| 1 |
efrain |
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 |
/**
|
|
|
31 |
* @type HTMLPurifier_TokenFactory
|
|
|
32 |
*/
|
|
|
33 |
private $factory;
|
|
|
34 |
|
|
|
35 |
public function __construct()
|
|
|
36 |
{
|
|
|
37 |
// setup the factory
|
|
|
38 |
parent::__construct();
|
|
|
39 |
$this->factory = new HTMLPurifier_TokenFactory();
|
|
|
40 |
}
|
|
|
41 |
|
|
|
42 |
/**
|
|
|
43 |
* @param string $html
|
|
|
44 |
* @param HTMLPurifier_Config $config
|
|
|
45 |
* @param HTMLPurifier_Context $context
|
|
|
46 |
* @return HTMLPurifier_Token[]
|
|
|
47 |
*/
|
|
|
48 |
public function tokenizeHTML($html, $config, $context)
|
|
|
49 |
{
|
|
|
50 |
$html = $this->normalize($html, $config, $context);
|
|
|
51 |
|
|
|
52 |
// attempt to armor stray angled brackets that cannot possibly
|
|
|
53 |
// form tags and thus are probably being used as emoticons
|
|
|
54 |
if ($config->get('Core.AggressivelyFixLt')) {
|
|
|
55 |
$char = '[^a-z!\/]';
|
|
|
56 |
$comment = "/<!--(.*?)(-->|\z)/is";
|
|
|
57 |
$html = preg_replace_callback($comment, array($this, 'callbackArmorCommentEntities'), $html);
|
|
|
58 |
do {
|
|
|
59 |
$old = $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
|
|
|
63 |
}
|
|
|
64 |
|
|
|
65 |
// preprocess html, essential for UTF-8
|
|
|
66 |
$html = $this->wrapHTML($html, $config, $context);
|
|
|
67 |
|
|
|
68 |
$doc = new DOMDocument();
|
|
|
69 |
$doc->encoding = 'UTF-8'; // theoretically, the above has this covered
|
|
|
70 |
|
|
|
71 |
$options = 0;
|
|
|
72 |
if ($config->get('Core.AllowParseManyTags') && defined('LIBXML_PARSEHUGE')) {
|
|
|
73 |
$options |= LIBXML_PARSEHUGE;
|
|
|
74 |
}
|
| 1441 |
ariadna |
75 |
if ($config->get('Core.RemoveBlanks') && defined('LIBXML_NOBLANKS')) {
|
|
|
76 |
$options |= LIBXML_NOBLANKS;
|
|
|
77 |
}
|
| 1 |
efrain |
78 |
|
|
|
79 |
set_error_handler(array($this, 'muteErrorHandler'));
|
|
|
80 |
// loadHTML() fails on PHP 5.3 when second parameter is given
|
|
|
81 |
if ($options) {
|
|
|
82 |
$doc->loadHTML($html, $options);
|
|
|
83 |
} else {
|
|
|
84 |
$doc->loadHTML($html);
|
|
|
85 |
}
|
|
|
86 |
restore_error_handler();
|
|
|
87 |
|
|
|
88 |
$body = $doc->getElementsByTagName('html')->item(0)-> // <html>
|
|
|
89 |
getElementsByTagName('body')->item(0); // <body>
|
|
|
90 |
|
|
|
91 |
$div = $body->getElementsByTagName('div')->item(0); // <div>
|
|
|
92 |
$tokens = array();
|
|
|
93 |
$this->tokenizeDOM($div, $tokens, $config);
|
|
|
94 |
// If the div has a sibling, that means we tripped across
|
|
|
95 |
// a premature </div> tag. So remove the div we parsed,
|
|
|
96 |
// and then tokenize the rest of body. We can't tokenize
|
|
|
97 |
// the sibling directly as we'll lose the tags in that case.
|
|
|
98 |
if ($div->nextSibling) {
|
|
|
99 |
$body->removeChild($div);
|
|
|
100 |
$this->tokenizeDOM($body, $tokens, $config);
|
|
|
101 |
}
|
|
|
102 |
return $tokens;
|
|
|
103 |
}
|
|
|
104 |
|
|
|
105 |
/**
|
|
|
106 |
* Iterative function that tokenizes a node, putting it into an accumulator.
