<?php
        /////////////////////////////////////////////////////////
        //
        //  AHNAR CSA3004 APT University of Malta
        //  Ian Bugeja 2005
        //
        //////////////////////////////////////////////////////////

        require_once 'HTML_Parser.php';
        require_once 'HTTP_Connect.php';
        require_once 'AHNAR_Tools.php';
        session_start();
        header("Cache-control: private"); // IE 6 Fix.
        $starttime = microtime(true); //TO count script execution

        $url = $_REQUEST["ahnarurl"] or AHNAR_Tools::redError(0, "Missing url...please specify ?ahnarurl= parameter with URL to script.");
        if (stripos($url, 'ahnar.iannet.org') !== false) AHNAR_Tools::redError(0, " Sorry you cannot visit AHNAR from within");
        $queryget = HTTP_Connect::sort_query($_GET);
        $querypost = HTTP_Connect::sort_query($_POST);
        trim($querypost);

        echo "<!--";
        do
        {
                if ((stripos($url, 'http://') === false) && (stripos($url, 'https://') === false)) $url = 'http://' . $url;
                $url = str_replace('*amp;', '&', $url);
                $url = trim($url);
                $urlparsed = parse_url($url);
                $scheme = $urlparsed["scheme"]; if ($scheme == '') $scheme = 'http'; elseif ($scheme == 'https') AHNAR_TOOLS::stopError(1, "Sorry https not allowed.");
                $host = $urlparsed["host"]; if ($host == '') AHNAR_TOOLS::stopError(1, "$host incorrect host");
                $port = $urlparsed["port"]; if ($port == '') $port = 80;
                $path = $urlparsed["path"]; if ($path == '') $path = '/';
                $query = $queryget; //$urlparsed["query"];
                if ($query != '')
                        $retrieve = $path.'?'.$query;
                else
                {
                        $retrieve = $path;
                        if (stripos($url, "?") !== false) //when moved
                        {
                                $mp = stripos($url, "?");
                                $retrieve = $retrieve . substr($url, $mp, strlen($url)-$mp);
                        }
                }
                $retrieve = str_replace(" ", "%20", $retrieve);

                $fnd = strpos($path, '.');
                if ($fnd !== false)
                {
                        $fnd = strrpos($path, '/');
                        $path1 = substr($path, 0, $fnd);
                }
                else
                        $path1 = $path;
                $nurl = 'http://'.$urlparsed['host'].$urlparsed['port'].$path1;

                $cookiestring = "";
                if (isset($_SESSION['cookies']))
                foreach($_SESSION['cookies'] as $cookurl => $cookie)   //Cookie Handling
                {
                        if ($cookurl == $urlparsed["host"])
                        {
                                $cookiestring = HTTP_Connect::sort_cookie($cookie);
                                break;
                        }
                }

                $loopflag = false;
                $httpcon = new HTTP_Connect();
                /////////////////////////////////////////////////////////////
                //Construct HTTP GET/POST Header----------------------------
                //===========================================================
                ///HEAD
                        if ($querypost == "")
                                $reqheader = "GET $retrieve HTTP/1.1\r\n";
                        else
                                $reqheader = "POST $retrieve HTTP/1.1\r\n";
                ///TYPE
                        $reqheader .= "Accept: image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, application/vnd.ms-excel, application/vnd.ms-powerpoint, application/msword, application/x-shockwave-flash, */*\r\n";
                ///LANGUAGE
                        $reqheader .= "Accept-Language: en-us\r\n";
                ///Encoding
                        //$reqheader .= "Accept-Encoding: deflate\r\n";
                ///HOST
                        $reqheader .= "Host: $host\r\n";
                ///COOKIE
                        if ($cookiestring != "") $reqheader .= "Cookie: ".$cookiestring."\r\n";
                ///USER AGENT
                        $reqheader .= "User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; Maxthon; .NET CLR 1.1.4322)\r\n";
                ///CONTENT LENGTH
                        if ($querypost != "") //Case of POST
                        {
                                $reqheader .= "Content-Length: ".strlen($querypost)."\r\n";
                                $reqheader .= "Content-Type: application/x-www-form-urlencoded\r\n";
                        }
                //CONNECTION TYPE
                        $reqheader .= "Connection: Close\r\n\r\n";
                //POST BODY
                        if ($querypost != "")
                                $reqheader .= $querypost;
                $contents = $httpcon->http_request($host, $port, $reqheader); //send request and get response

