3 * Performs transformations of HTML by wrapping around libxml2 and working
4 * around its countless bugs.
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation; either version 2 of the License, or
9 * (at your option) any later version.
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
16 * You should have received a copy of the GNU General Public License along
17 * with this program; if not, write to the Free Software Foundation, Inc.,
18 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
19 * http://www.gnu.org/copyleft/gpl.html
30 private $itemsToRemove = array();
31 private $elementsToFlatten = array();
32 protected $removeMedia = false;
37 * @param string $html Text to process
39 public function __construct( $html ) {
44 * Turns a chunk of HTML into a proper document
48 public static function wrapHTML( $html ) {
49 return '<!doctype html><html><head></head><body>' . $html . '</body></html>';
53 * Override this in descendant class to modify HTML after it has been converted from DOM tree
54 * @param string $html HTML to process
55 * @return string Processed HTML
57 protected function onHtmlReady( $html ) {
62 * @return DOMDocument DOM to manipulate
64 public function getDoc() {
66 $html = mb_convert_encoding( $this->html
, 'HTML-ENTITIES', 'UTF-8' );
68 // Workaround for bug that caused spaces before references
69 // to disappear during processing:
70 // https://bugzilla.wikimedia.org/show_bug.cgi?id=53086
72 // Please replace with a better fix if one can be found.
73 $html = str_replace( ' <', ' <', $html );
75 libxml_use_internal_errors( true );
76 $loader = libxml_disable_entity_loader();
77 $this->doc
= new DOMDocument();
78 $this->doc
->strictErrorChecking
= false;
79 $this->doc
->loadHTML( $html );
80 libxml_disable_entity_loader( $loader );
81 libxml_use_internal_errors( false );
82 $this->doc
->encoding
= 'UTF-8';
88 * Sets whether images/videos/sounds should be removed from output
91 public function setRemoveMedia( $flag = true ) {
92 $this->removeMedia
= $flag;
96 * Adds one or more selector of content to remove. A subset of CSS selector
97 * syntax is supported:
104 * @param array|string $selectors Selector(s) of stuff to remove
106 public function remove( $selectors ) {
107 $this->itemsToRemove
= array_merge( $this->itemsToRemove
, (array)$selectors );
111 * Adds one or more element name to the list to flatten (remove tag, but not its content)
112 * Can accept undelimited regexes
114 * Note this interface may fail in surprising unexpected ways due to usage of regexes,
115 * so should not be relied on for HTML markup security measures.
117 * @param array|string $elements Name(s) of tag(s) to flatten
119 public function flatten( $elements ) {
120 $this->elementsToFlatten
= array_merge( $this->elementsToFlatten
, (array)$elements );
124 * Instructs the formatter to flatten all tags
126 public function flattenAllTags() {
127 $this->flatten( '[?!]?[a-z0-9]+' );
131 * Removes content we've chosen to remove. The text of the removed elements can be
132 * extracted with the getText method.
133 * @return array Array of removed DOMElements
135 public function filterContent() {
136 $removals = $this->parseItemsToRemove();
138 // Bail out early if nothing to do
139 if ( array_reduce( $removals,
140 function ( $carry, $item ) {
141 return $carry && !$item;
148 $doc = $this->getDoc();
152 // You can't remove DOMNodes from a DOMNodeList as you're iterating
153 // over them in a foreach loop. It will seemingly leave the internal
154 // iterator on the foreach out of wack and results will be quite
155 // strange. Though, making a queue of items to remove seems to work.
156 $domElemsToRemove = array();
157 foreach ( $removals['TAG'] as $tagToRemove ) {
158 $tagToRemoveNodes = $doc->getElementsByTagName( $tagToRemove );
159 foreach ( $tagToRemoveNodes as $tagToRemoveNode ) {
160 if ( $tagToRemoveNode ) {
161 $domElemsToRemove[] = $tagToRemoveNode;
165 $removed = $this->removeElements( $domElemsToRemove );
167 // Elements with named IDs
168 $domElemsToRemove = array();
169 foreach ( $removals['ID'] as $itemToRemove ) {
170 $itemToRemoveNode = $doc->getElementById( $itemToRemove );
171 if ( $itemToRemoveNode ) {
172 $domElemsToRemove[] = $itemToRemoveNode;
175 $removed = array_merge( $removed, $this->removeElements( $domElemsToRemove ) );
178 $domElemsToRemove = array();
179 $xpath = new DOMXPath( $doc );
180 foreach ( $removals['CLASS'] as $classToRemove ) {
181 $elements = $xpath->query( '//*[contains(@class, "' . $classToRemove . '")]' );
183 /** @var $element DOMElement */
184 foreach ( $elements as $element ) {
185 $classes = $element->getAttribute( 'class' );
186 if ( preg_match( "/\b$classToRemove\b/", $classes ) && $element->parentNode
) {
187 $domElemsToRemove[] = $element;
191 $removed = array_merge( $removed, $this->removeElements( $domElemsToRemove ) );
193 // Tags with CSS Classes
194 foreach ( $removals['TAG_CLASS'] as $classToRemove ) {
195 $parts = explode( '.', $classToRemove );
197 $elements = $xpath->query(
198 '//' . $parts[0] . '[@class="' . $parts[1] . '"]'
200 $removed = array_merge( $removed, $this->removeElements( $elements ) );
207 * Removes a list of elelments from DOMDocument
208 * @param array|DOMNodeList $elements
209 * @return array Array of removed elements
211 private function removeElements( $elements ) {
213 if ( $elements instanceof DOMNodeList
) {
215 foreach ( $elements as $element ) {
219 /** @var $element DOMElement */
220 foreach ( $list as $element ) {
221 if ( $element->parentNode
) {
222 $element->parentNode
->removeChild( $element );
229 * libxml in its usual pointlessness converts many chars to entities - this function
230 * perfoms a reverse conversion
231 * @param string $html
234 private function fixLibXML( $html ) {
235 static $replacements;
236 if ( !$replacements ) {
237 // We don't include rules like '"' => '&quot;' because entities had already been
238 // normalized by libxml. Using this function with input not sanitized by libxml is UNSAFE!
