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