                if (isset($httpcon->resheader['set-cookie'])) //Add Cookie to SESSION
                        $_SESSION['cookies'][$urlparsed['host']] = $httpcon->resheader['set-cookie'];

                if ($httpcon->error[0] == 'nohost')
                {
                        echo "--><html><head><link rel='stylesheet' href='ahnarstyle.css'></head><body class='ahnarbody'>";
                        echo "Host not found...Please specify a correct host next time. Go [<a href='javascript:history.back()'>Back</a>]<br>";
                        echo "<a href='http://ahnar.dyndns.org/'>AHNAR</a><br>";
                        die("Error: ".$httpcon->error[1]." Error Number: ".$httpcon->error[2]."</body></html>");
                }
                elseif ($httpcon->error[0] == 'connerr')
                {
                        echo "--><html><head><link rel='stylesheet' href='ahnarstyle.css'></head><body class='ahnarbody'>";
                        echo "Error in connection...please try again... Go [<a href='javascript:history.back()'>Back</a>]<br>";
                        echo "<a href='http://ahnar.dyndns.org/'>AHNAR</a><br>";
                        die("Error: ".$httpcon->error[1]." Error Number: ".$httpcon->error[2]."</body></html>");
                }
                elseif ($httpcon->error[0] == 'nocontent') //content other than text/html, text/plain
                {
                        $newcontents = "<body class='ahnarbody'>";
                        $newcontents .= "<h1 class='ahnarlarge'>AHNAR&nbsp;&nbsp;&nbsp;&nbsp;<span class='ahnarmedium'>[<a href='javascript:history.back()'>Back</a>]</span><hr style='color:white'></h1>";
                        $newcontents .= "Incorrect Content Type.. ".$httpcon->resheader['content-type']." not supported by AHNAR<br>";
                        $newcontents .= "Contents of the <a href='$url' target='_blank'>URL</a> is also below:<br>";
                        $newcontents .= "<iframe src='$url' width='95%' height='80%'>IFrame not supported</iframe>";
                        $newcontents .= "<br><br><div class='ahnarsmall'>Ian Bugeja 2005, University of Malta, CSA3004 APT</div>";
                }
                elseif ($httpcon->error[0] == 'moved')
                {       $querypost = "";
                        $moved = $httpcon->resheader['location'];
                        if (stripos($moved, 'http://') === false) //not found thus must reside part of domain
                        {
                                if ($moved[0] == '/') $url = $host.$moved;
                                elseif ($moved[0] == '.') $url = $host.substr($moved, 2, strlen($moved)-2);
                                else
                                {
                                        if ($url[strlen($url)-1] == '/')
                                                $url = $url . $moved;
                                        else
                                                $url = $url.'/'.$moved;
                                }
                        }
                        else
                                $url = $moved;

                        $loopflag = true;
                }
                elseif ($httpcon->error[0] == 'server')
                {
                        $rse = $urlparsed['host'];
                        die("-->$rse Server error..Page Cannot be retrieved");
                }
                else
                {
                        if (($_REQUEST['ahnarback'] != 'true') && ($httpcon->error[0] != 'nocontent'))
                        {
                                if (isset($_SESSION['visited']))
                                {
                                        if (end($_SESSION['visited']) != $url)
                                                $_SESSION['visited'][] = $url; //store visited URL in Session
                                }
                                else
                                        $_SESSION['visited'][] = $url; //store visited URL in Session
                        }
                        //
                        //HTML PARSING-------------------------------------------
                        //============
                        $htmlparser = new HTML_Parser('', $nurl);
                        $newcontents = $htmlparser->html_parse($contents);

                        $_SESSION['titles'][] = trim($htmlparser->pagetitle); //store page title in Session
                }
        }
        while ($loopflag == true);
 ?>
-->
<?php
        if ($htmlparser->docutype != '') //<!DOCTYPE tag
                echo $htmlparser->docutype;
        else
        {
?>
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<?php   } ?>

<!-- Page Processed on:<?php echo date("l dS of F Y h:i:s A"); ?>-->
<!-- URL Retrieved: <?php echo $url; ?>-->
<html>

