library/HTMLPurifier/Lexer/DOMLex.php

   1 <?php
   2
   3 /**
   4  * Parser that uses PHP 5's DOM extension (part of the core).
   5  *
   6  * In PHP 5, the DOM XML extension was revamped into DOM and added to the core.
   7  * It gives us a forgiving HTML parser, which we use to transform the HTML
   8  * into a DOM, and then into the tokens.  It is blazingly fast (for large
   9  * documents, it performs twenty times faster than
  10  * HTMLPurifier_Lexer_DirectLex,and is the default choice for PHP 5.
  11  *
  12  * @note Any empty elements will have empty tokens associated with them, even if
  13  * this is prohibited by the spec. This is cannot be fixed until the spec
  14  * comes into play.
  15  *
  16  * @note PHP's DOM extension does not actually parse any entities, we use
  17  *       our own function to do that.
  18  *
  19  * @warning DOM tends to drop whitespace, which may wreak havoc on indenting.
  20  *          If this is a huge problem, due to the fact that HTML is hand
  21  *          edited and you are unable to get a parser cache that caches the
  22  *          the output of HTML Purifier while keeping the original HTML lying
  23  *          around, you may want to run Tidy on the resulting output or use
  24  *          HTMLPurifier_DirectLex
  25  */
  26
  27 class HTMLPurifier_Lexer_DOMLex extends HTMLPurifier_Lexer
  28 {
  29
  30     private $factory;
  31
  32     public function __construct() {
  33         // setup the factory
  34         parent::__construct();
  35         $this->factory = new HTMLPurifier_TokenFactory();
  36     }
  37
  38     public function tokenizeHTML($html, $config, $context) {
  39
  40         $html = $this->normalize($html, $config, $context);
  41
  42         // attempt to armor stray angled brackets that cannot possibly
  43         // form tags and thus are probably being used as emoticons
  44         if ($config->get('Core.AggressivelyFixLt')) {
  45             $char = '[^a-z!\/]';
  46             $comment = "/<!--(.*?)(-->|\z)/is";
  47             $html = preg_replace_callback($comment, array($this, 'callbackArmorCommentEntities'), $html);
  48             do {
  49                 $old = $html;
  50                 $html = preg_replace("/<($char)/i", '&lt;\\1', $html);
  51             } while ($html !== $old);
  52             $html = preg_replace_callback($comment, array($this, 'callbackUndoCommentSubst'), $html); // fix comments
  53         }
  54
  55         // preprocess html, essential for UTF-8
  56         $html = $this->wrapHTML($html, $config, $context);
  57
  58         $doc = new DOMDocument();
  59         $doc->encoding = 'UTF-8'; // theoretically, the above has this covered
  60
  61         set_error_handler(array($this, 'muteErrorHandler'));
  62         $doc->loadHTML($html);
  63         restore_error_handler();
  64
  65         $tokens = array();
  66         $this->tokenizeDOM(
  67             $doc->getElementsByTagName('html')->item(0)-> // <html>
  68                   getElementsByTagName('body')->item(0)-> //   <body>
  69                   getElementsByTagName('div')->item(0)    //     <div>
  70             , $tokens);
  71         return $tokens;
  72     }
  73
  74     /**
  75      * Iterative function that tokenizes a node, putting it into an accumulator.
  76      * To iterate is human, to recurse divine - L. Peter Deutsch
  77      * @param $node     DOMNode to be tokenized.
  78      * @param $tokens   Array-list of already tokenized tokens.
  79      * @returns Tokens of node appended to previously passed tokens.
  80      */
  81     protected function tokenizeDOM($node, &$tokens) {
  82
  83         $level = 0;
  84         $nodes = array($level => array($node));
  85         $closingNodes = array();
  86         do {
  87             while (!empty($nodes[$level])) {
  88                 $node = array_shift($nodes[$level]); // FIFO
  89                 $collect = $level > 0 ? true : false;
  90                 $needEndingTag = $this->createStartNode($node, $tokens, $collect);
  91                 if ($needEndingTag) {
  92                     $closingNodes[$level][] = $node;
  93                 }
  94                 if ($node->childNodes && $node->childNodes->length) {
  95                     $level++;
  96                     $nodes[$level] = array();
  97                     foreach ($node->childNodes as $childNode) {
  98                         array_push($nodes[$level], $childNode);
  99                     }
 100                 }
 101             }
 102             $level--;
 103             if ($level && isset($closingNodes[$level])) {
 104                 while($node = array_pop($closingNodes[$level])) {
 105                     $this->createEndNode($node, $tokens);
 106                 }
 107             }
 108         } while ($level > 0);
 109     }
 110
 111     /**
 112      * @param $node  DOMNode to be tokenized.
 113      * @param $tokens   Array-list of already tokenized tokens.
 114      * @param $collect  Says whether or start and close are collected, set to
 115      *                    false at first recursion because it's the implicit DIV
 116      *                    tag you're dealing with.
