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 = [];
  31         private $elementsToFlatten = [];
  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                         // DOMDocument::loadHTML apparently isn't very good with encodings, so
  67                         // convert input to ASCII by encoding everything above 128 as entities.
  68                         if ( function_exists( 'mb_convert_encoding' ) ) {
  69                                 $html = mb_convert_encoding( $this->html, 'HTML-ENTITIES', 'UTF-8' );
  70                         } else {
  71                                 $html = preg_replace_callback( '/[\x{80}-\x{10ffff}]/u', function ( $m ) {
  72                                         return '&#' . UtfNormal\Utils::utf8ToCodepoint( $m[0] ) . ';';
  73                                 }, $this->html );
  74                         }
  75
  76                         // Workaround for bug that caused spaces before references
  77                         // to disappear during processing: https://phabricator.wikimedia.org/T55086
  78                         // TODO: Please replace with a better fix if one can be found.
  79                         $html = str_replace( ' <', '&#32;<', $html );
  80
  81                         libxml_use_internal_errors( true );
  82                         $loader = libxml_disable_entity_loader();
  83                         $this->doc = new DOMDocument();
  84                         $this->doc->strictErrorChecking = false;
  85                         $this->doc->loadHTML( $html );
  86                         libxml_disable_entity_loader( $loader );
  87                         libxml_use_internal_errors( false );
  88                         $this->doc->encoding = 'UTF-8';
  89                 }
  90                 return $this->doc;
  91         }
  92
  93         /**
  94          * Sets whether images/videos/sounds should be removed from output
  95          * @param bool $flag
  96          */
  97         public function setRemoveMedia( $flag = true ) {
  98                 $this->removeMedia = $flag;
  99         }
 100
 101         /**
 102          * Adds one or more selector of content to remove. A subset of CSS selector
 103          * syntax is supported:
 104          *
 105          *   <tag>
 106          *   <tag>.class
 107          *   .<class>
 108          *   #<id>
 109          *
 110          * @param array|string $selectors Selector(s) of stuff to remove
 111          */
 112         public function remove( $selectors ) {
 113                 $this->itemsToRemove = array_merge( $this->itemsToRemove, (array)$selectors );
 114         }
 115
 116         /**
 117          * Adds one or more element name to the list to flatten (remove tag, but not its content)
 118          * Can accept undelimited regexes
 119          *
 120          * Note this interface may fail in surprising unexpected ways due to usage of regexes,
 121          * so should not be relied on for HTML markup security measures.
 122          *
 123          * @param array|string $elements Name(s) of tag(s) to flatten
 124          */
 125         public function flatten( $elements ) {
 126                 $this->elementsToFlatten = array_merge( $this->elementsToFlatten, (array)$elements );
 127         }
 128
 129         /**
 130          * Instructs the formatter to flatten all tags
 131          */
 132         public function flattenAllTags() {
 133                 $this->flatten( '[?!]?[a-z0-9]+' );
 134         }
 135
 136         /**
 137          * Removes content we've chosen to remove.  The text of the removed elements can be
 138          * extracted with the getText method.
 139          * @return array Array of removed DOMElements
 140          */
 141         public function filterContent() {
 142                 $removals = $this->parseItemsToRemove();
 143
 144                 // Bail out early if nothing to do
 145                 if ( array_reduce( $removals,
 146                         function ( $carry, $item ) {
 147                                 return $carry && !$item;
 148                         },
 149                         true
 150                 ) ) {
 151                         return [];
 152                 }
 153
 154                 $doc = $this->getDoc();
 155
 156                 // Remove tags
 157
 158                 // You can't remove DOMNodes from a DOMNodeList as you're iterating
 159                 // over them in a foreach loop. It will seemingly leave the internal
 160                 // iterator on the foreach out of wack and results will be quite
 161                 // strange. Though, making a queue of items to remove seems to work.
