rdf.class.php

基于Fase 4 RDF/RSS语法分析器

        } else {
            unset($this->_display_opt);
            return false;
        }
    }
    
    /**
    * This Method allows you to define the width of the table that holds the representation of the rdf/rss file.
    *
    * @access    public
	* @author    Stefan Saasen <[e-mail]>
    * @param     int  $width  attribute width in tag <table>
	* @see		 _table_width
    */            
    function set_table_width( $width = 400 )
    {
        $this->_table_width = $width;
        return true;
    }
    
    /**
    * This Method returns an assocative Array with available Options.
    *
    * The Keys are the Name of the Options to be set. 
    * The Values are  short Description of available Options.
    *
    * @access    public
	* @author    Stefan Saasen <[e-mail]>
    * @return    array  $options
	* @see		 _display_opt
    */        
    function get_Options()
    {
        $options = array(   "image"=>"If 'image' is set to \"hidden\" no image provided by the RDF Publisher will be displayed.",
                            "channel"=>"If 'channel' is set to \"hidden\" the Channel Meta Data (i.e the Title and the short description regarding the RDF Publisher will not be displayed",
                            "textinput"=>"If set to \"hidden\" no Input Form will be displayed",
                            "build"=>"If set to \"hidden\" the Build Date (if provided) of the RDF File will not be displayed",
                            "cache_update"=>"If set to \"hidden\" the Update Date/Time of the cached Rdf File will not be displayed");
        return $options;
    }    

    /**
    * This Method returns the Content of the RDF File in one string. The String actually holds the whole XML Document.
    *
    * @access    public
	* @author    Stefan Saasen <[e-mail]>
    * @param     string $rdf    RDF File (Location)
    * @return    string XML Presentation of parsed RDF File
	* @see		 _cached_file, _remote_file, _cache_dir, _refresh, _update_cache()
    */                    
    function cache()
    {
        // checks if the cache directory already exists
        // if not, the cache directory will be created
        if(!$this->_cache_dir_ok) {
            $this->_create_cache_dir();
        }
        if($this->_use_dynamic_display == true) {
            $this->_cached_file = md5("dynamic".$this->_remote_file);
            $this->_cache_type = "normal";
        } else {
            $this->_cached_file = md5($this->_remote_file);
            $this->_cache_type = "fast";            
        }
        
        $_cache_f = $this->_cache_dir.$this->_cached_file;

        if ( (!file_exists($_cache_f)) || (filemtime($_cache_f) < $this->_refresh) || (filesize($_cache_f) == 0)) {
        // We have to parse the remote file
        $this->_use_cached_file = false;
            // --> we want to provide proper Information for Use in 
            // get_cache_update_time()
            clearstatcache();
            if($this->_use_dynamic_display == true) {
                $_rdf = @implode(" ", $this->_rdf_data()); // -> proxy
                if(!$_rdf) {
                    $this->_throw_exception( $this->_remote_file." is not available" );
                }
                $this->_parse_xRDF( $_rdf );
                $this->_update_cache( $_rdf );
                $data = $this->_output;
            } else {
                $_rdf = @implode(" ", $this->_rdf_data()); // -> proxy
                if(!$_rdf) {
                    $this->_throw_exception( $this->_remote_file." is not available" );
                }                
                $this->_parse_xRDF( $_rdf );
                $this->_update_cache( $this->_output );
                $data = $this->_output;
            }
        } else {
        // we can use the cached file
        $this->_use_cached_file = true;        
            if($this->_use_dynamic_display == true) {
                $this->_parse_xRDF( implode(" ", file($_cache_f)) );
                $data = $this->_output;
            } else {        
                $data = @implode(" ", file($_cache_f));
            }
        }
        return trim($data);
    }   // END cache()

