<?php
//本文件用于扩展采集结果
//下载居于http 1.1协议的防盗链图片
//------------------------------------
function DownImageKeep($gurl,$rfurl,$filename,$gcookie="",$JumpCount=0,$maxtime=30){
$urlinfos = GetHostInfo($gurl);
$ghost = $urlinfos['host'];
$gquery = $urlinfos['query'];
if($gcookie=="") $gcookie = RefurlCookie($rfurl);
$sessionQuery = "GET $gquery HTTP/1.1\r\n";
$sessionQuery .= "Host: $ghost\r\n";
$sessionQuery .= "Referer: $rfurl\r\n";
$sessionQuery .= "Accept: */*\r\n";
$sessionQuery .= "User-Agent: Mozilla/4.0 (compatible; MSIE 5.00; Windows 98)\r\n";
if($gcookie!=""&&!ereg("[\r\n]",$gcookie)) $sessionQuery .= $gcookie."\r\n";
$sessionQuery .= "Connection: Keep-Alive\r\n\r\n";
$errno = "";
$errstr = "";
$m_fp = fsockopen($ghost, 80, $errno, $errstr,10);
fwrite($m_fp,$sessionQuery);
$lnum = 0;
//获取详细应答头
$m_httphead = Array();
$httpstas = explode(" ",fgets($m_fp,256));
$m_httphead["http-edition"] = trim($httpstas[0]);
$m_httphead["http-state"] = trim($httpstas[1]);
while(!feof($m_fp)){
$line = trim(fgets($m_fp,256));
if($line == "" || $lnum>100) break;
$hkey = "";
$hvalue = "";
$v = 0;
for($i=0;$i<strlen($line);$i++){
if($v==1) $hvalue .= $line[$i];
if($line[$i]==":") $v = 1;
if($v==0) $hkey .= $line[$i];
}
$hkey = trim($hkey);
if($hkey!="") $m_httphead[strtolower($hkey)] = trim($hvalue);
}
//分析返回记录
if(ereg("^3",$m_httphead["http-state"])){
if(isset($m_httphead["location"]) && $JumpCount<3){
$JumpCount++;
DownImageKeep($gurl,$rfurl,$filename,$gcookie,$JumpCount);
}
else{ return false; }
}
if(!ereg("^2",$m_httphead["http-state"])){
return false;
}
if(!isset($m_httphead)) return false;
$contentLength = $m_httphead['content-length'];
//保存文件
$fp = fopen($filename,"w") or die("写入文件：{$filename} 失败！");
$i=0;
$okdata = "";
$starttime = time();
while(!feof($m_fp)){
$okdata .= fgetc($m_fp);
$i++;
//超时结束
if(time()-$starttime>$maxtime) break;
//到达指定大小结束
if($i >= $contentLength) break;
}
if($okdata!="") fwrite($fp,$okdata);
fclose($fp);
if($okdata==""){
@unlink($filename);
fclose($m_fp);
return false;
}
fclose($m_fp);
return true;
}
//获得某页面返回的Cookie信息
//----------------------------
function RefurlCookie($gurl){
$urlinfos = GetHostInfo($gurl);
$ghost = $urlinfos['host'];
$gquery = $urlinfos['query'];
$sessionQuery = "GET $gquery HTTP/1.1\r\n";
$sessionQuery .= "Host: $ghost\r\n";
$sessionQuery .= "Accept: */*\r\n";
$sessionQuery .= "User-Agent: Mozilla/4.0 (compatible; MSIE 5.00; Windows 98)\r\n";
$sessionQuery .= "Connection: Close\r\n\r\n";
$errno = "";
$errstr = "";
$m_fp = fsockopen($ghost, 80, $errno, $errstr,10);
fwrite($m_fp,$sessionQuery);
$lnum = 0;
//获取详细应答头
$gcookie = "";
while(!feof($m_fp)){
$line = trim(fgets($m_fp,256));
if($line == "" || $lnum>100) break;
else{
if(eregi("^cookie",$line)){
$gcookie = $line;
break;
}
}
}
fclose($m_fp);
return $gcookie;
}
//获得网址的host和query部份
//-------------------------------------
function GetHostInfo($gurl){
$gurl = eregi_replace("^http://","",$gurl);
$garr['host'] = eregi_replace("/(.*)$","",$gurl);
$garr['query'] = "/".eregi_replace("^([^/]*)/","",$gurl);
return $garr;
}
//HTML里的图片转DEDE格式
//-----------------------------------
function TurnImageTag($ttx){
preg_match_all('/src="(.+?)"/is',$ttx,$match);
for($i=0;$i<count($match[1]);$i++){
$tx .="{dede:img text='' }".$match[1][$i]." {/dede:img}"."\r\n";
}
$ttx="{dede:pagestyle maxwidth='800' ddmaxwidth='150' row='3' col='3' value='3'/}\r\n".$tx;
return $ttx;
}
?>
|
帝国论坛 → 帝国网站管理系统交流区 → 采集使用交流 → 教大家个方法，采集防盗链的站的图片
顶部 
加入收藏夹
2007-09-10 03:35:00 
已设置保密
回复
引用
报告
编辑
删除
