解析url

// parse url 解析 url c c++ 代码 rfc 2068

 

#include <stdio.h>
#include <string.h>
#include <stdlib.h>

// 解析url，作为示例，很多情况没考虑，比如说user,pass之类的
int parse_url(char *url, char **serverstrp, int *portp, char **pathstrp)
{
    char buf[256];
    int serverlen, numread=0;

    // 跳过"http://" ，例"http://www.a.cn"->"www.a.cn"
    url = url+7;

    // 输入url直到遇到'/'和':'符号
    sscanf(url, "%255[^/:]", buf);
    serverlen = strlen(buf);
    *serverstrp = (char *)malloc(serverlen+1);
    strcpy(*serverstrp, buf);
    if(url[serverlen]==':')
    {
        // 获取port
        sscanf(&url[serverlen+1], "%d%n", portp, &numread);
        /* add one to go PAST it */
       
        numread++;
    }
    else
    {
        *portp = 80;
    }
    /* the path is a pointer into the rest of url */
   
    *pathstrp = &url[serverlen+numread];
    return 0;
}

int main()
{
    char url[256] = "http://www.a.cn:80//index.html";
    char pathstr[256] = {'/0'};
    char szserverstrp[256] = {'/0'};
    char* serverstrp = szserverstrp;
    char* pathstrp = &pathstr[0];
    int port;

    parse_url(url,&serverstrp,&port,&pathstrp);
    printf("%s/n%s/n%d/n%s/n",url,serverstrp,port,pathstrp);

    return 0;
}