Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.32% |
1 / 313 |
|
4.76% |
1 / 21 |
CRAP | |
0.00% |
0 / 1 |
| DOMNormalizer | |
0.32% |
1 / 313 |
|
4.76% |
1 / 21 |
23035.26 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
6 | |||
| similar | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
132 | |||
| mergable | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| swappable | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| firstChild | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| isInsertedContent | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
20 | |||
| rewriteablePair | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
30 | |||
| addDiffMarks | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
182 | |||
| merge | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
30 | |||
| swap | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
6 | |||
| hoistLinks | |
0.00% |
0 / 26 |
|
0.00% |
0 / 1 |
156 | |||
| stripIfEmpty | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
12 | |||
| moveTrailingSpacesOut | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
56 | |||
| stripBRs | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
20 | |||
| stripBidiCharsAroundCategories | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
72 | |||
| moveFormatTagOutsideATag | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
156 | |||
| normalizeNode | |
0.00% |
0 / 62 |
|
0.00% |
0 / 1 |
1122 | |||
| normalizeSiblingPair | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
30 | |||
| processSubtree | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
42 | |||
| processNode | |
0.00% |
0 / 34 |
|
0.00% |
0 / 1 |
156 | |||
| normalize | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Html2Wt; |
| 5 | |
| 6 | use Wikimedia\Assert\Assert; |
| 7 | use Wikimedia\Assert\UnreachableException; |
| 8 | use Wikimedia\Parsoid\DOM\DocumentFragment; |
| 9 | use Wikimedia\Parsoid\DOM\Element; |
| 10 | use Wikimedia\Parsoid\DOM\Node; |
| 11 | use Wikimedia\Parsoid\DOM\Text; |
| 12 | use Wikimedia\Parsoid\Utils\ContentUtils; |
| 13 | use Wikimedia\Parsoid\Utils\DiffDOMUtils; |
| 14 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 15 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 16 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 17 | use Wikimedia\Parsoid\Utils\PHPUtils; |
| 18 | use Wikimedia\Parsoid\Utils\WTUtils; |
| 19 | use Wikimedia\Parsoid\Wikitext\Consts; |
| 20 | |
| 21 | /* |
| 22 | * Tag minimization |
| 23 | * ---------------- |
| 24 | * Minimize a pair of tags in the dom tree rooted at node. |
| 25 | * |
| 26 | * This function merges adjacent nodes of the same type |
| 27 | * and swaps nodes where possible to enable further merging. |
| 28 | * |
| 29 | * See examples below: |
| 30 | * |
| 31 | * 1. <b>X</b><b>Y</b> |
| 32 | * ==> <b>XY</b> |
| 33 | * |
| 34 | * 2. <i>A</i><b><i>X</i></b><b><i>Y</i></b><i>Z</i> |
| 35 | * ==> <i>A<b>XY</b>Z</i> |
| 36 | * |
| 37 | * 3. <a href="Football">Foot</a><a href="Football">ball</a> |
| 38 | * ==> <a href="Football">Football</a> |
| 39 | */ |
| 40 | |
| 41 | /** |
| 42 | * DOM normalization. |
| 43 | * |
| 44 | * DOM normalizations are performed after DOMDiff is run. |
| 45 | * So, normalization routines should update diff markers appropriately. |
| 46 | */ |
| 47 | class DOMNormalizer { |
| 48 | |
| 49 | private const IGNORABLE_ATTRS = [ |
| 50 | 'data-parsoid', 'id', 'title', DOMDataUtils::DATA_OBJECT_ATTR_NAME |
| 51 | ]; |
| 52 | private const HTML_IGNORABLE_ATTRS = [ 'data-parsoid', DOMDataUtils::DATA_OBJECT_ATTR_NAME ]; |
| 53 | |
| 54 | private static $specializedAttribHandlers; |
| 55 | |
| 56 | /** @var bool */ |
| 57 | private $inInsertedContent; |
| 58 | |
| 59 | /** @var SerializerState */ |
| 60 | private $state; |
| 61 | |
| 62 | public function __construct( SerializerState $state ) { |
| 63 | if ( !self::$specializedAttribHandlers ) { |
| 64 | self::$specializedAttribHandlers = [ |
| 65 | 'data-mw' => static function ( $nodeA, $dmwA, $nodeB, $dmwB ) { |
| 66 | return $dmwA == $dmwB; |
| 67 | } |
| 68 | ]; |
| 69 | } |
| 70 | |
| 71 | $this->state = $state; |
| 72 | |
| 73 | $this->inInsertedContent = false; |
| 74 | } |
| 75 | |
| 76 | private static function similar( Node $a, Node $b ): bool { |
