PHP fsockopen 简单抓取网页内容 [转]

 1 function  get_page_content( $url ){
 2 $url   =   eregi_replace ( ' ^http:// ' ,   '' ,   $url );
 3 $temp   =   explode ( ' / ' ,   $url );
 4 $host   =   array_shift ( $temp );
 5 $path   =   ' / ' . implode ( ' / ' ,   $temp );
 6 $temp   =   explode ( ' : ' ,   $host );
 7 $host   =   $temp [ 0 ];
 8 $port   =   isset ( $temp [ 1 ])  ?   $temp [ 1 ]  :   80 ;
 9 $fp   =  @ fsockopen ( $host ,   $port ,   & $errno ,   & $errstr ,   30 );
10 if  ( $fp ){
11 @ fputs ( $fp ,  “GET  $path  HTTP / 1.1 \r\nHost :   $host \r\nAccept :   * /* \r\nReferer:$url\r\nUser-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)\r\nConnection: Close\r\n\r\n”);
12 }
13 $Content = ”;
14 while ($str = @fread($fp, 4096)){
15 $Content .= $str;
16 }
17 @fclose($fp);
18 //重定向
19 if(preg_match(”/^HTTP\/\d.\d 301 Moved Permanently/is”,$Content)){
20 if(preg_match(”/Location:(.*?)\r\n/is”,$Content,$murl)){
21 return get_page_content($murl[1]);
22 }
23 }
24 //读取内容
25 if(preg_match(”/^HTTP\/\d.\d 200 OK/is”,$Content)){
26 preg_match(”/Content-Type:(.*?)\r\n/is”,$Content,$murl);
27 $contentType=trim($murl[1]);
28 $Content=explode(”\r\n\r\n”,$Content,2);
29 $Content=$Content[1];
30 }
31 return $Content;
32 }
33
34