includes/HtmlFormatter.php

   1 <?php
   2 /**
   3  * Performs transformations of HTML by wrapping around libxml2 and working
   4  * around its countless bugs.
   5  *
   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.
  10  *
  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.
  15  *
  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
  20  *
  21  * @file
  22  */
  23 class HtmlFormatter {
  24         /**
  25          * @var DOMDocument
  26          */
  27         private $doc;
  28
  29         private $html;
  30         private $itemsToRemove = array();
  31         private $elementsToFlatten = array();
  32         protected $removeMedia = false;
  33
  34         /**
  35          * Constructor
  36          *
  37          * @param string $html: Text to process
  38          */
  39         public function __construct( $html ) {
  40                 $this->html = $html;
  41         }
  42
  43         /**
  44          * Turns a chunk of HTML into a proper document
  45          * @param string $html
  46          * @return string
  47          */
  48         public static function wrapHTML( $html ) {
  49                 return '<!doctype html><html><head></head><body>' . $html . '</body></html>';
  50         }
  51
  52         /**
  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
  56          */
  57         protected function onHtmlReady( $html ) {
  58                 return $html;
  59         }
  60
  61         /**
  62          * @return DOMDocument: DOM to manipulate
  63          */
  64         public function getDoc() {
  65                 if ( !$this->doc ) {
  66                         $html = mb_convert_encoding( $this->html, 'HTML-ENTITIES', 'UTF-8' );
  67
  68                         // Workaround for bug that caused spaces before references
  69                         // to disappear during processing:
  70                         // https://bugzilla.wikimedia.org/show_bug.cgi?id=53086
  71                         //
  72                         // Please replace with a better fix if one can be found.
  73                         $html = str_replace( ' <', '&#32;<', $html );
  74
  75                         libxml_use_internal_errors( true );
  76                         $this->doc = new DOMDocument();
  77                         $this->doc->strictErrorChecking = false;
  78                         $this->doc->loadHTML( $html );
  79                         libxml_use_internal_errors( false );
  80                         $this->doc->encoding = 'UTF-8';
  81                 }
  82                 return $this->doc;
  83         }
  84
  85         /**
  86          * Sets whether images/videos/sounds should be removed from output
  87          * @param bool $flag
  88          */
  89         public function setRemoveMedia( $flag = true ) {
  90                 $this->removeMedia = $flag;
  91         }
  92
  93         /**
  94          * Adds one or more selector of content to remove. A subset of CSS selector
  95          * syntax is supported:
  96          *
  97          *   <tag>
  98          *   <tag>.class
  99          *   .<class>
 100          *   #<id>
 101          *
 102          * @param Array|string $selectors: Selector(s) of stuff to remove
 103          */
 104         public function remove( $selectors ) {
 105                 $this->itemsToRemove = array_merge( $this->itemsToRemove, (array)$selectors );
 106         }
 107
 108         /**
 109          * Adds one or more element name to the list to flatten (remove tag, but not its content)
 110          * Can accept undelimited regexes
 111          *
 112          * Note this interface may fail in surprising unexpected ways due to usage of regexes,
 113          * so should not be relied on for HTML markup security measures.
 114          *
 115          * @param Array|string $elements: Name(s) of tag(s) to flatten
 116          */
 117         public function flatten( $elements ) {
 118                 $this->elementsToFlatten = array_merge( $this->elementsToFlatten, (array)$elements );
 119         }
 120
 121         /**
 122          * Instructs the formatter to flatten all tags
 123          */
 124         public function flattenAllTags() {
 125                 $this->flatten( '[?!]?[a-z0-9]+' );
 126         }
 127
 128         /**
 129          * Removes content we've chosen to remove
 130          */
 131         public function filterContent() {
 132                 wfProfileIn( __METHOD__ );
 133                 $removals = $this->parseItemsToRemove();
 134
 135                 if ( !$removals ) {
 136                         wfProfileOut( __METHOD__ );
 137                         return;
 138                 }
 139
 140                 $doc = $this->getDoc();
 141
 142                 // Remove tags
 143
 144                 // You can't remove DOMNodes from a DOMNodeList as you're iterating
 145                 // over them in a foreach loop. It will seemingly leave the internal
 146                 // iterator on the foreach out of wack and results will be quite
 147                 // strange. Though, making a queue of items to remove seems to work.
