<?
$t1 = (double)microtime();
// ------------------------------------------------------------------------- //
// Description du script.                                                    //
// ------------------------------------------------------------------------- //
// Auteur: Damien MARTRAIRE                                                  //
// Email:  Send from http://www.atypix.com/contact.php3                       //
// Web:    http://www.atypix.com/                                              //
// Fichier du 15/05/2001, Mise à jour du 18/11/2003                          //
// ------------------------------------------------------------------------- //



$color1="#EEEEEE";    //content table color
$color2="#996699";    //title table color
//$proxy="proxy";    //si proxy
//$proxy_port=8080;    //si proxy


/** Classe PageAnalyse */
class PageAnalyse
{
    var $url="";            //url du fichier
    var $page="";            //contenu de la page à analyser
    var $xml="";            //buffer de sortie XML
    var $error="";            //buffer avec les erreurs
    var $content="";        //buffer de sortie du code HTML
    var $header=array();    //tableau avec les en tete http recu avec fsock
    var $statut="";            //etat de la connection 200=ok
    var $date="";            //date de la page
    var $connexion;            //type de connexion "fread" ou "fsock"



    /*****************************************************
    ** Constructor                                          **
    *****************************************************/

    function PageAnalyse($url)
    {
        global $action,$connexion;
        $this->connexion=$connexion;
        $this->url=$url;
        $this->beginTable();
        $this->insertTitle("Defi PHP n°3","a10",0);
        $this->insertLine("<a href='http://www.atypix.com/contact.php3'>Send from http://www.atypix.com/contact.php3</a>");
        $this->insertLine("<a href='http://www.atypix.com'>www.ATYPIX.com</a>");
        $this->insertLine("Fichier du 15/05/2001, Mise à jour du 18/11/2003");
        $this->printForm();
        if($url)
        {
            $this->connect();
            if($this->headProcess() && !$action )
            {
                $this->menu();
                $this->insertTitle("Header de la page","a20");
                $this->getHead();
                $this->insertTitle("Données de la page","a30");
                $this->getInfosPage();
                $this->insertTitle("Métas de la page","a40");
                $this->getMetas();
                $this->insertTitle("Liens de la page","a50");
                $this->getLinks();
                $this->insertTitle("Images de la page","a60");
                $this->getImages();
                $this->printXml();
                //$this->writeXml("url.xml"); write the XML in a file
            }
            else if($action=="source")
            {
                $this->insertTitle("Code source de la page $this->url","a80");
                $this->insertLine(nl2br(htmlentities($this->page)),"");
            }
            else
            {
                $this->insertTitle("Erreur de connexion sur $this->url","a80");
                if($this->connexion=="fread")
                {
                    $text="ERREUR: Fichier introuvable";
                    $this->insertLine($text,$this->error);
                }
                else
                {
                    $text="ERREUR: $this->error - Code HTTP";
                    $this->insertLine($text,$this->statut);
                }
            }
        }

        $this->insertTitle("Fin de l'analyse","a70");
        $this->closeTable();
    }


    /*****************************************************
    ** Output buffer management                              **
    *****************************************************/

    /** Handle the error buffer */
    function error($text)
    {
        $this->error.=$text."<br>\n";
    }

    /** Put a begin table balise in output buffer */
    function beginTable()
    {
        $this->content.="<table border=\"0\" cellspacing=\"1\" ";
        $this->content.="cellpadding=\"2\" width=\"100%\">\n";
        $this->content.="<tr><td width=\"50%\">";
        $this->content.="</td><td width=\"50%\"></td></tr>\n";
    }

    /** Put text in output buffer */
    function insertText($text)
    {
        $this->content.="$text";
    }

    /** Put a begin table balise in output buffer */
    function closeTable()
    {
        $this->content.="</table>\n\n";
    }

    /** Put a table title ligne in output buffer */
    function insertTitle($text,$num,$br="1")
    {
        if($br) //insert une ligne vide avant
        {
            $this->content.="<tr><td colspan=\"2\">&nbsp;</td></tr>\n";
        }
        $this->content.="<tr><td class=\"title\" colspan=\"2\">\n";