    /**
    * This Methods creates the Cache Directory if the specified Directory does not exist.
    *
    * @access    private
	* @author    Stefan Saasen <[e-mail]>
    * @param     string $dir Path to Directory.
	* @return 	 boolean
	* @see		 _cache_dir, _cache_dir_ok
    */        
    function _create_cache_dir()
    {
        if(!@is_dir($this->_cache_dir)) {
            $arr = explode("/", $this->_cache_dir);
            $c = count($arr);
            if($arr[0]=="") {
                $path = "/";
            }
            for($i = 0;$i<$c;$i++)
            {
                if($arr[$i]!="") {
                    $path .= $arr[$i]."/";
                    if(!@is_dir($path)) {
                       if(!@mkdir($path, 0777)) { 
                            $this->_throw_exception("failed to create directory:<b>".$this->_cache_dir."</b>.<br /><br />Exception on Line: ".__LINE__);
                        return false;
                        }                    
                    }
                }
            }
            $this->_cache_dir_ok = true;
            return true;
        } else {
            $this->_cache_dir_ok = true;
            return true;    
        }
    }   // END _create_cache_dir()
    
    /**
    * This Method updates the cached RDF Files and synchronises them with their remote Counterparts.
    *
    * @access    private
	* @author    Stefan Saasen <[e-mail]>
    * @param     string $rdf    RDF File (Location)
	* @see		 _cache_dir, _cached_file, _throw_exception()
    */                    
    function _update_cache( $content = "" )
    {
             $_local = @fopen( $this->_cache_dir.$this->_cached_file, "w" );
             if(!$_local) { 
                $this->_throw_exception( "Cannot open ".$this->_cache_dir.$this->_cached_file."<br /><br />Exception at Line: ".__LINE__ ); 
                return false;
             }
             if(!fwrite( $_local, $content)) { 
                $this->_throw_exception( "Cannot write to: ".$this->_cached_file."<br /><br />Exception at Line: ".__LINE__ ); 
                return false;
             }             
             fclose( $_local ); 
             return true;
    }   // END _update_cache()
    
    /**
    * This Method returns the Date/Time of last Cache Update of the actually parsed RDF File.
	*
    * @access    public
	* @author    Stefan Saasen <[e-mail]>
	* @return 	 string Date/Time of last Update
	* @see		 _cache_dir, _cached_file
    */    
    function get_cache_update_time()
    {   
            return (file_exists($this->_cache_dir.$this->_cached_file))?date("d.m.Y H:i:s", filemtime($this->_cache_dir.$this->_cached_file)):"Cachemiss";
    }   // END get_cache_update_time()

    /**
    * This Method returns the Type of Cache that was used ('normal' or 'fast')
    *
    * @access    public
	* @author    Stefan Saasen <[e-mail]>
    * @param     string $rdf    RDF File (Location)
    * @return    string Displays RDF Content ( using _display() )
	* @see		 _remote_file, cache()
    */    
    function get_CacheType()
    {
        return $this->_cache_type;
    }
    
    /**
    * Returns true if cached file was used, otherwise false
    *
    * @access    public
	* @author    Stefan Saasen <[e-mail]>
    * @return    array  $options
	* @see		 _use_cached_file
    */        
    function is_cachedFile()
    {
        return $this->_use_cached_file;
    }
    
    /**
    * This Method deletes all the cached Files.
    *
    * Please keep in mind to use this method just as a 'manual garbage collection'
    * You should cache the rss/rdf files locally to avoid unnecessary traffic. 
    * (Both for your visitors and the Publisher)
	*
    * @access    public
	* @author    Stefan Saasen <[e-mail]>
	* @see		 _cache_dir
    */    
    function clear_cache()
    {
        $dir = dir($this->_cache_dir);
        while($file=$dir->read()) {
            if($file!="." && $file!="..")  {
                if(!@unlink($dir->path.$file)) {
                    $this->_throw_exception( 
                    "Unable to unlink ".$dir->path.$file
                    ."<br /><br />Exception at Line: ".__LINE__ ); 
                    return false;
                }
            }
        }
        $dir->close();
        return true;
    }   // END clear_cache()	

    /**
    * Cuts the String $string after $str_len and adds "... "
    *
    * @access   private
    * @param    string  $string String to be shortened
    * @param    int     $str_len length of the returned String (overall length including "... ")
    * @return   string  Cut String
    */
    function _cut_string( $string, $str_len = "30" )
    {
        if(strlen(trim($string))>$str_len) {
        $string = substr( trim($string) , 0, $str_len - 4);
        $string .= " ...";
        }
        return $string;
    }   // END _cut_string()
    