|
|
|
107 |
* To iterate is human, to recurse divine - L. Peter Deutsch
|
|
|
108 |
* @param DOMNode $node DOMNode to be tokenized.
|
|
|
109 |
* @param HTMLPurifier_Token[] $tokens Array-list of already tokenized tokens.
|
|
|
110 |
*/
|
|
|
111 |
protected function tokenizeDOM($node, &$tokens, $config)
|
|
|
112 |
{
|
|
|
113 |
$level = 0;
|
|
|
114 |
$nodes = array($level => new HTMLPurifier_Queue(array($node)));
|
|
|
115 |
$closingNodes = array();
|
|
|
116 |
do {
|
|
|
117 |
while (!$nodes[$level]->isEmpty()) {
|
|
|
118 |
$node = $nodes[$level]->shift(); // FIFO
|
|
|
119 |
$collect = $level > 0 ? true : false;
|
|
|
120 |
$needEndingTag = $this->createStartNode($node, $tokens, $collect, $config);
|
|
|
121 |
if ($needEndingTag) {
|
|
|
122 |
$closingNodes[$level][] = $node;
|
|
|
123 |
}
|
|
|
124 |
if ($node->childNodes && $node->childNodes->length) {
|
|
|
125 |
$level++;
|
|
|
126 |
$nodes[$level] = new HTMLPurifier_Queue();
|
|
|
127 |
foreach ($node->childNodes as $childNode) {
|
|
|
128 |
$nodes[$level]->push($childNode);
|
|
|
129 |
}
|
|
|
130 |
}
|
|
|
131 |
}
|
|
|
132 |
$level--;
|
|
|
133 |
if ($level && isset($closingNodes[$level])) {
|
|
|
134 |
while ($node = array_pop($closingNodes[$level])) {
|
|
|
135 |
$this->createEndNode($node, $tokens);
|
|
|
136 |
}
|
|
|
137 |
}
|
|
|
138 |
} while ($level > 0);
|
|
|
139 |
}
|
|
|
140 |
|
|
|
141 |
/**
|
|
|
142 |
* Portably retrieve the tag name of a node; deals with older versions
|
|
|
143 |
* of libxml like 2.7.6
|
|
|
144 |
* @param DOMNode $node
|
|
|
145 |
*/
|
|
|
146 |
protected function getTagName($node)
|
|
|
147 |
{
|
|
|
148 |
if (isset($node->tagName)) {
|
|
|
149 |
return $node->tagName;
|
|
|
150 |
} else if (isset($node->nodeName)) {
|
|
|
151 |
return $node->nodeName;
|
|
|
152 |
} else if (isset($node->localName)) {
|
|
|
153 |
return $node->localName;
|
|
|
154 |
}
|
|
|
155 |
return null;
|
|
|
156 |
}
|
|
|
157 |
|
|
|
158 |
/**
|
|
|
159 |
* Portably retrieve the data of a node; deals with older versions
|
|
|
160 |
* of libxml like 2.7.6
|
|
|
161 |
* @param DOMNode $node
|
|
|
162 |
*/
|
|
|
163 |
protected function getData($node)
|
|
|
164 |
{
|
|
|
165 |
if (isset($node->data)) {
|
|
|
166 |
return $node->data;
|
|
|
167 |
} else if (isset($node->nodeValue)) {
|
|
|
168 |
return $node->nodeValue;
|
|
|
169 |
} else if (isset($node->textContent)) {
|
|
|
170 |
return $node->textContent;
|
|
|
171 |
}
|
|
|
172 |
return null;
|
|
|
173 |
}
|
|
|
174 |
|
|
|
175 |
|
|
|
176 |
/**
|
|
|
177 |
* @param DOMNode $node DOMNode to be tokenized.
|
|
|
178 |
* @param HTMLPurifier_Token[] $tokens Array-list of already tokenized tokens.
|
|
|
179 |
* @param bool $collect Says whether or start and close are collected, set to
|
|
|
180 |
* false at first recursion because it's the implicit DIV
|
|
|
181 |
* tag you're dealing with.
|
|
|
182 |
* @return bool if the token needs an endtoken
|
|
|
183 |
* @todo data and tagName properties don't seem to exist in DOMNode?