 117      * @returns bool if the token needs an endtoken
 118      */
 119     protected function createStartNode($node, &$tokens, $collect) {
 120         // intercept non element nodes. WE MUST catch all of them,
 121         // but we're not getting the character reference nodes because
 122         // those should have been preprocessed
 123         if ($node->nodeType === XML_TEXT_NODE) {
 124             $tokens[] = $this->factory->createText($node->data);
 125             return false;
 126         } elseif ($node->nodeType === XML_CDATA_SECTION_NODE) {
 127             // undo libxml's special treatment of <script> and <style> tags
 128             $last = end($tokens);
 129             $data = $node->data;
 130             // (note $node->tagname is already normalized)
 131             if ($last instanceof HTMLPurifier_Token_Start && ($last->name == 'script' || $last->name == 'style')) {
 132                 $new_data = trim($data);
 133                 if (substr($new_data, 0, 4) === '<!--') {
 134                     $data = substr($new_data, 4);
 135                     if (substr($data, -3) === '-->') {
 136                         $data = substr($data, 0, -3);
 137                     } else {
 138                         // Highly suspicious! Not sure what to do...
 139                     }
 140                 }
 141             }
 142             $tokens[] = $this->factory->createText($this->parseData($data));
 143             return false;
 144         } elseif ($node->nodeType === XML_COMMENT_NODE) {
 145             // this is code is only invoked for comments in script/style in versions
 146             // of libxml pre-2.6.28 (regular comments, of course, are still
 147             // handled regularly)
 148             $tokens[] = $this->factory->createComment($node->data);
 149             return false;
 150         } elseif (
 151             // not-well tested: there may be other nodes we have to grab
 152             $node->nodeType !== XML_ELEMENT_NODE
 153         ) {
 154             return false;
 155         }
 156
 157         $attr = $node->hasAttributes() ? $this->transformAttrToAssoc($node->attributes) : array();
 158
 159         // We still have to make sure that the element actually IS empty
 160         if (!$node->childNodes->length) {
 161             if ($collect) {
 162                 $tokens[] = $this->factory->createEmpty($node->tagName, $attr);
 163             }
 164             return false;
 165         } else {
 166             if ($collect) {
 167                 $tokens[] = $this->factory->createStart(
 168                     $tag_name = $node->tagName, // somehow, it get's dropped
 169                     $attr
 170                 );
 171             }
 172             return true;
 173         }
 174     }
 175
 176     protected function createEndNode($node, &$tokens) {
 177         $tokens[] = $this->factory->createEnd($node->tagName);
 178     }
 179
 180
 181     /**
 182      * Converts a DOMNamedNodeMap of DOMAttr objects into an assoc array.
 183      *
 184      * @param $attribute_list DOMNamedNodeMap of DOMAttr objects.
 185      * @returns Associative array of attributes.
 186      */
 187     protected function transformAttrToAssoc($node_map) {
 188         // NamedNodeMap is documented very well, so we're using undocumented
 189         // features, namely, the fact that it implements Iterator and
 190         // has a ->length attribute
 191         if ($node_map->length === 0) return array();
 192         $array = array();
 193         foreach ($node_map as $attr) {
 194             $array[$attr->name] = $attr->value;
 195         }
 196         return $array;
 197     }
 198
 199     /**
 200      * An error handler that mutes all errors
 201      */
 202     public function muteErrorHandler($errno, $errstr) {}
 203
 204     /**
 205      * Callback function for undoing escaping of stray angled brackets
 206      * in comments
 207      */
 208     public function callbackUndoCommentSubst($matches) {
 209         return '<!--' . strtr($matches[1], array('&amp;'=>'&','&lt;'=>'<')) . $matches[2];
 210     }
 211
 212     /**
 213      * Callback function that entity-izes ampersands in comments so that
 214      * callbackUndoCommentSubst doesn't clobber them
 215      */
 216     public function callbackArmorCommentEntities($matches) {
 217         return '<!--' . str_replace('&', '&amp;', $matches[1]) . $matches[2];
 218     }
 219
 220     /**
 221      * Wraps an HTML fragment in the necessary HTML
 222      */
 223     protected function wrapHTML($html, $config, $context) {
 224         $def = $config->getDefinition('HTML');
 225         $ret = '';
 226
 227         if (!empty($def->doctype->dtdPublic) || !empty($def->doctype->dtdSystem)) {
 228             $ret .= '<!DOCTYPE html ';
 229             if (!empty($def->doctype->dtdPublic)) $ret .= 'PUBLIC "' . $def->doctype->dtdPublic . '" ';
 230             if (!empty($def->doctype->dtdSystem)) $ret .= '"' . $def->doctype->dtdSystem . '" ';
 231             $ret .= '>';
 232         }
 233
 234         $ret .= '<html><head>';
 235         $ret .= '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />';
 236         // No protection if $html contains a stray </div>!
 237         $ret .= '</head><body><div>'.$html.'</div></body></html>';
 238         return $ret;
 239     }
 240
 241 }
 242
 243 // vim: et sw=4 sts=4