        $this->content.="<table width=\"100%\"><tr><td class=\"title\">";
        $this->content.="<a name=\"$num\">\n$text&nbsp;\n";
        $this->content.="</td><td align=\"right\">";
        $this->content.="<a href=\"#top\" class=\"title\">Haut</a>\n";
        $this->content.="</td></tr></table>";

        $this->content.="</td>\n</td></tr>\n";
    }

    /** Put a table content ligne in output buffer */
    function insertLine($text,$value="")
    {
        $this->content.="<tr>\n";
        if($value)
        {
            $this->content.="<td class=\"content\">$text&nbsp;</td>\n";
            $this->content.="<td class=\"content\">$value&nbsp;</td>\n";
        }
        else
        {
            $this->content.="<td class=\"content\" colspan=\"2\">\n";
            $this->content.="$text&nbsp;</td>\n";
        }
        $this->content.="</tr>\n";
    }

    /** Put a table content ligne in output buffer */
    function insertXML($tag,$attribut="",$value)
    {

        if($attribut=="")
        {
            $this->xml.="<".strtoupper($tag).">";
        }else
        {
            $this->xml.="<".strtoupper($tag)." $attribut>";
        }
        $this->xml.=$value;
        $this->xml.="</".strtoupper($tag).">\n";
    }


    /*****************************************************
    ** Output buffer print management                     **
    *****************************************************/

    /** print a menu */
    function menu()
    {
        global $PHP_SELF;
        $text ="&nbsp;<b>: : :</b>&nbsp; Menu &nbsp;<b>: : :</b>&nbsp;";
        $text.="&nbsp;<a href=\"#a20\">En tête</a>&nbsp; | ";
        $text.="&nbsp;<a href=\"#a30\">informations</a>&nbsp; | ";
        $text.="&nbsp;<a href=\"#a40\">métas</a>&nbsp; | ";
        $text.="&nbsp;<a href=\"#a50\">liens</a>&nbsp; | ";
        $text.="&nbsp;<a href=\"#a60\">images</a>&nbsp; | ";
        $text.="&nbsp;<a href=\"#xml\">format XML</a>&nbsp; | ";
        $text.="&nbsp;<a href=\"".basename($PHP_SELF);
        $text.="?action=source&url=$this->url\">";
        $text.="sources</a>&nbsp;";
        $this->insertLine($text,"");
    }

    /** PRINT the url input form */
    function printForm()
    {
        global $PHP_SELF;
        $this->insertText("<form action=\"".basename($PHP_SELF));
        $this->insertText("\" method=\"post\">\n");
        $this->insertTitle("URL à analyser","a100");
        $text ="<input type=\"text\" name=\"url\" size=\"35\" ";
        $text.="maxlength=\"256\" value=\"";
        $text.=($this->url)?$this->url:"http://www.atypix.com";
        $text.="\">&nbsp;\nConnexion par&nbsp;";
        $text.="<input type=\"radio\" name=\"connexion\" value=\"fsock\" ";
        if($this->connexion=="fsock" || !$this->connexion)$text.="checked";
        $text.=">Socket&nbsp;\n";
        $text.="<input type=\"radio\" name=\"connexion\" value=\"fread\" ";
        if($this->connexion=="fread")$text.="checked";
        $text.=">Fichier&nbsp;\n";
        $text.="<input type=\"submit\" value=\"Analyser\" value=\"";
        $text.="$this->url\">\n";
        $text.="<input type=\"hidden\" value=\"submit\" value=\"1";
        $text.="$this->url\">\n";
        $this->insertLine($text,"");
        $this->insertText("</form>\n\n");
    }

    /** PRINT the HTML content of this analys page */
    function printPage()
    {
        print $this->content;
    }


    /** PRINT the XML content of this analys page */
    function printXml()
    {
        $this->xml = str_replace("&amp;nbsp;","",$this->xml);
        $this->insertTitle("Analyse de $this->url au format XML","xml");
        $this->insertLine(nl2br(htmlentities($this->xml)));
    }