|
|
|
184 |
*/
|
|
|
185 |
protected function createStartNode($node, &$tokens, $collect, $config)
|
|
|
186 |
{
|
|
|
187 |
// intercept non element nodes. WE MUST catch all of them,
|
|
|
188 |
// but we're not getting the character reference nodes because
|
|
|
189 |
// those should have been preprocessed
|
|
|
190 |
if ($node->nodeType === XML_TEXT_NODE) {
|
|
|
191 |
$data = $this->getData($node); // Handle variable data property
|
|
|
192 |
if ($data !== null) {
|
|
|
193 |
$tokens[] = $this->factory->createText($data);
|
|
|
194 |
}
|
|
|
195 |
return false;
|
|
|
196 |
} elseif ($node->nodeType === XML_CDATA_SECTION_NODE) {
|
|
|
197 |
// undo libxml's special treatment of <script> and <style> tags
|
|
|
198 |
$last = end($tokens);
|
|
|
199 |
$data = $node->data;
|
|
|
200 |
// (note $node->tagname is already normalized)
|
|
|
201 |
if ($last instanceof HTMLPurifier_Token_Start && ($last->name == 'script' || $last->name == 'style')) {
|
|
|
202 |
$new_data = trim($data);
|
|
|
203 |
if (substr($new_data, 0, 4) === '<!--') {
|
|
|
204 |
$data = substr($new_data, 4);
|
|
|
205 |
if (substr($data, -3) === '-->') {
|
|
|
206 |
$data = substr($data, 0, -3);
|
|
|
207 |
} else {
|
|
|
208 |
// Highly suspicious! Not sure what to do...
|
|
|
209 |
}
|
|
|
210 |
}
|
|
|
211 |
}
|
|
|
212 |
$tokens[] = $this->factory->createText($this->parseText($data, $config));
|
|
|
213 |
return false;
|
|
|
214 |
} elseif ($node->nodeType === XML_COMMENT_NODE) {
|
|
|
215 |
// this is code is only invoked for comments in script/style in versions
|
|
|
216 |
// of libxml pre-2.6.28 (regular comments, of course, are still
|
|
|
217 |
// handled regularly)
|
|
|
218 |
$tokens[] = $this->factory->createComment($node->data);
|
|
|
219 |
return false;
|
|
|
220 |
} elseif ($node->nodeType !== XML_ELEMENT_NODE) {
|
|
|
221 |
// not-well tested: there may be other nodes we have to grab
|
|
|
222 |
return false;
|
|
|
223 |
}
|
|
|
224 |
$attr = $node->hasAttributes() ? $this->transformAttrToAssoc($node->attributes) : array();
|
|
|
225 |
$tag_name = $this->getTagName($node); // Handle variable tagName property
|
|
|
226 |
if (empty($tag_name)) {
|
|
|
227 |
return (bool) $node->childNodes->length;
|
|
|
228 |
}
|
|
|
229 |
// We still have to make sure that the element actually IS empty
|
|
|
230 |
if (!$node->childNodes->length) {
|
|
|
231 |
if ($collect) {
|
|
|
232 |
$tokens[] = $this->factory->createEmpty($tag_name, $attr);
|
|
|
233 |
}
|
|
|
234 |
return false;
|
|
|
235 |
} else {
|
|
|
236 |
if ($collect) {
|
|
|
237 |
$tokens[] = $this->factory->createStart($tag_name, $attr);
|
|
|
238 |
}
|
|
|
239 |
return true;
|
|
|
240 |
}
|
|
|
241 |
}
|
|
|
242 |
|
|
|
243 |
/**
|
|
|
244 |
* @param DOMNode $node
|
|
|
245 |
* @param HTMLPurifier_Token[] $tokens
|
|
|
246 |
*/
|
|
|
247 |
protected function createEndNode($node, &$tokens)
|
|
|
248 |
{
|
|
|
249 |
$tag_name = $this->getTagName($node); // Handle variable tagName property
|
|
|
250 |
$tokens[] = $this->factory->createEnd($tag_name);
|
|
|
251 |
}
|
|
|
252 |
|
|
|
253 |
/**
|
|
|
254 |
* Converts a DOMNamedNodeMap of DOMAttr objects into an assoc array.
|
|
|
255 |
*
|
|
|
256 |
* @param DOMNamedNodeMap $node_map DOMNamedNodeMap of DOMAttr objects.
|
|
|
257 |
* @return array Associative array of attributes.