239 $replacements = new ReplacementArray( array(
240 '"' => '&quot;',
241 '&' => '&amp;',
242 '<' => '&lt;',
243 '>' => '&gt;',
246 $html = $replacements->replace( $html );
247 $html = mb_convert_encoding( $html, 'UTF-8', 'HTML-ENTITIES' );
252 * Performs final transformations and returns resulting HTML. Note that if you want to call this
253 * both without an element and with an element you should call it without an element first. If you
254 * specify the $element in the method it'll change the underlying dom and you won't be able to get
257 * @param DOMElement|string|null $element ID of element to get HTML from or
258 * false to get it from the whole tree
259 * @return string Processed HTML
261 public function getText( $element = null ) {
264 wfProfileIn( __METHOD__
. '-dom' );
265 if ( $element !== null && !( $element instanceof DOMElement
) ) {
266 $element = $this->doc
->getElementById( $element );
269 $body = $this->doc
->getElementsByTagName( 'body' )->item( 0 );
270 $nodesArray = array();
271 foreach ( $body->childNodes
as $node ) {
272 $nodesArray[] = $node;
274 foreach ( $nodesArray as $nodeArray ) {
275 $body->removeChild( $nodeArray );
277 $body->appendChild( $element );
279 $html = $this->doc
->saveHTML();
280 wfProfileOut( __METHOD__
. '-dom' );
282 wfProfileIn( __METHOD__
. '-fixes' );
283 $html = $this->fixLibXml( $html );
284 if ( wfIsWindows() ) {
285 // Cleanup for CRLF misprocessing of unknown origin on Windows.
287 // If this error continues in the future, please track it down in the
288 // XML code paths if possible and fix there.
289 $html = str_replace( ' ', '', $html );
291 wfProfileOut( __METHOD__
. '-fixes' );
295 // Remove stuff added by wrapHTML()
296 $html = preg_replace( '/<!--.*?-->|^.*?<body>|<\/body>.*$/s', '', $html );
297 $html = $this->onHtmlReady( $html );
299 wfProfileIn( __METHOD__
. '-flatten' );
300 if ( $this->elementsToFlatten
) {
301 $elements = implode( '|', $this->elementsToFlatten
);
302 $html = preg_replace( "#</?($elements)\\b[^>]*>#is", '', $html );
304 wfProfileOut( __METHOD__
. '-flatten' );
310 * Helper function for parseItemsToRemove(). This function extracts the selector type
311 * and the raw name of a selector from a CSS-style selector string and assigns those
312 * values to parameters passed by reference. For example, if given '#toc' as the
313 * $selector parameter, it will assign 'ID' as the $type and 'toc' as the $rawName.
314 * @param string $selector CSS selector to parse
315 * @param string $type The type of selector (ID, CLASS, TAG_CLASS, or TAG)
316 * @param string $rawName The raw name of the selector
317 * @return bool Whether the selector was successfully recognised
318 * @throws MWException
320 protected function parseSelector( $selector, &$type, &$rawName ) {
321 if ( strpos( $selector, '.' ) === 0 ) {
323 $rawName = substr( $selector, 1 );
324 } elseif ( strpos( $selector, '#' ) === 0 ) {
326 $rawName = substr( $selector, 1 );
327 } elseif ( strpos( $selector, '.' ) !== 0 && strpos( $selector, '.' ) !== false ) {
329 $rawName = $selector;
330 } elseif ( strpos( $selector, '[' ) === false && strpos( $selector, ']' ) === false ) {
332 $rawName = $selector;
334 throw new MWException( __METHOD__
. "(): unrecognized selector '$selector'" );
341 * Transforms CSS-style selectors into an internal representation suitable for
342 * processing by filterContent()
345 protected function parseItemsToRemove() {
350 'TAG_CLASS' => array(),
353 foreach ( $this->itemsToRemove
as $itemToRemove ) {
356 if ( $this->parseSelector( $itemToRemove, $type, $rawName ) ) {
357 $removals[$type][] = $rawName;
361 if ( $this->removeMedia
) {
362 $removals['TAG'][] = 'img';
363 $removals['TAG'][] = 'audio';
364 $removals['TAG'][] = 'video';