   <head>
      <title>AHNAR <?php if ($htmlparser->pagetitle != "") echo '[ '.$htmlparser->pagetitle.' ]'; ?></title>
      <meta http-equiv="generator" content="AHNAR">
      <?php if ($htmlparser->stshlink != '') echo $htmlparser->stshlink;
            if ($httpcon->error[0] == 'nocontent') echo "<link rel='stylesheet' href='ahnarstyle.css'>";
      ?>
   </head>

<?php
   if ($htmlparser->conframes == false)  //page does not contain frames
   if (strlen($htmlparser->words) >= 50)   //page contents > 50
   {
      if ($httpcon->error[0] != '') echo "<body class='ahnarbody'>";
      elseif ($htmlparser->bodytag != "") echo $htmlparser->bodytag;
      else echo "<body style='text-align: left'>";
      //Leave empty line for header
      echo "&nbsp;<br>";
      echo "<div style='position:absolute; top:0px; left:0px'><nobr><span style=\"background-color: Background; color: white; font-family: Tahoma; font-size: x-small\">";
      echo "&nbsp;<strong>AHNAR</strong> ---- (header loading....)&nbsp;</span></nobr></div>";

   } else echo $htmlparser->bodytag;
?>

<!--
<?php
        //KEYWORD PROCESSING (common words)
        //if page is requested as back do not do any keyword processing
        //this also applies if contents (ie tags removed) is less than 100
        if (($_REQUEST['ahnarback'] != 'true') && (strlen($htmlparser->words) >= 100))
        {
            require_once 'AdaptiveHyp.php';
            $adapt = new AdaptiveHyp();

            if ($url != $_SESSION['visited'][count($_SESSION['visited'])-2]) //do not age/reupdate keyword if page refresh
            {
                //Reset Keywords in Session
                if (isset($_SESSION['keywords'])) //keywords in session exists
                foreach($_SESSION['keywords'] as &$okey)
                {
                        $okey[3] = $okey[3] + 1;  //age the keyword
                        $okey[2] = 0; //updated last is 0
                }


                //Remove old keywords with no importance
                $visit = count($_SESSION['visited']);
                if (isset($_SESSION['keywords']))
                {
                        for ($i=0; $i < count($_SESSION['keywords']); $i++)
                        {
                                if ($_SESSION['keywords'][$i][1] < $visit) //occurrance
                                if ($_SESSION['keywords'][$i][3] > $visit/2) //age
                                {
                                        array_splice($_SESSION['keywords'], $i, 1);  //REMOVE
                                }
                        }

                        if (count($_SESSION['keywords']) > 25) //keep below 25 approx
                        {
                                require_once "AHNAR_Tools.php";
                                $ahtools = new AHNAR_Tools();
                                $Okeywords = $ahtools->KeywordSort($_SESSION['keywords']);

                                for (; count($Okeywords) >= 25; )
                                        $Okeywords = array_pop($Okeywords);

                                $_SESSION['keywords'] = $Okeywords;
                        }
                }

                //get popular keywords
                $wordcount = substr_count($htmlparser->words, " ");  //compute approx number of words present
                if ($wordcount >= 1000) $maxa = 8;        //assign how much popular keywords to get
                elseif ($wordcount >= 800) $maxa = 7;
                elseif ($wordcount >= 500) $maxa = 5;
                elseif ($wordcount >= 200) $maxa = 3;
                else $maxa = 0;
                if ($maxa > 0)
                        $Okeywords = $adapt->popular($htmlparser->words, $_SESSION['keywords'], $maxa);
                else
                        unset($Okeywords);
                unset($_SESSION['latestkeys']);  //empty latest keys

                //Merge found keywords into Session
                if (isset($Okeywords)) //if keywords have been found
                foreach($Okeywords as $k)
                {
                        $kflag = false;

                        if (isset($_SESSION['keywords'])) //there exist some keywords in session
                        foreach($_SESSION['keywords'] as &$okey)
                        {
                                if (AdaptiveHyp::equal($okey[0], $k[0])) //if ($okey[5] == $k[5]) //compare
                                {       //MERGE into SESSION
                                        $kflag = true;
                                        $okey[1] = $okey[1] + $k[1];
                                        $okey[2] = 1; //marks that updated last
                                        $okey[4] = $okey[4] + 1; //updated times
                                        $_SESSION['latestkeys'][] = $k;
                                        break;
                                }
                        }

                        if ($kflag == false) //not found thus add to list
                        {       //ADD to SESSION
                                $_SESSION['keywords'][] = $k;
                                $_SESSION['latestkeys'][] = $k;
                        }
                }
            }


