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     protected $last_token_was_empty;
  30
  31     private $parent_handler;
  32     private $stack = array();
  33
  34     public function tokenizeHTML($string, $config, $context) {
  35
  36         $this->tokens = array();
  37         $this->last_token_was_empty = false;
  38
  39         $string = $this->normalize($string, $config, $context);
  40
  41         $this->parent_handler = set_error_handler(array($this, 'muteStrictErrorHandler'));
  42
  43         $parser = new XML_HTMLSax3();
  44         $parser->set_object($this);
  45         $parser->set_element_handler('openHandler','closeHandler');
  46         $parser->set_data_handler('dataHandler');
  47         $parser->set_escape_handler('escapeHandler');
  48
  49         // doesn't seem to work correctly for attributes
  50         $parser->set_option('XML_OPTION_ENTITIES_PARSED', 1);
  51
  52         $parser->parse($string);
  53
  54         restore_error_handler();
  55
  56         return $this->tokens;
  57
  58     }
  59
  60     /**
  61      * Open tag event handler, interface is defined by PEAR package.
  62      */
  63     public function openHandler(&$parser, $name, $attrs, $closed) {
  64         // entities are not resolved in attrs
  65         foreach ($attrs as $key => $attr) {
  66             $attrs[$key] = $this->parseData($attr);
  67         }
  68         if ($closed) {
  69             $this->tokens[] = new HTMLPurifier_Token_Empty($name, $attrs);
  70             $this->last_token_was_empty = true;
  71         } else {
  72             $this->tokens[] = new HTMLPurifier_Token_Start($name, $attrs);
  73         }
  74         $this->stack[] = $name;
  75         return true;
  76     }
  77
  78     /**
  79      * Close tag event handler, interface is defined by PEAR package.
  80      */
  81     public function closeHandler(&$parser, $name) {
  82         // HTMLSax3 seems to always send empty tags an extra close tag
  83         // check and ignore if you see it:
  84         // [TESTME] to make sure it doesn't overreach
  85         if ($this->last_token_was_empty) {
  86             $this->last_token_was_empty = false;
  87             return true;
  88         }
  89         $this->tokens[] = new HTMLPurifier_Token_End($name);
  90         if (!empty($this->stack)) array_pop($this->stack);
  91         return true;
  92     }
  93
  94     /**
  95      * Data event handler, interface is defined by PEAR package.
  96      */
  97     public function dataHandler(&$parser, $data) {
  98         $this->last_token_was_empty = false;
  99         $this->tokens[] = new HTMLPurifier_Token_Text($data);
 100         return true;
 101     }
 102
 103     /**
 104      * Escaped text handler, interface is defined by PEAR package.
 105      */
 106     public function escapeHandler(&$parser, $data) {
 107         if (strpos($data, '--') === 0) {
 108             // remove trailing and leading double-dashes
 109             $data = substr($data, 2);
 110             if (strlen($data) >= 2 && substr($data, -2) == "--") {
 111                 $data = substr($data, 0, -2);
 112             }
 113             if (isset($this->stack[sizeof($this->stack) - 1]) &&
 114                 $this->stack[sizeof($this->stack) - 1] == "style") {
 115                 $this->tokens[] = new HTMLPurifier_Token_Text($data);
 116             } else {
 117                 $this->tokens[] = new HTMLPurifier_Token_Comment($data);
 118             }
 119             $this->last_token_was_empty = false;
 120         }
 121         // CDATA is handled elsewhere, but if it was handled here:
 122         //if (strpos($data, '[CDATA[') === 0) {
 123         //    $this->tokens[] = new HTMLPurifier_Token_Text(
 124         //        substr($data, 7, strlen($data) - 9) );
 125         //}
 126         return true;
 127     }
 128
 129     /**
 130      * An error handler that mutes strict errors
 131      */
 132     public function muteStrictErrorHandler($errno, $errstr, $errfile=null, $errline=null, $errcontext=null) {
 133         if ($errno == E_STRICT) return;
 134         return call_user_func($this->parent_handler, $errno, $errstr, $errfile, $errline, $errcontext);
 135     }
 136
 137 }
 138
 139 // vim: et sw=4 sts=4