htmlpurifier: HTMLPurifier/Lexer/DOMLex.php Quellcode

00001 <?php
00002 
00027 class HTMLPurifier_Lexer_DOMLex extends HTMLPurifier_Lexer
00028 {
00029 
00030     private $factory;
00031 
00032     public function __construct() {
00033         // setup the factory
00034         parent::__construct();
00035         $this->factory = new HTMLPurifier_TokenFactory();
00036     }
00037 
00038     public function tokenizeHTML($html, $config, $context) {
00039 
00040         $html = $this->normalize($html, $config, $context);
00041 
00042         // attempt to armor stray angled brackets that cannot possibly
00043         // form tags and thus are probably being used as emoticons
00044         if ($config->get('Core.AggressivelyFixLt')) {
00045             $char = '[^a-z!\/]';
00046             $comment = "/<!--(.*?)(-->|\z)/is";
00047             $html = preg_replace_callback($comment, array($this, 'callbackArmorCommentEntities'), $html);
00048             do {
00049                 $old = $html;
00050                 $html = preg_replace("/<($char)/i", '&lt;\\1', $html);
00051             } while ($html !== $old);
00052             $html = preg_replace_callback($comment, array($this, 'callbackUndoCommentSubst'), $html); // fix comments
00053         }
00054 
00055         // preprocess html, essential for UTF-8
00056         $html = $this->wrapHTML($html, $config, $context);
00057 
00058         $doc = new DOMDocument();
00059         $doc->encoding = 'UTF-8'; // theoretically, the above has this covered
00060 
00061         set_error_handler(array($this, 'muteErrorHandler'));
00062         $doc->loadHTML($html);
00063         restore_error_handler();
00064 
00065         $tokens = array();
00066         $this->tokenizeDOM(
00067             $doc->getElementsByTagName('html')->item(0)-> // <html>
00068                   getElementsByTagName('body')->item(0)-> //   <body>
00069                   getElementsByTagName('div')->item(0)    //     <div>
00070             , $tokens);
00071         return $tokens;
00072     }
00073 
00084     protected function tokenizeDOM($node, &$tokens, $collect = false) {
00085 
00086         // intercept non element nodes. WE MUST catch all of them,
00087         // but we're not getting the character reference nodes because
00088         // those should have been preprocessed
00089         if ($node->nodeType === XML_TEXT_NODE) {
00090             $tokens[] = $this->factory->createText($node->data);
00091             return;
00092         } elseif ($node->nodeType === XML_CDATA_SECTION_NODE) {
00093             // undo libxml's special treatment of <script> and <style> tags
00094             $last = end($tokens);
00095             $data = $node->data;
00096             // (note $node->tagname is already normalized)
00097             if ($last instanceof HTMLPurifier_Token_Start && ($last->name == 'script' || $last->name == 'style')) {
00098                 $new_data = trim($data);
00099                 if (substr($new_data, 0, 4) === '<!--') {
00100                     $data = substr($new_data, 4);
00101                     if (substr($data, -3) === '-->') {
00102                         $data = substr($data, 0, -3);
00103                     } else {
00104                         // Highly suspicious! Not sure what to do...
00105                     }
00106                 }
00107             }
00108             $tokens[] = $this->factory->createText($this->parseData($data));
00109             return;
00110         } elseif ($node->nodeType === XML_COMMENT_NODE) {
00111             // this is code is only invoked for comments in script/style in versions
00112             // of libxml pre-2.6.28 (regular comments, of course, are still
00113             // handled regularly)
00114             $tokens[] = $this->factory->createComment($node->data);
00115             return;
00116         } elseif (
00117             // not-well tested: there may be other nodes we have to grab
00118             $node->nodeType !== XML_ELEMENT_NODE
00119         ) {
00120             return;
00121         }
00122 
00123         $attr = $node->hasAttributes() ?
00124             $this->transformAttrToAssoc($node->attributes) :
00125             array();
00126 
00127         // We still have to make sure that the element actually IS empty
00128         if (!$node->childNodes->length) {
00129             if ($collect) {
00130                 $tokens[] = $this->factory->createEmpty($node->tagName, $attr);
00131             }
00132         } else {
00133             if ($collect) { // don't wrap on first iteration
00134                 $tokens[] = $this->factory->createStart(
00135                     $tag_name = $node->tagName, // somehow, it get's dropped
00136                     $attr
00137                 );
00138             }
00139             foreach ($node->childNodes as $node) {
00140                 // remember, it's an accumulator. Otherwise, we'd have
00141                 // to use array_merge
00142                 $this->tokenizeDOM($node, $tokens, true);
00143             }
00144             if ($collect) {
00145                 $tokens[] = $this->factory->createEnd($tag_name);
00146             }
00147         }
00148 
00149     }
00150 
00157     protected function transformAttrToAssoc($node_map) {
00158         // NamedNodeMap is documented very well, so we're using undocumented
00159         // features, namely, the fact that it implements Iterator and
00160         // has a ->length attribute
00161         if ($node_map->length === 0) return array();
00162         $array = array();
00163         foreach ($node_map as $attr) {
00164             $array[$attr->name] = $attr->value;
00165         }
00166         return $array;
00167     }
00168 
00172     public function muteErrorHandler($errno, $errstr) {}
00173 
00178     public function callbackUndoCommentSubst($matches) {
00179         return '<!--' . strtr($matches[1], array('&amp;'=>'&','&lt;'=>'<')) . $matches[2];
00180     }
00181 
00186     public function callbackArmorCommentEntities($matches) {
00187         return '<!--' . str_replace('&', '&amp;', $matches[1]) . $matches[2];
00188     }
00189 
00193     protected function wrapHTML($html, $config, $context) {
00194         $def = $config->getDefinition('HTML');
00195         $ret = '';
00196 
00197         if (!empty($def->doctype->dtdPublic) || !empty($def->doctype->dtdSystem)) {
00198             $ret .= '<!DOCTYPE html ';
00199             if (!empty($def->doctype->dtdPublic)) $ret .= 'PUBLIC "' . $def->doctype->dtdPublic . '" ';
00200             if (!empty($def->doctype->dtdSystem)) $ret .= '"' . $def->doctype->dtdSystem . '" ';
00201             $ret .= '>';
00202         }
00203 
00204         $ret .= '<html><head>';
00205         $ret .= '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />';
00206         // No protection if $html contains a stray </div>!
00207         $ret .= '</head><body><div>'.$html.'</div></body></html>';
00208         return $ret;
00209     }
00210 
00211 }
00212 
00213 // vim: et sw=4 sts=4


Copyright © 2003 - 2009 MyOOS [Shopsystem]. All rights reserved. MyOOS [Shopsystem] is Free Software released under the GNU/GPL License. Webmaster: info@r23.de (Impressum)