            //---------------------------------
            //Suggestion of Links to Follow----
            //=================================
            if (count($_SESSION['visited']) > 2)  //SUGGEST ONLY IF 2 OR MORE HAVE BEEN VISITED
            {
                $adapttags1 = $adapt->adaptKeyLinks($htmlparser->ataglist, $htmlparser->atagcount, $htmlparser->middleatext, $_SESSION['keywords'], $newcontents);
                $adapttags2 = $adapt->adaptLinks($htmlparser->ataglist, $htmlparser->atagcount, $newcontents);
                //merge the 2 above arrays removing duplicates
                if ((isset($adapttags1)) && isset($adapttags2))
                {
                        foreach($adapttags1 as $ad1k => $ad1v)
                        {
                                $adapttags2[$ad1k] = $ad1v;
                        }
                        $adapttags = $adapttags2;
                }
                elseif (isset($adapttags1))
                        $adapttags = $adapttags1;
                elseif (isset($adapttags2))
                        $adapttags = $adapttags2;

                //print_r($adapttags); print_r($htmlparser->middleatext);
                if (isset($adapttags))
                {
                        $adaptcount = count($adapttags);
                        foreach($adapttags as $kadt => $adt)    //out of adaptable find those to adapt for sure
                        {
                                $adaptflag = true;
                                if (isset($sureadapt))          //if already suggested a link to same location
                                foreach($sureadapt as $suk => $suv)
                                {
                                        if (trim($htmlparser->adest[$suk], "/ ") == trim($htmlparser->adest[$kadt], "/ "))
                                        {
                                                $adaptflag = false; //do not suggest
                                        }
                                }


                                if ($adaptflag == false)
                                {
                                        //do not suggest (already suggested link to same location above check)
                                }
                                elseif (stripos($htmlparser->middleatext[$kadt], "@") !== false) { /*email should not be suggested*/ }
                                elseif ($htmlparser->middleatext[$kadt] == "") { /*neither empty*/ }
                                elseif (AdaptiveHyp::commonlink($htmlparser->middleatext[$kadt]))
                                {
                                        //common link...do nothing
                                }
                                elseif (stripos($htmlparser->adest[$kadt], '#') !== false)
                                {
                                        if (AdaptiveHyp::inPageLink($htmlparser->ataglist, $htmlparser->middleatext, $htmlparser->adest[$kadt], $newcontents, $_SESSION['keywords']))
                                        {
                                                $sureadapt[$kadt] = $adt;    //keyword near link
                                                $adaptcolor[$kadt] = 0;
                                        }
                                }
                                elseif (trim($htmlparser->adest[$kadt], "/ ") == trim($url, "/ "))
                                {
                                        //points to same page...do nothing
                                }
                                elseif (AdaptiveHyp::linkList($adt, $newcontents))
                                {       //echo $htmlparser->middleatext[$kadt];
                                        //link list
                                }
                                elseif (AdaptiveHyp::beforeAfterLink($adt, $newcontents, $_SESSION['keywords']))
                                {
                                       $sureadapt[$kadt] = $adt;    //keyword near link
                                       $adaptcolor[$kadt] = 1;
                                }
                                elseif (AdaptiveHyp::textLink($htmlparser->middleatext[$kadt], $_SESSION['keywords']))
                                {
                                       $sureadapt[$kadt] = $adt;    //link text matches keyword
                                       $adaptcolor[$kadt] = 1;
                                }
                                elseif (AdaptiveHyp::implink($htmlparser->middleatext[$kadt]))
                                {
                                       $sureadapt[$kadt] = $adt;
                                       $adaptcolor[$kadt] = 2;
                                }
                        }

                        //Further Filter on Suggestion   ----------------FURTHER FILTER
                        if (isset($sureadapt))
                              foreach($sureadapt as $ksadt => &$sadt)
                              {       //usefull when page contains a number of links with same text
                                    foreach($sureadapt as $ksadt2 => &$sadt2)
                                    {
                                          if ($ksadt != $ksadt2)
                                          if (AdaptiveHyp::implink($htmlparser->middleatext[$ksadt]) == false)
                                          if (AdaptiveHyp::equal($htmlparser->middleatext[$ksadt], $htmlparser->middleatext[$ksadt2]))
                                          {
                                              $sadt = 0;  $sadt2 = 0; //remove some of them
                                          }
                                    }
                              }


