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