用有穷自动机解析Cookie.doc

用有穷自动机解析Cookie
以前我们开发一个代理转换服务器，它位于移动设备(或者移动设备网关, 如WAP网关）与WEB服务器之间，移动设备通过我们的代理转换服务器去访问WEB服务器，代理转换服务器会把取到的网页（HTML）转换成移动设备支持的网页（如XHTML/CHTML/HDML/WML等),然后发给设备。这样，移动设备就可以访问现有的WEB服务了。
后来，客户要求加一个功能：大多数网站都要求浏览器支持Cookie，但很多移动设备的浏览器没有这个功能，因为这个原因，设备访问不了这些网站。客户要求代理转换服务器能够解决这个问题。
在代理转换服务器解中解决这个问题并不困难，可以这样实现：当设备通过代理转换服务器访问WEB服务器时，如果在WEB服务器的响应中带有Set-Cookie，则代理转换服务器把它存起来，在这个设备后续访问时, 代理转换服务器会从数据库中把Cookie取出来，加到请求里，并发送给WEB服务器，这就模拟了设备支持Cookie的功能。
实现时，我们需要把Cookie解析出来，并加上SessionID，查找时要根据SessionID和Domain等条件进行匹配，满足条件才加入到请求头中。同时要定期检查Cookie是否过期，若过期了，则从数据库中删除。这里涉及到Cookie的解析和重组，当然这非常简单，我们用一个小程序演示如何用DFA解析Cookie。
Cookie的格式为:
“NAME=VALUE; expires=DATE;path=PATH; domain=DOMAIN_NAME; Secure”
一个Cookie可以看作一个一个的{KEY, VALUE}对，KEY和Value之间用’=’分隔，{ KEY, VALUE }对之间用’;’分隔，有的Value是可选的，如Secure。
DFA的描述如下：
字母表：ASCII码
状态：KEY前、KEY中，VALUE前，VALUE中和接受状态。KEY前状态同时是起始状态。
转换规则：
KEY前状态下读到非空格，进入在KEY中。
KEY中状态下读到’=’，进入VALUE前状态。
KEY中状态下读到’;’，进入KEY前状态。
VALUE前状态下读到非空格，进入VALUE中状态。
VALUE中状态下读到’;’， 进入KEY前。
KEY前状态、KEY中状态, VALUE前和VALUE中状态下读到’/0’，进入结束状态。
每当读到一个完整的{KEY, VALUE}，就调用一个回调函数去构建Cookie。
实现代码如下：
//Build: gcc -DDEBUG -g parsecookie.c -o pc.exe
#include <stdio.h>
#include <stdlib.h>
#include <assert.h>
#include <string.h>

typedef int (COOKIE_BUILDER)(void* pBuildCtx, const char* key, const char* value);
//"Set-Cookie: name=value; Expires=date; domain=domainname;path=path; secure;"
int ParseCookie(const char* pszCookie, void* pBuildCtx, COOKIE_BUILDER Builder)
{
         enum
         {
                  STAT_BEFORE_KEY,
                  STAT_IN_KEY,
                  STAT_BEFORE_VALUE,
                  STAT_IN_VALUE
         }eState = STAT_BEFORE_KEY;
         char* key = NULL;
         char* value = NULL;
         char* pszWorkBuff = NULL;
         char* p = NULL;
         assert(pszCookie != NULL && Builder != NULL);
         pszWorkBuff = (char*)strdup(pszCookie);
         p = pszWorkBuff;
         while(*p != '/0')
         {
                  switch(eState)
                   {
                            case STAT_BEFORE_KEY:
                                     {
                                               if(*p != ' ')
                                               {
                                                        eState = STAT_IN_KEY;
                                                        key = p;
                                               }
                                               break;
                                     }
                            case STAT_IN_KEY:
                                     {
                                               if(*p == '=')
                                               {
                                                        *p = '/0';
                                                        eState = STAT_BEFORE_VALUE;
                                               }
                                               else if(*p == ';')
                                               {
                                                        *p = '/0';
                                                        eState = STAT_BEFORE_KEY;
                                                        Builder(pBuildCtx, key, NULL);
                                                        key = NULL;
                                                        value = NULL;
                                               }
                                               break;
                                     }
                            case STAT_BEFORE_VALUE:
                                     {
                                               if(*p != ' ')
                                               {
                                                        eState = STAT_IN_VALUE;
                                                        value = p;
                                               }
                                               break;
                                     }
                            case STAT_IN_VALUE:
                                     {
                                               if(*p == ';')
                                               {
                                                        *p = '/0';
                                                        eState = STAT_BEFORE_KEY;
                                                        Builder(pBuildCtx, key, value);
                                                        key = NULL;
                                                        value = NULL;                                                    
                                               }
                                               break;
                                     }
                            default:break;
                   }
                   p++;
         }
          free(pszWorkBuff );
         return 0;
}

int DumpCookieBuilder(void* pBuildCtx, const char* key, const char* value)
{
         printf("%s=%s/n", key, value);
         return 0;
}

int main(int argc, char* argv[])
{
         int nRet = 0;
         if(argc == 2)
         {
                   nRet = ParseCookie(argv[1], NULL, DumpCookieBuilder);
         }
         else
         {
                  printf("Usage: %s [str]/n", argv[0]);
                  printf("for example: /"name=value; Expires=date; domain=domainname;path=path; secure;/"/n");
                   nRet = 0;
         }
         return nRet;
}