includes/tidy/RemexCompatMunger.php

   1 <?php
   2
   3 namespace MediaWiki\Tidy;
   4
   5 use RemexHtml\HTMLData;
   6 use RemexHtml\Serializer\Serializer;
   7 use RemexHtml\Serializer\SerializerNode;
   8 use RemexHtml\Tokenizer\Attributes;
   9 use RemexHtml\Tokenizer\PlainAttributes;
  10 use RemexHtml\TreeBuilder\TreeBuilder;
  11 use RemexHtml\TreeBuilder\TreeHandler;
  12 use RemexHtml\TreeBuilder\Element;
  13
  14 /**
  15  * @internal
  16  */
  17 class RemexCompatMunger implements TreeHandler {
  18         private static $onlyInlineElements = [
  19                 "a" => true,
  20                 "abbr" => true,
  21                 "acronym" => true,
  22                 "applet" => true,
  23                 "b" => true,
  24                 "basefont" => true,
  25                 "bdo" => true,
  26                 "big" => true,
  27                 "br" => true,
  28                 "button" => true,
  29                 "cite" => true,
  30                 "code" => true,
  31                 "del" => true,
  32                 "dfn" => true,
  33                 "em" => true,
  34                 "font" => true,
  35                 "i" => true,
  36                 "iframe" => true,
  37                 "img" => true,
  38                 "input" => true,
  39                 "ins" => true,
  40                 "kbd" => true,
  41                 "label" => true,
  42                 "legend" => true,
  43                 "map" => true,
  44                 "object" => true,
  45                 "param" => true,
  46                 "q" => true,
  47                 "rb" => true,
  48                 "rbc" => true,
  49                 "rp" => true,
  50                 "rt" => true,
  51                 "rtc" => true,
  52                 "ruby" => true,
  53                 "s" => true,
  54                 "samp" => true,
  55                 "select" => true,
  56                 "small" => true,
  57                 "span" => true,
  58                 "strike" => true,
  59                 "strong" => true,
  60                 "sub" => true,
  61                 "sup" => true,
  62                 "textarea" => true,
  63                 "tt" => true,
  64                 "u" => true,
  65                 "var" => true,
  66                 // Those defined in tidy.conf
  67                 "video" => true,
  68                 "audio" => true,
  69                 "bdi" => true,
  70                 "data" => true,
  71                 "time" => true,
  72                 "mark" => true,
  73         ];
  74
  75         private static $formattingElements = [
  76                 'a' => true,
  77                 'b' => true,
  78                 'big' => true,
  79                 'code' => true,
  80                 'em' => true,
  81                 'font' => true,
  82                 'i' => true,
  83                 'nobr' => true,
  84                 's' => true,
  85                 'small' => true,
  86                 'strike' => true,
  87                 'strong' => true,
  88                 'tt' => true,
  89                 'u' => true,
  90         ];
  91
  92         /** @var Serializer */
  93         private $serializer;
  94
  95         /** @var bool */
  96         private $trace;
  97
  98         /**
  99          * @param Serializer $serializer
 100          * @param bool $trace
 101          */
 102         public function __construct( Serializer $serializer, $trace = false ) {
 103                 $this->serializer = $serializer;
 104                 $this->trace = $trace;
 105         }
 106
 107         public function startDocument( $fragmentNamespace, $fragmentName ) {
 108                 $this->serializer->startDocument( $fragmentNamespace, $fragmentName );
 109                 $root = $this->serializer->getRootNode();
 110                 $root->snData = new RemexMungerData;
 111                 $root->snData->needsPWrapping = true;
 112         }
 113
 114         public function endDocument( $pos ) {
 115                 $this->serializer->endDocument( $pos );
 116         }
 117
 118         private function getParentForInsert( $preposition, $refElement ) {
 119                 if ( $preposition === TreeBuilder::ROOT ) {
 120                         return [ $this->serializer->getRootNode(), null ];
 121                 } elseif ( $preposition === TreeBuilder::BEFORE ) {
 122                         $refNode = $refElement->userData;
 123                         return [ $this->serializer->getParentNode( $refNode ), $refNode ];
 124                 } else {
 125                         $refNode = $refElement->userData;
 126                         $refData = $refNode->snData;
 127                         if ( $refData->currentCloneElement ) {
 128                                 // Follow a chain of clone links if necessary
 129                                 $origRefData = $refData;
 130                                 while ( $refData->currentCloneElement ) {
 131                                         $refElement = $refData->currentCloneElement;
 132                                         $refNode = $refElement->userData;
 133                                         $refData = $refNode->snData;
 134                                 }
 135                                 // Cache the end of the chain in the requested element
 136                                 $origRefData->currentCloneElement = $refElement;
 137                         } elseif ( $refData->childPElement ) {
 138                                 $refElement = $refData->childPElement;
 139                                 $refNode = $refElement->userData;
 140                         }
 141                         return [ $refNode, $refNode ];
 142                 }
 143         }
 144
 145         /**
 146          * Insert a p-wrapper
 147          *
 148          * @param SerializerNode $parent
 149          * @param int $sourceStart
 150          * @return SerializerNode
 151          */
 152         private function insertPWrapper( SerializerNode $parent, $sourceStart ) {
 153                 $pWrap = new Element( HTMLData::NS_HTML, 'mw:p-wrap', new PlainAttributes );
 154                 $this->serializer->insertElement( TreeBuilder::UNDER, $parent, $pWrap, false,
 155                         $sourceStart, 0 );
 156                 $data = new RemexMungerData;
 157                 $data->isPWrapper = true;
 158                 $data->wrapBaseNode = $parent;
 159                 $pWrap->userData->snData = $data;
 160                 $parent->snData->childPElement = $pWrap;
 161                 return $pWrap->userData;
 162         }
 163
 164         public function characters( $preposition, $refElement, $text, $start, $length,
 165                 $sourceStart, $sourceLength
 166         ) {
 167                 $isBlank = strspn( $text, "\t\n\f\r ", $start, $length ) === $length;
 168
 169                 list( $parent, $refNode ) = $this->getParentForInsert( $preposition, $refElement );
 170                 $parentData = $parent->snData;
 171
 172                 if ( $preposition === TreeBuilder::UNDER ) {
 173                         if ( $parentData->needsPWrapping && !$isBlank ) {
 174                                 // Add a p-wrapper for bare text under body/blockquote
 175                                 $refNode = $this->insertPWrapper( $refNode, $sourceStart );
 176                                 $parent = $refNode;
 177                                 $parentData = $parent->snData;
 178                         } elseif ( $parentData->isSplittable && !$parentData->ancestorPNode ) {
 179                                 // The parent is splittable and in block mode, so split the tag stack
 180                                 $refNode = $this->splitTagStack( $refNode, true, $sourceStart );
 181                                 $parent = $refNode;
 182                                 $parentData = $parent->snData;
 183                         }
 184                 }
 185
 186                 if ( !$isBlank ) {
 187                         // Non-whitespace characters detected
 188                         $parentData->nonblankNodeCount++;
 189                 }
 190                 $this->serializer->characters( $preposition, $refNode, $text, $start,
 191                         $length, $sourceStart, $sourceLength );
 192         }
 193
 194         private function trace( $msg ) {
 195                 if ( $this->trace ) {
 196                         wfDebug( "[RCM] $msg" );
 197                 }
 198         }
 199
 200         /**
 201          * Insert or reparent an element. Create p-wrappers or split the tag stack
 202          * as necessary.
 203          *
 204          * Consider the following insertion locations. The parent may be:
 205          *
 206          *   - A: A body or blockquote (!!needsPWrapping)
 207          *   - B: A p-wrapper (!!isPWrapper)
 208          *   - C: A descendant of a p-wrapper (!!ancestorPNode)
 209          *     - CS: With splittable formatting elements in the stack region up to
 210          *       the p-wrapper
 211          *     - CU: With one or more unsplittable elements in the stack region up
 212          *       to the p-wrapper
 213          *   - D: Not a descendant of a p-wrapper (!ancestorNode)
 214          *     - DS: With splittable formatting elements in the stack region up to
 215          *       the body or blockquote
 216          *     - DU: With one or more unsplittable elements in the stack region up
 217          *       to the body or blockquote
 218          *
 219          * And consider that we may insert two types of element:
 220          *   - b: block
 221          *   - i: inline
 222          *
 223          * We handle the insertion as follows:
 224          *
 225          *   - A/i: Create a p-wrapper, insert under it
 226          *   - A/b: Insert as normal
 227          *   - B/i: Insert as normal
 228          *   - B/b: Close the p-wrapper, insert under the body/blockquote (wrap
 229          *     base) instead)
 230          *   - C/i: Insert as normal
 231          *   - CS/b: Split the tag stack, insert the block under cloned formatting
 232          *     elements which have the wrap base (the parent of the p-wrap) as
 233          *     their ultimate parent.