| 77 | if ( DOMCompat::nodeName( $a ) === 'a' ) { |
| 78 | // FIXME: Similar to 1ce6a98, DiffDOMUtils::nextNonDeletedSibling is being |
| 79 | // used in this file where maybe DiffDOMUtils::nextNonSepSibling belongs. |
| 80 | return $a instanceof Element && $b instanceof Element && |
| 81 | DiffUtils::attribsEquals( $a, $b, self::IGNORABLE_ATTRS, self::$specializedAttribHandlers ); |
| 82 | } else { |
| 83 | $aIsHtml = WTUtils::isLiteralHTMLNode( $a ); |
| 84 | $bIsHtml = WTUtils::isLiteralHTMLNode( $b ); |
| 85 | // TODO (Anomie) |
| 86 | // It looks like $ignorableAttrs is only used when $aIsHtml is true. |
| 87 | // Or is that the fixme referred to in the comment below? |
| 88 | $ignorableAttrs = $aIsHtml ? self::HTML_IGNORABLE_ATTRS : self::IGNORABLE_ATTRS; |
| 89 | |
| 90 | // FIXME: For non-HTML I/B tags, we seem to be dropping all attributes |
| 91 | // in our tag handlers (which seems like a bug). Till that is fixed, |
| 92 | // we'll preserve existing functionality here. |
| 93 | return ( !$aIsHtml && !$bIsHtml ) || |
| 94 | ( $aIsHtml && $bIsHtml && |
| 95 | $a instanceof Element && $b instanceof Element && |
| 96 | DiffUtils::attribsEquals( $a, $b, $ignorableAttrs, self::$specializedAttribHandlers ) ); |
| 97 | } |
| 98 | } |
| 99 | |
| 100 | /** |
| 101 | * Can a and b be merged into a single node? |
| 102 | * @param Node $a |
| 103 | * @param Node $b |
| 104 | * @return bool |
| 105 | */ |
| 106 | private static function mergable( Node $a, Node $b ): bool { |
| 107 | return DOMCompat::nodeName( $a ) === DOMCompat::nodeName( $b ) && self::similar( $a, $b ); |
| 108 | } |
| 109 | |
| 110 | /** |
| 111 | * Can a and b be combined into a single node |
| 112 | * if we swap a and a.firstChild? |
| 113 | * |
| 114 | * For example: A='<b><i>x</i></b>' b='<i>y</i>' => '<i><b>x</b>y</i>'. |
| 115 | * @param Node $a |
| 116 | * @param Node $b |
| 117 | * @return bool |
| 118 | */ |
| 119 | private static function swappable( Node $a, Node $b ): bool { |
| 120 | return DiffDOMUtils::numNonDeletedChildNodes( $a ) === 1 |
| 121 | && self::similar( $a, DiffDOMUtils::firstNonDeletedChild( $a ) ) |
| 122 | && self::mergable( DiffDOMUtils::firstNonDeletedChild( $a ), $b ); |
| 123 | } |
| 124 | |
| 125 | private static function firstChild( Node $node, bool $rtl ): ?Node { |
| 126 | return $rtl ? DiffDOMUtils::lastNonDeletedChild( $node ) : DiffDOMUtils::firstNonDeletedChild( $node ); |
| 127 | } |
| 128 | |
| 129 | private function isInsertedContent( Node $node ): bool { |
| 130 | while ( true ) { |
| 131 | if ( DiffUtils::hasInsertedDiffMark( $node ) ) { |
| 132 | return true; |
| 133 | } |
| 134 | if ( DOMUtils::atTheTop( $node ) ) { |
| 135 | return false; |
| 136 | } |
| 137 | $node = $node->parentNode; |
| 138 | } |
| 139 | } |
| 140 | |
| 141 | private function rewriteablePair( Node $a, Node $b ): bool { |
| 142 | if ( isset( Consts::$WTQuoteTags[DOMCompat::nodeName( $a )] ) ) { |
| 143 | // For <i>/<b> pair, we need not check whether the node being transformed |
| 144 | // are new / edited, etc. since these minimization scenarios can |
| 145 | // never show up in HTML that came from parsed wikitext. |
| 146 | // |
| 147 | // <i>..</i><i>..</i> can never show up without a <nowiki/> in between. |
| 148 | // Similarly for <b>..</b><b>..</b> and <b><i>..</i></b><i>..</i>. |
| 149 | // |
| 150 | // This is because a sequence of 4 quotes is not parsed as ..</i><i>.. |
| 151 | // Neither is a sequence of 7 quotes parsed as ..</i></b><i>.. |
| 152 | // |
| 153 | // So, if we see a minimizable pair of nodes, it is because the HTML |
| 154 | // didn't originate from wikitext OR the HTML has been subsequently edited. |
| 155 | // In both cases, we want to transform the DOM. |
| 156 | |
| 157 | return isset( Consts::$WTQuoteTags[DOMCompat::nodeName( $b )] ); |
| 158 | } elseif ( DOMCompat::nodeName( $a ) === 'a' ) { |
| 159 | // For <a> tags, we require at least one of the two tags |
| 160 | // to be a newly created element. |
| 161 | return DOMCompat::nodeName( $b ) === 'a' && ( WTUtils::isNewElt( $a ) || WTUtils::isNewElt( $b ) ); |
| 162 | } |
| 163 | return false; |
| 164 | } |
| 165 | |
| 166 | public function addDiffMarks( Node $node, string $mark, bool $dontRecurse = false ): void { |
| 167 | if ( !$this->state->selserMode || DiffUtils::hasDiffMark( $node, $mark ) ) { |
| 168 | return; |
| 169 | } |
| 170 | |
| 171 | // Don't introduce nested inserted markers |
| 172 | if ( $this->inInsertedContent && $mark === DiffMarkers::INSERTED ) { |
| 173 | return; |
| 174 | } |
| 175 | |
| 176 | $env = $this->state->getEnv(); |
| 177 | |
| 178 | // Newly added elements don't need diff marks |
| 179 | if ( !WTUtils::isNewElt( $node ) ) { |
| 180 | DiffUtils::addDiffMark( $node, $env, $mark ); |
| 181 | if ( $mark === DiffMarkers::INSERTED || $mark === DiffMarkers::DELETED ) { |
| 182 | DiffUtils::addDiffMark( $node->parentNode, $env, DiffMarkers::CHILDREN_CHANGED ); |
| 183 | } |
| 184 | } |
| 185 | |
| 186 | if ( $dontRecurse ) { |
| 187 | return; |
| 188 | } |
| 189 | |
| 190 | // Walk up the subtree and add 'subtree-changed' markers |
| 191 | $node = $node->parentNode; |
| 192 | while ( $node instanceof Element && !DOMUtils::atTheTop( $node ) ) { |
| 193 | if ( DiffUtils::hasDiffMark( $node, DiffMarkers::SUBTREE_CHANGED ) ) { |
| 194 | return; |
| 195 | } |
| 196 | if ( !WTUtils::isNewElt( $node ) ) { |
| 197 | DiffUtils::addDiffMark( $node, $env, DiffMarkers::SUBTREE_CHANGED ); |
| 198 | } |
| 199 | $node = $node->parentNode; |
| 200 | } |
| 201 | } |
| 202 | |
| 203 | /** |
| 204 | * Transfer all of b's children to a and delete b. |
| 205 | * @param Element $a |
| 206 | * @param Element $b |
| 207 | * @return Element |
| 208 | */ |
| 209 | public function merge( Element $a, Element $b ): Element { |
| 210 | $sentinel = $b->firstChild; |
| 211 | |
| 212 | // Migrate any intermediate nodes (usually 0 / 1 diff markers) |
| 213 | // present between a and b to a |
| 214 | $next = $a->nextSibling; |
| 215 | if ( $next !== $b ) { |
| 216 | $a->appendChild( $next ); |
| 217 | } |
| 218 | |
| 219 | // The real work of merging |
| 220 | DOMUtils::migrateChildren( $b, $a ); |
| 221 | $b->parentNode->removeChild( $b ); |
| 222 | |
| 223 | // Normalize the node to merge any adjacent text nodes |
| 224 | DOMCompat::normalize( $a ); |
| 225 | |
| 226 | // Update diff markers |
| 227 | $this->addDiffMarks( $a->parentNode, DiffMarkers::CHILDREN_CHANGED ); // $b was removed |
| 228 | $this->addDiffMarks( $a, DiffMarkers::CHILDREN_CHANGED ); // $a got more children |
| 229 | if ( !DOMUtils::isRemoved( $sentinel ) ) { |
| 230 | // Nodes starting at 'sentinal' were inserted into 'a' |
| 231 | // b, which was a's sibling was deleted |
| 232 | // Only addDiffMarks to sentinel, if it is still part of the dom |
| 233 | // (and hasn't been deleted by the call to a.normalize() ) |
| 234 | if ( $sentinel->parentNode ) { |
| 235 | $this->addDiffMarks( $sentinel, DiffMarkers::MOVED, true ); |
| 236 | } |
| 237 | } |
| 238 | if ( $a->nextSibling ) { |
| 239 | // FIXME: Hmm .. there is an API hole here |
| 240 | // about ability to add markers after last child |
| 241 | $this->addDiffMarks( $a->nextSibling, DiffMarkers::MOVED, true ); |
| 242 | } |
| 243 | |
| 244 | return $a; |
| 245 | } |
| 246 | |
| 247 | /** |
| 248 | * b is a's sole non-deleted child. Switch them around. |
| 249 | * @param Element $a |
| 250 | * @param Element $b |
| 251 | * @return Element |
| 252 | */ |
| 253 | public function swap( Element $a, Element $b ): Element { |
| 254 | DOMUtils::migrateChildren( $b, $a ); |
| 255 | $a->parentNode->insertBefore( $b, $a ); |
| 256 | $b->appendChild( $a ); |
| 257 | |
| 258 | // Mark a's subtree, a, and b as all having moved |
| 259 | if ( $a->firstChild !== null ) { |
| 260 | $this->addDiffMarks( $a->firstChild, DiffMarkers::MOVED, true ); |
| 261 | } |
| 262 | $this->addDiffMarks( $a, DiffMarkers::MOVED, true ); |
| 263 | $this->addDiffMarks( $b, DiffMarkers::MOVED, true ); |
| 264 | $this->addDiffMarks( $a, DiffMarkers::CHILDREN_CHANGED, true ); |
| 265 | $this->addDiffMarks( $b, DiffMarkers::CHILDREN_CHANGED, true ); |
| 266 | $this->addDiffMarks( $b->parentNode, DiffMarkers::CHILDREN_CHANGED ); |
| 267 | |
| 268 | return $b; |
| 269 | } |
| 270 | |
| 271 | public function hoistLinks( Element $node, bool $rtl ): void { |
| 272 | $sibling = self::firstChild( $node, $rtl ); |
| 273 | $hasHoistableContent = false; |
| 274 | |
| 275 | while ( $sibling ) { |
| 276 | $next = $rtl |
| 277 | ? DiffDOMUtils::previousNonDeletedSibling( $sibling ) |
| 278 | : DiffDOMUtils::nextNonDeletedSibling( $sibling ); |
| 279 | if ( !DiffDOMUtils::isContentNode( $sibling ) ) { |
| 280 | // Nothing to do, continue. |
| 281 | } elseif ( !WTUtils::isRenderingTransparentNode( $sibling ) || |
| 282 | WTUtils::isEncapsulationWrapper( $sibling ) |
| 283 | ) { |
| 284 | // Don't venture into templated content |
| 285 | break; |
| 286 | } else { |
| 287 | $hasHoistableContent = true; |
| 288 | } |
| 289 | $sibling = $next; |
| 290 | } |
| 291 | |
| 292 | if ( $hasHoistableContent ) { |
| 293 | // soak up all the non-content nodes (exclude sibling) |
| 294 | $move = self::firstChild( $node, $rtl ); |
| 295 | $firstNode = $move; |
| 296 | while ( $move !== $sibling ) { |
| 297 | $refnode = $rtl ? DiffDOMUtils::nextNonDeletedSibling( $node ) : $node; |
| 298 | $node->parentNode->insertBefore( $move, $refnode ); |
| 299 | $move = self::firstChild( $node, $rtl ); |
| 300 | } |
| 301 | |
| 302 | // and drop any leading whitespace |
| 303 | if ( $sibling instanceof Text ) { |
| 304 | $sibling->nodeValue = $rtl ? rtrim( $sibling->nodeValue ) : ltrim( $sibling->nodeValue ); |
| 305 | } |
| 306 | |
| 307 | // Update diff markers |
| 308 | $this->addDiffMarks( $firstNode, DiffMarkers::MOVED, true ); |
| 309 | if ( $sibling ) { |
| 310 | $this->addDiffMarks( $sibling, DiffMarkers::MOVED, true ); |
| 311 | } |
| 312 | $this->addDiffMarks( $node, DiffMarkers::CHILDREN_CHANGED, true ); |
| 313 | $this->addDiffMarks( $node->parentNode, DiffMarkers::CHILDREN_CHANGED ); |
| 314 | } |
| 315 | } |
| 316 | |
| 317 | public function stripIfEmpty( Element $node ): ?Node { |
| 318 | $next = DiffDOMUtils::nextNonDeletedSibling( $node ); |
| 319 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 320 | $autoInserted = isset( $dp->autoInsertedStart ) || isset( $dp->autoInsertedEnd ); |
| 321 | |
| 322 | $strippable = |
| 323 | DiffDOMUtils::nodeEssentiallyEmpty( $node, false ); |
| 324 | // Ex: "<a..>..</a><b></b>bar" |
| 325 | // From [[Foo]]<b/>bar usage found on some dewiki pages. |
| 326 | // FIXME: Should we enable this? |
| 327 | // !( false /* used to be rt-test mode */ && ( $dp->stx ?? null ) === 'html' ); |
| 328 | |
| 329 | if ( $strippable ) { |
| 330 | // Update diff markers (before the deletion) |
| 331 | $this->addDiffMarks( $node, DiffMarkers::DELETED, true ); |
| 332 | $node->parentNode->removeChild( $node ); |
| 333 | return $next; |
| 334 | } else { |
| 335 | return $node; |
| 336 | } |
| 337 | } |
| 338 | |
| 339 | public function moveTrailingSpacesOut( Node $node ): void { |
| 340 | $next = DiffDOMUtils::nextNonDeletedSibling( $node ); |
| 341 | $last = DiffDOMUtils::lastNonDeletedChild( $node ); |
| 342 | $matches = null; |
| 343 | if ( $last instanceof Text && |
| 344 | preg_match( '/\s+$/D', $last->nodeValue, $matches ) > 0 |
| 345 | ) { |
| 346 | $trailing = $matches[0]; |
| 347 | $last->nodeValue = substr( $last->nodeValue, 0, -strlen( $trailing ) ); |
| 348 | // Try to be a little smarter and drop the spaces if possible. |
| 349 | if ( $next && ( !( $next instanceof Text ) || !preg_match( '/^\s+/', $next->nodeValue ) ) ) { |
| 350 | if ( !( $next instanceof Text ) ) { |
| 351 | $txt = $node->ownerDocument->createTextNode( '' ); |
| 352 | $node->parentNode->insertBefore( $txt, $next ); |
| 353 | $next = $txt; |
| 354 | } |
| 355 | $next->nodeValue = $trailing . $next->nodeValue; |
| 356 | // next (a text node) is new / had new content added to it |
| 357 | $this->addDiffMarks( $next, DiffMarkers::INSERTED, true ); |
| 358 | } |
| 359 | $this->addDiffMarks( $last, DiffMarkers::INSERTED, true ); |
| 360 | $this->addDiffMarks( $node->parentNode, DiffMarkers::CHILDREN_CHANGED ); |
| 361 | } |
| 362 | } |
| 363 | |
| 364 | public function stripBRs( Element $node ): void { |
| 365 | $child = $node->firstChild; |
| 366 | while ( $child ) { |
| 367 | $next = $child->nextSibling; |
| 368 | if ( DOMCompat::nodeName( $child ) === 'br' ) { |
| 369 | // replace <br/> with a single space |
| 370 | $node->removeChild( $child ); |
| 371 | $node->insertBefore( $node->ownerDocument->createTextNode( ' ' ), $next ); |
| 372 | } elseif ( $child instanceof Element ) { |
| 373 | $this->stripBRs( $child ); |
| 374 | } |
| 375 | $child = $next; |
| 376 | } |
| 377 | } |
| 378 | |
| 379 | /** |
| 380 | * FIXME see |
| 381 | * https://gerrit.wikimedia.org/r/#/c/mediawiki/services/parsoid/+/500975/7/src/Html2Wt/DOMNormalizer.php@423 |
| 382 | * @param Node $node |
| 383 | * @return Node|null |
| 384 | */ |
| 385 | public function stripBidiCharsAroundCategories( Node $node ): ?Node { |
| 386 | if ( !( $node instanceof Text ) || |
| 387 | ( !WTUtils::isCategoryLink( $node->previousSibling ) && |
| 388 | !WTUtils::isCategoryLink( $node->nextSibling ) ) |
| 389 | ) { |
| 390 | // Not a text node and not adjacent to a category link |
| 391 | return $node; |
| 392 | } |
| 393 | |
| 394 | $next = $node->nextSibling; |
| 395 | if ( !$next || WTUtils::isCategoryLink( $next ) ) { |
| 396 | // The following can leave behind an empty text node. |
| 397 | $oldLength = strlen( $node->nodeValue ); |
| 398 | $node->nodeValue = preg_replace( |
| 399 | '/([\x{200e}\x{200f}]+\n)?[\x{200e}\x{200f}]+$/uD', |
| 400 | '', |
| 401 | $node->nodeValue |
| 402 | ); |
| 403 | $newLength = strlen( $node->nodeValue ); |
| 404 | |
| 405 | if ( $oldLength !== $newLength ) { |
| 406 | // Log changes for editors benefit |
| 407 | $this->state->getEnv()->log( 'warn/html2wt/bidi', |
| 408 | 'LRM/RLM unicode chars stripped around categories' |
| 409 | ); |
| 410 | } |
| 411 | |
| 412 | if ( $newLength === 0 ) { |
| 413 | // Remove empty text nodes to keep DOM in normalized form |
| 414 | $ret = DiffDOMUtils::nextNonDeletedSibling( $node ); |
| 415 | $node->parentNode->removeChild( $node ); |
| 416 | $this->addDiffMarks( $node, DiffMarkers::DELETED ); |
| 417 | return $ret; |
| 418 | } |
| 419 | |
| 420 | // Treat modified node as having been newly inserted |
| 421 | $this->addDiffMarks( $node, DiffMarkers::INSERTED ); |
| 422 | } |
| 423 | return $node; |
| 424 | } |
| 425 | |
| 426 | /** |
| 427 | * When an A tag is encountered, if there are format tags inside, move them outside |
| 428 | * Also merge a single sibling A tag that is mergable |
| 429 | * The link href and text must match for this normalization to take effect |
| 430 | * |
| 431 | * @param Element $node |
| 432 | * @return Node|null |
| 433 | */ |
| 434 | public function moveFormatTagOutsideATag( Element $node ): ?Node { |
| 435 | if ( DOMCompat::nodeName( $node ) !== 'a' ) { |
| 436 | return $node; |
| 437 | } |
| 438 | $sibling = DiffDOMUtils::nextNonDeletedSibling( $node ); |
| 439 | if ( $sibling ) { |
| 440 | $this->normalizeSiblingPair( $node, $sibling ); |
| 441 | } |
| 442 | |
| 443 | $firstChild = DiffDOMUtils::firstNonDeletedChild( $node ); |
| 444 | $fcNextSibling = null; |
| 445 | if ( $firstChild ) { |
| 446 | $fcNextSibling = DiffDOMUtils::nextNonDeletedSibling( $firstChild ); |
| 447 | } |
| 448 | |
| 449 | if ( !$node->hasAttribute( 'href' ) ) { |
| 450 | return $node; |
| 451 | } |
| 452 | $nodeHref = DOMCompat::getAttribute( $node, 'href' ) ?? ''; |
| 453 | |
| 454 | // If there are no tags to swap, we are done |
| 455 | if ( $firstChild instanceof Element && |
| 456 | // No reordering possible with multiple children |
| 457 | $fcNextSibling === null && |
| 458 | // Do not normalize WikiLinks with these attributes |
| 459 | !$firstChild->hasAttribute( 'color' ) && |
| 460 | !$firstChild->hasAttribute( 'style' ) && |
| 461 | !$firstChild->hasAttribute( 'class' ) && |
| 462 | // Compare textContent to the href, noting that this matching doesn't handle all |
| 463 | // possible simple-wiki-link scenarios that isSimpleWikiLink in link handler tackles |
| 464 | $node->textContent === PHPUtils::stripPrefix( $nodeHref, './' ) |
| 465 | ) { |
| 466 | for ( |
| 467 | $child = DiffDOMUtils::firstNonDeletedChild( $node ); |
| 468 | DOMUtils::isFormattingElt( $child ); |
| 469 | $child = DiffDOMUtils::firstNonDeletedChild( $node ) |
| 470 | ) { |
| 471 | '@phan-var Element $child'; // @var Element $child |
| 472 | $this->swap( $node, $child ); |
| 473 | } |
| 474 | return $firstChild; |
| 475 | } |
| 476 | |
| 477 | return $node; |
| 478 | } |
| 479 | |
| 480 | /** |
| 481 | * Wikitext normalizations implemented right now: |
| 482 | * |
| 483 | * 1. Tag minimization (I/B tags) in normalizeSiblingPair |
| 484 | * 2. Strip empty headings and style tags |
| 485 | * 3. Force SOL transparent links to serialize before/after heading |
| 486 | * 4. Trailing spaces are migrated out of links |
