suche script liefert doppelte ergebnisse

<?php
// Domain-Name für die Anzeige in den Resultaten



$DomainName = "http://".GetParam("SERVER_NAME", "S");



// Root-Verzeichnis für den Zugriff auf die Dateien am Server

// Beispiel: $RootDir=dirname("/htdocs");

$RootDir = dirname(GetParam("SCRIPT_FILENAME", "S"));



// Dateiname für die Protokollierung der Suchbegriffe

// (kein Dateiname zum deaktivieren der Protokollierung)

// z.B. "../../files/logs/search_words.log"

$SearchWordLog = "";



// Länge der Textfragmente um die Fundstellen (in Zeichen)

$Found_Piece_Len = 50;



// Erlaubtes Verzeichnis

// Z.B.: $AllowedDirs = $RootDir."/content";

$AllowedDir = array($RootDir."",$RootDir."");



// Erlaubte Erweiterungen als Array, getrennt mit einem Beistrich

// Z.B.: $AllowedExts=array(".php",".php3",".php4",".htm",".html",".ihtml",".shtm",".shtml",".txt");

$AllowedExts = array(".php",".html",".shtml");



// Deutsche Umlaute dekodieren (dadurch wird beispielsweise "&auml;" zu "ä")

$ActivateUmlaut = true;



// Links zu den gefundenen Seiten in einem neuen Tab/Fenster öffnen

$LinkTargetBlank = false;



// *****************************************************************************



?>





<!--[if IE]><style type="text/css"> body {word-wrap: break-word;}</style><![endif]-->





        <!-- Search form -->

        <?php

  $SearchTerm = FormatSearchString(stripslashes(GetParam("q", "P")));

  if (!$SearchTerm) $SearchTerm = FormatSearchString(stripslashes(GetParam("q", "G")));

