获取网页源码

此代码暂不能处理乱码问题，但获得的网页源码可以直接显示在终端上或定位到终端上后用firefox查看

*  -*- C -*-
 *
 * gethtml.c -     Note: http头结束是\r\n\r\n  
 * 参考 ：
 * Author          : heixia
 * Created On      : Sat Mar 29 01:33:11 2008
 *
 */

#include
#include
#include
#include
#include
#include
#include

#define  BUFSIZE 128

int main(int argc,char *argv[  ]  ){
     char   *url,*pHost,host[ BUFSIZE ],GET[ BUFSIZE ];
     int    sockfd;
     int    len,result;
     struct sockaddr_in address;
     struct hostent *hostinfo;
     struct servent *servinfo;
     char   buffer[ BUFSIZE ],header[ BUFSIZE ]="";

     if( argc == 1 ){
          fprintf( stderr,"./a.out netaddress.\n." );
          exit( 1 );
     }
     else
          url = argv[ 1 ];

     /*
      *        分离url中的主机地址和相对路径
      */

     for (pHost = url; *pHost != '/' && *pHost != '\0'; pHost++ )
          ;
     if ( (int)(pHost - url) == strlen(url) )
          strcpy(GET, "/");
     else
          strcpy(GET, pHost);
     *pHost = '\0';
     strcpy(host, url);
     printf("%s\n%s\n", host, GET);

     hostinfo = gethostbyname( host );
     if( !hostinfo ){
          fprintf( stderr,"can't get inof for host: %s\n",host );
          exit( 1 );
     }

     sockfd = socket( AF_INET,SOCK_STREAM,0 );

     address.sin_family = AF_INET;
     address.sin_port = htons( 80 );
     address.sin_addr = *( struct in_addr * )*hostinfo->h_addr_list;
     len = sizeof( address );

     strcat( header,"GET " );
     strcat( header,GET);
     strcat( header," " );
     strcat( header,"HTTP/1.1\r\n");
     strcat( header,"Host: ");
     strcat( header,host);
     strcat( header,"\r\nConnection: close\r\n\r\n");
    
     //建立连接
     result = connect( sockfd,( struct sockaddr * )&address,len );
     if( result == -1 ){
          perror( "oops:getdate" );
          exit( 1 );
     }

     write( sockfd,header,strlen( header ) );
    
     while( read( sockfd,buffer,sizeof( buffer )) > 0){
          printf( "%s",buffer );
          memset( buffer,0,sizeof( buffer ) );
     }

     close( sockfd );
    
     exit( 0 );
}