| 487 | * 5. Space is added before escapable prefixes in table cells |
| 488 | * 6. Strip <br/> from headings |
| 489 | * 7. Strip bidi chars around categories |
| 490 | * 8. When an A tag is encountered, if there are format tags inside, move them outside |
| 491 | * |
| 492 | * The return value from this function should respect the |
| 493 | * following contract: |
| 494 | * - if input node is unmodified, return it. |
| 495 | * - if input node is modified, return the new node |
| 496 | * that it transforms into. |
| 497 | * If you return a node other than this, normalizations may not |
| 498 | * apply cleanly and may be skipped. |
| 499 | * |
| 500 | * @param Node $node |
| 501 | * @return Node|null the normalized node |
| 502 | */ |
| 503 | public function normalizeNode( Node $node ): ?Node { |
| 504 | $nodeName = DOMCompat::nodeName( $node ); |
| 505 | |
| 506 | if ( $this->state->getEnv()->getSiteConfig()->scrubBidiChars() ) { |
| 507 | // Strip bidirectional chars around categories |
| 508 | // Note that this is being done everywhere, |
| 509 | // not just in selser mode |
| 510 | $next = $this->stripBidiCharsAroundCategories( $node ); |
| 511 | if ( $next !== $node ) { |
| 512 | return $next; |
| 513 | } |
| 514 | } |
| 515 | |
| 516 | // Skip unmodified content |
| 517 | if ( $this->state->selserMode && !DOMUtils::atTheTop( $node ) && |
| 518 | !$this->inInsertedContent && |
| 519 | !DiffUtils::hasDiffMarkers( $node ) && |
| 520 | // If orig-src is not valid, this in effect becomes |
| 521 | // an edited node and needs normalizations applied to it. |
| 522 | WTSUtils::origSrcValidInEditedContext( $this->state, $node ) |
| 523 | ) { |
| 524 | return $node; |
| 525 | } |
| 526 | |
| 527 | // Headings |
| 528 | if ( DOMUtils::isHeading( $node ) ) { |
| 529 | '@phan-var Element $node'; // @var Element $node |
| 530 | $this->hoistLinks( $node, false ); |
| 531 | $this->hoistLinks( $node, true ); |
| 532 | $this->stripBRs( $node ); |
| 533 | |
| 534 | return $this->stripIfEmpty( $node ); |
| 535 | |
| 536 | // Quote tags |
| 537 | } elseif ( isset( Consts::$WTQuoteTags[$nodeName] ) ) { |
| 538 | '@phan-var Element $node'; // @var Element $node |
| 539 | return $this->stripIfEmpty( $node ); |
| 540 | |
| 541 | // Anchors |
| 542 | } elseif ( $nodeName === 'a' ) { |
| 543 | '@phan-var Element $node'; // @var Element $node |
| 544 | $next = DiffDOMUtils::nextNonDeletedSibling( $node ); |
| 545 | // We could have checked for !mw:ExtLink but in |
| 546 | // the case of links without any annotations, |
| 547 | // the positive test is semantically safer than the |
| 548 | // negative test. |
| 549 | if ( DOMUtils::hasRel( $node, 'mw:WikiLink' ) && |
| 550 | $this->stripIfEmpty( $node ) !== $node |
| 551 | ) { |
| 552 | return $next; |
| 553 | } |
| 554 | $this->moveTrailingSpacesOut( $node ); |
| 555 | |
| 556 | return $this->moveFormatTagOutsideATag( $node ); |
| 557 | |
| 558 | // Table cells |
| 559 | } elseif ( $nodeName === 'td' ) { |
| 560 | '@phan-var Element $node'; // @var Element $node |
| 561 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 562 | // * HTML <td>s won't have escapable prefixes |
| 563 | // * First cell should always be checked for escapable prefixes |
| 564 | // * Second and later cells in a wikitext td row (with stx='row' flag) |
| 565 | // won't have escapable prefixes. |
| 566 | $stx = $dp->stx ?? null; |
| 567 | if ( $stx === 'html' || |
| 568 | ( DiffDOMUtils::firstNonSepChild( $node->parentNode ) !== $node && $stx === 'row' ) ) { |
| 569 | return $node; |
| 570 | } |
| 571 | |
| 572 | $first = DiffDOMUtils::firstNonDeletedChild( $node ); |
| 573 | // Emit a space before escapable prefix |
| 574 | // This is preferable to serializing with a nowiki. |
| 575 | if ( $first instanceof Text && strspn( $first->nodeValue, '-+}', 0, 1 ) ) { |
| 576 | $first->nodeValue = ' ' . $first->nodeValue; |
| 577 | $this->addDiffMarks( $first, DiffMarkers::INSERTED, true ); |
| 578 | } |
| 579 | |
| 580 | return $node; |
| 581 | |
| 582 | // Font tags without any attributes |
| 583 | } elseif ( |
| 584 | $node instanceof Element && $nodeName === 'font' && |
| 585 | DOMDataUtils::noAttrs( $node ) |
| 586 | ) { |
| 587 | $next = DiffDOMUtils::nextNonDeletedSibling( $node ); |
