lib/RssParser.php

   1 <?php // -*-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     var $title = "";
  48     var $author = "";
  49     var $pubDate = "";
  50     var $link  = "";
  51     var $description = "";
  52     var $inside_item = false;
  53     var $list_items = false;
  54     var $item  = array();
  55     var $items;
  56     var $channel;
  57     var $divers = "";
  58     var $date = "";
  59
  60     function tag_open($parser, $name, $attrs=''){
  61         global $current_tag, $current_attrs;
  62
  63         $current_tag = $name;
  64         $current_attrs = $attrs;
  65         if ($name == "ITEM")
  66             $this->inside_item = true;
  67         elseif ($name == "ITEMS")
  68             $this->list_items = true;
  69         elseif ($name == "IMAGE")
  70             $this->inside_item = true;
  71     }
  72
  73     function tag_close($parser, $tagName, $attrs=''){
  74         global $current_tag;
  75
  76         if ($tagName == "ITEM") {
  77             if (empty($this->items)) {
  78                 $this->items = array();
  79                 $GLOBALS['rss_parser_items'] =& $this->items;
  80             } elseif (!empty($this->items[0]['link']) and $this->items[0]['title'] == '') {
  81                 // override the initial <items> list with detailed <item>'s
  82                 $this->items = array();
  83                 $GLOBALS['rss_parser_items'] =& $this->items;
  84             }
  85             $this->items[] = array("title"       => $this->item['TITLE'],
  86                                    "author"      => $this->item['AUTHOR'],
  87                                    "pubDate"     => $this->item['PUBDATE'],
  88                                    "description" => @$this->item['DESCRIPTION'],
  89                                    "link"        => $this->item['LINK']);
  90             $this->item = array("TITLE"       => "",
  91                                 "DESCRIPTION" => "",
  92                                 "LINK"        => "");
  93             $this->inside_item = false;
  94         } elseif ($tagName == "IMAGE") {
  95             $this->item = array("TITLE"       => "",
  96                                 "DESCRIPTION" => "",
  97                                 "LINK"        => "");
  98             $this->inside_item = false;
  99         } elseif ($tagName == "CHANNEL") {
 100             $this->channel = array("title" => $this->title,
 101                                    "description" => $this->description,
 102                                    "link" => $this->link,
 103                                    "date" => $this->date,
 104                                    "divers" => $this->divers);
 105             $GLOBALS['rss_parser_channel'] =& $this->channel;
 106             $this->title       = "";
 107             $this->description = "";
 108             $this->link        = "";
 109             $this->divers      = "";
 110             $this->date        = "";
 111         } elseif ($tagName == "ITEMS") {
 112             $GLOBALS['rss_parser_items'] =& $this->items;
 113             $this->item = array("TITLE"       => "",
 114                                 "DESCRIPTION" => "",
 115                                 "LINK"        => "");
 116             $this->list_items = false;
 117         }
 118     }
 119
 120     function cdata($parser, $data){
 121         global $current_tag, $current_attrs;
 122
 123         if ($this->inside_item) {
 124             if (empty($this->item[$current_tag]))
 125                 $this->item[$current_tag] = '';
 126             if ($current_tag == 'LINK') {
 127                 if (trim($data))
 128                     $this->item[$current_tag] = trim($data);
 129             } else {
 130                 $this->item[$current_tag] .= trim($data);
 131             }
 132         } elseif ($this->list_items) {
 133             if ($current_tag == 'RDF:LI') {
 134                 // FIXME: avoid duplicates. cdata called back 4x per RDF:LI
 135                 if ($this->items[count($this->items)-1]['link'] != @$current_attrs['RDF:RESOURCE'])
 136                     $this->items[] = array('link' => @$current_attrs['RDF:RESOURCE'],
 137                                            'title' => '');
 138             }
 139         } else {
 140             switch ($current_tag) {
 141             case "TITLE":
 142                 if (trim($data))
 143                     $this->title .= " " . trim($data);
 144                 break;
 145             case "DESCRIPTION":
 146                 if (trim($data))
 147                     $this->description .= trim($data);
 148                 break;
 149             case "LINK":
 150                 if (trim($data))
 151                     $this->link = trim($data);
 152                 break;
 153             case "DC:DATE":
 154                 if (trim($data))
 155                     $this->date .= " " . trim($data);
 156             default:
 157                 if (trim($data))
 158                     $this->divers .= " " . $current_tag."/".$data;
 159                 break;
 160             }
 161         }
 162     }
 163
 164     function parse($content, $is_final = true) {
 165         xml_parse($this->_parser, $content, $is_final) or
 166             trigger_error(sprintf("XML error: %s at line %d",
 167                                   xml_error_string(xml_get_error_code($this->_parser)),
 168                                   xml_get_current_line_number($this->_parser)),
 169                           E_USER_WARNING);
 170         //OO workaround: parser object looses its params. we have to store them in globals
 171         if ($is_final) {
 172             if (empty($this->items)) {
 173                 $this->items   = @$GLOBALS['rss_parser_items'];
 174                 $this->channel = @$GLOBALS['rss_parser_channel'];
 175             }
 176             unset($GLOBALS['rss_parser_items']);
 177             unset($GLOBALS['rss_parser_channel']);
 178         }
 179     }
 180 }
 181
 182 // Local Variables:
 183 // mode: php
 184 // tab-width: 8
 185 // c-basic-offset: 4
 186 // c-hanging-comment-ender-p: nil
 187 // indent-tabs-mode: nil
 188 // End: