lib/RssParser.php

   1 <?php
   2
   3 /**
   4  * Simple RSSParser Class
   5  * Based on Duncan Gough RSSParser class
   6  * Copyleft Arnaud Fontaine
   7  * Licence : GPL
   8  * See lib/plugin/RssFeed.php and lib/XmlParser.php
   9  *
  10  * The myth of RSS compatibility:
  11  *   http://diveintomark.org/archives/2004/02/04/incompatible-rss
  12  */
  13
  14 /*
  15  * This file is part of PhpWiki.
  16  *
  17  * PhpWiki is free software; you can redistribute it and/or modify
  18  * it under the terms of the GNU General Public License as published by
  19  * the Free Software Foundation; either version 2 of the License, or
  20  * (at your option) any later version.
  21  *
  22  * PhpWiki is distributed in the hope that it will be useful,
  23  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  24  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  25  * GNU General Public License for more details.
  26  *
  27  * You should have received a copy of the GNU General Public License along
  28  * with PhpWiki; if not, write to the Free Software Foundation, Inc.,
  29  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  30  */
  31
  32 /**
  33  * 2004-04-09 16:30:50 rurban:
  34  *   added fsockopen allow_url_fopen = Off workaround
  35  * 2004-04-12 20:04:12 rurban:
  36  *   fixes for IMAGE element (sf.net)
  37  * 2005-04-10 11:17:35 rurban
  38  *   certain RSS dont contain <item> tags to describe the list of <items>
  39  *     http://ws.audioscrobbler.com/rdf/ for example
  40  */
  41
  42 require_once 'lib/XmlParser.php';
  43
  44 class RSSParser
  45     extends XmlParser
  46 {
  47
  48     public $title = "";
  49     public $author = "";
  50     public $pubDate = "";
  51     public $link = "";
  52     public $description = "";
  53     public $inside_item = false;
  54     public $list_items = false;
  55     public $item = array();
  56     public $items;
  57     public $channel;
  58     public $divers = "";
  59     public $date = "";
  60
  61     function tag_open($parser, $name, $attrs = '')
  62     {
  63         global $current_tag, $current_attrs;
  64
  65         $current_tag = $name;
  66         $current_attrs = $attrs;
  67         if ($name == "ITEM")
  68             $this->inside_item = true;
  69         elseif ($name == "ITEMS")
  70             $this->list_items = true; elseif ($name == "IMAGE")
  71             $this->inside_item = true;
  72     }
  73
  74     function tag_close($parser, $tagName, $attrs = '')
  75     {
  76         global $current_tag;
  77
  78         if ($tagName == "ITEM") {
  79             if (empty($this->items)) {
  80                 $this->items = array();
  81                 $GLOBALS['rss_parser_items'] =& $this->items;
  82             } elseif (!empty($this->items[0]['link']) and $this->items[0]['title'] == '') {
  83                 // override the initial <items> list with detailed <item>'s
  84                 $this->items = array();
  85                 $GLOBALS['rss_parser_items'] =& $this->items;
  86             }
  87             $this->items[] = array("title" => $this->item['TITLE'],
  88                 "author" => $this->item['AUTHOR'],
  89                 "pubDate" => $this->item['PUBDATE'],
  90                 "description" => @$this->item['DESCRIPTION'],
  91                 "link" => $this->item['LINK']);
  92             $this->item = array("TITLE" => "",
  93                 "DESCRIPTION" => "",
  94                 "LINK" => "");
  95             $this->inside_item = false;
  96         } elseif ($tagName == "IMAGE") {
  97             $this->item = array("TITLE" => "",
  98                 "DESCRIPTION" => "",
  99                 "LINK" => "");
 100             $this->inside_item = false;
 101         } elseif ($tagName == "CHANNEL") {
 102             $this->channel = array("title" => $this->title,
 103                 "description" => $this->description,
 104                 "link" => $this->link,
 105                 "date" => $this->date,
 106                 "divers" => $this->divers);
 107             $GLOBALS['rss_parser_channel'] =& $this->channel;
 108             $this->title = "";
 109             $this->description = "";
 110             $this->link = "";
 111             $this->divers = "";
 112             $this->date = "";
 113         } elseif ($tagName == "ITEMS") {
 114             $GLOBALS['rss_parser_items'] =& $this->items;
 115             $this->item = array("TITLE" => "",
 116                 "DESCRIPTION" => "",
 117                 "LINK" => "");
 118             $this->list_items = false;
 119         }
 120     }
 121
 122     function cdata($parser, $data)
 123     {
 124         global $current_tag, $current_attrs;
 125
 126         if ($this->inside_item) {
 127             if (empty($this->item[$current_tag]))
 128                 $this->item[$current_tag] = '';
 129             if ($current_tag == 'LINK') {
 130                 if (trim($data))
 131                     $this->item[$current_tag] = trim($data);
 132             } else {
 133                 $this->item[$current_tag] .= trim($data);
 134             }
 135         } elseif ($this->list_items) {
 136             if ($current_tag == 'RDF:LI') {
 137                 // FIXME: avoid duplicates. cdata called back 4x per RDF:LI
 138                 if ($this->items[count($this->items) - 1]['link'] != @$current_attrs['RDF:RESOURCE'])
 139                     $this->items[] = array('link' => @$current_attrs['RDF:RESOURCE'],
 140                         'title' => '');
 141             }
 142         } else {
 143             switch ($current_tag) {
 144                 case "TITLE":
 145                     if (trim($data))
 146                         $this->title .= " " . trim($data);
 147                     break;
 148                 case "DESCRIPTION":
 149                     if (trim($data))
 150                         $this->description .= trim($data);
 151                     break;
 152                 case "LINK":
 153                     if (trim($data))
 154                         $this->link = trim($data);
 155                     break;
 156                 case "DC:DATE":
 157                     if (trim($data))
 158                         $this->date .= " " . trim($data);
 159                 default:
 160                     if (trim($data))
 161                         $this->divers .= " " . $current_tag . "/" . $data;
 162                     break;
 163             }
 164         }
 165     }
 166
 167     function parse($content, $is_final = true)
 168     {
 169         xml_parse($this->_parser, $content, $is_final) or
 170             trigger_error(sprintf("XML error: %s at line %d",
 171                     xml_error_string(xml_get_error_code($this->_parser)),
 172                     xml_get_current_line_number($this->_parser)),
 173                 E_USER_WARNING);
 174         //OO workaround: parser object looses its params. we have to store them in globals
 175         if ($is_final) {
 176             if (empty($this->items)) {
 177                 $this->items = @$GLOBALS['rss_parser_items'];
 178                 $this->channel = @$GLOBALS['rss_parser_channel'];
 179             }
 180             unset($GLOBALS['rss_parser_items']);
 181             unset($GLOBALS['rss_parser_channel']);
 182         }
 183     }
 184 }
 185
 186 // Local Variables:
 187 // mode: php
 188 // tab-width: 8
 189 // c-basic-offset: 4
 190 // c-hanging-comment-ender-p: nil
 191 // indent-tabs-mode: nil
 192 // End: