Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 489 |
|
0.00% |
0 / 24 |
CRAP | |
0.00% |
0 / 1 |
| DOMRangeBuilder | |
0.00% |
0 / 489 |
|
0.00% |
0 / 24 |
32220 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
2 | |||
| updateDSRForFirstRangeNode | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
42 | |||
| getRangeEndDSR | |
0.00% |
0 / 26 |
|
0.00% |
0 / 1 |
90 | |||
| getRangeId | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDOMRange | |
0.00% |
0 / 74 |
|
0.00% |
0 / 1 |
342 | |||
| getStartConsideringFosteredContent | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
20 | |||
| stripStartMeta | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| findToplevelEnclosingRange | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
12 | |||
| recordTemplateInfo | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
30 | |||
| introducesCycle | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| rangesOverlap | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| findTopLevelNonOverlappingRanges | |
0.00% |
0 / 109 |
|
0.00% |
0 / 1 |
1122 | |||
| findFirstTemplatedNode | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
30 | |||
| ensureElementsInRange | |
0.00% |
0 / 20 |
|
0.00% |
0 / 1 |
30 | |||
| findEncapTarget | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
42 | |||
| encapsulateTemplates | |
0.00% |
0 / 92 |
|
0.00% |
0 / 1 |
1332 | |||
| addNodeRange | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getNodeRanges | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| findWrappableMetaRanges | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| findWrappableTemplateRangesRecursive | |
0.00% |
0 / 44 |
|
0.00% |
0 / 1 |
600 | |||
| matchMetaType | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| verifyTplInfoExpectation | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| execute | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| findEnclosingRange | |
0.00% |
0 / 22 |
|
0.00% |
0 / 1 |
30 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Wt2Html\DOM\Processors; |
| 5 | |
| 6 | use Error; |
| 7 | use SplObjectStorage; |
| 8 | use Wikimedia\Assert\Assert; |
| 9 | use Wikimedia\Assert\UnreachableException; |
| 10 | use Wikimedia\Parsoid\Config\Env; |
| 11 | use Wikimedia\Parsoid\Core\DomSourceRange; |
| 12 | use Wikimedia\Parsoid\Core\ElementRange; |
| 13 | use Wikimedia\Parsoid\DOM\Document; |
| 14 | use Wikimedia\Parsoid\DOM\Element; |
| 15 | use Wikimedia\Parsoid\DOM\Node; |
| 16 | use Wikimedia\Parsoid\DOM\Text; |
| 17 | use Wikimedia\Parsoid\NodeData\DataParsoid; |
| 18 | use Wikimedia\Parsoid\NodeData\TempData; |
| 19 | use Wikimedia\Parsoid\NodeData\TemplateInfo; |
| 20 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 21 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 22 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 23 | use Wikimedia\Parsoid\Utils\PHPUtils; |
| 24 | use Wikimedia\Parsoid\Utils\Utils; |
| 25 | use Wikimedia\Parsoid\Utils\WTUtils; |
| 26 | use Wikimedia\Parsoid\Wt2Html\Frame; |
| 27 | |
| 28 | /** |
| 29 | * Template encapsulation happens in three steps. |
| 30 | * |
| 31 | * 1. findWrappableTemplateRanges |
| 32 | * |
| 33 | * Locate start and end metas. Walk upwards towards the root from both and |
| 34 | * find a common ancestor A. The subtree rooted at A is now effectively the |
| 35 | * scope of the dom template ouput. |
| 36 | * |
| 37 | * 2. findTopLevelNonOverlappingRanges |
| 38 | * |
| 39 | * Mark all nodes in a range and walk up to root from each range start to |
| 40 | * determine overlaps, nesting. Merge overlapping and nested ranges to find |
| 41 | * the subset of top-level non-overlapping ranges which will be wrapped as |
| 42 | * individual units. |
| 43 | * |
| 44 | * 3. encapsulateTemplates |
| 45 | * |
| 46 | * For each non-overlapping range, |
| 47 | * - compute a data-mw according to the DOM spec |
| 48 | * - replace the start / end meta markers with transclusion type and data-mw |
| 49 | * on the first DOM node |
| 50 | * - add about ids on all top-level nodes of the range |
| 51 | * |
| 52 | * This is a simple high-level overview of the 3 steps to help understand this |
| 53 | * code. |
| 54 | * |
| 55 | * FIXME: At some point, more of the details should be extracted and documented |
| 56 | * in pseudo-code as an algorithm. |
| 57 | * @module |
| 58 | */ |
| 59 | class DOMRangeBuilder { |
| 60 | |
| 61 | private const MAP_TBODY_TR = [ |
| 62 | 'tbody' => true, |
| 63 | 'tr' => true |
| 64 | ]; |
| 65 | |
| 66 | /** @var Document */ |
| 67 | private $document; |
| 68 | |
| 69 | /** @var Frame */ |
| 70 | private $frame; |
| 71 | |
| 72 | /** @var Env */ |
| 73 | protected $env; |
| 74 | |
| 75 | /** @var SplObjectStorage */ |
| 76 | protected $nodeRanges; |
| 77 | |
| 78 | /** @var array<string|CompoundTemplateInfo>[] */ |
| 79 | private $compoundTpls = []; |
| 80 | |
| 81 | /** Are we generating spec 3.x HTML for parser functions */ |
| 82 | private bool $v3PFOutput; |
| 83 | |
| 84 | /** @var string */ |
| 85 | protected $traceType; |
| 86 | |
| 87 | public function __construct( |
| 88 | Document $document, Frame $frame |
| 89 | ) { |
| 90 | $this->document = $document; |
| 91 | $this->frame = $frame; |
| 92 | $this->env = $frame->getEnv(); |
| 93 | $this->nodeRanges = new SplObjectStorage; |
| 94 | $this->traceType = "tplwrap"; |
| 95 | // @phan-suppress-next-line PhanDeprecatedFunction |
| 96 | $this->v3PFOutput = (bool)$this->env->getSiteConfig()->getMWConfigValue( |
| 97 | 'ParsoidExperimentalParserFunctionOutput' |
| 98 | ); |
| 99 | } |
| 100 | |
| 101 | protected function updateDSRForFirstRangeNode( Element $target, Element $source ): void { |
| 102 | $srcDP = DOMDataUtils::getDataParsoid( $source ); |
| 103 | $tgtDP = DOMDataUtils::getDataParsoid( $target ); |
| 104 | |
| 105 | // Since TSRs on template content tokens are cleared by the |
| 106 | // template handler, all computed dsr values for template content |
| 107 | // is always inferred from top-level content values and is safe. |
| 108 | // So, do not overwrite a bigger end-dsr value. |
| 109 | if ( isset( $srcDP->dsr ) && $srcDP->dsr->end !== null && |
| 110 | isset( $tgtDP->dsr ) && $tgtDP->dsr->end !== null && |
| 111 | $tgtDP->dsr->end > $srcDP->dsr->end |
| 112 | ) { |
| 113 | $tgtDP->dsr->start = $srcDP->dsr->start ?? null; |
| 114 | } else { |
| 115 | $tgtDP->dsr = clone $srcDP->dsr; |
| 116 | $tgtDP->src = $srcDP->src ?? null; |
| 117 | } |
| 118 | } |
| 119 | |
| 120 | /** |
| 121 | * Get the DSR of the end of a DOMRange |
| 122 | * |
| 123 | * @param DOMRangeInfo $range |
| 124 | * @return DomSourceRange|null |
| 125 | */ |
| 126 | private static function getRangeEndDSR( DOMRangeInfo $range ): ?DomSourceRange { |
| 127 | $endNode = $range->end; |
| 128 | if ( $endNode instanceof Element ) { |
| 129 | return DOMDataUtils::getDataParsoid( $endNode )->dsr ?? null; |
| 130 | } else { |
| 131 | // In the rare scenario where the last element of a range is not an ELEMENT, |
| 132 | // extrapolate based on DSR of first leftmost sibling that is an ELEMENT. |
| 133 | // We don't try any harder than this for now. |
| 134 | $offset = 0; |
| 135 | $n = $endNode->previousSibling; |
| 136 | while ( $n && !( $n instanceof Element ) ) { |
| 137 | if ( $n instanceof Text ) { |
| 138 | $offset += strlen( $n->nodeValue ); |
| 139 | } else { |
| 140 | // A comment |
| 141 | // @phan-suppress-next-line PhanTypeMismatchArgumentSuperType |
| 142 | $offset += WTUtils::decodedCommentLength( $n ); |
| 143 | } |
| 144 | $n = $n->previousSibling; |
| 145 | } |
| 146 | |
| 147 | $dsr = null; |
| 148 | if ( $n ) { |
| 149 | /** |
| 150 | * The point of the above loop is to ensure we're working |
| 151 | * with a Element if there is an $n. |
| 152 | * |
| 153 | * @var Element $n |
| 154 | */ |
| 155 | '@phan-var Element $n'; |
| 156 | $dsr = DOMDataUtils::getDataParsoid( $n )->dsr ?? null; |
| 157 | } |
| 158 | |
| 159 | if ( $dsr && is_int( $dsr->end ?? null ) ) { |
| 160 | $len = $endNode instanceof Text |
| 161 | ? strlen( $endNode->nodeValue ) |
| 162 | // A comment |
| 163 | // @phan-suppress-next-line PhanTypeMismatchArgumentSuperType |
| 164 | : WTUtils::decodedCommentLength( $endNode ); |
| 165 | $dsr = new DomSourceRange( |
| 166 | $dsr->end + $offset, |
| 167 | $dsr->end + $offset + $len, |
| 168 | null, |
| 169 | null |
| 170 | ); |
| 171 | } |
| 172 | |
| 173 | return $dsr; |
| 174 | } |
| 175 | } |
| 176 | |
| 177 | /** |
| 178 | * Returns the range ID of a node - in the case of templates, its "about" attribute. |
| 179 | * @param Element $node |
| 180 | * @return string |
| 181 | */ |
| 182 | protected function getRangeId( Element $node ): string { |
| 183 | return DOMCompat::getAttribute( $node, "about" ); |
| 184 | } |
| 185 | |
| 186 | /** |
| 187 | * Find the common DOM ancestor of two DOM nodes. |
| 188 | * |
| 189 | * @param Element $startMeta |
| 190 | * @param Element $endMeta |
| 191 | * @param Element $endElem |
| 192 | * @return DOMRangeInfo |
| 193 | */ |
| 194 | private function getDOMRange( |
| 195 | Element $startMeta, Element $endMeta, Element $endElem |
| 196 | ) { |
| 197 | $range = $this->findEnclosingRange( $startMeta, $endMeta, $endElem ); |
| 198 | $startsInFosterablePosn = DOMUtils::isFosterablePosition( $range->start ); |
| 199 | $next = $range->start->nextSibling; |
| 200 | |
| 201 | // Detect empty content and handle them! |
| 202 | if ( WTUtils::isTplMarkerMeta( $range->start ) && $next === $endElem ) { |
| 203 | Assert::invariant( |
| 204 | $range->start === $range->startElem, |
| 205 | "Expected startElem to be same as range.start" |
| 206 | ); |
| 207 | if ( $startsInFosterablePosn ) { |
| 208 | // Expand range! |
| 209 | $range->start = $range->end = $range->start->parentNode; |
| 210 | $startsInFosterablePosn = false; |
| 211 | } else { |
| 212 | $emptySpan = $this->document->createElement( 'span' ); |
| 213 | $range->start->parentNode->insertBefore( $emptySpan, $endElem ); |
| 214 | } |
| 215 | |
| 216 | // Handle unwrappable content in fosterable positions |
| 217 | // and expand template range, if required. |
| 218 | // NOTE: Template marker meta tags are translated from comments |
| 219 | // *after* the DOM has been built which is why they can show up in |
| 220 | // fosterable positions in the DOM. |
| 221 | } elseif ( $startsInFosterablePosn && |
| 222 | ( !( $range->start instanceof Element ) || |
| 223 | ( WTUtils::isTplMarkerMeta( $range->start ) && |
| 224 | ( !( $next instanceof Element ) || WTUtils::isTplMarkerMeta( $next ) ) ) |
| 225 | ) |
| 226 | ) { |
| 227 | $rangeStartParent = $range->start->parentNode; |
| 228 | |
| 229 | // If we are in a table in a foster-element position, then all non-element |
| 230 | // nodes will be white-space and comments. Skip over all of them and find |
| 231 | // the first table content node. |
| 232 | $noWS = true; |
| 233 | $nodesToMigrate = []; |
| 234 | $newStart = $range->start; |
| 235 | $n = $range->start instanceof Element ? $next : $range->start; |
| 236 | while ( !( $n instanceof Element ) ) { |
| 237 | if ( $n instanceof Text ) { |
| 238 | $noWS = false; |
| 239 | } |
| 240 | $nodesToMigrate[] = $n; |
| 241 | $n = $n->nextSibling; |
| 242 | $newStart = $n; |
| 243 | } |
| 244 | |
| 245 | // As long as $newStart is a tr/tbody or we don't have whitespace |
| 246 | // migrate $nodesToMigrate into $newStart. Pushing whitespace into |
| 247 | // th/td/caption can change display semantics. |
| 248 | if ( $newStart && ( $noWS || isset( self::MAP_TBODY_TR[DOMCompat::nodeName( $newStart )] ) ) ) { |
| 249 | /** |
| 250 | * The point of the above loop is to ensure we're working |
| 251 | * with a Element if there is a $newStart. |
| 252 | * |
| 253 | * @var Element $newStart |
| 254 | */ |
| 255 | '@phan-var Element $newStart'; |
| 256 | $insertPosition = $newStart->firstChild; |
| 257 | foreach ( $nodesToMigrate as $n ) { |
| 258 | $newStart->insertBefore( $n, $insertPosition ); |
| 259 | } |
| 260 | $range->start = $newStart; |
| 261 | // Update dsr to point to original start |
| 262 | $this->updateDSRForFirstRangeNode( $range->start, $range->startElem ); |
| 263 | } else { |
| 264 | // If not, we are forced to expand the template range. |
| 265 | $range->start = $range->end = $rangeStartParent; |
| 266 | } |
| 267 | } |
| 268 | |
| 269 | // Ensure range->start is an element node since we want to |
| 270 | // add/update the data-parsoid attribute to it. |
| 271 | if ( !( $range->start instanceof Element ) ) { |
| 272 | $span = $this->document->createElement( 'span' ); |
| 273 | $range->start->parentNode->insertBefore( $span, $range->start ); |
| 274 | $span->appendChild( $range->start ); |
| 275 | $range->start = $span; |
| 276 | $this->updateDSRForFirstRangeNode( $range->start, $range->startElem ); |
| 277 | } |
| 278 | |
| 279 | $range->start = $this->getStartConsideringFosteredContent( $range->start ); |
| 280 | |
| 281 | // Use the negative test since it doesn't mark the range as flipped |
| 282 | // if range.start === range.end |
| 283 | if ( !DOMUtils::inSiblingOrder( $range->start, $range->end ) ) { |
| 284 | // In foster-parenting situations, the end-meta tag (and hence range.end) |
| 285 | // can show up before the range.start which would be the table itself. |
| 286 | // So, we record this info for later analysis. |
| 287 | $range->flipped = true; |
| 288 | } |
| 289 | |
| 290 | $this->env->log( |
| 291 | "trace/{$this->traceType}/findranges", |
| 292 | static function () use ( &$range ) { |
| 293 | $msg = ''; |
| 294 | $dp1 = DOMDataUtils::getDataParsoid( $range->start ); |
| 295 | $dp2 = DOMDataUtils::getDataParsoid( $range->end ); |
| 296 | $tmp1 = $dp1->tmp; |
| 297 | $tmp2 = $dp2->tmp; |
| 298 | $dp1->tmp = null; |
| 299 | $dp2->tmp = null; |
| 300 | $msg .= "\n----------------------------------------------"; |
| 301 | $msg .= "\nFound range : " . $range->id . '; flipped? ' . ( (string)$range->flipped ) . |
| 302 | '; offset: ' . $range->startOffset; |
| 303 | $msg .= "\nstart-elem : " . DOMCompat::getOuterHTML( $range->startElem ) . '; DP: ' . |
| 304 | PHPUtils::jsonEncode( DOMDataUtils::getDataParsoid( $range->startElem ) ); |
| 305 | $msg .= "\nend-elem : " . DOMCompat::getOuterHTML( $range->endElem ) . '; DP: ' . |
| 306 | PHPUtils::jsonEncode( DOMDataUtils::getDataParsoid( $range->endElem ) ); |
| 307 | $msg .= "\nstart : [TAG_ID " . ( $tmp1->tagId ?? '?' ) . ']: ' . |
| 308 | DOMCompat::getOuterHTML( $range->start ) . |
| 309 | '; DP: ' . PHPUtils::jsonEncode( $dp1 ); |
| 310 | $msg .= "\nend : [TAG_ID " . ( $tmp2->tagId ?? '?' ) . ']: ' . |
| 311 | DOMCompat::getOuterHTML( $range->end ) . |
| 312 | '; DP: ' . PHPUtils::jsonEncode( $dp2 ); |
| 313 | $msg .= "\n----------------------------------------------"; |
| 314 | $dp1->tmp = $tmp1; |
| 315 | $dp2->tmp = $tmp2; |
| 316 | return $msg; |
| 317 | } |
| 318 | ); |
| 319 | |
| 320 | return $range; |
| 321 | } |
| 322 | |
| 323 | /** |
| 324 | * Returns the current node if it's not just after fostered content, the first node |
| 325 | * of fostered content otherwise. |
| 326 | * @param Node $node |
| 327 | * @return Node |
| 328 | */ |
| 329 | protected function getStartConsideringFosteredContent( Node $node ): Node { |
| 330 | if ( DOMCompat::nodeName( $node ) === 'table' ) { |
| 331 | // If we have any fostered content, include it as well. |
| 332 | for ( $previousSibling = $node->previousSibling; |
| 333 | $previousSibling instanceof Element && |
| 334 | !empty( DOMDataUtils::getDataParsoid( $previousSibling )->fostered ); |
| 335 | $previousSibling = $node->previousSibling |
| 336 | ) { |
| 337 | $node = $previousSibling; |
| 338 | } |
| 339 | } |
| 340 | return $node; |
| 341 | } |
| 342 | |
| 343 | private static function stripStartMeta( Element $meta ): void { |
| 344 | if ( DOMCompat::nodeName( $meta ) === 'meta' ) { |
| 345 | $meta->parentNode->removeChild( $meta ); |
| 346 | } else { |
| 347 | // Remove mw:* from the typeof. |
| 348 | $type = DOMCompat::getAttribute( $meta, 'typeof' ); |
| 349 | if ( $type !== null ) { |
| 350 | $type = preg_replace( '/(?:^|\s)mw:[^\/]*(\/\S+|(?=$|\s))/D', '', $type ); |
| 351 | $meta->setAttribute( 'typeof', $type ); |
| 352 | } |
| 353 | } |
| 354 | } |
| 355 | |
| 356 | private static function findToplevelEnclosingRange( |
| 357 | array $nestingInfo, ?string $startId |
| 358 | ): ?string { |
| 359 | // Walk up the implicit nesting tree to find the |
| 360 | // top-level range within which rId is nested. |
| 361 | // No cycles can exist since they have been suppressed. |
| 362 | $visited = []; |
| 363 | $rId = $startId; |
| 364 | while ( isset( $nestingInfo[$rId] ) ) { |
| 365 | if ( isset( $visited[$rId] ) ) { |
| 366 | throw new Error( "Found a cycle in tpl-range nesting where there shouldn't have been one." ); |
| 367 | } |
| 368 | $visited[$rId] = true; |
| 369 | $rId = $nestingInfo[$rId]; |
| 370 | } |
| 371 | return $rId; |
| 372 | } |
| 373 | |
| 374 | /** |
| 375 | * Add a template to $this->compoundTpls |
| 376 | * |
| 377 | * @param string $compoundTplId |
| 378 | * @param DOMRangeInfo $range |
| 379 | * @param TemplateInfo $templateInfo |
| 380 | */ |
| 381 | private function recordTemplateInfo( |
| 382 | string $compoundTplId, DOMRangeInfo $range, TemplateInfo $templateInfo |
| 383 | ): void { |
| 384 | $this->compoundTpls[$compoundTplId] ??= []; |
| 385 | |
| 386 | // Record template args info along with any intervening wikitext |
| 387 | // between templates that are part of the same compound structure. |
| 388 | /** @var array $tplArray */ |
| 389 | $tplArray = &$this->compoundTpls[$compoundTplId]; |
| 390 | $dp = DOMDataUtils::getDataParsoid( $range->startElem ); |
| 391 | $dsr = $dp->dsr; |
| 392 | |
| 393 | if ( count( $tplArray ) > 0 ) { |
| 394 | $prevTplInfo = PHPUtils::lastItem( $tplArray ); |
| 395 | if ( $prevTplInfo->dsr->end < $dsr->start ) { |
| 396 | $width = $dsr->start - $prevTplInfo->dsr->end; |
| 397 | $tplArray[] = PHPUtils::safeSubstr( |
| 398 | $this->frame->getSrcText(), $prevTplInfo->dsr->end, $width ); |
| 399 | } |
| 400 | } |
| 401 | |
| 402 | if ( !empty( $dp->unwrappedWT ) ) { |
| 403 | $tplArray[] = (string)$dp->unwrappedWT; |
| 404 | } |
| 405 | |
| 406 | // Get rid of src-offsets since they aren't needed anymore. |
| 407 | foreach ( $templateInfo->paramInfos as $pi ) { |
| 408 | $pi->srcOffsets = null; |
| 409 | } |
| 410 | $tplArray[] = new CompoundTemplateInfo( |
| 411 | $dsr, $templateInfo, DOMUtils::hasTypeOf( $range->startElem, 'mw:Param' ) |
| 412 | ); |
| 413 | } |
| 414 | |
| 415 | /** |
| 416 | * Determine whether adding the given range would introduce a cycle in the |
| 417 | * subsumedRanges graph. |
| 418 | * |
| 419 | * Nesting cycles with multiple ranges can show up because of foster |
| 420 | * parenting scenarios if they are not detected and suppressed. |
| 421 | * |
| 422 | * @param string $start The ID of the new range |
| 423 | * @param string $end The ID of the other range |
| 424 | * @param string[] $subsumedRanges The subsumed ranges graph, encoded as an |
| 425 | * array in which each element maps one string range ID to another range ID |
| 426 | * @return bool |
| 427 | */ |
| 428 | private static function introducesCycle( string $start, string $end, array $subsumedRanges ): bool { |
| 429 | $visited = [ $start => true ]; |
| 430 | $elt = $subsumedRanges[$end] ?? null; |
| 431 | while ( $elt ) { |
| 432 | if ( !empty( $visited[$elt] ) ) { |
| 433 | return true; |
| 434 | } |
| 435 | $elt = $subsumedRanges[$elt] ?? null; |
| 436 | } |
| 437 | return false; |
| 438 | } |
| 439 | |
| 440 | /** |
| 441 | * Determine whether DOM ranges overlap. |
| 442 | * |
| 443 | * The `inSiblingOrder` check here is sufficient to determine overlaps |
| 444 | * because the algorithm in `findWrappableTemplateRanges` will put the |
| 445 | * start/end elements for intersecting ranges on the same plane and prev/ |
| 446 | * curr are in textual order (which translates to dom order). |
| 447 | * |
| 448 | * @param DOMRangeInfo $prev |
| 449 | * @param DOMRangeInfo $curr |
| 450 | * @return bool |
| 451 | */ |
| 452 | private static function rangesOverlap( DOMRangeInfo $prev, DOMRangeInfo $curr ): bool { |
| 453 | $prevEnd = ( !$prev->flipped ) ? $prev->end : $prev->start; |
| 454 | $currStart = ( !$curr->flipped ) ? $curr->start : $curr->end; |
| 455 | return DOMUtils::inSiblingOrder( $currStart, $prevEnd ); |
| 456 | } |
| 457 | |
| 458 | /** |
| 459 | * Identify the elements of $tplRanges that are non-overlapping. |
| 460 | * Record template info in $this->compoundTpls as we go. |
| 461 | * |
| 462 | * @param Node $docRoot |
| 463 | * @param DOMRangeInfo[] $tplRanges The potentially overlapping ranges |
| 464 | * @return DOMRangeInfo[] The non-overlapping ranges |
| 465 | */ |
| 466 | public function findTopLevelNonOverlappingRanges( Node $docRoot, array $tplRanges ): array { |
| 467 | // For each node, assign an attribute that is a record of all |
| 468 | // tpl ranges it belongs to at the top-level. |
| 469 | foreach ( $tplRanges as $r ) { |
| 470 | $n = !$r->flipped ? $r->start : $r->end; |
| 471 | $e = !$r->flipped ? $r->end : $r->start; |
| 472 | |
| 473 | while ( $n ) { |
| 474 | if ( $n instanceof Element ) { |
| 475 | $this->addNodeRange( $n, $r ); |
| 476 | // Done |
| 477 | if ( $n === $e ) { |
| 478 | break; |
| 479 | } |
| 480 | } |
| 481 | |
| 482 | $n = $n->nextSibling; |
| 483 | } |
| 484 | } |
| 485 | |
| 486 | // In the first pass over `numRanges` below, `subsumedRanges` is used to |
| 487 | // record purely the nested ranges. However, in the second pass, we also |
| 488 | // add the relationships between overlapping ranges so that |
| 489 | // `findToplevelEnclosingRange` can use that information to add `argInfo` |
| 490 | // to the right `compoundTpls`. This scenario can come up when you have |
| 491 | // three ranges, 1 intersecting with 2 but not 3, and 3 nested in 2. |
| 492 | $subsumedRanges = []; |
| 493 | |
| 494 | // For each range r:(s, e), walk up from s --> docRoot and if any of |
| 495 | // these nodes have tpl-ranges (besides r itself) assigned to them, |
| 496 | // then r is nested in those other templates and can be ignored. |
| 497 | foreach ( $tplRanges as $r ) { |
| 498 | $n = $r->start; |
| 499 | |
| 500 | while ( $n !== $docRoot ) { |
| 501 | $ranges = $this->getNodeRanges( $n ); |
| 502 | if ( $ranges ) { |
| 503 | if ( $n !== $r->start ) { |
| 504 | // 'r' is nested for sure |
| 505 | // Record the outermost range in which 'r' is nested. |
| 506 | $outermostId = null; |
| 507 | $outermostOffset = null; |
| 508 | foreach ( $ranges as $rangeId => $range ) { |
| 509 | if ( $outermostId === null |
| 510 | || $range->startOffset < $outermostOffset |
| 511 | ) { |
| 512 | $outermostId = $rangeId; |
| 513 | $outermostOffset = $range->startOffset; |
| 514 | } |
| 515 | } |
| 516 | $subsumedRanges[$r->id] = (string)$outermostId; |
| 517 | break; |
| 518 | } else { |
| 519 | // n === r.start |
| 520 | // |
| 521 | // We have to make sure this is not an overlap scenario. |
| 522 | // Find the ranges that r.start and r.end belong to and |
| 523 | // compute their intersection. If this intersection has |
| 524 | // another tpl range besides r itself, we have a winner! |
| 525 | // |
| 526 | // The code below does the above check efficiently. |
| 527 | $eTpls = $this->getNodeRanges( $r->end ); |
| 528 | $foundNesting = false; |
| 529 | |
| 530 | foreach ( $ranges as $otherId => $other ) { |
| 531 | // - Don't record nesting cycles. |
| 532 | // - Record the outermost range in which 'r' is nested in. |
| 533 | if ( $otherId !== $r->id && |
| 534 | !empty( $eTpls[$otherId] ) && |
| 535 | // When we have identical ranges, pick the range with |
| 536 | // the larger offset to be subsumed. |
| 537 | ( $r->start !== $other->start || |
| 538 | $r->end !== $other->end || |
| 539 | $other->startOffset < $r->startOffset |
| 540 | ) && |
| 541 | !self::introducesCycle( $r->id, (string)$otherId, $subsumedRanges ) |
| 542 | ) { |
| 543 | $foundNesting = true; |
| 544 | if ( !isset( $subsumedRanges[$r->id] ) || |
| 545 | $other->startOffset < $ranges[$subsumedRanges[$r->id]]->startOffset |
| 546 | ) { |
| 547 | $subsumedRanges[$r->id] = (string)$otherId; |
| 548 | } |
| 549 | } |
| 550 | } |
| 551 | |
| 552 | if ( $foundNesting ) { |
| 553 | // 'r' is nested |
| 554 | break; |
| 555 | } |
| 556 | } |
| 557 | } |
| 558 | |
| 559 | // Move up |
| 560 | $n = $n->parentNode; |
| 561 | } |
| 562 | } |
| 563 | |
| 564 | // Sort by start offset in source wikitext |
| 565 | usort( $tplRanges, static function ( $r1, $r2 ) { |
| 566 | return $r1->startOffset - $r2->startOffset; |
| 567 | } ); |
| 568 | |
| 569 | // Since the tpl ranges are sorted in textual order (by start offset), |
| 570 | // it is sufficient to only look at the most recent template to see |
| 571 | // if the current one overlaps with the previous one. |
| 572 | // |
| 573 | // This works because we've already identify nested ranges and can ignore them. |
| 574 | |
| 575 | $newRanges = []; |
| 576 | $prev = null; |
| 577 | |
| 578 | foreach ( $tplRanges as $r ) { |
| 579 | $endTagToRemove = null; |
| 580 | $startTagToStrip = null; |
| 581 | |
| 582 | // Extract tplargInfo |
| 583 | $tmp = DOMDataUtils::getDataParsoid( $r->startElem )->getTemp(); |
| 584 | $templateInfo = $tmp->tplarginfo ?? null; |
| 585 | |
| 586 | $this->verifyTplInfoExpectation( $templateInfo, $tmp ); |
| 587 | |
| 588 | $this->env->log( "trace/{$this->traceType}/merge", static function () use ( &$DOMDataUtils, &$r ) { |
| 589 | $msg = ''; |
| 590 | $dp1 = DOMDataUtils::getDataParsoid( $r->start ); |
| 591 | $dp2 = DOMDataUtils::getDataParsoid( $r->end ); |
| 592 | $tmp1 = $dp1->tmp; |
| 593 | $tmp2 = $dp2->tmp; |
| 594 | $dp1->tmp = null; |
| 595 | $dp2->tmp = null; |
| 596 | $msg .= "\n##############################################"; |
| 597 | $msg .= "\nrange " . $r->id . '; r-start-elem: ' . DOMCompat::getOuterHTML( $r->startElem ) . |
| 598 | '; DP: ' . PHPUtils::jsonEncode( DOMDataUtils::getDataParsoid( $r->startElem ) ); |
| 599 | $msg .= "\nrange " . $r->id . '; r-end-elem: ' . DOMCompat::getOuterHTML( $r->endElem ) . |
| 600 | '; DP: ' . PHPUtils::jsonEncode( DOMDataUtils::getDataParsoid( $r->endElem ) ); |
| 601 | $msg .= "\nrange " . $r->id . '; r-start: [TAG_ID ' . ( $tmp1->tagId ?? '?' ) . ']: ' . |
| 602 | DOMCompat::getOuterHTML( $r->start ) . '; DP: ' . PHPUtils::jsonEncode( $dp1 ); |
| 603 | $msg .= "\nrange " . $r->id . '; r-end: [TAG_ID ' . ( $tmp2->tagId ?? '?' ) . ']: ' . |
| 604 | DOMCompat::getOuterHTML( $r->end ) . '; DP: ' . PHPUtils::jsonEncode( $dp2 ); |
| 605 | $msg .= "\n----------------------------------------------"; |
| 606 | $dp1->tmp = $tmp1; |
| 607 | $dp2->tmp = $tmp2; |
| 608 | return $msg; |
| 609 | } ); |
| 610 | |
| 611 | $enclosingRangeId = self::findToplevelEnclosingRange( |
| 612 | $subsumedRanges, |
| 613 | $subsumedRanges[$r->id] ?? null |
| 614 | ); |
| 615 | if ( $enclosingRangeId ) { |
| 616 | $this->env->log( "trace/{$this->traceType}/merge", '--nested in ', $enclosingRangeId, '--' ); |
| 617 | |
| 618 | // Nested -- ignore r |
| 619 | $startTagToStrip = $r->startElem; |
| 620 | $endTagToRemove = $r->endElem; |
| 621 | if ( $templateInfo ) { |
| 622 | // 'r' is nested in 'enclosingRange' at the top-level |
| 623 | // So, enclosingRange gets r's argInfo |
| 624 | $this->recordTemplateInfo( $enclosingRangeId, $r, $templateInfo ); |
| 625 | } |
| 626 | } elseif ( $prev && self::rangesOverlap( $prev, $r ) ) { |
| 627 | // In the common case, in overlapping scenarios, r.start is |
| 628 | // identical to prev.end. However, in fostered content scenarios, |
| 629 | // there can true overlap of the ranges. |
| 630 | $this->env->log( "trace/{$this->traceType}/merge", '--overlapped--' ); |
| 631 | |
| 632 | // See comment above, where `subsumedRanges` is defined. |
| 633 | $subsumedRanges[$r->id] = $prev->id; |
| 634 | |
| 635 | // Overlapping ranges. |
| 636 | // r is the regular kind |
| 637 | // Merge r with prev |
| 638 | |
| 639 | // Note that if a table comes from a template, a foster box isn't |
| 640 | // emitted so the enclosure isn't guaranteed. In pathological |
| 641 | // cases, like where the table end tag isn't emitted, we can still |
| 642 | // end up with flipped ranges if the template end marker gets into |
| 643 | // a fosterable position (which can still happen despite being |
| 644 | // emitted as a comment). |
| 645 | Assert::invariant( !$r->flipped, |
| 646 | 'Flipped range should have been enclosed.' |
| 647 | ); |
| 648 | |
| 649 | $startTagToStrip = $r->startElem; |
| 650 | $endTagToRemove = $prev->endElem; |
| 651 | |
| 652 | $prev->end = $r->end; |
| 653 | $prev->endElem = $r->endElem; |
| 654 | if ( WTUtils::isMarkerAnnotation( $r->endElem ) ) { |
| 655 | $endDataMw = DOMDataUtils::getDataMw( $r->endElem ); |
| 656 | $endDataMw->rangeId = $r->id; |
| 657 | $prev->extendedByOverlapMerge = true; |
| 658 | } |
| 659 | |
| 660 | // Update compoundTplInfo |
| 661 | if ( $templateInfo ) { |
| 662 | $this->recordTemplateInfo( $prev->id, $r, $templateInfo ); |
| 663 | } |
| 664 | } else { |
| 665 | $this->env->log( "trace/{$this->traceType}/merge", '--normal--' ); |
| 666 | |
| 667 | // Default -- no overlap |
| 668 | // Emit the merged range |
| 669 | $newRanges[] = $r; |
| 670 | $prev = $r; |
| 671 | |
| 672 | // Update compoundTpls |
| 673 | if ( $templateInfo ) { |
| 674 | $this->recordTemplateInfo( $r->id, $r, $templateInfo ); |
| 675 | } |
| 676 | } |
| 677 | |
| 678 | if ( $endTagToRemove ) { |
| 679 | // Remove start and end meta-tags |
| 680 | // Not necessary to remove the start tag, but good to cleanup |
| 681 | $endTagToRemove->parentNode->removeChild( $endTagToRemove ); |
| 682 | self::stripStartMeta( $startTagToStrip ); |
| 683 | } |
| 684 | } |
| 685 | |
| 686 | return $newRanges; |
| 687 | } |
| 688 | |
| 689 | /** |
| 690 | * Note that the case of nodeName varies with DOM implementation. This |
| 691 | * method currently forces the name nodeName to uppercase. In the future |
| 692 | * we can/should switch to using the "native" case of the DOM |
| 693 | * implementation; we do a case-insensitive match (by converting the result |
| 694 | * to the "native" case of the DOM implementation) in |
| 695 | * EncapsulatedContentHandler when this value is used. |
| 696 | * @param DOMRangeInfo $range |
| 697 | * @return string|null nodeName with an optional "_$stx" suffix. |
| 698 | */ |
| 699 | private static function findFirstTemplatedNode( DOMRangeInfo $range ): ?string { |
| 700 | $firstNode = $range->start; |
| 701 | |
| 702 | // Skip tpl marker meta |
| 703 | if ( WTUtils::isTplMarkerMeta( $firstNode ) ) { |
| 704 | $firstNode = $firstNode->nextSibling; |
| 705 | } |
| 706 | |
| 707 | // Walk past fostered nodes since they came from within a table |
| 708 | // Note that this is not foolproof because in some scenarios, |
| 709 | // fostered content is not marked up. Ex: when a table is templated, |
| 710 | // and content from the table is fostered. |
| 711 | $dp = DOMDataUtils::getDataParsoid( $firstNode ); |
| 712 | while ( !empty( $dp->fostered ) ) { |
| 713 | $firstNode = $firstNode->nextSibling; |
| 714 | /** @var Element $firstNode */ |
| 715 | DOMUtils::assertElt( $firstNode ); |
| 716 | $dp = DOMDataUtils::getDataParsoid( $firstNode ); |
| 717 | } |
| 718 | |
| 719 | // FIXME: It is harder to use META as a node name since this is a generic |
| 720 | // placeholder for a whole bunch of things each of which has its own |
| 721 | // newline constraint requirements. So, for now, I am skipping that |
| 722 | // can of worms to prevent confusing the serializer with an overloaded |
| 723 | // tag name. |
| 724 | if ( DOMCompat::nodeName( $firstNode ) === 'meta' ) { |
| 725 | return null; |
| 726 | } |
| 727 | |
| 728 | // FIXME spec-compliant values would be upper-case, this is just a workaround |
| 729 | // for current PHP DOM implementation and could be removed in the future |
| 730 | // See discussion in the method comment above. |
| 731 | $nodeName = mb_strtoupper( DOMCompat::nodeName( $firstNode ), "UTF-8" ); |
| 732 | |
| 733 | return !empty( $dp->stx ) ? $nodeName . '_' . $dp->stx : $nodeName; |
| 734 | } |
| 735 | |
| 736 | /** |
| 737 | * Encapsulation requires adding about attributes on the top-level |
| 738 | * nodes of the range. This requires them to all be Elements. |
| 739 | * |
| 740 | * @param DOMRangeInfo $range |
| 741 | */ |
| 742 | private function ensureElementsInRange( DOMRangeInfo $range ): void { |
| 743 | $n = $range->start; |
| 744 | $e = $range->end; |
| 745 | $about = DOMCompat::getAttribute( $range->startElem, 'about' ); |
| 746 | while ( $n ) { |
| 747 | $next = $n->nextSibling; |
| 748 | if ( $n instanceof Element ) { |
| 749 | $n->setAttribute( 'about', $about ); |
| 750 | } elseif ( DOMUtils::isFosterablePosition( $n ) ) { |
| 751 | // NOTE: There cannot be any non-IEW text in fosterable position |
| 752 | // since the HTML tree builder would already have fostered it out. |
| 753 | // So, any non-element node found here is safe to delete since: |
| 754 | // (a) this has no rendering output impact, and |
| 755 | // (b) data-mw captures template output => we don't need |
| 756 | // to preserve this for html2wt either. Removing this |
| 757 | // lets us preserve DOM range continuity. |
| 758 | $n->parentNode->removeChild( $n ); |
| 759 | } else { |
| 760 | // Add a span wrapper to let us add about-ids to represent |
| 761 | // the DOM range as a contiguous chain of DOM nodes. |
| 762 | $span = $this->document->createElement( 'span' ); |
| 763 | $span->setAttribute( 'about', $about ); |
| 764 | $dp = new DataParsoid; |
| 765 | $dp->setTempFlag( TempData::WRAPPER ); |
| 766 | DOMDataUtils::setDataParsoid( $span, $dp ); |
| 767 | $n->parentNode->replaceChild( $span, $n ); |
| 768 | $span->appendChild( $n ); |
| 769 | $n = $span; |
| 770 | } |
| 771 | |
| 772 | if ( $n === $e ) { |
| 773 | break; |
| 774 | } |
| 775 | |
| 776 | $n = $next; |
| 777 | } |
| 778 | } |
| 779 | |
| 780 | /** |
| 781 | * Find the first element to be encapsulated. |
| 782 | * Skip past marker metas and non-elements (which will all be IEW |
| 783 | * in fosterable positions in a table). |
| 784 | * |
| 785 | * @param DOMRangeInfo $range |
| 786 | * @return Element |
| 787 | */ |
| 788 | private static function findEncapTarget( DOMRangeInfo $range ): Element { |
| 789 | $encapTgt = $range->start; |
| 790 | '@phan-var Node $encapTgt'; |
| 791 | |
| 792 | // Skip template-marker meta-tags. |
| 793 | while ( WTUtils::isTplMarkerMeta( $encapTgt ) || |
| 794 | !( $encapTgt instanceof Element ) |
| 795 | ) { |
| 796 | // Detect unwrappable template and bail out early. |
| 797 | if ( $encapTgt === $range->end || |
| 798 | ( !( $encapTgt instanceof Element ) && |
| 799 | !DOMUtils::isFosterablePosition( $encapTgt ) |
| 800 | ) |
| 801 | ) { |
| 802 | throw new Error( 'Cannot encapsulate transclusion. Start=' . |
| 803 | DOMCompat::getOuterHTML( $range->startElem ) ); |
| 804 | } |
| 805 | $encapTgt = $encapTgt->nextSibling; |
| 806 | } |
| 807 | |
| 808 | '@phan-var Element $encapTgt'; |
| 809 | return $encapTgt; |
| 810 | } |
| 811 | |
| 812 | /** |
| 813 | * Add markers to the DOM around the non-overlapping ranges. |
| 814 | * |
| 815 | * @param DOMRangeInfo[] $nonOverlappingRanges |
| 816 | */ |
| 817 | private function encapsulateTemplates( array $nonOverlappingRanges ): void { |
| 818 | foreach ( $nonOverlappingRanges as $i => $range ) { |
| 819 | |
| 820 | // We should never have flipped overlapping ranges, and indeed that's |
| 821 | // asserted in `findTopLevelNonOverlappingRanges`. Flipping results |
| 822 | // in either completely nested ranges, or non-intersecting ranges. |
| 823 | // |
| 824 | // If the table causing the fostering is not transcluded, we emit a |
| 825 | // foster box and wrap the whole table+fb in metas, producing nested |
| 826 | // ranges. For ex, |
| 827 | // |
| 828 | // <table> |
| 829 | // {{1x|<div>}} |
| 830 | // |
| 831 | // The tricky part is when the table *is* transcluded, and we omit the |
| 832 | // foster box. The common case (for some definition of common) might |
| 833 | // be like, |
| 834 | // |
| 835 | // {{1x|<table>}} |
| 836 | // {{1x|<div>}} |
| 837 | // |
| 838 | // Here, #mwt1 leaves a table open and the end meta from #mwt2 is |
| 839 | // fostered, since it gets closed into the div. The range for #mwt1 |
| 840 | // is the entire table, which thankfully contains #mwt2, so we still |
| 841 | // have the expected entire nesting. Any tricks to extend the range |
| 842 | // of #mwt2 beyond the table (so that we have an overlapping range) will |
| 843 | // inevitably result in the end meta not being fostered, and we avoid |
| 844 | // this situation altogether. |
| 845 | // |
| 846 | // The very edgy case is as follows, |
| 847 | // |
| 848 | // {{1x|<table><div>}}</div> |
| 849 | // {{1x|<div>}} |
| 850 | // |
| 851 | // where both end metas are fostered. Ignoring that we don't even |
| 852 | // roundtrip the first transclusion properly on its own, here we have |
| 853 | // a flipped range where, since the end meta for the first range was |
| 854 | // also fostered, the ranges still don't overlap. |
| 855 | |
| 856 | // FIXME: The code below needs to be aware of flipped ranges. |
| 857 | |
| 858 | $this->ensureElementsInRange( $range ); |
| 859 | |
| 860 | $tplArray = $this->compoundTpls[$range->id] ?? null; |
| 861 | Assert::invariant( (bool)$tplArray, 'No parts for template range!' ); |
| 862 | |
| 863 | $encapTgt = self::findEncapTarget( $range ); |
| 864 | $encapValid = false; |
| 865 | $encapDP = DOMDataUtils::getDataParsoid( $encapTgt ); |
| 866 | |
| 867 | // Update type-of (always even if tpl-encap below will fail). |
| 868 | // This ensures that VE will still "edit-protect" this template |
| 869 | // and not allow its content to be edited directly. |
| 870 | $startElem = $range->startElem; |
| 871 | if ( $startElem !== $encapTgt ) { |
| 872 | $t1 = DOMCompat::getAttribute( $startElem, 'typeof' ); |
| 873 | if ( $t1 !== null ) { |
| 874 | foreach ( array_reverse( explode( ' ', $t1 ) ) as $t ) { |
| 875 | DOMUtils::addTypeOf( $encapTgt, $t, true ); |
| 876 | } |
| 877 | } |
| 878 | } |
| 879 | |
| 880 | /* ---------------------------------------------------------------- |
| 881 | * We'll attempt to update dp1.dsr to reflect the entire range of |
| 882 | * the template. This relies on a couple observations: |
| 883 | * |
| 884 | * 1. In the common case, dp2.dsr->end will be > dp1.dsr->end |
| 885 | * If so, new range = dp1.dsr->start, dp2.dsr->end |
| 886 | * |
| 887 | * 2. But, foster parenting can complicate this when range.end is a table |
| 888 | * and range.start has been fostered out of the table (range.end). |
| 889 | * But, we need to verify this assumption. |
| 890 | * |
| 891 | * 2a. If dp2.dsr->start is smaller than dp1.dsr->start, this is a |
| 892 | * confirmed case of range.start being fostered out of range.end. |
| 893 | * |
| 894 | * 2b. If dp2.dsr->start is unknown, we rely on fostered flag on |
| 895 | * range.start, if any. |
| 896 | * ---------------------------------------------------------------- */ |
| 897 | $dp1 = DOMDataUtils::getDataParsoid( $range->start ); |
| 898 | $dp1DSR = isset( $dp1->dsr ) ? clone $dp1->dsr : null; |
| 899 | $dp2DSR = self::getRangeEndDSR( $range ); |
| 900 | |
| 901 | if ( $dp1DSR ) { |
| 902 | if ( $dp2DSR ) { |
| 903 | // Case 1. above |
| 904 | if ( $dp2DSR->end > $dp1DSR->end ) { |
| 905 | $dp1DSR->end = $dp2DSR->end; |
| 906 | } |
| 907 | |
| 908 | // Case 2. above |
| 909 | $endDsr = $dp2DSR->start; |
| 910 | if ( DOMCompat::nodeName( $range->end ) === 'table' && |
| 911 | $endDsr !== null && |
| 912 | ( $endDsr < $dp1DSR->start || !empty( $dp1->fostered ) ) |
| 913 | ) { |
| 914 | $dp1DSR->start = $endDsr; |
| 915 | } |
| 916 | } |
| 917 | |
| 918 | // encapsulation possible only if dp1.dsr is valid |
| 919 | $encapValid = Utils::isValidDSR( $dp1DSR ) && |
| 920 | $dp1DSR->end >= $dp1DSR->start; |
| 921 | } |
| 922 | |
| 923 | if ( $encapValid ) { |
| 924 | // Find transclusion info from the array (skip past a wikitext element) |
| 925 | /** @var CompoundTemplateInfo $firstTplInfo */ |
| 926 | $firstTplInfo = is_string( $tplArray[0] ) ? $tplArray[1] : $tplArray[0]; |
| 927 | |
| 928 | // Add any leading wikitext |
| 929 | if ( $firstTplInfo->dsr->start > $dp1DSR->start ) { |
| 930 | // This gap in dsr (between the final encapsulated content, and the |
| 931 | // content that actually came from a template) is indicative of this |
| 932 | // being a mixed-template-content-block and/or multi-template-content-block |
| 933 | // scenario. |
| 934 | // |
| 935 | // In this case, record the name of the first node in the encapsulated |
| 936 | // content. During html -> wt serialization, newline constraints for |
| 937 | // this entire block has to be determined relative to this node. |
| 938 | $ftn = self::findFirstTemplatedNode( $range ); |
| 939 | if ( $ftn !== null ) { |
| 940 | $encapDP->firstWikitextNode = $ftn; |
| 941 | } |
| 942 | $width = $firstTplInfo->dsr->start - $dp1DSR->start; |
| 943 | array_unshift( |
| 944 | $tplArray, |
| 945 | PHPUtils::safeSubstr( $this->frame->getSrcText(), $dp1DSR->start, $width ) |
| 946 | ); |
| 947 | } |
| 948 | |
| 949 | // Add any trailing wikitext |
| 950 | /** @var CompoundTemplateInfo $lastTplInfo */ |
| 951 | $lastTplInfo = PHPUtils::lastItem( $tplArray ); |
| 952 | if ( $lastTplInfo->dsr->end < $dp1DSR->end ) { |
| 953 | $width = $dp1DSR->end - $lastTplInfo->dsr->end; |
| 954 | $tplArray[] = PHPUtils::safeSubstr( $this->frame->getSrcText(), $lastTplInfo->dsr->end, $width ); |
| 955 | } |
| 956 | |
| 957 | // Map the array of { dsr: .. , args: .. } objects to just the args property |
| 958 | $infoIndex = 0; |
| 959 | $parts = []; |
| 960 | $pi = []; |
| 961 | foreach ( $tplArray as $a ) { |
| 962 | if ( is_string( $a ) ) { |
| 963 | $parts[] = $a; |
| 964 | } elseif ( $a instanceof CompoundTemplateInfo ) { |
| 965 | // Remember the position of the transclusion relative |
| 966 | // to other transclusions. Should match the index of |
| 967 | // the corresponding private metadata in $templateInfos. |
| 968 | $a->info->i = $infoIndex++; |
| 969 | $a->info->type = 'template'; |
| 970 | if ( $a->isParam ) { |
| 971 | $a->info->type = 'templatearg'; |
| 972 | } elseif ( $a->info->func ) { |
| 973 | $a->info->type = $this->v3PFOutput ? |
| 974 | 'v3parserfunction' : 'parserfunction'; |
| 975 | } |
| 976 | $parts[] = $a->info; |
| 977 | // FIXME: we throw away the array keys and rebuild them |
| 978 | // again in WikitextSerializer |
| 979 | $pi[] = array_values( $a->info->paramInfos ); |
| 980 | } |
| 981 | } |
| 982 | |
| 983 | if ( !is_string( $parts[0] ) && $parts[0]->type === 'v3parserfunction' ) { |
| 984 | DOMUtils::addTypeOf( $encapTgt, 'mw:ParserFunction', false ); |
| 985 | } |
| 986 | |
| 987 | // Set up dsr->start, dsr->end, and data-mw on the target node |
| 988 | // Avoid clobbering existing (ex: extension) data-mw information (T214241) |
| 989 | $encapDataMw = DOMDataUtils::getDataMw( $encapTgt ); |
| 990 | $encapDataMw->parts = $parts; |
| 991 | DOMDataUtils::setDataMw( $encapTgt, $encapDataMw ); |
| 992 | $encapDP->pi = $pi; |
| 993 | |
| 994 | // Special case when mixed-attribute-and-content templates are |
| 995 | // involved. This information is reliable and comes from the |
| 996 | // AttributeExpander and gets around the problem of unmarked |
| 997 | // fostered content that findFirstTemplatedNode runs into. |
| 998 | $firstWikitextNode = DOMDataUtils::getDataParsoid( |
| 999 | $range->startElem |
| 1000 | )->firstWikitextNode ?? null; |
| 1001 | if ( empty( $encapDP->firstWikitextNode ) && $firstWikitextNode ) { |
| 1002 | $encapDP->firstWikitextNode = $firstWikitextNode; |
| 1003 | } |
| 1004 | } else { |
| 1005 | $errors = [ 'Do not have necessary info. to encapsulate Tpl: ' . $i ]; |
| 1006 | $errors[] = 'Start Elt : ' . DOMCompat::getOuterHTML( $startElem ); |
| 1007 | $errors[] = 'End Elt : ' . DOMCompat::getOuterHTML( $range->endElem ); |
| 1008 | $errors[] = 'Start DSR : ' . PHPUtils::jsonEncode( $dp1DSR ?? 'no-start-dsr' ); |
| 1009 | $errors[] = 'End DSR : ' . PHPUtils::jsonEncode( $dp2DSR ?? [] ); |
| 1010 | $this->env->log( 'error', implode( "\n", $errors ) ); |
| 1011 | } |
| 1012 | |
| 1013 | // Make DSR range zero-width for fostered templates after |
| 1014 | // setting up data-mw. However, since template encapsulation |
| 1015 | // sometimes captures both fostered content as well as the table |
| 1016 | // from which it was fostered from, in those scenarios, we should |
| 1017 | // leave DSR info untouched. |
| 1018 | // |
| 1019 | // SSS FIXME: |
| 1020 | // 1. Should we remove the fostered flag from the entire |
| 1021 | // encapsulated block if we dont set dsr width range to zero |
| 1022 | // since only part of the block is fostered, not the entire |
| 1023 | // encapsulated block? |
| 1024 | // |
| 1025 | // 2. In both cases, should we mark these uneditable by adding |
| 1026 | // mw:Placeholder to the typeof? |
| 1027 | if ( !empty( $dp1->fostered ) ) { |
| 1028 | $encapDataMw = DOMDataUtils::getDataMw( $encapTgt ); |
| 1029 | if ( !$encapDataMw || |
| 1030 | !$encapDataMw->parts || |
| 1031 | count( $encapDataMw->parts ) === 1 |
| 1032 | ) { |
| 1033 | $dp1DSR->end = $dp1DSR->start; |
| 1034 | } |
| 1035 | } |
| 1036 | |
| 1037 | // Update DSR after fostering-related fixes are done. |
| 1038 | if ( $encapValid ) { |
| 1039 | // encapInfo.dp points to DOMDataUtils.getDataParsoid(encapInfo.target) |
| 1040 | // and all updates below update properties in that object tree. |
| 1041 | if ( empty( $encapDP->dsr ) ) { |
| 1042 | $encapDP->dsr = $dp1DSR; |
| 1043 | } else { |
| 1044 | $encapDP->dsr->start = $dp1DSR->start; |
| 1045 | $encapDP->dsr->end = $dp1DSR->end; |
| 1046 | } |
| 1047 | $encapDP->src = $encapDP->dsr->substr( |
| 1048 | $this->frame->getSrcText() |
| 1049 | ); |
| 1050 | } |
| 1051 | |
| 1052 | // Remove startElem (=range.startElem) if a meta. If a meta, |
| 1053 | // it is guaranteed to be a marker meta added to mark the start |
| 1054 | // of the template. |
| 1055 | if ( WTUtils::isTplMarkerMeta( $startElem ) ) { |
| 1056 | $startElem->parentNode->removeChild( $startElem ); |
| 1057 | } |
| 1058 | |
| 1059 | $range->endElem->parentNode->removeChild( $range->endElem ); |
| 1060 | } |
| 1061 | } |
| 1062 | |
| 1063 | /** |
| 1064 | * Attach a range to a node. |
| 1065 | * |
| 1066 | * @param Element $node |
| 1067 | * @param DOMRangeInfo $range |
| 1068 | */ |
| 1069 | private function addNodeRange( Element $node, DOMRangeInfo $range ): void { |
| 1070 | // With the native DOM extension, normally you assume that DOMNode |
| 1071 | // objects are temporary -- you get a new DOMNode every time you |
| 1072 | // traverse the DOM. But by retaining a reference in the |
| 1073 | // SplObjectStorage, we ensure that the DOMNode object stays live while |
| 1074 | // the pass is active. Then its address can be used as an index. |
| 1075 | if ( !isset( $this->nodeRanges[$node] ) ) { |
| 1076 | // We have to use an object as the data because |
| 1077 | // SplObjectStorage::offsetGet() does not provide an lval. |
| 1078 | $this->nodeRanges[$node] = new DOMRangeInfoArray; |
| 1079 | } |
| 1080 | $this->nodeRanges[$node]->ranges[$range->id] = $range; |
| 1081 | } |
| 1082 | |
| 1083 | /** |
| 1084 | * Get the ranges attached to this node, indexed by range ID. |
| 1085 | * |
| 1086 | * @param Element $node |
| 1087 | * @return DOMRangeInfo[]|null |
| 1088 | */ |
| 1089 | private function getNodeRanges( Element $node ): ?array { |
| 1090 | return $this->nodeRanges[$node]->ranges ?? null; |
| 1091 | } |
| 1092 | |
| 1093 | /** |
| 1094 | * Recursively walk the DOM tree. Find wrappable template ranges and return them. |
| 1095 | * |
| 1096 | * @param Node $rootNode |
| 1097 | * @return DOMRangeInfo[] |
| 1098 | */ |
| 1099 | protected function findWrappableMetaRanges( Node $rootNode ): array { |
| 1100 | $tpls = []; |
| 1101 | $tplRanges = []; |
| 1102 | $this->findWrappableTemplateRangesRecursive( $rootNode, $tpls, $tplRanges ); |
| 1103 | return $tplRanges; |
| 1104 | } |
| 1105 | |
| 1106 | /** |
| 1107 | * Recursive helper for findWrappableTemplateRanges() |
| 1108 | * |
| 1109 | * @param Node $rootNode |
| 1110 | * @param ElementRange[] &$tpls Template start and end elements by ID |
| 1111 | * @param DOMRangeInfo[] &$tplRanges Template range info |
| 1112 | */ |
| 1113 | private function findWrappableTemplateRangesRecursive( |
| 1114 | Node $rootNode, array &$tpls, array &$tplRanges |
| 1115 | ): void { |
| 1116 | $elem = $rootNode->firstChild; |
| 1117 | |
| 1118 | while ( $elem ) { |
| 1119 | // get the next sibling before doing anything since |
| 1120 | // we may delete elem as part of encapsulation |
| 1121 | $nextSibling = $elem->nextSibling; |
| 1122 | |
| 1123 | if ( $elem instanceof Element ) { |
| 1124 | $metaType = $this->matchMetaType( $elem ); |
| 1125 | |
| 1126 | // Ignore templates without tsr. |
| 1127 | // |
| 1128 | // These are definitely nested in other templates / extensions |
| 1129 | // and need not be wrapped themselves since they |
| 1130 | // can never be edited directly. |
| 1131 | // |
| 1132 | // NOTE: We are only testing for tsr presence on the start-elem |
| 1133 | // because wikitext errors can lead to parse failures and no tsr |
| 1134 | // on end-meta-tags. |
| 1135 | // |
| 1136 | // Ex: "<ref>{{1x|bar}}<!--bad-></ref>" |
| 1137 | if ( $metaType !== null && |
| 1138 | ( !empty( DOMDataUtils::getDataParsoid( $elem )->tsr ) || |
| 1139 | str_ends_with( $metaType, '/End' ) |
| 1140 | ) |
| 1141 | ) { |
| 1142 | $about = $this->getRangeId( $elem ); |
| 1143 | $tpl = $tpls[$about] ?? null; |
| 1144 | // Is this a start marker? |
| 1145 | if ( !str_ends_with( $metaType, '/End' ) ) { |
| 1146 | if ( $tpl ) { |
| 1147 | $tpl->startElem = $elem; |
| 1148 | // content or end marker existed already |
| 1149 | if ( !empty( $tpl->endElem ) ) { |
| 1150 | // End marker was foster-parented. |
| 1151 | // Found actual start tag. |
| 1152 | $tplRanges[] = $this->getDOMRange( |
| 1153 | $elem, $tpl->endElem, $tpl->endElem ); |
| 1154 | } else { |
| 1155 | // should not happen! |
| 1156 | throw new UnreachableException( "start found after content for $about." ); |
| 1157 | } |
| 1158 | } else { |
| 1159 | $tpl = new ElementRange; |
| 1160 | $tpl->startElem = $elem; |
| 1161 | $tpls[$about] = $tpl; |
| 1162 | } |
| 1163 | } else { |
| 1164 | // elem is the end-meta tag |
| 1165 | if ( $tpl ) { |
| 1166 | /* ------------------------------------------------------------ |
| 1167 | * Special case: In some cases, the entire template content can |
| 1168 | * get fostered out of a table, not just the start/end marker. |
| 1169 | * |
| 1170 | * Simplest example: |
| 1171 | * |
| 1172 | * {| |
| 1173 | * {{1x|foo}} |
| 1174 | * |} |
| 1175 | * |
| 1176 | * More complex example: |
| 1177 | * |
| 1178 | * {| |
| 1179 | * {{1x| |
| 1180 | * a |
| 1181 | * b |
| 1182 | * |
| 1183 | * c |
| 1184 | * }} |
| 1185 | * |} |
| 1186 | * |
| 1187 | * Since meta-tags don't normally get fostered out, this scenario |
| 1188 | * only arises when the entire content including meta-tags was |
| 1189 | * wrapped in p-tags. So, we look to see if: |
| 1190 | * 1. the end-meta-tag's parent has a table sibling, |
| 1191 | * 2. the start meta's parent is marked as fostered. |
| 1192 | * If so, we recognize this as an adoption scenario and fix up |
| 1193 | * DSR of start-meta-tag's parent to include the table's DSR. |
| 1194 | * ------------------------------------------------------------*/ |
| 1195 | $sm = $tpl->startElem; |
| 1196 | |
| 1197 | // TODO: this should only happen in fairly specific cases of the |
| 1198 | // annotation processing and should eventually be handled properly. |
| 1199 | // In the meantime, we create and log an exception to have an idea |
| 1200 | // of the amplitude of the problem. |
| 1201 | if ( $sm === null ) { |
| 1202 | throw new RangeBuilderException( 'No start tag found for the range' ); |
| 1203 | } |
| 1204 | $em = $elem; |
| 1205 | $ee = $em; |
| 1206 | $tbl = $em->parentNode->nextSibling; |
| 1207 | |
| 1208 | // Dont get distracted by a newline node -- skip over it |
| 1209 | // Unsure why it shows up occasionally |
| 1210 | if ( $tbl && $tbl instanceof Text && $tbl->nodeValue === "\n" ) { |
| 1211 | $tbl = $tbl->nextSibling; |
| 1212 | } |
| 1213 | |
| 1214 | $dp = !DOMUtils::atTheTop( $sm->parentNode ) ? |
| 1215 | DOMDataUtils::getDataParsoid( $sm->parentNode ) : null; |
| 1216 | if ( $tbl && DOMCompat::nodeName( $tbl ) === 'table' && !empty( $dp->fostered ) ) { |
| 1217 | '@phan-var Element $tbl'; /** @var Element $tbl */ |
| 1218 | $tblDP = DOMDataUtils::getDataParsoid( $tbl ); |
| 1219 | if ( isset( $dp->tsr ) && $dp->tsr->start !== null && $dp->tsr->start !== null && |
| 1220 | isset( $tblDP->dsr ) && $tblDP->dsr->start !== null && $tblDP->dsr->start === null |
| 1221 | ) { |
| 1222 | $tblDP->dsr->start = $dp->tsr->start; |
| 1223 | } |
| 1224 | $tbl->setAttribute( 'about', $about ); // set about on elem |
| 1225 | $ee = $tbl; |
| 1226 | } |
| 1227 | $tplRanges[] = $this->getDOMRange( $sm, $em, $ee ); |
| 1228 | } else { |
| 1229 | // The end tag can appear before the start tag if it is fostered out |
| 1230 | // of the table and the start tag is not. |
| 1231 | // It can even technically happen that both tags are fostered out of |
| 1232 | // a table and that the range is flipped: while the fostered content of |
| 1233 | // single table is fostered in-order, the ordering might change |
| 1234 | // across tables if the tags are not initially fostered by the same |
| 1235 | // table. |
| 1236 | $tpl = new ElementRange; |
| 1237 | $tpl->endElem = $elem; |
| 1238 | $tpls[$about] = $tpl; |
| 1239 | } |
| 1240 | } |
| 1241 | } else { |
| 1242 | $this->findWrappableTemplateRangesRecursive( $elem, $tpls, $tplRanges ); |
| 1243 | } |
| 1244 | } |
| 1245 | |
| 1246 | $elem = $nextSibling; |
| 1247 | } |
| 1248 | } |
| 1249 | |
| 1250 | /** |
| 1251 | * Returns the meta type of the element if it exists and matches the type expected by the |
| 1252 | * current class, null otherwise |
| 1253 | * @param Element $elem the element to check |
| 1254 | * @return string|null |
| 1255 | */ |
| 1256 | protected function matchMetaType( Element $elem ): ?string { |
| 1257 | // for this class we're interested in the template type |
| 1258 | return WTUtils::matchTplType( $elem ); |
| 1259 | } |
| 1260 | |
| 1261 | protected function verifyTplInfoExpectation( ?TemplateInfo $templateInfo, TempData $tmp ): void { |
| 1262 | if ( !$templateInfo ) { |
| 1263 | // An assertion here is probably an indication that we're |
| 1264 | // mistakenly doing template wrapping in a nested context. |
| 1265 | Assert::invariant( $tmp->getFlag( TempData::FROM_FOSTER ), 'Template range without arginfo.' ); |
| 1266 | } |
| 1267 | } |
| 1268 | |
| 1269 | public function execute( Node $root ): void { |
| 1270 | $tplRanges = $this->findWrappableMetaRanges( $root ); |
| 1271 | if ( count( $tplRanges ) > 0 ) { |
| 1272 | $nonOverlappingRanges = $this->findTopLevelNonOverlappingRanges( $root, $tplRanges ); |
| 1273 | $this->encapsulateTemplates( $nonOverlappingRanges ); |
| 1274 | } |
| 1275 | } |
| 1276 | |
| 1277 | /** |
| 1278 | * Creates a range that encloses $startMeta and $endMeta |
| 1279 | * |
| 1280 | * @param Element $startMeta |
| 1281 | * @param Element $endMeta |
| 1282 | * @param ?Element $endElem |
| 1283 | * @return DOMRangeInfo |
| 1284 | */ |
| 1285 | protected function findEnclosingRange( |
| 1286 | Element $startMeta, Element $endMeta, ?Element $endElem = null |
| 1287 | ): DOMRangeInfo { |
| 1288 | $range = new DOMRangeInfo( |
| 1289 | Utils::stripParsoidIdPrefix( $this->getRangeId( $startMeta ) ), |
| 1290 | DOMDataUtils::getDataParsoid( $startMeta )->tsr->start, |
| 1291 | $startMeta, |
| 1292 | $endMeta |
| 1293 | ); |
| 1294 | |
| 1295 | // Find common ancestor of startMeta and endElem |
| 1296 | $startAncestors = DOMUtils::pathToRoot( $startMeta ); |
| 1297 | $elem = $endElem ?? $endMeta; |
| 1298 | $parentNode = $elem->parentNode; |
| 1299 | while ( $parentNode && $parentNode->nodeType !== XML_DOCUMENT_NODE ) { |
| 1300 | $i = array_search( $parentNode, $startAncestors, true ); |
| 1301 | if ( $i === 0 ) { |
| 1302 | throw new UnreachableException( |
| 1303 | 'The startMeta cannot be the common ancestor.' |
| 1304 | ); |
| 1305 | } elseif ( $i > 0 ) { |
| 1306 | $range->start = $startAncestors[$i - 1]; |
| 1307 | $range->end = $elem; |
| 1308 | break; |
| 1309 | } |
| 1310 | $elem = $parentNode; |
| 1311 | $parentNode = $elem->parentNode; |
| 1312 | } |
| 1313 | |
| 1314 | return $range; |
| 1315 | } |
| 1316 | } |