 148                 $domElemsToRemove = array();
 149                 foreach ( $removals['TAG'] as $tagToRemove ) {
 150                         $tagToRemoveNodes = $doc->getElementsByTagName( $tagToRemove );
 151                         foreach ( $tagToRemoveNodes as $tagToRemoveNode ) {
 152                                 if ( $tagToRemoveNode ) {
 153                                         $domElemsToRemove[] = $tagToRemoveNode;
 154                                 }
 155                         }
 156                 }
 157
 158                 $this->removeElements( $domElemsToRemove );
 159
 160                 // Elements with named IDs
 161                 $domElemsToRemove = array();
 162                 foreach ( $removals['ID'] as $itemToRemove ) {
 163                         $itemToRemoveNode = $doc->getElementById( $itemToRemove );
 164                         if ( $itemToRemoveNode ) {
 165                                 $domElemsToRemove[] = $itemToRemoveNode;
 166                         }
 167                 }
 168                 $this->removeElements( $domElemsToRemove );
 169
 170                 // CSS Classes
 171                 $domElemsToRemove = array();
 172                 $xpath = new DOMXpath( $doc );
 173                 foreach ( $removals['CLASS'] as $classToRemove ) {
 174                         $elements = $xpath->query( '//*[contains(@class, "' . $classToRemove . '")]' );
 175
 176                         /** @var $element DOMElement */
 177                         foreach ( $elements as $element ) {
 178                                 $classes = $element->getAttribute( 'class' );
 179                                 if ( preg_match( "/\b$classToRemove\b/", $classes ) && $element->parentNode ) {
 180                                         $domElemsToRemove[] = $element;
 181                                 }
 182                         }
 183                 }
 184                 $this->removeElements( $domElemsToRemove );
 185
 186                 // Tags with CSS Classes
 187                 foreach ( $removals['TAG_CLASS'] as $classToRemove ) {
 188                         $parts = explode( '.', $classToRemove );
 189
 190                         $elements = $xpath->query(
 191                                 '//' . $parts[0] . '[@class="' . $parts[1] . '"]'
 192                         );
 193
 194                         $this->removeElements( $elements );
 195                 }
 196
 197                 wfProfileOut( __METHOD__ );
 198         }
 199
 200         /**
 201          * Removes a list of elelments from DOMDocument
 202          * @param array|DOMNodeList $elements
 203          */
 204         private function removeElements( $elements ) {
 205                 $list = $elements;
 206                 if ( $elements instanceof DOMNodeList ) {
 207                         $list = array();
 208                         foreach ( $elements as $element ) {
 209                                 $list[] = $element;
 210                         }
 211                 }
 212                 /** @var $element DOMElement */
 213                 foreach ( $list as $element ) {
 214                         if ( $element->parentNode ) {
 215                                 $element->parentNode->removeChild( $element );
 216                         }
 217                 }
 218         }
 219
 220         /**
 221          * libxml in its usual pointlessness converts many chars to entities - this function
 222          * perfoms a reverse conversion
 223          * @param string $html
 224          * @return string
 225          */
 226         private function fixLibXML( $html ) {
 227                 wfProfileIn( __METHOD__ );
 228                 static $replacements;
 229                 if ( ! $replacements ) {
 230                         // We don't include rules like '&#34;' => '&amp;quot;' because entities had already been
 231                         // normalized by libxml. Using this function with input not sanitized by libxml is UNSAFE!
