lib/RssParser.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id: RssParser.php,v 1.4 2004-04-18 01:11:51 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  * 2004-04-12 20:04:12 rurban:
  30  *   fixes for IMAGE element (sf.net)
  31  */
  32 class RSSParser {
  33
  34     var $title = "";
  35     var $link  = "";
  36     var $description = "";
  37     var $inside_item = false;
  38     var $item  = array();
  39     var $items = array();
  40     var $channel = array();
  41     var $divers = "";
  42     var $date = "";
  43
  44     function startElement($parser, $name, $attrs=''){
  45         global $current_tag;
  46
  47         $current_tag = $name;
  48         if ($name == "ITEM")
  49             $this->inside_item = true;
  50         elseif ($name == "IMAGE")
  51             $this->inside_item = true;
  52     } // startElement
  53
  54     function endElement($parser, $tagName, $attrs=''){
  55         global $current_tag;
  56
  57         if ($tagName == "ITEM") {
  58             $this->items[] = array("title"       => $this->item['TITLE'],
  59                                    "description" => $this->item['DESCRIPTION'],
  60                                    "link"        => $this->item['LINK']);
  61             $this->item['TITLE']       = "";
  62             $this->item['DESCRIPTION'] = "";
  63             $this->item['LINK']        = "";
  64             $this->inside_item = false;
  65         } elseif ($tagName == "IMAGE") {
  66             $this->item['TITLE']       = "";
  67             $this->item['DESCRIPTION'] = "";
  68             $this->item['LINK']        = "";
  69             $this->inside_item = false;
  70         } elseif ($tagName == "CHANNEL") {
  71             $this->channel = array("title" => $this->title,
  72                                    "description" => $this->description,
  73                                    "link" => $this->link,
  74                                    "date" => $this->date,
  75                                    "divers" => $this->divers);
  76             $this->title       = "";
  77             $this->description = "";
  78             $this->link        = "";
  79             $this->divers      = "";
  80             $this->date        = "";
  81         }
  82     }
  83
  84     function characterData($parser, $data){
  85         global $current_tag;
  86
  87         if ($this->inside_item) {
  88             if (empty($this->item[$current_tag]))
  89                 $this->item[$current_tag] = '';
  90             if ($current_tag == 'LINK') {
  91                 if (trim($data))
  92                     $this->item[$current_tag] = trim($data);
  93             } else {
  94                 $this->item[$current_tag] .= trim($data);
  95             }
  96         } else {
  97             switch ($current_tag) {
  98             case "TITLE":
  99                 if (trim($data))
 100                     $this->title .= " " . trim($data);
 101                 break;
 102             case "DESCRIPTION":
 103                 if (trim($data))
 104                     $this->description .= trim($data);
 105                 break;
 106             case "LINK":
 107                 if (trim($data))
 108                     $this->link = trim($data);
 109                 break;
 110             case "DC:DATE":
 111                 if (trim($data))
 112                     $this->date .= " " . trim($data);
 113             default:
 114                 if (trim($data))
 115                     $this->divers .= " " . $current_tag."/".$data;
 116                 break;
 117             }
 118         }
 119     } // characterData
 120
 121     function parse_results($xml_parser, $rss_parser, $file, $debug=false)   {
 122         xml_set_object($xml_parser, &$rss_parser);
 123         xml_set_element_handler($xml_parser, "startElement", "endElement");
 124         xml_set_character_data_handler($xml_parser, "characterData");
 125
 126         if (ini_get('allow_url_fopen')) {
 127             $fp = fopen("$file","r") or die("Error reading XML file, $file");
 128             while ($data = fread($fp, 4096))  {
 129                 xml_parse($xml_parser, $data, feof($fp)) or
 130                     die(sprintf("XML error: %s at line %d",
 131                                 xml_error_string(xml_get_error_code($xml_parser)),
 132                                 xml_get_current_line_number($xml_parser)));
 133             }
 134             fclose($fp);
 135         } else {
 136             // other url_fopen workarounds: curl, socket (http 80 only)
 137             require_once("lib/HttpClient.php");
 138             $bits = parse_url($file);
 139             $host = $bits['host'];
 140             $port = isset($bits['port']) ? $bits['port'] : 80;
 141             $path = isset($bits['path']) ? $bits['path'] : '/';
 142             if (isset($bits['query'])) {
 143                 $path .= '?'.$bits['query'];
 144             }
 145             $client = new HttpClient($host, $port);
 146             $client->use_gzip = false;
 147             if ($debug) $client->debug = true;
 148             if (!$client->get($path)) {
 149                 $data = false;
 150             } else {
 151                 $data = $client->getContent();
 152             }
 153             xml_parse($xml_parser, $data, true) or
 154                 die(sprintf("XML error: %s at line %d",
 155                             xml_error_string(xml_get_error_code($xml_parser)),
 156                             xml_get_current_line_number($xml_parser)));
 157         }
 158         xml_parser_free($xml_parser);
 159     }
 160 }
 161
 162 // $Log: not supported by cvs2svn $
 163
 164 // For emacs users
 165 // Local Variables:
 166 // mode: php
 167 // tab-width: 8
 168 // c-basic-offset: 4
 169 // c-hanging-comment-ender-p: nil
 170 // indent-tabs-mode: nil
 171 // End:
 172 ?>