lib/RssParser.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id: RssParser.php,v 1.11 2005-04-10 10:24:58 rurban Exp $');
   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
  28  along with PhpWiki; if not, write to the Free Software
  29  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  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     var $title = "";
  48     var $link  = "";
  49     var $description = "";
  50     var $inside_item = false;
  51     var $list_items = false;
  52     var $item  = array();
  53     var $items;
  54     var $channel;
  55     var $divers = "";
  56     var $date = "";
  57
  58     function tag_open($parser, $name, $attrs=''){
  59         global $current_tag, $current_attrs;
  60
  61         $current_tag = $name;
  62         $current_attrs = $attrs;
  63         if ($name == "ITEM")
  64             $this->inside_item = true;
  65         elseif ($name == "ITEMS")
  66             $this->list_items = true;
  67         elseif ($name == "IMAGE")
  68             $this->inside_item = true;
  69     }
  70
  71     function tag_close($parser, $tagName, $attrs=''){
  72         global $current_tag;
  73
  74         if ($tagName == "ITEM") {
  75             if (empty($this->items)) {
  76                 $this->items = array();
  77                 $GLOBALS['rss_parser_items'] =& $this->items;
  78             } elseif (!empty($this->items[0]['link']) and $this->items[0]['title'] == '') {
  79                 // override the initial <items> list with detailed <item>'s
  80                 $this->items = array();
  81                 $GLOBALS['rss_parser_items'] =& $this->items;
  82             }
  83             $this->items[] = array("title"       => $this->item['TITLE'],
  84                                    "description" => @$this->item['DESCRIPTION'],
  85                                    "link"        => $this->item['LINK']);
  86             $this->item = array("TITLE"       => "",
  87                                 "DESCRIPTION" => "",
  88                                 "LINK"        => "");
  89             $this->inside_item = false;
  90         } elseif ($tagName == "IMAGE") {
  91             $this->item = array("TITLE"       => "",
  92                                 "DESCRIPTION" => "",
  93                                 "LINK"        => "");
  94             $this->inside_item = false;
  95         } elseif ($tagName == "CHANNEL") {
  96             $this->channel = array("title" => $this->title,
  97                                    "description" => $this->description,
  98                                    "link" => $this->link,
  99                                    "date" => $this->date,
 100                                    "divers" => $this->divers);
 101             $GLOBALS['rss_parser_channel'] =& $this->channel;
 102             $this->title       = "";
 103             $this->description = "";
 104             $this->link        = "";
 105             $this->divers      = "";
 106             $this->date        = "";
 107         } elseif ($tagName == "ITEMS") {
 108             $GLOBALS['rss_parser_items'] =& $this->items;
 109             $this->item = array("TITLE"       => "",
 110                                 "DESCRIPTION" => "",
 111                                 "LINK"        => "");
 112             $this->list_items = false;
 113         }
 114     }
 115
 116     function cdata($parser, $data){
 117         global $current_tag, $current_attrs;
 118
 119         if ($this->inside_item) {
 120             if (empty($this->item[$current_tag]))
 121                 $this->item[$current_tag] = '';
 122             if ($current_tag == 'LINK') {
 123                 if (trim($data))
 124                     $this->item[$current_tag] = trim($data);
 125             } else {
 126                 $this->item[$current_tag] .= trim($data);
 127             }
 128         } elseif ($this->list_items) {
 129             if ($current_tag == 'RDF:LI') {
 130                 // FIXME: avoid duplicates. cdata called back 4x per RDF:LI
 131                 if ($this->items[count($this->items)-1]['link'] != @$current_attrs['RDF:RESOURCE'])
 132                     $this->items[] = array('link' => @$current_attrs['RDF:RESOURCE'],
 133                                            'title' => '');
 134             }
 135         } else {
 136             switch ($current_tag) {
 137             case "TITLE":
 138                 if (trim($data))
 139                     $this->title .= " " . trim($data);
 140                 break;
 141             case "DESCRIPTION":
 142                 if (trim($data))
 143                     $this->description .= trim($data);
 144                 break;
 145             case "LINK":
 146                 if (trim($data))
 147                     $this->link = trim($data);
 148                 break;
 149             case "DC:DATE":
 150                 if (trim($data))
 151                     $this->date .= " " . trim($data);
 152             default:
 153                 if (trim($data))
 154                     $this->divers .= " " . $current_tag."/".$data;
 155                 break;
 156             }
 157         }
 158     }
 159
 160     function parse($content, $is_final = true) {
 161         xml_parse($this->_parser, $content, $is_final) or
 162             trigger_error(sprintf("XML error: %s at line %d",
 163                                   xml_error_string(xml_get_error_code($this->_parser)),
 164                                   xml_get_current_line_number($this->_parser)),
 165                           E_USER_WARNING);
 166         //OO workaround: parser object looses its params. we have to store them in globals
 167         if ($is_final) {
 168             if (empty($this->items)) {
 169                 $this->items   = @$GLOBALS['rss_parser_items'];
 170                 $this->channel = @$GLOBALS['rss_parser_channel'];
 171             }
 172             unset($GLOBALS['rss_parser_items']);
 173             unset($GLOBALS['rss_parser_channel']);
 174         }
 175     }
 176 }
 177
 178 // $Log: not supported by cvs2svn $
 179 // Revision 1.10  2005/01/22 11:45:09  rurban
 180 // docs
 181 //
 182 // Revision 1.9  2004/06/08 21:12:02  rurban
 183 // is_final fix for incremental parsing
 184 //
 185 // Revision 1.8  2004/06/08 21:03:20  rurban
 186 // updated RssParser for XmlParser quirks (store parser object params in globals)
 187 //
 188 // Revision 1.7  2004/05/24 17:31:31  rurban
 189 // new XmlParser and HtmlParser, RssParser based on that.
 190 //
 191 // Revision 1.6  2004/05/18 16:18:36  rurban
 192 // AutoSplit at subpage seperators
 193 // RssFeed stability fix for empty feeds or broken connections
 194 //
 195 // Revision 1.5  2004/04/26 20:44:34  rurban
 196 // locking table specific for better databases
 197 //
 198 // Revision 1.4  2004/04/18 01:11:51  rurban
 199 // more numeric pagename fixes.
 200 // fixed action=upload with merge conflict warnings.
 201 // charset changed from constant to global (dynamic utf-8 switching)
 202 //
 203
 204 // For emacs users
 205 // Local Variables:
 206 // mode: php
 207 // tab-width: 8
 208 // c-basic-offset: 4
 209 // c-hanging-comment-ender-p: nil
 210 // indent-tabs-mode: nil
 211 // End:
 212 ?>