新浪新闻小偷1.<html><head><meta http-equiv="Refresh" content="60"><META HTTP-EQUIV="Pragma" CONTENT="no-cache"><meta http-equiv="Content-Type" content="text/html"; charset="??????"><title>新浪_新闻抓取程序</title></head><body><center><img src="pic.gif"></center><?$host="127.0.0.1"; // MYSQL 主机名$namesql="????"; // MYSQL 用户名$passsql="????"; // MYSQL 密码$lib="news"; // 数据库名$table="news"; // 数据库表名$filename="http://news.sina.com.cn/news1000/index.shtml"; // 抓取的新闻页$keytop="新闻开始"; // 新闻开始关键词$keybottom="新闻结束"; // 新闻结束关键词$conn=mysql_connect($host,$namesql,$passsql);mysql_select_db($lib,$conn);$fp=fopen($filename,"r",1);$flag=0; $add=""; $found=0; $end=0; $i=0; $temp[4]="";while(!$end==1){while(!$flag==1){$word=fgetc($fp);$add=$add.$word;if($word=="<" and strlen($add)==1){ $flag=0; }if($word=="<" and strlen($add)!=1){ $flag=1; $add=substr($add,0,strlen($add)-1); }if($word==">"){ $flag=1; } }if(strchr($add,$keytop)){ $found=1; }if(strchr($add,$keybottom)){ $found=0; $end=1; }if(((strchr($add,"<")==True and strchr($add,"href")==True) or strchr($add,"<")==False) and $found==1){ $text[$i]=$add; $i++; } if($word=="<" and $flag==1){ $add=$word; $flag=0; } else { $add=""; $flag=0; } }fclose($fp);for($i=1;$i<sizeof($text)-1;$i+=5){ $temp[1]=$text[$i+1]; $temp[2]=$text[$i+2]; $temp[3]=$text[$i+3]; $temp[4]=$text[$i+4];$ins="select * from $table where TITLE="$temp[3]"";$list=mysql_query($ins,$conn); @$count=mysql_num_rows($list);if($count==0){$ins="insert into $table(TYPE,URL,TITLE,DATE) values ("$temp[1]","$temp[2]","$temp[3]","$temp[4]")";$list=mysql_query($ins,$conn); } }mysql_close($conn);?></body></html>2.<? $handle = fopen("http://news.sina.com.cn/news1000/","r"); $sign = 0; while(!feof($handle)) { $message = fgets($handle,512); if($sign == 1) { print("$message"); } if(ereg("新闻开始",$message,$result)) { $sign = 1; } else if(ereg("新闻结束",$message,$result)) { $sign = 0; } } fclose($handle); ?>
