Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 281 |
|
0.00% |
0 / 10 |
CRAP | |
0.00% |
0 / 1 |
| TableFixups | |
0.00% |
0 / 281 |
|
0.00% |
0 / 10 |
11990 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| stripDoubleTDs | |
0.00% |
0 / 19 |
|
0.00% |
0 / 1 |
110 | |||
| isSimpleTemplatedSpan | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| fillDSRGap | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| hoistTransclusionInfo | |
0.00% |
0 / 55 |
|
0.00% |
0 / 1 |
182 | |||
| collectAttributishContent | |
0.00% |
0 / 38 |
|
0.00% |
0 / 1 |
210 | |||
| reparseTemplatedAttributes | |
0.00% |
0 / 29 |
|
0.00% |
0 / 1 |
72 | |||
| combineAttrsWithPreviousCell | |
0.00% |
0 / 37 |
|
0.00% |
0 / 1 |
72 | |||
| getReparseType | |
0.00% |
0 / 35 |
|
0.00% |
0 / 1 |
600 | |||
| handleTableCellTemplates | |
0.00% |
0 / 62 |
|
0.00% |
0 / 1 |
702 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Wt2Html\DOM\Handlers; |
| 5 | |
| 6 | use Wikimedia\Assert\Assert; |
| 7 | use Wikimedia\Parsoid\Config\Env; |
| 8 | use Wikimedia\Parsoid\Core\Sanitizer; |
| 9 | use Wikimedia\Parsoid\DOM\Comment; |
| 10 | use Wikimedia\Parsoid\DOM\Element; |
| 11 | use Wikimedia\Parsoid\DOM\Node; |
| 12 | use Wikimedia\Parsoid\DOM\Text; |
| 13 | use Wikimedia\Parsoid\NodeData\DataMw; |
| 14 | use Wikimedia\Parsoid\NodeData\TempData; |
| 15 | use Wikimedia\Parsoid\NodeData\TemplateInfo; |
| 16 | use Wikimedia\Parsoid\Utils\DiffDOMUtils; |
| 17 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 18 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 19 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 20 | use Wikimedia\Parsoid\Utils\DTState; |
| 21 | use Wikimedia\Parsoid\Utils\PHPUtils; |
| 22 | use Wikimedia\Parsoid\Utils\Utils; |
| 23 | use Wikimedia\Parsoid\Utils\WTUtils; |
| 24 | use Wikimedia\Parsoid\Wt2Html\Frame; |
| 25 | use Wikimedia\Parsoid\Wt2Html\PegTokenizer; |
| 26 | |
| 27 | /** |
| 28 | * Provides DOMTraverser visitors that fix template-induced interrupted table cell parsing |
| 29 | * by recombining table cells and/or reparsing table cell content as attributes. |
| 30 | * - stripDoubleTDs |
| 31 | * - handleTableCellTemplates |
| 32 | */ |
| 33 | class TableFixups { |
| 34 | /** |
| 35 | * @var PegTokenizer |
| 36 | */ |
| 37 | private $tokenizer; |
| 38 | |
| 39 | public function __construct( Env $env ) { |
| 40 | /** |
| 41 | * Set up some helper objects for reparseTemplatedAttributes |
| 42 | */ |
| 43 | |
| 44 | /** |
| 45 | * Actually the regular tokenizer, but we'll use |
| 46 | * tokenizeTableCellAttributes only. |
| 47 | */ |
| 48 | $this->tokenizer = new PegTokenizer( $env ); |
| 49 | } |
| 50 | |
| 51 | /** |
| 52 | * DOM visitor that strips the double td for this test case: |
| 53 | * ``` |
| 54 | * |{{1x|{{!}} Foo}} |
| 55 | * ``` |
| 56 | * |
| 57 | * @see https://phabricator.wikimedia.org/T52603 |
| 58 | * @param Element $node |
| 59 | * @param Frame $frame |
| 60 | * @return bool|Node |
| 61 | */ |
| 62 | public function stripDoubleTDs( Element $node, Frame $frame ) { |
| 63 | $nextNode = $node->nextSibling; |
| 64 | if ( !WTUtils::isLiteralHTMLNode( $node ) && |
| 65 | $nextNode instanceof Element && |
