Namespaces

Classes

Interfaces

Exceptions

Functions

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", '&lt;\\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: $options |= LIBXML_PARSEHUGE;
74: }
75:
76: set_error_handler(array($this, 'muteErrorHandler'));
77: // loadHTML() fails on PHP 5.3 when second parameter is given
78: if ($options) {
79: $doc->loadHTML($html, $options);
80: } else {
81: $doc->loadHTML($html);
82: }
83: restore_error_handler();
84:
85: $body = $doc->getElementsByTagName('html')->item(0)-> // <html>
86: getElementsByTagName('body')->item(0); // <body>
87:
88: $div = $body->getElementsByTagName('div')->item(0); // <div>
89: $tokens = array();
90: $this->tokenizeDOM($div, $tokens, $config);
91: // If the div has a sibling, that means we tripped across
92: // a premature </div> tag. So remove the div we parsed,
93: // and then tokenize the rest of body. We can't tokenize
94: // the sibling directly as we'll lose the tags in that case.
95: if ($div->nextSibling) {
96: $body->removeChild($div);
97: $this->tokenizeDOM($body, $tokens, $config);
98: }
99: return $tokens;
100: }
101:
102: /**
103: * Iterative function that tokenizes a node, putting it into an accumulator.
104: * To iterate is human, to recurse divine - L. Peter Deutsch
105: * @param DOMNode $node DOMNode to be tokenized.
106: * @param HTMLPurifier_Token[] $tokens Array-list of already tokenized tokens.
107: * @return HTMLPurifier_Token of node appended to previously passed tokens.
108: */
109: protected function tokenizeDOM($node, &$tokens, $config)
110: {
111: $level = 0;
112: $nodes = array($level => new HTMLPurifier_Queue(array($node)));
113: $closingNodes = array();
114: do {
115: while (!$nodes[$level]->isEmpty()) {
116: $node = $nodes[$level]->shift(); // FIFO
117: $collect = $level > 0 ? true : false;
118: $needEndingTag = $this->createStartNode($node, $tokens, $collect, $config);
119: if ($needEndingTag) {
120: $closingNodes[$level][] = $node;
121: }
122: if ($node->childNodes && $node->childNodes->length) {
123: $level++;
124: $nodes[$level] = new HTMLPurifier_Queue();
125: foreach ($node->childNodes as $childNode) {
126: $nodes[$level]->push($childNode);
127: }
128: }
129: }
130: $level--;
131: if ($level && isset($closingNodes[$level])) {
132: while ($node = array_pop($closingNodes[$level])) {
133: $this->createEndNode($node, $tokens);
134: }
135: }
136: } while ($level > 0);
137: }
138:
139: /**
140: * Portably retrieve the tag name of a node; deals with older versions
141: * of libxml like 2.7.6
142: * @param DOMNode $node
143: */
144: protected function getTagName($node)
145: {
146: if (isset($node->tagName)) {
147: return $node->tagName;
148: } else if (isset($node->nodeName)) {
149: return $node->nodeName;
150: } else if (isset($node->localName)) {
151: return $node->localName;
152: }
153: return null;
154: }
155:
156: /**
157: * Portably retrieve the data of a node; deals with older versions
158: * of libxml like 2.7.6
159: * @param DOMNode $node
160: */
161: protected function getData($node)
162: {
163: if (isset($node->data)) {
164: return $node->data;
165: } else if (isset($node->nodeValue)) {
166: return $node->nodeValue;
167: } else if (isset($node->textContent)) {
168: return $node->textContent;
169: }
170: return null;
171: }
172:
173:
174: /**
175: * @param DOMNode $node DOMNode to be tokenized.
176: * @param HTMLPurifier_Token[] $tokens Array-list of already tokenized tokens.
177: * @param bool $collect Says whether or start and close are collected, set to
178: * false at first recursion because it's the implicit DIV
179: * tag you're dealing with.
180: * @return bool if the token needs an endtoken
181: * @todo data and tagName properties don't seem to exist in DOMNode?
182: */
183: protected function createStartNode($node, &$tokens, $collect, $config)
184: {
185: // intercept non element nodes. WE MUST catch all of them,
186: // but we're not getting the character reference nodes because
187: // those should have been preprocessed
188: if ($node->nodeType === XML_TEXT_NODE) {
189: $data = $this->getData($node); // Handle variable data property
190: if ($data !== null) {
191: $tokens[] = $this->factory->createText($data);
192: }
193: return false;
194: } elseif ($node->nodeType === XML_CDATA_SECTION_NODE) {
195: // undo libxml's special treatment of <script> and <style> tags
196: $last = end($tokens);
197: $data = $node->data;
198: // (note $node->tagname is already normalized)
