library/HTMLPurifier/Lexer/PEARSax3.php

   1 <?php
   2
   3 /**
   4  * Proof-of-concept lexer that uses the PEAR package XML_HTMLSax3 to parse HTML.
   5  *
   6  * PEAR, not suprisingly, also has a SAX parser for HTML.  I don't know
   7  * very much about implementation, but it's fairly well written.  However, that
   8  * abstraction comes at a price: performance. You need to have it installed,
   9  * and if the API changes, it might break our adapter. Not sure whether or not
  10  * it's UTF-8 aware, but it has some entity parsing trouble (in all areas,
  11  * text and attributes).
  12  *
  13  * Quite personally, I don't recommend using the PEAR class, and the defaults
  14  * don't use it. The unit tests do perform the tests on the SAX parser too, but
  15  * whatever it does for poorly formed HTML is up to it.
  16  *
  17  * @todo Generalize so that XML_HTMLSax is also supported.
  18  *
  19  * @warning Entity-resolution inside attributes is broken.
  20  */
  21
  22 class HTMLPurifier_Lexer_PEARSax3 extends HTMLPurifier_Lexer
  23 {
  24
  25     /**
  26      * Internal accumulator array for SAX parsers.
  27      */
  28     protected $tokens = array();
  29
  30     private $parent_handler;
  31
  32     public function tokenizeHTML($string, $config, $context) {
  33
  34         $this->tokens = array();
  35
  36         $string = $this->normalize($string, $config, $context);
  37
  38         $this->parent_handler = set_error_handler(array($this, 'muteStrictErrorHandler'));
  39
  40         $parser = new XML_HTMLSax3();
  41         $parser->set_object($this);
  42         $parser->set_element_handler('openHandler','closeHandler');
  43         $parser->set_data_handler('dataHandler');
  44         $parser->set_escape_handler('escapeHandler');
  45
  46         // doesn't seem to work correctly for attributes
  47         $parser->set_option('XML_OPTION_ENTITIES_PARSED', 1);
  48
  49         $parser->parse($string);
  50
  51         restore_error_handler();
  52
  53         return $this->tokens;
  54
  55     }
  56
  57     /**
  58      * Open tag event handler, interface is defined by PEAR package.
  59      */
  60     public function openHandler(&$parser, $name, $attrs, $closed) {
  61         // entities are not resolved in attrs
  62         foreach ($attrs as $key => $attr) {
  63             $attrs[$key] = $this->parseData($attr);
  64         }
  65         if ($closed) {
  66             $this->tokens[] = new HTMLPurifier_Token_Empty($name, $attrs);
  67         } else {
  68             $this->tokens[] = new HTMLPurifier_Token_Start($name, $attrs);
  69         }
  70         return true;
  71     }
  72
  73     /**
  74      * Close tag event handler, interface is defined by PEAR package.
  75      */
  76     public function closeHandler(&$parser, $name) {
  77         // HTMLSax3 seems to always send empty tags an extra close tag
  78         // check and ignore if you see it:
  79         // [TESTME] to make sure it doesn't overreach
  80         if ($this->tokens[count($this->tokens)-1] instanceof HTMLPurifier_Token_Empty) {
  81             return true;
  82         }
  83         $this->tokens[] = new HTMLPurifier_Token_End($name);
  84         return true;
  85     }
  86
  87     /**
  88      * Data event handler, interface is defined by PEAR package.
  89      */
  90     public function dataHandler(&$parser, $data) {
  91         $this->tokens[] = new HTMLPurifier_Token_Text($data);
  92         return true;
  93     }
  94
  95     /**
  96      * Escaped text handler, interface is defined by PEAR package.
  97      */
  98     public function escapeHandler(&$parser, $data) {
  99         if (strpos($data, '--') === 0) {
 100             $this->tokens[] = new HTMLPurifier_Token_Comment($data);
 101         }
 102         // CDATA is handled elsewhere, but if it was handled here:
 103         //if (strpos($data, '[CDATA[') === 0) {
 104         //    $this->tokens[] = new HTMLPurifier_Token_Text(
 105         //        substr($data, 7, strlen($data) - 9) );
 106         //}
 107         return true;
 108     }
 109
 110     /**
 111      * An error handler that mutes strict errors
 112      */
 113     public function muteStrictErrorHandler($errno, $errstr, $errfile=null, $errline=null, $errcontext=null) {
 114         if ($errno == E_STRICT) return;
 115         return call_user_func($this->parent_handler, $errno, $errstr, $errfile, $errline, $errcontext);
 116     }
 117
 118 }
 119
 120 // vim: et sw=4 sts=4