    /** PRINT the errors of this analys */
    function printError()
    {
        print $this->error;
    }


    /*****************************************************
    ** URL and file management                              **
    *****************************************************/

    function writeXml($url)
    {
        $file = fopen($url,"w");
        if($file)
        {
            fwrite($file,$this->xml);
            fclose($file);
        }else
        {
            $this->error("Fichier impossible à ouvrir");
        }
    }

    /** FSOCK connection for distant url */
    function connectSocket($url,$port="80")
    {
        global $proxy, $proxy_port;
        if($proxy)
        {
            $fp = fsockopen($proxy,$proxy_port);
        }
        else
        {
            $array=explode("/",$url);
            $fp = fsockopen("$array[2]",$port);
        }
        if(!$fp)
        {
            $this->error("Page non accessible par socket");
        }
        else
        {
            fputs($fp, "GET $this->url/ HTTP/1.0\r\n\r\n");
            while (!feof($fp)) $this->page.=fgets($fp,64000);
            fclose($fp);
        }
    }

    /** FOPEN connection for local url */
    function connectFile($url)
    {
        if(eregi("^(http[s]?\:\/\/)(.*)",$url))
        {
            $fp = @fopen($url,"r");
            if(!$fp)
            {
                $this->error("Page non accessible par ouverture de fichier");
            }
            else
            {
                while(!feof($fp)) 
                { 
                    $page.=fread($fp, 1024); 
                } 
                $this->page=$page;
                fclose($fp);
            }
        }else
        {
            $this->error("Le fichier doit être sur un serveur Internet");
        }
    }

    /** CONNEXION management */
    function connect()
    {
        switch($this->connexion)
        {
            case "fread":
            $this->connectFile($this->url);
            break;

            case "fsock":
            $this->connectSocket($this->url);
            break;

            default:
            $this->connectSocket($this->url);
            break;
        }
    }


    /*****************************************************
    ** HTTP Header management                              **
    *****************************************************/

    /** return the statut to continue or not in case of HTTP code */
    function httpCode()
    {
        switch(1)
        {
            case($this->statut==302):     //redirection
            for ($i=1;$this->header[$i];$i++)
            {
                $location=split(":",$this->header[$i],2);
                if(eregi("location",$location[0]))
                {
                    $redirect=trim($location[1]);
                }
            }
            $url =basename($PHP_SELF);
            $url.="?url=$redirect&from=$this->url";
            header("Location: $url");
            return 1;
            break;

            case($this->statut==401):    //Authentification requise
            $this->statut.="&nbsp;Authentification requise";
            return 0;
            break;

            case($this->statut==403):    //Acces non autorise
            $this->statut.="&nbsp;Accès non autorisé";
            return 0;
            break;

            case($this->statut==404):    //Page inexistante
            $this->statut.="&nbsp;Page inexistante";
            return 0;
            break;

            case($this->statut==503):    //serveur non valide
            $this->statut.="&nbsp;URL non valide";
            return 0;
            break;

            case($this->statut >=200 && $this->statut <400): //ok
            $this->statut.="&nbsp;ok";
            return 1;
            break;

            default:
            return 0;
            break;
        }
    }

    /** parse the HTTP header and return the statut */
    function headProcess()
    {
        if($this->page!="")
        {
            if($this->connexion=="fsock")
            {
                $html_begin=strpos($this->page, "<");
                $this->header=substr($this->page, 0, $html_begin);
                $this->page=substr($this->page, $html_begin);
                $this->header=explode("\n",$this->header);
                for ($i=1;$this->header[$i];$i++)
                {
                    if(strlen($this->header[$i])<=1)
                    {
                        unset($this->header[$i]);
                    }
                }
                $statut=explode(" ",$this->header[0]);
                $this->statut=$statut[1];
                return 1;
            }
            else
            {
                return 1;
            }
        }
        else return 0;
    }