 162                 $domElemsToRemove = [];
 163                 foreach ( $removals['TAG'] as $tagToRemove ) {
 164                         $tagToRemoveNodes = $doc->getElementsByTagName( $tagToRemove );
 165                         foreach ( $tagToRemoveNodes as $tagToRemoveNode ) {
 166                                 if ( $tagToRemoveNode ) {
 167                                         $domElemsToRemove[] = $tagToRemoveNode;
 168                                 }
 169                         }
 170                 }
 171                 $removed = $this->removeElements( $domElemsToRemove );
 172
 173                 // Elements with named IDs
 174                 $domElemsToRemove = [];
 175                 foreach ( $removals['ID'] as $itemToRemove ) {
 176                         $itemToRemoveNode = $doc->getElementById( $itemToRemove );
 177                         if ( $itemToRemoveNode ) {
 178                                 $domElemsToRemove[] = $itemToRemoveNode;
 179                         }
 180                 }
 181                 $removed = array_merge( $removed, $this->removeElements( $domElemsToRemove ) );
 182
 183                 // CSS Classes
 184                 $domElemsToRemove = [];
 185                 $xpath = new DOMXPath( $doc );
 186                 foreach ( $removals['CLASS'] as $classToRemove ) {
 187                         $elements = $xpath->query( '//*[contains(@class, "' . $classToRemove . '")]' );
 188
 189                         /** @var $element DOMElement */
 190                         foreach ( $elements as $element ) {
 191                                 $classes = $element->getAttribute( 'class' );
 192                                 if ( preg_match( "/\b$classToRemove\b/", $classes ) && $element->parentNode ) {
 193                                         $domElemsToRemove[] = $element;
 194                                 }
 195                         }
 196                 }
 197                 $removed = array_merge( $removed, $this->removeElements( $domElemsToRemove ) );
 198
 199                 // Tags with CSS Classes
 200                 foreach ( $removals['TAG_CLASS'] as $classToRemove ) {
 201                         $parts = explode( '.', $classToRemove );
 202
 203                         $elements = $xpath->query(
 204                                 '//' . $parts[0] . '[@class="' . $parts[1] . '"]'
 205                         );
 206                         $removed = array_merge( $removed, $this->removeElements( $elements ) );
 207                 }
 208
 209                 return $removed;
 210         }
 211
 212         /**
 213          * Removes a list of elelments from DOMDocument
 214          * @param array|DOMNodeList $elements
 215          * @return array Array of removed elements
 216          */
 217         private function removeElements( $elements ) {
 218                 $list = $elements;
 219                 if ( $elements instanceof DOMNodeList ) {
 220                         $list = [];
 221                         foreach ( $elements as $element ) {
 222                                 $list[] = $element;
 223                         }
 224                 }
 225                 /** @var $element DOMElement */
 226                 foreach ( $list as $element ) {
 227                         if ( $element->parentNode ) {
 228                                 $element->parentNode->removeChild( $element );
 229                         }
 230                 }
 231                 return $list;
 232         }
 233
 234         /**
 235          * libxml in its usual pointlessness converts many chars to entities - this function
 236          * perfoms a reverse conversion
 237          * @param string $html
 238          * @return string
 239          */
 240         private function fixLibXML( $html ) {
 241                 static $replacements;
 242                 if ( !$replacements ) {
 243                         // We don't include rules like '&#34;' => '&amp;quot;' because entities had already been
 244                         // normalized by libxml. Using this function with input not sanitized by libxml is UNSAFE!
 245                         $replacements = new ReplacementArray( [
 246                                 '&quot;' => '&amp;quot;',
 247                                 '&amp;' => '&amp;amp;',
 248                                 '&lt;' => '&amp;lt;',
 249                                 '&gt;' => '&amp;gt;',
 250                         ] );
 251                 }
 252                 $html = $replacements->replace( $html );
 253
 254                 if ( function_exists( 'mb_convert_encoding' ) ) {
 255                         // Just in case the conversion in getDoc() above used named
 256                         // entities that aren't known to html_entity_decode().
 257                         $html = mb_convert_encoding( $html, 'UTF-8', 'HTML-ENTITIES' );
 258                 } else {
 259                         $html = html_entity_decode( $html, ENT_COMPAT, 'utf-8' );
 260                 }
 261                 return $html;
 262         }
 263
 264         /**
 265          * Performs final transformations and returns resulting HTML.  Note that if you want to call this
 266          * both without an element and with an element you should call it without an element first.  If you
 267          * specify the $element in the method it'll change the underlying dom and you won't be able to get
 268          * it back.