| 66 | DOMCompat::nodeName( $nextNode ) === 'td' && |
| 67 | !WTUtils::isLiteralHTMLNode( $nextNode ) && |
| 68 | DiffDOMUtils::nodeEssentiallyEmpty( $node ) && ( |
| 69 | // Since typeof will not be set for nested template, |
| 70 | // use a hacky work-around for nested templates. |
| 71 | DOMUtils::hasTypeOf( $nextNode, 'mw:Transclusion' ) || |
| 72 | preg_match( '/^{{.*?}}$/D', DOMDataUtils::getDataParsoid( $nextNode )->src ?? '' ) |
| 73 | ) |
| 74 | ) { |
| 75 | // Update the dsr. Since we are coalescing the first |
| 76 | // node with the second (or, more precisely, deleting |
| 77 | // the first node), we have to update the second DSR's |
| 78 | // starting point and start tag width. |
| 79 | $nodeDSR = DOMDataUtils::getDataParsoid( $node )->dsr ?? null; |
| 80 | $nextNodeDP = DOMDataUtils::getDataParsoid( $nextNode ); |
| 81 | |
| 82 | if ( $nodeDSR && !empty( $nextNodeDP->dsr ) ) { |
| 83 | $nextNodeDP->dsr->start = $nodeDSR->start; |
| 84 | } |
| 85 | |
| 86 | $dataMW = DOMDataUtils::getDataMw( $nextNode ); |
| 87 | $nodeSrc = WTUtils::getWTSource( $frame, $node ); |
| 88 | $dataMW->parts ??= []; |
| 89 | array_unshift( $dataMW->parts, $nodeSrc ); |
| 90 | |
| 91 | // Delete the duplicated <td> node. |
| 92 | $node->parentNode->removeChild( $node ); |
| 93 | // This node was deleted, so don't continue processing on it. |
| 94 | return $nextNode; |
| 95 | } |
| 96 | |
| 97 | return true; |
| 98 | } |
| 99 | |
| 100 | private function isSimpleTemplatedSpan( Node $node ): bool { |
| 101 | return DOMCompat::nodeName( $node ) === 'span' && |
| 102 | DOMUtils::hasTypeOf( $node, 'mw:Transclusion' ) && |
| 103 | DOMUtils::allChildrenAreTextOrComments( $node ); |
| 104 | } |
| 105 | |
| 106 | /** |
| 107 | * @param list<string|TemplateInfo> &$parts |
| 108 | * @param Frame $frame |
| 109 | * @param int $offset1 |
| 110 | * @param int $offset2 |
| 111 | */ |
| 112 | private function fillDSRGap( array &$parts, Frame $frame, int $offset1, int $offset2 ): void { |
| 113 | if ( $offset1 < $offset2 ) { |
| 114 | $parts[] = PHPUtils::safeSubstr( $frame->getSrcText(), $offset1, $offset2 - $offset1 ); |
| 115 | } |
| 116 | } |
| 117 | |
| 118 | /** |
| 119 | * Hoist transclusion information from cell content / attributes |
| 120 | * onto the cell itself. |
| 121 | */ |
| 122 | private function hoistTransclusionInfo( |
| 123 | Frame $frame, array $transclusions, Element $td, DTState $dtState |
| 124 | ): void { |
| 125 | // Initialize dsr for $td |
| 126 | // In `handleTableCellTemplates`, we're creating a cell w/o dsr info. |
| 127 | $tdDp = DOMDataUtils::getDataParsoid( $td ); |
| 128 | if ( !Utils::isValidDSR( $tdDp->dsr ?? null ) ) { |
| 129 | $tplDp = DOMDataUtils::getDataParsoid( $transclusions[0] ); |
| 130 | Assert::invariant( Utils::isValidDSR( $tplDp->dsr ?? null ), 'Expected valid DSR' ); |
| 131 | $tdDp->dsr = clone $tplDp->dsr; |
| 132 | } |
| 133 | |
| 134 | // Build up $parts, $pi to set up the combined transclusion info on $td. |
| 135 | // Note that content for all but the last template has been swallowed into |
| 136 | // the attributes of $td. |
| 137 | $parts = []; |
| 138 | $pi = []; |
| 139 | $lastTpl = null; |
| 140 | $prevDp = null; |