    /**
    * this Method implements simple Garbage Collection
    *
    * @access    private
	* @author    Stefan Saasen <[e-mail]>
	* @see		 _cache_dir, gc_probability, gc_maxlifetime
    */            
    function _garbage_collection()
    {
        srand((double) microtime() * 1000000);
        if (rand(1, 100) <= $this->gc_probability) {
            $dir = dir($this->_cache_dir);
            while($file=$dir->read()) {
                if($file!="." AND $file!=".." AND filemtime($dir->path.$file) <= time() - $this->_refresh )  {
                @unlink($dir->path.$file);
                }
        }
        $dir->close();
        }   // END if
    }    

    /* ==== Proxy/Auth methods ==== */
    
   /**
    * this method sets a proxy server
    *
    * @access    public
    * @param     string $phost Proxy Host
    * @param     string $pport Prox Port
    * @author    Marco Kraus <[e-mail]>
    */
    function set_proxy($phost, $pport)
    {
     $this->_use_proxy = true;

     if ($phost != "")
        $this->_phost = $phost;

     if ($pport != "")
        $this->_pport = $pport;
    }

    /**
    * this method sets a proxy server authentification
    *
    * @access    public
    * @param     string $pname Username
    * @param     string $ppaswd Password
    * @author    Marco Kraus <[e-mail]>
    */
    function set_proxy_auth( $pname, $ppasswd )
    {
     $this->_use_proxy_auth = true;

     if ($pname != "")
        $this->_pname = $pname;

     if ($ppasswd != "")
        $this->_ppasswd = $ppasswd;
    }


   /**
    * gets _remote_file into an array
    *
    * needed, cause if you use a proxy, you have to open 
    * a raw-tcp-socket to get the data
    *
    * @access    private
    * @author    Marco Kraus <[e-mail]>
    * @return array
    * @see _use_proxy, cache()
    */
    function _rdf_data()
    {
      if ( $this->_use_proxy == true )
      {
       // we need a raw socket here to connect to proxy
       $fp = fsockopen($this->_phost,$this->_pport);

       if (!$fp) {
           $this->_throw_exception( $this->_remote_file." is not available with proxy" );
       } else {
        if ( $this->_use_proxy_auth == true ) {
           fputs($fp, "GET ".$this->_remote_file." HTTP/1.0\r\n\r\n");
           } else {
           fputs($fp, "GET ".$this->_remote_file." HTTP/1.0\r\nProxy-Authorization: Basic ".base64_encode("$this->_pname:$this->_ppasswd") ."\r\n\r\n");
           }
        }


       for ( $i = 0; !feof ($fp) ; $i++)
       {
          $usable_data[$i] = "";
          $usable_data[$i] = fgets($fp,4096);

        // PARSE HEADER ---- first line has to be <?xml, second rdf or rss, and third is blank

        // strstr did not fit (ask Rasmus why), so we compare each character   
            if ( ($usable_data[$i][0] == "<" ) &&
               ($usable_data[$i][1] == "?" ) &&
               ($usable_data[$i][2] == "x" ) &&
               ($usable_data[$i][3] == "m" ) &&
               ($usable_data[$i][4] == "l" ) ) {
                    $usable_data[0] = $usable_data[$i]; // save current field
                      $i = 1; // just reset array to start
              }

        // there seems to be proxystuff after the <?xml....we delete this
            if ( (
               ($usable_data[$i][0] == "<" ) &&
               ($usable_data[$i][1] == "r" ) &&
               ($usable_data[$i][2] == "d" ) &&
               ($usable_data[$i][3] == "f" ) &&
               ($usable_data[$i][4] == ":" ) 
               )
               ||
               (
               ($usable_data[$i][0] == "<" ) &&
               ($usable_data[$i][1] == "r" ) &&
               ($usable_data[$i][2] == "s" ) &&
               ($usable_data[$i][3] == "s" )
               ) 
            ) {

                $usable_data[1] = $usable_data[$i]; // save current field
                $usable_data[2] = "\n";
                $i = 2; // just reset array to start
          }
       }

       fclose($fp);
       return $usable_data;
     } else {
        return (file($this->_remote_file));
     }
   }    // END _rdf_data()
}   // END class
?>