framework/main/classes/parser/xml/class_XmlParser.php

   1 <?php
   2 // Own namespace
   3 namespace Org\Mxchange\CoreFramework\Parser\Xml;
   4
   5 // Import framework stuff
   6 use Org\Mxchange\CoreFramework\Generic\FrameworkInterface;
   7 use Org\Mxchange\CoreFramework\Middleware\Debug\DebugMiddleware;
   8 use Org\Mxchange\CoreFramework\Parser\BaseParser;
   9 use Org\Mxchange\CoreFramework\Parser\Parseable;
  10 use Org\Mxchange\CoreFramework\Template\CompileableTemplate;
  11
  12 // Import SPL stuff
  13 use \InvalidArgumentException;
  14
  15 /**
  16  * A Xml Parser class
  17  *
  18  * @author              Roland Haeder <webmaster@shipsimu.org>
  19  * @version             0.0.0
  20  * @copyright   Copyright (c) 2007, 2008 Roland Haeder, 2009 - 2023 Core Developer Team
  21  * @license             GNU GPL 3.0 or any newer version
  22  * @link                http://www.shipsimu.org
  23  *
  24  * This program is free software: you can redistribute it and/or modify
  25  * it under the terms of the GNU General Public License as published by
  26  * the Free Software Foundation, either version 3 of the License, or
  27  * (at your option) any later version.
  28  *
  29  * This program is distributed in the hope that it will be useful,
  30  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  31  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  32  * GNU General Public License for more details.
  33  *
  34  * You should have received a copy of the GNU General Public License
  35  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  36  */
  37 class XmlParser extends BaseParser implements Parseable {
  38         /**
  39          * Protected constructor
  40          *
  41          * @return      void
  42          */
  43         private function __construct () {
  44                 // Call parent constructor
  45                 parent::__construct(__CLASS__);
  46         }
  47
  48         /**
  49          * Creates an instance of the class Parser and prepares it for usage
  50          *
  51          * @param       $templateInstance       A CompileableTemplate instance
  52          * @return      $parserInstance         An instance of this parser
  53          */
  54         public static final function createXmlParser (CompileableTemplate $templateInstance) {
  55                 // Get a new instance
  56                 $parserInstance = new XmlParser();
  57
  58                 // Set the template instance
  59                 $parserInstance->setTemplateInstance($templateInstance);
  60
  61                 // Return the prepared instance
  62                 return $parserInstance;
  63         }
  64
  65         /**
  66          * Parses the given XML content
  67          *
  68          * @param       $content        Valid XML content
  69          * @return      void
  70          * @throws      XmlParserException      If an XML error was found
  71          */
  72         public function parseXmlContent (string $content) {
  73                 // Convert all to UTF8
  74                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('XML-PARSER: content()=%d - CALLED!', strlen($content)));
  75                 if (empty($content)) {
  76                         // No empty content
  77                         throw new InvalidArgumentException('content is empty', FrameworkInterface::EXCEPTION_INVALID_ARGUMENT);
  78                 } elseif (function_exists('recode')) {
  79                         // Recode found, so use it
  80                         $content = recode('html..utf8', $content);
  81                 } elseif (function_exists('mb_convert_encoding')) {
  82                         // Use mb_convert_encoding()
  83                         $content = mb_convert_encoding($content, 'UTF-8', 'auto');
  84                 } else {
  85                         // @TODO We need to find a fallback solution here
  86                         DebugMiddleware::getSelfInstance()->partialStub('Cannot find recode/mbstring extension!');
  87                 }
  88
  89                 // Get an XML parser
  90                 $xmlParser = xml_parser_create();
  91
  92                 // Force case-folding to on
  93                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('XML-PARSER: Created xmlParser=%s ...', $xmlParser));
  94                 xml_parser_set_option($xmlParser, XML_OPTION_CASE_FOLDING, TRUE);
  95
  96                 // Set UTF-8
  97                 xml_parser_set_option($xmlParser, XML_OPTION_TARGET_ENCODING, 'UTF-8');
  98
  99                 // Set object
 100                 xml_set_object($xmlParser, $this->getTemplateInstance());
 101
 102                 // Set handler call-backs
 103                 xml_set_element_handler($xmlParser, 'startElement', 'finishElement');
 104                 xml_set_character_data_handler($xmlParser, 'characterHandler');
 105
 106                 // Now parse the XML tree
 107                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('XML-PARSER: Invoking xml_parse(%s, content()=%d) ...', $xmlParser, strlen($content)));
 108                 if (!xml_parse($xmlParser, $content)) {
 109                         // Error found in XML
 110                         //* DEBUG: */ exit(__METHOD__ . ':<pre>'.htmlentities($content).'</pre>');
 111                         throw new XmlParserException([$this, $xmlParser], Parseable::EXCEPTION_XML_PARSER_ERROR);
 112                 }
 113
 114                 // Free the parser
 115                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('XML-PARSER: Freeing xmlParser=%s ...', $xmlParser));
 116                 xml_parser_free($xmlParser);
 117
 118                 // Trace message
 119                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('XML-PARSER: EXIT!');
 120         }
 121
 122 }