 234          *   - CU/b: Disable the p-wrap, by reparenting the currently open child
 235          *     of the p-wrap under the p-wrap's parent. Then insert the block as
 236          *     normal.
 237          *   - D/b: Insert as normal
 238          *   - DS/i: Split the tag stack, creating a new p-wrapper as the ultimate
 239          *     parent of the formatting elements thus cloned. The parent of the
 240          *     p-wrapper is the body or blockquote.
 241          *   - DU/i: Insert as normal
 242          *
 243          * FIXME: fostering ($preposition == BEFORE) is mostly done by inserting as
 244          * normal, the full algorithm is not followed.
 245          *
 246          * @param int $preposition
 247          * @param Element|SerializerNode|null $refElement
 248          * @param Element $element
 249          * @param bool $void
 250          * @param int $sourceStart
 251          * @param int $sourceLength
 252          */
 253         public function insertElement( $preposition, $refElement, Element $element, $void,
 254                 $sourceStart, $sourceLength
 255         ) {
 256                 list( $parent, $newRef ) = $this->getParentForInsert( $preposition, $refElement );
 257                 $parentData = $parent->snData;
 258                 $elementName = $element->htmlName;
 259
 260                 $inline = isset( self::$onlyInlineElements[$elementName] );
 261                 $under = $preposition === TreeBuilder::UNDER;
 262                 $elementToEnd = null;
 263
 264                 if ( $under && $parentData->isPWrapper && !$inline ) {
 265                         // [B/b] The element is non-inline and the parent is a p-wrapper,
 266                         // close the parent and insert into its parent instead
 267                         $this->trace( 'insert B/b' );
 268                         $newParent = $this->serializer->getParentNode( $parent );
 269                         $parent = $newParent;
 270                         $parentData = $parent->snData;
 271                         $pElement = $parentData->childPElement;
 272                         $parentData->childPElement = null;
 273                         $newRef = $refElement->userData;
 274                 } elseif ( $under && $parentData->isSplittable
 275                         && (bool)$parentData->ancestorPNode !== $inline
 276                 ) {
 277                         // [CS/b, DS/i] The parent is splittable and the current element is
 278                         // inline in block context, or if the current element is a block
 279                         // under a p-wrapper, split the tag stack.
 280                         $this->trace( $inline ? 'insert DS/i' : 'insert CS/b' );
 281                         $newRef = $this->splitTagStack( $newRef, $inline, $sourceStart );
 282                         $parent = $newRef;
 283                         $parentData = $parent->snData;
 284                 } elseif ( $under && $parentData->needsPWrapping && $inline ) {
 285                         // [A/i] If the element is inline and we are in body/blockquote,
 286                         // we need to create a p-wrapper
 287                         $this->trace( 'insert A/i' );
 288                         $newRef = $this->insertPWrapper( $newRef, $sourceStart );
 289                         $parent = $newRef;
 290                         $parentData = $parent->snData;
 291                 } elseif ( $parentData->ancestorPNode && !$inline ) {
 292                         // [CU/b] If the element is non-inline and (despite attempting to
 293                         // split above) there is still an ancestor p-wrap, disable that
 294                         // p-wrap
 295                         $this->trace( 'insert CU/b' );
 296                         $this->disablePWrapper( $parent, $sourceStart );
 297                 } else {
 298                         // [A/b, B/i, C/i, D/b, DU/i] insert as normal
 299                         $this->trace( 'insert normal' );
 300                 }
 301
 302                 // An element with element children is a non-blank element
 303                 $parentData->nonblankNodeCount++;
 304
 305                 // Insert the element downstream and so initialise its userData
 306                 $this->serializer->insertElement( $preposition, $newRef,
 307                         $element, $void, $sourceStart, $sourceLength );
 308
 309                 // Initialise snData
 310                 if ( !$element->userData->snData ) {
 311                         $elementData = $element->userData->snData = new RemexMungerData;
 312                 } else {
 313                         $elementData = $element->userData->snData;
 314                 }
 315                 if ( ( $parentData->isPWrapper || $parentData->isSplittable )
 316                         && isset( self::$formattingElements[$elementName] )
 317                 ) {
 318                         $elementData->isSplittable = true;
 319                 }
 320                 if ( $parentData->isPWrapper ) {
 321                         $elementData->ancestorPNode = $parent;