 269          *
 270          * @param DOMElement|string|null $element ID of element to get HTML from or
 271          *   false to get it from the whole tree
 272          * @return string Processed HTML
 273          */
 274         public function getText( $element = null ) {
 275
 276                 if ( $this->doc ) {
 277                         if ( $element !== null && !( $element instanceof DOMElement ) ) {
 278                                 $element = $this->doc->getElementById( $element );
 279                         }
 280                         if ( $element ) {
 281                                 $body = $this->doc->getElementsByTagName( 'body' )->item( 0 );
 282                                 $nodesArray = [];
 283                                 foreach ( $body->childNodes as $node ) {
 284                                         $nodesArray[] = $node;
 285                                 }
 286                                 foreach ( $nodesArray as $nodeArray ) {
 287                                         $body->removeChild( $nodeArray );
 288                                 }
 289                                 $body->appendChild( $element );
 290                         }
 291                         $html = $this->doc->saveHTML();
 292
 293                         $html = $this->fixLibXml( $html );
 294                         if ( wfIsWindows() ) {
 295                                 // Cleanup for CRLF misprocessing of unknown origin on Windows.
 296                                 // If this error continues in the future, please track it down in the
 297                                 // XML code paths if possible and fix there.
 298                                 $html = str_replace( '&#13;', '', $html );
 299                         }
 300                 } else {
 301                         $html = $this->html;
 302                 }
 303                 // Remove stuff added by wrapHTML()
 304                 $html = preg_replace( '/<!--.*?-->|^.*?<body>|<\/body>.*$/s', '', $html );
 305                 $html = $this->onHtmlReady( $html );
 306
 307                 if ( $this->elementsToFlatten ) {
 308                         $elements = implode( '|', $this->elementsToFlatten );
 309                         $html = preg_replace( "#</?($elements)\\b[^>]*>#is", '', $html );
 310                 }
 311
 312                 return $html;
 313         }
 314
 315         /**
 316          * Helper function for parseItemsToRemove(). This function extracts the selector type
 317          * and the raw name of a selector from a CSS-style selector string and assigns those
 318          * values to parameters passed by reference. For example, if given '#toc' as the
 319          * $selector parameter, it will assign 'ID' as the $type and 'toc' as the $rawName.
 320          * @param string $selector CSS selector to parse
 321          * @param string $type The type of selector (ID, CLASS, TAG_CLASS, or TAG)
 322          * @param string $rawName The raw name of the selector
 323          * @return bool Whether the selector was successfully recognised
 324          * @throws MWException
 325          */
 326         protected function parseSelector( $selector, &$type, &$rawName ) {
 327                 if ( strpos( $selector, '.' ) === 0 ) {
 328                         $type = 'CLASS';
 329                         $rawName = substr( $selector, 1 );
 330                 } elseif ( strpos( $selector, '#' ) === 0 ) {
 331                         $type = 'ID';
 332                         $rawName = substr( $selector, 1 );
 333                 } elseif ( strpos( $selector, '.' ) !== 0 && strpos( $selector, '.' ) !== false ) {
 334                         $type = 'TAG_CLASS';
 335                         $rawName = $selector;
 336                 } elseif ( strpos( $selector, '[' ) === false && strpos( $selector, ']' ) === false ) {
 337                         $type = 'TAG';
 338                         $rawName = $selector;
 339                 } else {
 340                         throw new MWException( __METHOD__ . "(): unrecognized selector '$selector'" );
 341                 }
 342
 343                 return true;
 344         }
 345
 346         /**
 347          * Transforms CSS-style selectors into an internal representation suitable for
 348          * processing by filterContent()
 349          * @return array
 350          */
 351         protected function parseItemsToRemove() {
 352                 $removals = [
 353                         'ID' => [],
 354                         'TAG' => [],
 355                         'CLASS' => [],
 356                         'TAG_CLASS' => [],
 357                 ];
 358
 359                 foreach ( $this->itemsToRemove as $itemToRemove ) {
 360                         $type = '';
 361                         $rawName = '';
 362                         if ( $this->parseSelector( $itemToRemove, $type, $rawName ) ) {
 363                                 $removals[$type][] = $rawName;
 364                         }
 365                 }
 366
 367                 if ( $this->removeMedia ) {
 368                         $removals['TAG'][] = 'img';
 369                         $removals['TAG'][] = 'audio';
 370                         $removals['TAG'][] = 'video';
 371                 }
 372
 373                 return $removals;
 374         }
 375 }