ILIAS  Release_5_0_x_branch Revision 61816
 All Data Structures Namespaces Files Functions Variables Groups Pages
DOMLex.php
Go to the documentation of this file.
1 <?php
2 
27 class HTMLPurifier_Lexer_DOMLex extends HTMLPurifier_Lexer
28 {
29 
33  private $factory;
34 
35  public function __construct()
36  {
37  // setup the factory
38  parent::__construct();
39  $this->factory = new HTMLPurifier_TokenFactory();
40  }
41 
48  public function tokenizeHTML($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  $doc->encoding = 'UTF-8'; // theoretically, the above has this covered
70 
71  set_error_handler(array($this, 'muteErrorHandler'));
72  $doc->loadHTML($html);
73  restore_error_handler();
74 
75  $tokens = array();
76  $this->tokenizeDOM(
77  $doc->getElementsByTagName('html')->item(0)-> // <html>
78  getElementsByTagName('body')->item(0)-> // <body>
79  getElementsByTagName('div')->item(0), // <div>
80  $tokens
81  );
82  return $tokens;
83  }
84 
92  protected function tokenizeDOM($node, &$tokens)
93  {
94  $level = 0;
95  $nodes = array($level => new HTMLPurifier_Queue(array($node)));
96  $closingNodes = array();
97  do {
98  while (!$nodes[$level]->isEmpty()) {
99  $node = $nodes[$level]->shift(); // FIFO
100  $collect = $level > 0 ? true : false;
101  $needEndingTag = $this->createStartNode($node, $tokens, $collect);
102  if ($needEndingTag) {
103  $closingNodes[$level][] = $node;
104  }
105  if ($node->childNodes && $node->childNodes->length) {
106  $level++;
107  $nodes[$level] = new HTMLPurifier_Queue();
108  foreach ($node->childNodes as $childNode) {
109  $nodes[$level]->push($childNode);
110  }
111  }
112  }
113  $level--;
114  if ($level && isset($closingNodes[$level])) {
115  while ($node = array_pop($closingNodes[$level])) {
116  $this->createEndNode($node, $tokens);
117  }
118  }
119  } while ($level > 0);
120  }
121 
131  protected function createStartNode($node, &$tokens, $collect)
132  {
133  // intercept non element nodes. WE MUST catch all of them,
134  // but we're not getting the character reference nodes because
135  // those should have been preprocessed
136  if ($node->nodeType === XML_TEXT_NODE) {
137  $tokens[] = $this->factory->createText($node->data);
138  return false;
139  } elseif ($node->nodeType === XML_CDATA_SECTION_NODE) {
140  // undo libxml's special treatment of <script> and <style> tags
141  $last = end($tokens);
142  $data = $node->data;
143  // (note $node->tagname is already normalized)
144  if ($last instanceof HTMLPurifier_Token_Start && ($last->name == 'script' || $last->name == 'style')) {
145  $new_data = trim($data);
146  if (substr($new_data, 0, 4) === '<!--') {
147  $data = substr($new_data, 4);
148  if (substr($data, -3) === '-->') {
149  $data = substr($data, 0, -3);
150  } else {
151  // Highly suspicious! Not sure what to do...
152  }
153  }
154  }
155  $tokens[] = $this->factory->createText($this->parseData($data));
156  return false;
157  } elseif ($node->nodeType === XML_COMMENT_NODE) {
158  // this is code is only invoked for comments in script/style in versions
159  // of libxml pre-2.6.28 (regular comments, of course, are still
160  // handled regularly)
161  $tokens[] = $this->factory->createComment($node->data);
162  return false;
163  } elseif ($node->nodeType !== XML_ELEMENT_NODE) {
164  // not-well tested: there may be other nodes we have to grab
165  return false;
166  }
167 
168  $attr = $node->hasAttributes() ? $this->transformAttrToAssoc($node->attributes) : array();
169 
170  // We still have to make sure that the element actually IS empty
171  if (!$node->childNodes->length) {
172  if ($collect) {
173  $tokens[] = $this->factory->createEmpty($node->tagName, $attr);
174  }
175  return false;
176  } else {
177  if ($collect) {
178  $tokens[] = $this->factory->createStart(
179  $tag_name = $node->tagName, // somehow, it get's dropped
180  $attr
181  );
182  }
183  return true;
184  }
185  }
186 
191  protected function createEndNode($node, &$tokens)
192  {
193  $tokens[] = $this->factory->createEnd($node->tagName);
194  }
195 
196 
203  protected function transformAttrToAssoc($node_map)
204  {
205  // NamedNodeMap is documented very well, so we're using undocumented
206  // features, namely, the fact that it implements Iterator and
207  // has a ->length attribute
208  if ($node_map->length === 0) {
209  return array();
210  }
211  $array = array();
212  foreach ($node_map as $attr) {
213  $array[$attr->name] = $attr->value;
214  }
215  return $array;
216  }
217 
223  public function muteErrorHandler($errno, $errstr)
224  {
225  }
226 
233  public function callbackUndoCommentSubst($matches)
234  {
235  return '<!--' . strtr($matches[1], array('&amp;' => '&', '&lt;' => '<')) . $matches[2];
236  }
237 
244  public function callbackArmorCommentEntities($matches)
245  {
246  return '<!--' . str_replace('&', '&amp;', $matches[1]) . $matches[2];
247  }
248 
256  protected function wrapHTML($html, $config, $context)
257  {
258  $def = $config->getDefinition('HTML');
259  $ret = '';
260 
261  if (!empty($def->doctype->dtdPublic) || !empty($def->doctype->dtdSystem)) {
262  $ret .= '<!DOCTYPE html ';
263  if (!empty($def->doctype->dtdPublic)) {
264  $ret .= 'PUBLIC "' . $def->doctype->dtdPublic . '" ';
265  }
266  if (!empty($def->doctype->dtdSystem)) {
267  $ret .= '"' . $def->doctype->dtdSystem . '" ';
268  }
269  $ret .= '>';
270  }
271 
272  $ret .= '<html><head>';
273  $ret .= '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />';
274  // No protection if $html contains a stray </div>!
275  $ret .= '</head><body><div>' . $html . '</div></body></html>';
276  return $ret;
277  }
278 }
279 
280 // vim: et sw=4 sts=4