 322                 } elseif ( $parentData->ancestorPNode ) {
 323                         $elementData->ancestorPNode = $parentData->ancestorPNode;
 324                 }
 325                 if ( $parentData->wrapBaseNode ) {
 326                         $elementData->wrapBaseNode = $parentData->wrapBaseNode;
 327                 } elseif ( $parentData->needsPWrapping ) {
 328                         $elementData->wrapBaseNode = $parent;
 329                 }
 330                 if ( $elementName === 'body'
 331                         || $elementName === 'blockquote'
 332                         || $elementName === 'html'
 333                 ) {
 334                         $elementData->needsPWrapping = true;
 335                 }
 336         }
 337
 338         /**
 339          * Clone nodes in a stack range and return the new parent
 340          *
 341          * @param SerializerNode $parentNode
 342          * @param bool $inline
 343          * @param int $pos The source position
 344          * @return SerializerNode
 345          */
 346         private function splitTagStack( SerializerNode $parentNode, $inline, $pos ) {
 347                 $parentData = $parentNode->snData;
 348                 $wrapBase = $parentData->wrapBaseNode;
 349                 $pWrap = $parentData->ancestorPNode;
 350                 if ( !$pWrap ) {
 351                         $cloneEnd = $wrapBase;
 352                 } else {
 353                         $cloneEnd = $parentData->ancestorPNode;
 354                 }
 355
 356                 $serializer = $this->serializer;
 357                 $node = $parentNode;
 358                 $root = $serializer->getRootNode();
 359                 $nodes = [];
 360                 $removableNodes = [];
 361                 while ( $node !== $cloneEnd ) {
 362                         $nextParent = $serializer->getParentNode( $node );
 363                         if ( $nextParent === $root ) {
 364                                 throw new \Exception( 'Did not find end of clone range' );
 365                         }
 366                         $nodes[] = $node;
 367                         if ( $node->snData->nonblankNodeCount === 0 ) {
 368                                 $removableNodes[] = $node;
 369                                 $nextParent->snData->nonblankNodeCount--;
 370                         }
 371                         $node = $nextParent;
 372                 }
 373
 374                 if ( $inline ) {
 375                         $pWrap = $this->insertPWrapper( $wrapBase, $pos );
 376                         $node = $pWrap;
 377                 } else {
 378                         if ( $pWrap ) {
 379                                 // End the p-wrap which was open, cancel the diversion
 380                                 $wrapBase->snData->childPElement = null;
 381                         }
 382                         $pWrap = null;
 383                         $node = $wrapBase;
 384                 }
 385
 386                 for ( $i = count( $nodes ) - 1; $i >= 0; $i-- ) {
 387                         $oldNode = $nodes[$i];
 388                         $oldData = $oldNode->snData;
 389                         $nodeParent = $node;
 390                         $element = new Element( $oldNode->namespace, $oldNode->name, $oldNode->attrs );
 391                         $this->serializer->insertElement( TreeBuilder::UNDER, $nodeParent,
 392                                 $element, false, $pos, 0 );
 393                         $oldData->currentCloneElement = $element;
 394
 395                         $newNode = $element->userData;
 396                         $newData = $newNode->snData = new RemexMungerData;
 397                         if ( $pWrap ) {
 398                                 $newData->ancestorPNode = $pWrap;
 399                         }
 400                         $newData->isSplittable = true;
 401                         $newData->wrapBaseNode = $wrapBase;
 402                         $newData->isPWrapper = $oldData->isPWrapper;
 403
 404                         $nodeParent->snData->nonblankNodeCount++;
 405
 406                         $node = $newNode;
 407                 }
 408                 foreach ( $removableNodes as $rNode ) {
 409                         $fakeElement = new Element( $rNode->namespace, $rNode->name, $rNode->attrs );
 410                         $fakeElement->userData = $rNode;
 411                         $this->serializer->removeNode( $fakeElement, $pos );
 412                 }
 413                 return $node;
 414         }
 415
 416         /**
 417          * Find the ancestor of $node which is a child of a p-wrapper, and
 418          * reparent that node so that it is placed after the end of the p-wrapper
 419          */
 420         private function disablePWrapper( SerializerNode $node, $sourceStart ) {
 421                 $nodeData = $node->snData;
 422                 $pWrapNode = $nodeData->ancestorPNode;
 423                 $newParent = $this->serializer->getParentNode( $pWrapNode );
 424                 if ( $pWrapNode !== $this->serializer->getLastChild( $newParent ) ) {
 425                         // Fostering or something? Abort!