|
|
|
258 |
*/
|
|
|
259 |
protected function transformAttrToAssoc($node_map)
|
|
|
260 |
{
|
|
|
261 |
// NamedNodeMap is documented very well, so we're using undocumented
|
|
|
262 |
// features, namely, the fact that it implements Iterator and
|
|
|
263 |
// has a ->length attribute
|
|
|
264 |
if ($node_map->length === 0) {
|
|
|
265 |
return array();
|
|
|
266 |
}
|
|
|
267 |
$array = array();
|
|
|
268 |
foreach ($node_map as $attr) {
|
|
|
269 |
$array[$attr->name] = $attr->value;
|
|
|
270 |
}
|
|
|
271 |
return $array;
|
|
|
272 |
}
|
|
|
273 |
|
|
|
274 |
/**
|
|
|
275 |
* An error handler that mutes all errors
|
|
|
276 |
* @param int $errno
|
|
|
277 |
* @param string $errstr
|
|
|
278 |
*/
|
|
|
279 |
public function muteErrorHandler($errno, $errstr)
|
|
|
280 |
{
|
|
|
281 |
}
|
|
|
282 |
|
|
|
283 |
/**
|
|
|
284 |
* Callback function for undoing escaping of stray angled brackets
|
|
|
285 |
* in comments
|
|
|
286 |
* @param array $matches
|
|
|
287 |
* @return string
|
|
|
288 |
*/
|
|
|
289 |
public function callbackUndoCommentSubst($matches)
|
|
|
290 |
{
|
|
|
291 |
return '<!--' . strtr($matches[1], array('&' => '&', '<' => '<')) . $matches[2];
|
|
|
292 |
}
|
|
|
293 |
|
|
|
294 |
/**
|
|
|
295 |
* Callback function that entity-izes ampersands in comments so that
|
|
|
296 |
* callbackUndoCommentSubst doesn't clobber them
|
|
|
297 |
* @param array $matches
|
|
|
298 |
* @return string
|
|
|
299 |
*/
|
|
|
300 |
public function callbackArmorCommentEntities($matches)
|
|
|
301 |
{
|
|
|
302 |
return '<!--' . str_replace('&', '&', $matches[1]) . $matches[2];
|
|
|
303 |
}
|
|
|
304 |
|
|
|
305 |
/**
|
|
|
306 |
* Wraps an HTML fragment in the necessary HTML
|
|
|
307 |
* @param string $html
|
|
|
308 |
* @param HTMLPurifier_Config $config
|
|
|
309 |
* @param HTMLPurifier_Context $context
|
|
|
310 |
* @return string
|
|
|
311 |
*/
|
|
|
312 |
protected function wrapHTML($html, $config, $context, $use_div = true)
|
|
|
313 |
{
|
|
|
314 |
$def = $config->getDefinition('HTML');
|
|
|
315 |
$ret = '';
|
|
|
316 |
|
|
|
317 |
if (!empty($def->doctype->dtdPublic) || !empty($def->doctype->dtdSystem)) {
|
|
|
318 |
$ret .= '<!DOCTYPE html ';
|
|
|
319 |
if (!empty($def->doctype->dtdPublic)) {
|
|
|
320 |
$ret .= 'PUBLIC "' . $def->doctype->dtdPublic . '" ';
|
|
|
321 |
}
|
|
|
322 |
if (!empty($def->doctype->dtdSystem)) {
|
|
|
323 |
$ret .= '"' . $def->doctype->dtdSystem . '" ';
|
|
|
324 |
}
|
|
|
325 |
$ret .= '>';
|
|
|
326 |
}
|
|
|
327 |
|
|
|
328 |
$ret .= '<html><head>';
|
|
|
329 |
$ret .= '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />';
|
|
|
330 |
// No protection if $html contains a stray </div>!
|
|
|
331 |
$ret .= '</head><body>';
|
|
|
332 |
if ($use_div) $ret .= '<div>';
|
|
|
333 |
$ret .= $html;
|
|
|
334 |
if ($use_div) $ret .= '</div>';
|
|
|
335 |
$ret .= '</body></html>';
|
|
|
336 |
return $ret;
|
|
|
337 |
}
|
|
|
338 |
}
|
|
|
339 |
|
|
|
340 |
// vim: et sw=4 sts=4
|