| 588 | DOMUtils::migrateChildren( $node, $node->parentNode, $node ); |
| 589 | $node->parentNode->removeChild( $node ); |
| 590 | |
| 591 | return $next; |
| 592 | } elseif ( $node instanceof Element && $nodeName === 'p' |
| 593 | && !WTUtils::isLiteralHTMLNode( $node ) ) { |
| 594 | $next = DiffDOMUtils::nextNonSepSibling( $node ); |
| 595 | // Normalization of <p></p>, <p><br/></p>, <p><meta/></p> and the like to avoid |
| 596 | // extraneous new lines |
| 597 | if ( DiffDOMUtils::hasNChildren( $node, 1 ) && |
| 598 | WTUtils::isMarkerAnnotation( $node->firstChild ) |
| 599 | ) { |
| 600 | // Converts <p><meta /></p> (where meta is an annotation tag) to <meta /> without |
| 601 | // the wrapping <p> (that would typically be added by VE) to avoid getting too many |
| 602 | // newlines. |
| 603 | $ann = $node->firstChild; |
| 604 | DOMUtils::migrateChildren( $node, $node->parentNode, $node ); |
| 605 | $node->parentNode->removeChild( $node ); |
| 606 | return $ann; |
| 607 | } elseif ( |
| 608 | // Don't apply normalization to <p></p> nodes that |
| 609 | // were generated through deletions or other normalizations. |
| 610 | // FIXME: This trick fails for non-selser mode since |
| 611 | // diff markers are only added in selser mode. |
| 612 | DiffDOMUtils::hasNChildren( $node, 0, true ) && |
| 613 | // FIXME: Also, skip if this is the only child. |
| 614 | // Eliminates spurious test failures in non-selser mode. |
| 615 | !DiffDOMUtils::hasNChildren( $node->parentNode, 1 ) |
| 616 | ) { |
| 617 | // T184755: Convert sequences of <p></p> nodes to sequences of |
| 618 | // <br/>, <p><br/>..other content..</p>, <p><br/><p/> to ensure |
| 619 | // they serialize to as many newlines as the count of <p></p> nodes. |
| 620 | // Also handles <p><meta/></p> case for annotations. |
| 621 | if ( $next && DOMCompat::nodeName( $next ) === 'p' && |
| 622 | !WTUtils::isLiteralHTMLNode( $next ) ) { |
| 623 | // Replace 'node' (<p></p>) with a <br/> and make it the |
| 624 | // first child of 'next' (<p>..</p>). If 'next' was actually |
| 625 | // a <p></p> (i.e. empty), 'next' becomes <p><br/></p> |
| 626 | // which will serialize to 2 newlines. |
| 627 | $br = $node->ownerDocument->createElement( 'br' ); |
| 628 | $next->insertBefore( $br, $next->firstChild ); |
| 629 | |
| 630 | // Avoid nested insertion markers |
| 631 | if ( !$this->isInsertedContent( $next ) ) { |
| 632 | $this->addDiffMarks( $br, DiffMarkers::INSERTED ); |
| 633 | } |
| 634 | |
| 635 | // Delete node |
| 636 | $this->addDiffMarks( $node->parentNode, DiffMarkers::DELETED ); |
| 637 | $node->parentNode->removeChild( $node ); |
| 638 | } |
| 639 | } else { |
| 640 | // We cannot merge the <br/> with 'next' because |
| 641 | // it is not a <p>..</p>. |
| 642 | } |
| 643 | return $next; |
| 644 | } |
| 645 | // Default |
| 646 | return $node; |
| 647 | } |
| 648 | |
| 649 | public function normalizeSiblingPair( Node $a, Node $b ): Node { |
| 650 | if ( !$this->rewriteablePair( $a, $b ) ) { |
| 651 | return $b; |
| 652 | } |
| 653 | |
| 654 | // Since 'a' and 'b' make a rewriteable tag-pair, we are good to go. |
| 655 | if ( self::mergable( $a, $b ) ) { |
| 656 | '@phan-var Element $a'; // @var Element $a |
| 657 | '@phan-var Element $b'; // @var Element $b |
| 658 | $a = $this->merge( $a, $b ); |
| 659 | // The new a's children have new siblings. So let's look |
| 660 | // at a again. But their grandkids haven't changed, |
| 661 | // so we don't need to recurse further. |
| 662 | $this->processSubtree( $a, false ); |
| 663 | return $a; |
| 664 | } |
| 665 | |
| 666 | if ( self::swappable( $a, $b ) ) { |
| 667 | '@phan-var Element $a'; // @var Element $a |
| 668 | '@phan-var Element $b'; // @var Element $b |
| 669 | $firstNonDeletedChild = DiffDOMUtils::firstNonDeletedChild( $a ); |
| 670 | '@phan-var Element $firstNonDeletedChild'; // @var Element $firstNonDeletedChild |
| 671 | $a = $this->merge( $this->swap( $a, $firstNonDeletedChild ), $b ); |
| 672 | // Again, a has new children, but the grandkids have already |
| 673 | // been minimized. |
| 674 | $this->processSubtree( $a, false ); |
| 675 | return $a; |
| 676 | } |
| 677 | |
| 678 | if ( self::swappable( $b, $a ) ) { |
| 679 | '@phan-var Element $a'; // @var Element $a |
| 680 | '@phan-var Element $b'; // @var Element $b |
