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