Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- <?php
- /*open source wayback machine indexer
- depends:
- https://github.com/internetarchive/wayback/tree/master/wayback-cdx-server
- //maker:
- $fi=fopen("wayback.index.php","w");
- fwrite($fi,file_get_contents("https://pastebin.com/raw/pEtsC7zE"));
- fclose($fi);
- echo "did?";
- //or:
- if(isset($_GET["a2"])&&isset($_GET["name"])){
- $fi=fopen($_GET["name"],"w");
- fwrite($fi,"http".$_GET["a2"]);
- fclose($fi);
- echo "did?";
- }elseif(isset($_GET["bak"])){
- echo file_get_contents($_GET["bak"]);
- }else
- echo "no url?";
- //and:
- ?a2=https://pastebin.com/raw/pEtsC7zE&name=wayback.index.php
- */
- $html='<!Doctype html><html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"><title>';
- if(isset($_SERVER["PATH_INFO"])&&$_SERVER["PATH_INFO"]=="/itall_ind.html"){
- $dats=scandir("./");
- $large="";
- foreach($dats as $fn){
- if(substr($fn,strlen($fn)-5)==".html"){
- $large.='<a href="/./'.$fn.'">'.$fn."</a><br>\r\n";
- }
- }
- echo $html."all</title></head><body><br>\r\n".$large."\r\n</body></html>";
- exit();
- }
- $url="";
- if(isset($_GET["url"]))
- $url=$_GET["url"];
- else{
- exit("error, no url param. make like so:<br>\r\n".$_SERVER["SCRIPT_URL"]."?url=pastebin.com/*&from=20191018");
- }
- if(isset($_GET["resumeKey"]))
- $hrk="&resumeKey=".$_GET["resumeKey"];
- else
- $hrk="";
- $date="";
- if(isset($_GET["from"]))
- $date.="&from=".$_GET["from"];
- if(isset($_GET["to"]))
- $date.="&to=".$_GET["to"];
- $dat=file_get_contents("https://web.archive.org/cdx/search/cdx?url=".$url."&limit=2500&filter=statuscode:(200|206)&collapse=urlkey".$date."&showResumeKey=true".$hrk);
- $dat=str_replace("\r\n","\n",$dat);
- $dat=str_replace("\n\r","\n",$dat);
- $dat=str_replace("\r","\n",$dat);
- $lines=explode("\n",$dat);
- $outp="";
- foreach($lines as $line){
- $parts=explode(" ", $line);
- $cp=count($parts);
- if($cp>2)
- $outp.='<a href="https://web.archive.org/web/'.$parts[1].'/'.$parts[2].'" >'.$parts[2].'</a> '.$parts[1]."<br>\r\n";
- elseif($cp==1&&$parts[0]){
- $outp.="<br>\r\n".'<a href="'.$_SERVER["SCRIPT_URL"].'?url='.$url.$date.'&resumeKey='.str_replace("+","%2B",$parts[0]).'" > Next page </a>';
- }
- }
- $t2=str_replace("/","_",$_SERVER["QUERY_STRING"]);
- $title="wayback.index.".preg_replace("/[^a-zA-Z0-9_\.]/","-",$t2)."_".dechex(time()).".html";
- $outp2=$html."wayback machine index</title></head><body>\r\n<br>".$title."<br>\r\n".$outp."<br><br></body></html>";
- $fi2=fopen($title,"w");
- fwrite($fi2,$outp2);
- fclose($fi2);
- $nl="\n";
- if(file_exists("wayback.index.sitemap.xml"))
- $smdat=file_get_contents("wayback.index.sitemap.xml");
- else{
- $smdat='<?xml version="1.0" encoding="UTF-8"?>'.$nl.'<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.$nl.'</urlset>';
- }
- $lel=strlen($smdat);
- if($lel>200) $lel-=200;
- else $lel=15;
- $eue=stripos($smdat,"</urlset>",$lel);
- $haz=substr($smdat,0,$eue).'<url><loc>'.$title.'</loc><lastmod>'.date(DATE_ATOM,time()).'</lastmod></url>'.$nl.substr($smdat,$eue);
- $fi3=fopen("wayback.index.sitemap.xml","w");
- fwrite($fi3,$haz);
- fclose($fi3);
- //finally done coding?
- echo $outp2;
- ?>
Add Comment
Please, Sign In to add comment