lib/RssParser.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id: RssParser.php,v 1.3 2004-04-12 17:32:19 rurban Exp $');
   3 /**
   4  * RSSParser Class, requires the expat extension
   5  * Based on Duncan Gough RSSParser class
   6  * Copyleft Arnaud Fontaine
   7  * Licence : GPL
   8
   9  This file is part of PhpWiki.
  10
  11  PhpWiki is free software; you can redistribute it and/or modify
  12  it under the terms of the GNU General Public License as published by
  13  the Free Software Foundation; either version 2 of the License, or
  14  (at your option) any later version.
  15
  16  PhpWiki is distributed in the hope that it will be useful,
  17  but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19  GNU General Public License for more details.
  20
  21  You should have received a copy of the GNU General Public License
  22  along with PhpWiki; if not, write to the Free Software
  23  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  24  */
  25
  26 /**
  27  * 2004-04-09 16:30:50 rurban:
  28  *   added fsockopen allow_url_fopen = Off workaround
  29  */
  30 class RSSParser {
  31
  32     var $title = "";
  33     var $link  = "";
  34     var $description = "";
  35     var $inside_item = false;
  36     var $item  = array();
  37     var $items = array();
  38     var $channel = array();
  39     var $divers = "";
  40     var $date = "";
  41
  42     function startElement($parser, $name, $attrs=''){
  43         global $current_tag;
  44
  45         $current_tag = $name;
  46         if ($name == "ITEM")
  47             $this->inside_item = true;
  48         elseif ($name == "IMAGE")
  49             $this->inside_item = true;
  50     } // startElement
  51
  52     function endElement($parser, $tagName, $attrs=''){
  53         global $current_tag;
  54
  55         if ($tagName == "ITEM") {
  56             $this->items[] = array("title"       => $this->item['TITLE'],
  57                                    "description" => $this->item['DESCRIPTION'],
  58                                    "link"        => $this->item['LINK']);
  59             $this->item['TITLE']       = "";
  60             $this->item['DESCRIPTION'] = "";
  61             $this->item['LINK']        = "";
  62             $this->inside_item = false;
  63         } elseif ($tagName == "IMAGE") {
  64             $this->item['TITLE']       = "";
  65             $this->item['DESCRIPTION'] = "";
  66             $this->item['LINK']        = "";
  67             $this->inside_item = false;
  68         } elseif ($tagName == "CHANNEL") {
  69             $this->channel = array("title" => $this->title,
  70                                    "description" => $this->description,
  71                                    "link" => $this->link,
  72                                    "date" => $this->date,
  73                                    "divers" => $this->divers);
  74             $this->title       = "";
  75             $this->description = "";
  76             $this->link        = "";
  77             $this->divers      = "";
  78             $this->date        = "";
  79         }
  80     }
  81
  82     function characterData($parser, $data){
  83         global $current_tag;
  84
  85         if ($this->inside_item) {
  86             if (empty($this->item[$current_tag]))
  87                 $this->item[$current_tag] = '';
  88             if ($current_tag == 'LINK') {
  89                 if (trim($data))
  90                     $this->item[$current_tag] = trim($data);
  91             } else {
  92                 $this->item[$current_tag] .= trim($data);
  93             }
  94         } else {
  95             switch ($current_tag) {
  96             case "TITLE":
  97                 if (trim($data))
  98                     $this->title .= " " . trim($data);
  99                 break;
 100             case "DESCRIPTION":
 101                 if (trim($data))
 102                     $this->description .= trim($data);
 103                 break;
 104             case "LINK":
 105                 if (trim($data))
 106                     $this->link = trim($data);
 107                 break;
 108             case "DC:DATE":
 109                 if (trim($data))
 110                     $this->date .= " " . trim($data);
 111             default:
 112                 if (trim($data))
 113                     $this->divers .= " " . $current_tag."/".$data;
 114                 break;
 115             }
 116         }
 117     } // characterData
 118
 119     function parse_results($xml_parser, $rss_parser, $file, $debug=false)   {
 120         xml_set_object($xml_parser, &$rss_parser);
 121         xml_set_element_handler($xml_parser, "startElement", "endElement");
 122         xml_set_character_data_handler($xml_parser, "characterData");
 123
 124         if (ini_get('allow_url_fopen')) {
 125             $fp = fopen("$file","r") or die("Error reading XML file, $file");
 126             while ($data = fread($fp, 4096))  {
 127                 xml_parse($xml_parser, $data, feof($fp)) or
 128                     die(sprintf("XML error: %s at line %d",
 129                                 xml_error_string(xml_get_error_code($xml_parser)),
 130                                 xml_get_current_line_number($xml_parser)));
 131             }
 132             fclose($fp);
 133         } else {
 134             // other url_fopen workarounds: curl, socket (http 80 only)
 135             require_once("lib/HttpClient.php");
 136             $bits = parse_url($file);
 137             $host = $bits['host'];
 138             $port = isset($bits['port']) ? $bits['port'] : 80;
 139             $path = isset($bits['path']) ? $bits['path'] : '/';
 140             if (isset($bits['query'])) {
 141                 $path .= '?'.$bits['query'];
 142             }
 143             $client = new HttpClient($host, $port);
 144             $client->use_gzip = false;
 145             if ($debug) $client->debug = true;
 146             if (!$client->get($path)) {
 147                 $data = false;
 148             } else {
 149                 $data = $client->getContent();
 150             }
 151             xml_parse($xml_parser, $data, true) or
 152                 die(sprintf("XML error: %s at line %d",
 153                             xml_error_string(xml_get_error_code($xml_parser)),
 154                             xml_get_current_line_number($xml_parser)));
 155         }
 156         xml_parser_free($xml_parser);
 157     }
 158 }
 159
 160 // For emacs users
 161 // Local Variables:
 162 // mode: php
 163 // tab-width: 8
 164 // c-basic-offset: 4
 165 // c-hanging-comment-ender-p: nil
 166 // indent-tabs-mode: nil
 167 // End:
 168 ?>