    /*****************************************************
    ** SEARCH and REGEX for item (link,img,..)              **
    *****************************************************/
    /** renvoi l'url de base avec les repertoires */
    function baseUrl($url)
    {
        $url=trim($url);
        if(substr($url,strlen($url)-1)!="/" && !$this->isLocal($url))
        {
            $url_array=explode("/",$url);
            if(ereg("\.",$url_array[count($url_array)-1])
            && count($url_array) > 3)
            {
                unset($url_array[count($url_array)-1]);
                $url=implode("/",$url_array);
            }
            $url.="/";
        }
        return $url;
    }

    /** IF LOCAL URL return 1 */
    function isLocal($url)
    {
        return !eregi("^(http[s]?|ftp|mailto):/{0,2}",trim($url));
    }

    /** return an array of balises (si lien renvoi jusqu'a "</a>") */
    function searchBalises($type)
    {
        $page=$this->page;
        while(stristr($page,"<$type"))
        {
            $page = stristr($page,"<$type");
            $end = strpos($page, ">");
            $balise = substr($page, 0, $end+1);
            $page = substr($page, $end+1);
            if($type=="a" || $type=="A")
            {

                $page_array = split("< */[aA] *>",$page,2);
                $text = eregi_replace("( *<[^>]*> *)","",$page_array[0]);
                $page = $page_array[1];
                $list[]=array("url"=>$balise,"alt"=>$text);
            }
            else
            {
                $list[]=$balise;
            }
        }
        return $list;
    }

    /** retourne la valeur d'un attibut donne dans une balise donnee */
    function getAttribut($name,$balise)
    {
        $exp1 ="(\" *([^\"]*) *\")";    // case "value"
        $exp2.="(' *([^']*) *')";        // case 'value'
        $exp3.="([^ \"'=<>]*)";            // case  value
        $exp  =" *($name) *= *($exp1|$exp2|$exp3) *";
        eregi($exp,$balise,$content);
        if($content[4] || $content[6])
        {
            $result=($content[4])?$content[4]:$content[6];
        }
        else
        {
            $result=$content[7];
        }
        return trim($result);
    }

    /** return a complete and valide url */
    function parseUrl($url)
    {
        global $PHP_SELF;
        $gurl=$this->baseUrl($this->url);
        $url=trim($url);
        switch(1)
        {
            case( ! $this->isLocal($url)):    //distant
            return $url;
            break;

            case(substr($url,0,3)=="../"):    //up ../x
            $array=explode("/",$gurl);
            unset($array[count($array)-1]);
            unset($array[count($array)-1]);
            $url2=substr($url,3);
            $url =implode("/",$array)."/".$url2;
            return $url;
            break;

            case(substr($url,0,2)=="./"):    //root ./x
            $array=explode("/",$gurl);
            $url2=substr($url,2);
            $url =$array[0]."//".$array[2]."/".$url2;
            return $url;
            break;

            case(substr($url,0,1)=="/"):    //root /x
            $array=explode("/",$gurl);
            $url2=substr($url,1);
            $url =$array[0]."//".$array[2]."/".$url2;
            return $url;
            break;

            default:
            return $gurl.$url;
            break;
        }
    }


    /*****************************************************
    ** ARRAY management                                      **
    *****************************************************/

    /** return first position of array element if exist, else -1 */
    function isIn2dArray($array,$value)
    {
        for($i=0;$array[$i];$i++)
        {
            if($value==$array[$i]["url"])
            {
                return $i;
            }
        }
        return -1;
    }

    /** insert an item in $array with $val1 and $val2, or sum it if exist */
    function insertArray($array,$val1,$val2)
    {
        if($val1)
        {
            $find=$this->isIn2dArray($array,$val1);
            if($find >=0)
            {
                $array[$find]["nb"]+=1;
                if(!$array[$find]["alt"] && $val2)
                {
                    $array[$find]["alt"]=$val2;
                }
            }
            else
            {
                $array2=array(
                "url"=>"$val1",
                "alt"=>"$val2",
                "nb"=>1);
                $array[]=$array2;
            }
        }
    }