                        if (count($sureadapt) < $htmlparser->atagcount * 0.05) //??? Problem SUGGESTING TOO MUCH
                        {
                                foreach($adapttags as $kadt => $adt)
                                {
                                     if (trim($htmlparser->adest[$kadt], "/ ") == trim($url, "/ "))
                                     {
                                        //url points to same page
                                     }
                                     elseif (AdaptiveHyp::commonlink($htmlparser->middleatext[$kadt]))
                                     {
                                        //common link...do nothing
                                     }
                                     elseif (AdaptiveHyp::textLink($htmlparser->middleatext[$kadt], $_SESSION['keywords'], 1))
                                     {
                                        $sureadapt[$kadt] = $adt;    //link text matches keyword
                                        $adaptcolor[$kadt] = 3;
                                     }
                                }
                        }

                }

                //Arrow to insert
                require_once "AHNAR_Tools.php";
                $arrow = AHNAR_Tools::getArrow(0);
                $arrowlen = strlen($arrow);
                $arrowcount = 0;

                //Modify Page contents by inserting Arrow
                if (isset($sureadapt))
                {
                        asort($sureadapt);
                        $sureadapt = array_unique($sureadapt); //?????? to check
                        foreach ($sureadapt as $ksadt => &$sadt)
                        {
                             if ($sadt != 0)
                             {
                                if (($adaptcolor[$ksadt] != 3) || ($htmlparser->atagcount*0.2 >= count($sureadapt)))
                                {
                                        $arrow = AHNAR_Tools::getArrow($adaptcolor[$ksadt]);
                                        $posl = $sadt + ($arrowlen * $arrowcount);
                                        $newcontents = substr($newcontents, 0, $posl). $arrow . substr($newcontents, $posl, strlen($newcontents)-($posl-1));
                                        $arrowcount++;
                                }
                             }
                        }
                }
            }
        }

?>

-->

        <?php echo $newcontents; ?>


<?php
      if ($htmlparser->conframes == false)
      if ($httpcon->error[0] != 'nocontent') ///insert AHNAR Header
      {
                echo "<div style='position:absolute; top:0px; left:0px'>";
                echo "<nobr>";
                echo "<span style='background-color: Background; color: white; font-family: Tahoma; font-size: x-small'>";
                        echo "&nbsp;<a style='background-color: Background; color: white' href='index.php'><strong>AHNAR</strong></a> -";
                        echo "<b>";
                        if (trim($htmlparser->pagetitle) == "")  //no pagetitle URL
                                echo '[ '.$url.' ]';
                        elseif (strlen($htmlparser->pagetitle) > 40) //LONG page title shorten
                                echo '[ '.substr($htmlparser->pagetitle, 0, 40).'... ]';
                        else
                                echo '[ '.$htmlparser->pagetitle.' ]';

                        echo "</b> -<i> Ian Bugeja 2005, University of Malta, CSA3004 APT </i>";

                        if (count($_SESSION['visited']) > 2) //see also available on more than 2 page visits
                                echo "[<a style=\"background-color: Background; color: white\" href=\"seealso.php\" target=\"_blank\">See also...</a>]";
                        else
                                echo "[<a style=\"background-color: Background; color: white\" href=\"javascript:alert('Sorry See Also... function disabled. Visit more than 2 pages please.')\">See also...</a>]";

                        //$ptit = str_replace(" ", "%20", $htmlparser->pagetitle);
                        //$ptit = str_replace("'", "", $ptit);
                        $ptit = urlencode($htmlparser->pagetitle);
                        echo "[<a style='background-color: Background; color: white' href='bookmark.php?ahurl=$url&ahtit=$ptit' target='_blank'>Bookmark</a>]";
                        echo "[<a style='background-color: Background; color: white' href='correct.php' target='_blank'>Correct</a>]";
                        echo "[<a style='background-color: Background; color: white' href='negative.php?ahneg=$url'>Negative</a>]";
                        echo "[<a style='background-color: Background; color: white' href='destroy.php' target='_top'>Reset</a>]&nbsp;";
                echo "</span>";
                echo "</nobr> </div>";
   } ?>

   </body>
</html>

<!--
<?php $endtime = microtime(true); $timetaken = $endtime - $starttime; echo "Execution Time: $timetaken"; ?>
-->