 232                         $replacements = new ReplacementArray( array(
 233                                 '&quot;' => '&amp;quot;',
 234                                 '&amp;' => '&amp;amp;',
 235                                 '&lt;' => '&amp;lt;',
 236                                 '&gt;' => '&amp;gt;',
 237                         ) );
 238                 }
 239                 $html = $replacements->replace( $html );
 240                 $html = mb_convert_encoding( $html, 'UTF-8', 'HTML-ENTITIES' );
 241                 wfProfileOut( __METHOD__ );
 242                 return $html;
 243         }
 244
 245         /**
 246          * Performs final transformations and returns resulting HTML
 247          *
 248          * @param DOMElement|string|null $element: ID of element to get HTML from or false to get it from the whole tree
 249          * @return string: Processed HTML
 250          */
 251         public function getText( $element = null ) {
 252                 wfProfileIn( __METHOD__ );
 253
 254                 if ( $this->doc ) {
 255                         if ( $element !== null && !( $element instanceof DOMElement ) ) {
 256                                 $element = $this->doc->getElementById( $element );
 257                         }
 258                         if ( $element ) {
 259                                 $body = $this->doc->getElementsByTagName( 'body' )->item( 0 );
 260                                 $nodesArray = array();
 261                                 foreach ( $body->childNodes as $node ) {
 262                                         $nodesArray[] = $node;
 263                                 }
 264                                 foreach ( $nodesArray as $nodeArray ) {
 265                                         $body->removeChild( $nodeArray );
 266                                 }
 267                                 $body->appendChild( $element );
 268                         }
 269                         $html = $this->doc->saveHTML();
 270                         $html = $this->fixLibXml( $html );
 271                 } else {
 272                         $html = $this->html;
 273                 }
 274                 if ( wfIsWindows() ) {
 275                         // Appears to be cleanup for CRLF misprocessing of unknown origin
 276                         // when running server on Windows platform.
 277                         //
 278                         // If this error continues in the future, please track it down in the
 279                         // XML code paths if possible and fix there.
 280                         $html = str_replace( '&#13;', '', $html );
 281                 }
 282                 $html = preg_replace( '/<!--.*?-->|^.*?<body>|<\/body>.*$/s', '', $html );
 283                 $html = $this->onHtmlReady( $html );
 284
 285                 if ( $this->elementsToFlatten ) {
 286                         $elements = implode( '|', $this->elementsToFlatten );
 287                         $html = preg_replace( "#</?($elements)\\b[^>]*>#is", '', $html );
 288                 }
 289
 290                 wfProfileOut( __METHOD__ );
 291                 return $html;
 292         }
 293
 294         /**
 295          * @param $selector: CSS selector to parse
 296          * @param $type
 297          * @param $rawName
 298          * @return bool: Whether the selector was successfully recognised
 299          */
 300         protected function parseSelector( $selector, &$type, &$rawName ) {
 301                 if ( strpos( $selector, '.' ) === 0 ) {
 302                         $type = 'CLASS';
 303                         $rawName = substr( $selector, 1 );
 304                 } elseif ( strpos( $selector, '#' ) === 0 ) {
 305                         $type = 'ID';
 306                         $rawName = substr( $selector, 1 );
 307                 } elseif ( strpos( $selector, '.' ) !== 0 &&
 308                         strpos( $selector, '.' ) !== false )
 309                 {
 310                         $type = 'TAG_CLASS';
 311                         $rawName = $selector;
 312                 } elseif ( strpos( $selector, '[' ) === false
 313                         && strpos( $selector, ']' ) === false )
 314                 {
 315                         $type = 'TAG';
 316                         $rawName = $selector;
 317                 } else {
 318                         throw new MWException( __METHOD__ . "(): unrecognized selector '$selector'" );
 319                 }
 320
 321                 return true;
 322         }
 323
 324         /**
 325          * Transforms CSS selectors into an internal representation suitable for processing
 326          * @return array
 327          */
 328         protected function parseItemsToRemove() {
 329                 wfProfileIn( __METHOD__ );
 330                 $removals = array(
 331                         'ID' => array(),
 332                         'TAG' => array(),
 333                         'CLASS' => array(),
 334                         'TAG_CLASS' => array(),
 335                 );
 336
 337                 foreach ( $this->itemsToRemove as $itemToRemove ) {
 338                         $type = '';
 339                         $rawName = '';
 340                         if ( $this->parseSelector( $itemToRemove, $type, $rawName ) ) {
 341                                 $removals[$type][] = $rawName;
 342                         }
 343                 }
 344
 345                 if ( $this->removeMedia ) {
 346                         $removals['TAG'][] = 'img';
 347                         $removals['TAG'][] = 'audio';
 348                         $removals['TAG'][] = 'video';
 349                 }
 350
 351                 wfProfileOut( __METHOD__ );
 352                 return $removals;
 353         }
 354 }