Open Journal Systems: classes/search/SearchHTMLParser.inc.php Source File

00001 <?php
00002 
00015 // $Id: SearchHTMLParser.inc.php,v 1.11 2008/07/01 01:16:10 asmecher Exp $
00016 
00017 
00018 import('search.SearchFileParser');
00019 import('core.String');
00020 
00021 class SearchHTMLParser extends SearchFileParser {
00022 
00023    function doRead() {
00024       // strip HTML tags from the read line
00025       $line = fgetss($this->fp, 4096);
00026 
00027       // convert HTML entities to valid UTF-8 characters
00028       $line = String::html2utf($line);
00029       
00030       // slightly (~10%) faster than above, but not quite as accurate, and requires html_entity_decode()
00031 //    $line = html_entity_decode($line, ENT_COMPAT, strtoupper(Config::getVar('i18n', 'client_charset')));
00032 
00033       return $line;
00034    }
00035 
00036 }
00037 
00038 ?>