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