    /** Compte les item d'un tableau avec le nombre de doublons */
    function countArray($array)
    {
        for ($i=0;$array[$i];$i++)
        {
            if($array[$i]["nb"]>1)
            {
                $nb+=$array[$i]["nb"];
            }
            else
            {
                $nb+=1;
            }
        }
        return $nb;
    }

    /** trie le tableau selon le nombre de doublons */
    function orderArray($array)
    {

    }

    /** gere l'affichage du nombre de balises */
    function showCount($type,$array)
    {
        if(count($array) >0)
        {
            $nb=$this->countArray($array);
            $text ="<b>$nb $type";
            if($nb>1)$text.="s";
            if($nb >= count($array))
            {
                $doublons = $nb - count($array);
            }
            else
            {
                $doublons="0";
            }
            $text.=" dont $doublons doublons</b>";
        }
        else
        {
            $text.="Aucun(e) $type trouvé dans cette page</b>";
        }
        $this->insertLine($text,"");
    }

    /** Print the given array of balises */
    function showArray($array,$type)
    {
        global $PHP_SELF;
        for ($i=0;$array[$i];$i++)
        {
            if($array[$i]["nb"]>1)
            {
                $nb="&nbsp;(".$array[$i]["nb"].")";
            }
            else
            {
                $nb="&nbsp;";
            }

            if($array[$i]["nb"] > 1)
            {
                $attribut="number=\"".$array[$i]["nb"]."\"";
            }else
            {
                $attribut="";
            }

            $alt=$array[$i]["alt"]."&nbsp;";
            $url=$array[$i]["url"];
            $xml=htmlentities($url);
            switch ($type)
            {
                case "meta":
                $this->insertLine($url.$nb,$alt);
                $alt=htmlentities($array[$i]["alt"]);
                if($attribut=="")
                {
                    $attribut="name=\"$xml\"";
                }else
                {
                    $attribut="$attribut name=\"$xml\"";
                }
                $this->insertXml("meta",$attribut,$alt);
                break;

                case "link":

                $link ="<a href=\"".basename($PHP_SELF)."?scan=";
                $link.=urlencode($url)."\"";
                $link.="target=\"_blank\">[scan] </a>";
                $link.="<a href=\"$url\" target=\"_blank\">";
                $link.="$url</a> $nb";
                $this->insertLine($link,$alt);
                $this->insertXml("link",$attribut,$xml);
                break;

                case "img":
                $link ="<a href=\"$url\" target=\"_blank\">";
                $link.="$url</a> $nb";
                $this->insertLine($link,$alt);
                $this->insertXml("image",$attribut,$xml);
                break;

                default:
                $this->insertLine($url.$nb,$alt);
                $this->insertXml("other",$attribut,$xml);
                break;
            }
        }
    }


    /*****************************************************
    ** HEAD page management                               **
    *****************************************************/
    function getHead()
    {
        if($this->connexion=="fsock")
        {
            $statut=explode(" ",$this->header[0]);
            $this->insertLine("Protocol",$statut[0]);
            $this->insertLine("Code etat",$statut[1]);
            $this->insertLine("Obtention de la page",$statut[2]);
            for ($i=1;$this->header[$i];$i++)
            {
                $statut=split(":",$this->header[$i],2);
                if(eregi("date",$statut[0]))
                {
                    $this->date=$statut[1];
                }
                else
                {
                    $this->insertLine($statut[0],$statut[1]);
                }
            }
        }
        else
        {
            $this->insertLine("Header","Pas de header en lecture de fichier");
        }
    }


    /************************************************************
    ** INFORMATIONS page management (indique les redirections) **
    ************************************************************/
    function getInfosPage()
    {
        global $from;
        if(eregi("<title>([^<]*)</title>",$this->page,$title))
        {
            $this->insertLine("Titre de la page",$title[1]);
        }
        $this->insertLine("Url",$this->url);
        if($from