| 681 | $firstNonDeletedChild = DiffDOMUtils::firstNonDeletedChild( $b ); |
| 682 | '@phan-var Element $firstNonDeletedChild'; // @var Element $firstNonDeletedChild |
| 683 | $a = $this->merge( $a, $this->swap( $b, $firstNonDeletedChild ) ); |
| 684 | // Again, a has new children, but the grandkids have already |
| 685 | // been minimized. |
| 686 | $this->processSubtree( $a, false ); |
| 687 | return $a; |
| 688 | } |
| 689 | |
| 690 | return $b; |
| 691 | } |
| 692 | |
| 693 | public function processSubtree( Node $node, bool $recurse ): void { |
| 694 | // Process the first child outside the loop. |
| 695 | $a = DiffDOMUtils::firstNonDeletedChild( $node ); |
| 696 | if ( !$a ) { |
| 697 | return; |
| 698 | } |
| 699 | |
| 700 | $a = $this->processNode( $a, $recurse ); |
| 701 | while ( $a ) { |
| 702 | // We need a pair of adjacent siblings for tag minimization. |
| 703 | $b = DiffDOMUtils::nextNonDeletedSibling( $a ); |
| 704 | if ( !$b ) { |
| 705 | return; |
| 706 | } |
| 707 | |
| 708 | // Process subtree rooted at 'b'. |
| 709 | $b = $this->processNode( $b, $recurse ); |
| 710 | |
| 711 | // If we skipped over a bunch of nodes in the middle, |
| 712 | // we no longer have a pair of adjacent siblings. |
| 713 | if ( $b && DiffDOMUtils::previousNonDeletedSibling( $b ) === $a ) { |
| 714 | // Process the pair. |
| 715 | $a = $this->normalizeSiblingPair( $a, $b ); |
| 716 | } else { |
| 717 | $a = $b; |
| 718 | } |
| 719 | } |
| 720 | } |
| 721 | |
| 722 | public function processNode( Node $node, bool $recurse ): ?Node { |
| 723 | // Normalize 'node' and the subtree rooted at 'node' |
| 724 | // recurse = true => recurse and normalize subtree |
| 725 | // recurse = false => assume the subtree is already normalized |
| 726 | |
| 727 | // Normalize node till it stabilizes |
| 728 | $next = null; |
| 729 | while ( true ) { |
| 730 | // Skip templated content |
| 731 | while ( $node && WTUtils::isFirstEncapsulationWrapperNode( $node ) ) { |
| 732 | $node = WTUtils::skipOverEncapsulatedContent( $node ); |
| 733 | } |
| 734 | |
| 735 | if ( !$node ) { |
| 736 | return null; |
| 737 | } |
| 738 | |
| 739 | // Set insertion marker |
| 740 | $insertedSubtree = DiffUtils::hasInsertedDiffMark( $node ); |
| 741 | if ( $insertedSubtree ) { |
| 742 | if ( $this->inInsertedContent ) { |
| 743 | // Dump debugging info |
| 744 | $options = [ 'storeDiffMark' => true, 'saveData' => true ]; |
| 745 | $dump = ContentUtils::dumpDOM( |
| 746 | DOMCompat::getBody( $node->ownerDocument ), |
| 747 | '-- DOM triggering nested inserted dom-diff flags --', |
| 748 | $options |
| 749 | ); |
| 750 | $this->state->getEnv()->log( 'error/html2wt/dom', |
| 751 | "--- Nested inserted dom-diff flags ---\n", |
| 752 | 'Node:', |
| 753 | $node instanceof Element ? ContentUtils::toXML( $node, $options ) : $node->textContent, |
| 754 | "\nNode's parent:", |
| 755 | ContentUtils::toXML( $node->parentNode, $options ), |
| 756 | $dump |
| 757 | ); |
| 758 | } |
| 759 | // FIXME: If this assert is removed, the above dumping code should |
| 760 | // either be removed OR fixed up to remove uses of ContentUtils.ppToXML |
| 761 | Assert::invariant( !$this->inInsertedContent, 'Found nested inserted dom-diff flags!' ); |
| 762 | $this->inInsertedContent = true; |
| 763 | } |
| 764 | |
| 765 | // Post-order traversal: Process subtree first, and current node after. |
| 766 | // This lets multiple normalizations take effect cleanly. |
| 767 | if ( $recurse && $node instanceof Element ) { |
| 768 | $this->processSubtree( $node, true ); |
| 769 | } |
| 770 | |
| 771 | $next = $this->normalizeNode( $node ); |
| 772 | |
| 773 | // Clear insertion marker |
| 774 | if ( $insertedSubtree ) { |
| 775 | $this->inInsertedContent = false; |
| 776 | } |
| 777 | |
| 778 | if ( $next === $node ) { |
| 779 | return $node; |
| 780 | } else { |
| 781 | $node = $next; |
| 782 | } |
| 783 | } |
| 784 | |
| 785 | // @phan-suppress-next-line PhanPluginUnreachableCode |
| 786 | throw new UnreachableException( 'Control should never get here!' ); |
| 787 | } |
| 788 | |
| 789 | /** |
| 790 | * @param Element|DocumentFragment $node |
| 791 | */ |
| 792 | public function normalize( Node $node ) { |
| 793 | $this->processNode( $node, true ); |
| 794 | } |
| 795 | } |