 426                         return;
 427                 }
 428
 429                 $nextParent = $node;
 430                 do {
 431                         $victim = $nextParent;
 432                         $victim->snData->ancestorPNode = null;
 433                         $nextParent = $this->serializer->getParentNode( $victim );
 434                 } while ( $nextParent !== $pWrapNode );
 435
 436                 // Make a fake Element to use in a reparenting operation
 437                 $victimElement = new Element( $victim->namespace, $victim->name, $victim->attrs );
 438                 $victimElement->userData = $victim;
 439
 440                 // Reparent
 441                 $this->serializer->insertElement( TreeBuilder::UNDER, $newParent, $victimElement,
 442                         false, $sourceStart, 0 );
 443
 444                 // Decrement nonblank node count
 445                 $pWrapNode->snData->nonblankNodeCount--;
 446
 447                 // Cancel the diversion so that no more elements are inserted under this p-wrap
 448                 $newParent->snData->childPElement = null;
 449         }
 450
 451         public function endTag( Element $element, $sourceStart, $sourceLength ) {
 452                 $data = $element->userData->snData;
 453                 if ( $data->childPElement ) {
 454                         $this->endTag( $data->childPElement, $sourceStart, 0 );
 455                 }
 456                 $this->serializer->endTag( $element, $sourceStart, $sourceLength );
 457                 $element->userData->snData = null;
 458                 $element->userData = null;
 459         }
 460
 461         public function doctype( $name, $public, $system, $quirks, $sourceStart, $sourceLength ) {
 462                 $this->serializer->doctype( $name, $public,  $system, $quirks,
 463                         $sourceStart, $sourceLength );
 464         }
 465
 466         public function comment( $preposition, $refElement, $text, $sourceStart, $sourceLength ) {
 467                 list( $parent, $refNode ) = $this->getParentForInsert( $preposition, $refElement );
 468                 $this->serializer->comment( $preposition, $refNode, $text,
 469                         $sourceStart, $sourceLength );
 470         }
 471
 472         public function error( $text, $pos ) {
 473                 $this->serializer->error( $text, $pos );
 474         }
 475
 476         public function mergeAttributes( Element $element, Attributes $attrs, $sourceStart ) {
 477                 $this->serializer->mergeAttributes( $element, $attrs, $sourceStart );
 478         }
 479
 480         public function removeNode( Element $element, $sourceStart ) {
 481                 $this->serializer->removeNode( $element, $sourceStart );
 482         }
 483
 484         public function reparentChildren( Element $element, Element $newParent, $sourceStart ) {
 485                 $self = $element->userData;
 486                 if ( $self->snData->childPElement ) {
 487                         // Reparent under the p-wrapper instead, so that e.g.
 488                         //   <blockquote><mw:p-wrap>...</mw:p-wrap></blockquote>
 489                         // becomes
 490                         //   <blockquote><mw:p-wrap><i>...</i></mw:p-wrap></blockquote>
 491
 492                         // The formatting element should not be the parent of the p-wrap.
 493                         // Without this special case, the insertElement() of the <i> below
 494                         // would be diverted into the p-wrapper, causing infinite recursion
 495                         // (T178632)
 496                         $this->reparentChildren( $self->snData->childPElement, $newParent, $sourceStart );
 497                         return;
 498                 }
 499
 500                 $children = $self->children;
 501                 $self->children = [];
 502                 $this->insertElement( TreeBuilder::UNDER, $element, $newParent, false, $sourceStart, 0 );
 503                 $newParentNode = $newParent->userData;
 504                 $newParentId = $newParentNode->id;
 505                 foreach ( $children as $child ) {
 506                         if ( is_object( $child ) ) {
 507                                 $this->trace( "reparent <{$child->name}>" );
 508                                 $child->parentId = $newParentId;
 509                         }
 510                 }
 511                 $newParentNode->children = $children;
 512         }
 513 }