?>









<?php
if($SearchTerm){
  // Protokollierung der Suchbegriffe
  if(file_exists($SearchWordLog)) {
    $fp=@fopen($SearchWordLog,"a");
    if($fp) {
      flock($fp,2);
      fputs($fp,$SearchTerm."\r\n",256);
      flock($fp,3);
      fclose($fp);
    }
  }
?>
<?
  echo "<h1>Resultate</h1>\n in mebra-mbh.de\n <p>";
 $files=ReadDirs($AllowedDir,$AllowedExts);
  $ResultCount=0;
  if($files && $SearchTerm){
    foreach($files as $f){
      if(SearchFile($f,$SearchTerm)){
        $f1 = str_replace(".php", ".php", $f);
        $fn=$f;
        if(!strpos($f1, "test") && !strpos($f1, "bak") && !strpos($f1, "old") && !strpos($f1, "search") && !strpos($f1, "suche")) {
        if(substr($f,0,strlen($RootDir))==$RootDir) $fn=$DomainName.substr($f1,strlen($RootDir));
        $ResultCount++;
        echo $ResultCount.". ";
        echo '<a href="'.$fn.'"';
        if ($LinkTargetBlank) echo ' target="_blank"';
        echo "><b>".$Site_Title."</b></a><br>\n";
        if($Meta_Description) echo $Meta_Description."<br>\n";
        echo "<span class=small><span class=grey>".$Site_Content."</span></span><br>\n";
        echo "<span class=grey>Datei:</span> <span class=green>".$fn."</span>";
        echo " - ";
        echo "<span class=grey>Gr&ouml;&szlig;e:</span> <span class=green>".round(filesize($f)/1024,2)." KB</span>";
        echo "<br>";
        echo "<br>\n";
        }
      }
    }
    clearstatcache();
  }
  echo "Es wurden <b>".$ResultCount."</b> Seiten gefunden.<br>\n";
}
function SearchFile($url,$search){
  global $Found_Piece_Len;
  global $ActivateUmlaut;
  global $RootDir;
  global $DomainName;
  global $Site_Title;
  global $Meta_Title;
  global $Site_Content;
  global $Meta_Description;
  global $Meta_Robots;
  $Site_Title="";
  $Meta_Title="";
  $Meta_Keywords="";
  $Site_Content="";
  $Meta_Description="";
  $Meta_Robots="";
  // *** Meta-Angaben ermitteln ***
  $gmtarray=get_meta_tags($url);
  while(list($key,$val)=each($gmtarray)){
    switch(strtolower($key)){
      case "title": $Meta_Title=$val; break;
      case "keywords": $Meta_Keywords=$val; break;
      case "description": $Meta_Description=$val; break;
      case "robots": $Meta_Robots=strtolower($val); break;
      case "revisit": $Meta_Revisit=strtolower($val); break;
      case "revisit-after": $Meta_RevisitAfter=strtolower($val); break;
    }
  }
  $Meta_Description="";
  // *** Dateiinhalt einlesen (bzw. Ausgabe bei PHP) ***
  $fp=@fopen($url,"r");
  if(!$fp) return false;
  $content="";
  while(!feof($fp)){
    $content.=fgets($fp,10240);
  }
  fclose($fp);
  $content = preg_replace("/<\?.*?\?>/s", "", $content);
  $content = preg_replace("/<script.*?\/script>/s", "", $content);
  $content = $Meta_Title." ".$Meta_Keywords." ".$Meta_Description." ".trim($content);
  // *** Seitentitel ermitteln ***
  $Site_Title = GetSiteTitle($content);
  if(!$Site_Title) $Site_Title = $Meta_Title;
  if(!$Site_Title) $Site_Title = basename($url);
  if (substr($content, 0, strlen($Site_Title)) == $Site_Title) $content = substr($content, strlen($Site_Title));
  $content = strip_tags($content);
  $content = str_replace("\n", " ", $content);
  $content = str_replace("\r", "", $content);
  $sc = " ".trim($content);
  while(strpos($sc,"  ")){
    $sc=str_replace("  "," ",$sc);
  }
 $content=$sc;
  // Deutsche Umlaute konvertieren
  if($ActivateUmlaut){
    $content=str_replace("&auml;","ä",$content);
    $content=str_replace("&ouml;","ö",$content);
    $content=str_replace("&uuml;","ü",$content);
    $content=str_replace("&Auml;","Ä",$content);
    $content=str_replace("&Ouml;","Ö",$content);
    $content=str_replace("&Uuml;","Ü",$content);
    $content=str_replace("&szlig;","ß",$content);
  }
  $content=str_replace("&euro;","€",$content);
  $content=str_replace("&nbsp;"," ",$content);
  // *** Suchen ***
  $found=false;
  $a=explode(" ",strtolower($search));
  $lowcontent=strtolower($content);
  $result_text="";
  foreach($a as $arg){
    $p0=strpos($lowcontent,$arg);
    if($p0>0){
      $p1=$Found_Piece_Len;
      $p2=$Found_Piece_Len;
      if (($p0-$p1) < 0) $p1=$p0;
      $result_text.="...".substr($content,$p0-$p1,$p1)."<b class=red>";
      $result_text.=substr($content,$p0,strlen($arg))."</b>".substr($content,$p0+strlen($arg),$p2);
      $found=true;
    }else{
      $found=false;
      break;
    }
  }
  if(!$found) return false;
  if($result_text) $result_text.="...";
  $Site_Content=$result_text;
  return true;
}
function FormatSearchString($search){
  $chars=".:-_,;!§$%&/()=#+*~'?\[|]{^°}²³µ@€ äöüß1234567890abcdefghijklmnopqrstuvwxyz";
  $s="";
  for($i=0;$i<strlen($search);$i++){
    $a=substr($search,$i,1);
    if(stristr($chars,$a)) $s.=$a;
  }
  while(strpos($s,"  ")){
    $s=str_replace("  "," ",$s);
  }
  return trim($s);
}
function ReadDirs($d1,$ExtArray){
  $fileArray=array();
  foreach($d1 as $d){
  if($hDir = opendir($d)){
    while($file=readdir($hDir)){
      if(!is_dir($d."/".$file)){
        // *** .*-Dateien (zB .htaccess) ignorieren ***
        if(substr($file,0,1)!="."){
                      foreach($ExtArray as $ext){
                        if(substr(strtolower($file),strlen($file)-strlen($ext),strlen($ext))==strtolower($ext)){
                                array_push($fileArray,$d."/".$file);
                                continue;
                        }
                      }
              }
      }
    }
    closedir($hDir);
  }
  }
  return $fileArray;
}
function GetSiteTitle($content)
{
  $p1=strpos(strtolower($content),"<title>");
  if(!$p1) return false;
  $p2=strpos(strtolower($content),"</title>",$p1);
  if(!$p2) return false;
  return trim(substr($content,$p1+7,$p2-$p1-7));
}
function GetParam($ParamName, $Method = "P", $DefaultValue = "") {
  if ($Method == "P") {
    if (isset($_POST[$ParamName])) return $_POST[$ParamName]; else return $DefaultValue;
  } else if ($Method == "G") {
    if (isset($_GET[$ParamName])) return $_GET[$ParamName]; else return $DefaultValue;
  } else if ($Method == "S") {
    if (isset($_SERVER[$ParamName])) return $_SERVER[$ParamName]; else return $DefaultValue;
  }
}
?>

