//was less than integral input (ie: less than 24 bits)
if(end)
{
if(j==1)
sb.append("==");
if(j==2)
sb.append(''='');
}
enc[0]=0;enc[1]=0;enc[2]=0;
j=0;
}
}
return sb.toString();
}
}
……