4 * Parser that uses PHP 5's DOM extension (part of the core).
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.
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
16 * @note PHP's DOM extension does not actually parse any entities, we use
17 * our own function to do that.
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
27 class HTMLPurifier_Lexer_DOMLex extends HTMLPurifier_Lexer
31 * @type HTMLPurifier_TokenFactory
35 public function __construct()
38 parent::__construct();
39 $this->factory = new HTMLPurifier_TokenFactory();
44 * @param HTMLPurifier_Config $config
45 * @param HTMLPurifier_Context $context
46 * @return HTMLPurifier_Token[]
48 public function tokenizeHTML($html, $config, $context)
50 $html = $this->normalize($html, $config, $context);
52 // attempt to armor stray angled brackets that cannot possibly
53 // form tags and thus are probably being used as emoticons
54 if ($config->get('Core.AggressivelyFixLt')) {
56 $comment = "/<!--(.*?)(-->|\z)/is";
57 $html = preg_replace_callback($comment, array($this, 'callbackArmorCommentEntities'), $html);
60 $html = preg_replace("/<($char)/i", '<\\1', $html);
61 } while ($html !== $old);
62 $html = preg_replace_callback($comment, array($this, 'callbackUndoCommentSubst'), $html); // fix comments
65 // preprocess html, essential for UTF-8
66 $html = $this->wrapHTML($html, $config, $context);
68 $doc = new DOMDocument();
69 $doc->encoding = 'UTF-8'; // theoretically, the above has this covered
71 set_error_handler(array($this, 'muteErrorHandler'));
72 $doc->loadHTML($html);
73 restore_error_handler();
77 $doc->getElementsByTagName('html')->item(0)-> // <html>
78 getElementsByTagName('body')->item(0), // <body>
85 * Iterative function that tokenizes a node, putting it into an accumulator.
86 * To iterate is human, to recurse divine - L. Peter Deutsch
87 * @param DOMNode $node DOMNode to be tokenized.
88 * @param HTMLPurifier_Token[] $tokens Array-list of already tokenized tokens.
89 * @return HTMLPurifier_Token of node appended to previously passed tokens.
91 protected function tokenizeDOM($node, &$tokens)
94 $nodes = array($level => new HTMLPurifier_Queue(array($node)));
95 $closingNodes = array();
97 while (!$nodes[$level]->isEmpty()) {
98 $node = $nodes[$level]->shift(); // FIFO
99 $collect = $level > 0 ? true : false;
100 $needEndingTag = $this->createStartNode($node, $tokens, $collect);
101 if ($needEndingTag) {
102 $closingNodes[$level][] = $node;
104 if ($node->childNodes && $node->childNodes->length) {
106 $nodes[$level] = new HTMLPurifier_Queue();
107 foreach ($node->childNodes as $childNode) {
108 $nodes[$level]->push($childNode);
113 if ($level && isset($closingNodes[$level])) {
114 while ($node = array_pop($closingNodes[$level])) {
115 $this->createEndNode($node, $tokens);
118 } while ($level > 0);
122 * @param DOMNode $node DOMNode to be tokenized.
123 * @param HTMLPurifier_Token[] $tokens Array-list of already tokenized tokens.
124 * @param bool $collect Says whether or start and close are collected, set to
125 * false at first recursion because it's the implicit DIV
126 * tag you're dealing with.
127 * @return bool if the token needs an endtoken
128 * @todo data and tagName properties don't seem to exist in DOMNode?
130 protected function createStartNode($node, &$tokens, $collect)
132 // intercept non element nodes. WE MUST catch all of them,
133 // but we're not getting the character reference nodes because
134 // those should have been preprocessed
135 if ($node->nodeType === XML_TEXT_NODE) {
136 $tokens[] = $this->factory->createText($node->data);
138 } elseif ($node->nodeType === XML_CDATA_SECTION_NODE) {
139 // undo libxml's special treatment of <script> and <style> tags
140 $last = end($tokens);
142 // (note $node->tagname is already normalized)
143 if ($last instanceof HTMLPurifier_Token_Start && ($last->name == 'script' || $last->name == 'style')) {
144 $new_data = trim($data);
145 if (substr($new_data, 0, 4) === '<!--') {
146 $data = substr($new_data, 4);
147 if (substr($data, -3) === '-->') {
148 $data = substr($data, 0, -3);
150 // Highly suspicious! Not sure what to do...
154 $tokens[] = $this->factory->createText($this->parseData($data));
156 } elseif ($node->nodeType === XML_COMMENT_NODE) {
157 // this is code is only invoked for comments in script/style in versions
158 // of libxml pre-2.6.28 (regular comments, of course, are still
159 // handled regularly)
160 $tokens[] = $this->factory->createComment($node->data);
162 } elseif ($node->nodeType !== XML_ELEMENT_NODE) {
163 // not-well tested: there may be other nodes we have to grab
167 $attr = $node->hasAttributes() ? $this->transformAttrToAssoc($node->attributes) : array();
169 // We still have to make sure that the element actually IS empty
170 if (!$node->childNodes->length) {
172 $tokens[] = $this->factory->createEmpty($node->tagName, $attr);
177 $tokens[] = $this->factory->createStart(
178 $tag_name = $node->tagName, // somehow, it get's dropped
187 * @param DOMNode $node
188 * @param HTMLPurifier_Token[] $tokens
190 protected function createEndNode($node, &$tokens)
192 $tokens[] = $this->factory->createEnd($node->tagName);
197 * Converts a DOMNamedNodeMap of DOMAttr objects into an assoc array.
199 * @param DOMNamedNodeMap $node_map DOMNamedNodeMap of DOMAttr objects.
200 * @return array Associative array of attributes.
202 protected function transformAttrToAssoc($node_map)
204 // NamedNodeMap is documented very well, so we're using undocumented
205 // features, namely, the fact that it implements Iterator and
206 // has a ->length attribute
207 if ($node_map->length === 0) {
211 foreach ($node_map as $attr) {
212 $array[$attr->name] = $attr->value;
218 * An error handler that mutes all errors
220 * @param string $errstr
222 public function muteErrorHandler($errno, $errstr)
227 * Callback function for undoing escaping of stray angled brackets
229 * @param array $matches
232 public function callbackUndoCommentSubst($matches)
234 return '<!--' . strtr($matches[1], array('&' => '&', '<' => '<')) . $matches[2];
238 * Callback function that entity-izes ampersands in comments so that
239 * callbackUndoCommentSubst doesn't clobber them
240 * @param array $matches
243 public function callbackArmorCommentEntities($matches)
245 return '<!--' . str_replace('&', '&', $matches[1]) . $matches[2];
249 * Wraps an HTML fragment in the necessary HTML
250 * @param string $html
251 * @param HTMLPurifier_Config $config
252 * @param HTMLPurifier_Context $context
255 protected function wrapHTML($html, $config, $context)
257 $def = $config->getDefinition('HTML');
260 if (!empty($def->doctype->dtdPublic) || !empty($def->doctype->dtdSystem)) {
261 $ret .= '<!DOCTYPE html ';
262 if (!empty($def->doctype->dtdPublic)) {
263 $ret .= 'PUBLIC "' . $def->doctype->dtdPublic . '" ';
265 if (!empty($def->doctype->dtdSystem)) {
266 $ret .= '"' . $def->doctype->dtdSystem . '" ';
271 $ret .= '<html><head>';
272 $ret .= '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />';
273 // No protection if $html contains a stray </div>!
274 $ret .= '</head><body>' . $html . '</body></html>';
279 // vim: et sw=4 sts=4