<?php
class Extensions {
    public function __construct() {
        $this->Allowed = ['html'];
    }
    public function Add( $extension ) : Extensions {
        if ( !in_array($extension, $this->Allowed) )
            $this->Allowed[] = $extension;
        return $this;
    }
    public function Remove( $extension ) : Extensions {
        if ( in_array($extension, $this->Allowed) ) {
            $iIndex = array_search( $extension, $this->Allowed );
            unset( $this->Allowed[$iIndex] );
            $this->Allowed = array_values( $this->Allowed );
        }
        return $this;
    }
    public function AddRange( $extensions ) : Extensions {
        if ( !is_array($extensions) )
            $extensions = [$extensions];
        $this->Allowed = array_merge( $this->Allowed, array_diff($extensions, $this->Allowed) );
        return $this;
    }
    public function IsValid( $extension ) : bool {
        return in_array( $extension, $this->Allowed );
    }
    public function Clear() : Extensions {
        $this->Allowed = [];
        return $this;
    }
} 
?>

<?php
class FileSearch {
    public function __construct( $base_dir = null ) {
        $this->BaseDir = is_null( $base_dir ) ? $base_dir : __DIR__;
        $this->Initialize();
    }
    public function CollectWhere( $search_term ) : FileSearch {
        if ( $this->Iterator == null )
            return $this;
        $this->SearchTerm = $search_term;
        while ( $this->Iterator->valid() ) {
            $this->Process( $this->Iterator->current() );
            $this->Iterator->next();
        }
        return $this;
    }
    public function GetData() {
        return $this->Results;
    }
    private function Initialize() {
        $this->AllowedExtensions = new Extensions;
        $this->AllowedExtensions->Clear()->AddRange( ['html', 'shtml', 'php'] );
        $this->Results = [];
        $this->Iterator = new RecursiveIteratorIterator(
                new RecursiveDirectoryIterator( $this->BaseDir )
            );
        $this->Iterator->rewind();
    }
    private function Process( $file_info ) {
        if ( !$file_info->isFile() || !$this->AllowedExtensions->IsValid($file_info->getExtension()) )
            return;
        $sFilePath = $file_info->getPath() . '/' . $file_info->getFileName();
        $aValues = $this->FetchMetaValues( file_get_contents($sFilePath) );
        if ( $aValues == null )
            return;
        $this->Results[$sFilePath][] = array_unique( $aValues );
    }
    private function FetchMetaValues( $content ) {
        $bHasResults = preg_match_all( '/<meta.*content\s*=\s*"(?P<found>[^"]+)/im', $content, $aMatches );
        if ( empty($aMatches['found']) )
            return null; 
        return array_filter( $aMatches['found'], function ($data) {
                return stripos( $data, $this->SearchTerm ) !== false;
            });
    }
} 
?>