199: if ($last instanceof HTMLPurifier_Token_Start && ($last->name == 'script' || $last->name == 'style')) {
200: $new_data = trim($data);
201: if (substr($new_data, 0, 4) === '<!--') {
202: $data = substr($new_data, 4);
203: if (substr($data, -3) === '-->') {
204: $data = substr($data, 0, -3);
205: } else {
206: // Highly suspicious! Not sure what to do...
207: }
208: }
209: }
210: $tokens[] = $this->factory->createText($this->parseText($data, $config));
211: return false;
212: } elseif ($node->nodeType === XML_COMMENT_NODE) {
213: // this is code is only invoked for comments in script/style in versions
214: // of libxml pre-2.6.28 (regular comments, of course, are still
215: // handled regularly)
216: $tokens[] = $this->factory->createComment($node->data);
217: return false;
218: } elseif ($node->nodeType !== XML_ELEMENT_NODE) {
219: // not-well tested: there may be other nodes we have to grab
220: return false;
221: }
222: $attr = $node->hasAttributes() ? $this->transformAttrToAssoc($node->attributes) : array();
223: $tag_name = $this->getTagName($node); // Handle variable tagName property
224: if (empty($tag_name)) {
225: return (bool) $node->childNodes->length;
226: }
227: // We still have to make sure that the element actually IS empty
228: if (!$node->childNodes->length) {
229: if ($collect) {
230: $tokens[] = $this->factory->createEmpty($tag_name, $attr);
231: }
232: return false;
233: } else {
234: if ($collect) {
235: $tokens[] = $this->factory->createStart($tag_name, $attr);
236: }
237: return true;
238: }
239: }
240:
241: /**
242: * @param DOMNode $node
243: * @param HTMLPurifier_Token[] $tokens
244: */
245: protected function createEndNode($node, &$tokens)
246: {
247: $tag_name = $this->getTagName($node); // Handle variable tagName property
248: $tokens[] = $this->factory->createEnd($tag_name);
249: }
250:
251: /**
252: * Converts a DOMNamedNodeMap of DOMAttr objects into an assoc array.
253: *
254: * @param DOMNamedNodeMap $node_map DOMNamedNodeMap of DOMAttr objects.
255: * @return array Associative array of attributes.
256: */
257: protected function transformAttrToAssoc($node_map)
258: {
259: // NamedNodeMap is documented very well, so we're using undocumented
260: // features, namely, the fact that it implements Iterator and
261: // has a ->length attribute
262: if ($node_map->length === 0) {
263: return array();
264: }
265: $array = array();
266: foreach ($node_map as $attr) {
267: $array[$attr->name] = $attr->value;
268: }
269: return $array;
270: }
271:
272: /**
273: * An error handler that mutes all errors
274: * @param int $errno
275: * @param string $errstr
276: */
277: public function muteErrorHandler($errno, $errstr)
278: {
279: }
280:
281: /**
282: * Callback function for undoing escaping of stray angled brackets
283: * in comments
284: * @param array $matches
285: * @return string
286: */
287: public function callbackUndoCommentSubst($matches)
288: {
289: return '<!--' . strtr($matches[1], array('&amp;' => '&', '&lt;' => '<')) . $matches[2];
290: }
291:
292: /**
293: * Callback function that entity-izes ampersands in comments so that
294: * callbackUndoCommentSubst doesn't clobber them
295: * @param array $matches
296: * @return string
297: */
298: public function callbackArmorCommentEntities($matches)
299: {
300: return '<!--' . str_replace('&', '&amp;', $matches[1]) . $matches[2];
301: }
302:
303: /**
304: * Wraps an HTML fragment in the necessary HTML
305: * @param string $html
306: * @param HTMLPurifier_Config $config
307: * @param HTMLPurifier_Context $context
308: * @return string
309: */
310: protected function wrapHTML($html, $config, $context, $use_div = true)
311: {
312: $def = $config->getDefinition('HTML');
313: $ret = '';
314:
315: if (!empty($def->doctype->dtdPublic) || !empty($def->doctype->dtdSystem)) {
316: $ret .= '<!DOCTYPE html ';
317: if (!empty($def->doctype->dtdPublic)) {
318: $ret .= 'PUBLIC "' . $def->doctype->dtdPublic . '" ';
319: }
320: if (!empty($def->doctype->dtdSystem)) {
321: $ret .= '"' . $def->doctype->dtdSystem . '" ';
322: }
323: $ret .= '>';
324: }
325:
326: $ret .= '<html><head>';
327: $ret .= '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />';
328: // No protection if $html contains a stray </div>!
329: $ret .= '</head><body>';
330: if ($use_div) $ret .= '<div>';
331: $ret .= $html;
332: if ($use_div) $ret .= '</div>';
333: $ret .= '</body></html>';
334: return $ret;
335: }
336: }
337:
338: // vim: et sw=4 sts=4
339:
API documentation generated by ApiGen v7.0.0-alpha.6