| 141 | |
| 142 | $index = 0; |
| 143 | foreach ( $transclusions as $i => $tpl ) { |
| 144 | $tplDp = DOMDataUtils::getDataParsoid( $tpl ); |
| 145 | Assert::invariant( Utils::isValidDSR( $tplDp->dsr ?? null ), 'Expected valid DSR' ); |
| 146 | |
| 147 | // Plug DSR gaps between transclusions |
| 148 | if ( !$prevDp ) { |
| 149 | $this->fillDSRGap( $parts, $frame, $tdDp->dsr->start, $tplDp->dsr->start ); |
| 150 | } else { |
| 151 | $this->fillDSRGap( $parts, $frame, $prevDp->dsr->end, $tplDp->dsr->start ); |
| 152 | } |
| 153 | |
| 154 | // Assimilate $tpl's data-mw and data-parsoid pi info |
| 155 | $dmw = DOMDataUtils::getDataMw( $tpl ); |
| 156 | foreach ( $dmw->parts ?? [] as $part ) { |
| 157 | // Template index is relative to other transclusions. |
| 158 | // This index is used to extract whitespace information from |
| 159 | // data-parsoid and that array only includes info for templates. |
| 160 | // So skip over strings here. |
| 161 | if ( !is_string( $part ) ) { |
| 162 | // Cloning is strictly not needed here, but mimicking |
| 163 | // code in WrapSectionsState.php |
| 164 | $part = clone $part; |
| 165 | $part->i = $index++; |
| 166 | } |
| 167 | $parts[] = $part; |
| 168 | } |
| 169 | PHPUtils::pushArray( $pi, $tplDp->pi ?? [ [] ] ); |
| 170 | DOMDataUtils::setDataMw( $tpl, null ); |
| 171 | |
| 172 | $lastTpl = $tpl; |
| 173 | $prevDp = $tplDp; |
| 174 | } |
| 175 | |
| 176 | $aboutId = DOMCompat::getAttribute( $lastTpl, 'about' ); |
| 177 | |
| 178 | // Hoist transclusion information to $td. |
| 179 | $td->setAttribute( 'typeof', 'mw:Transclusion' ); |
| 180 | $td->setAttribute( 'about', $aboutId ); |
| 181 | |
| 182 | // Add wikitext for the table cell content following $lastTpl |
| 183 | $this->fillDSRGap( $parts, $frame, $prevDp->dsr->end, $tdDp->dsr->end ); |
| 184 | |
| 185 | // Save the new data-mw on the td |
| 186 | $dmw = new DataMw( [] ); |
| 187 | $dmw->parts = $parts; |
| 188 | DOMDataUtils::setDataMw( $td, $dmw ); |
| 189 | $tdDp->pi = $pi; |
| 190 | |
| 191 | // td wraps everything now. |
| 192 | // Remove template encapsulation from here on. |
| 193 | // This simplifies the problem of analyzing the <td> |
| 194 | // for additional fixups (|| Boo || Baz) by potentially |
| 195 | // invoking 'reparseTemplatedAttributes' on split cells |
| 196 | // with some modifications. |
| 197 | $child = $lastTpl; |
| 198 | |
| 199 | // Transclusions may be nested in elements in some ugly wikitext so |
| 200 | // make sure we're starting at a direct descendant of td |
| 201 | while ( $child->parentNode !== $td ) { |
| 202 | $child = $child->parentNode; |
| 203 | } |
| 204 | |
| 205 | while ( $child ) { |
| 206 | if ( |
| 207 | DOMCompat::nodeName( $child ) === 'span' && |
| 208 | DOMCompat::getAttribute( $child, 'about' ) === $aboutId |
| 209 | ) { |
| 210 | // Remove the encapsulation attributes. If there are no more attributes left, |
| 211 | // the span wrapper is useless and can be removed. |
| 212 | $child->removeAttribute( 'about' ); |
| 213 | $child->removeAttribute( 'typeof' ); |
| 214 | if ( DOMDataUtils::noAttrs( $child ) ) { |
| 215 | $next = $child->firstChild ?: $child->nextSibling; |
| 216 | DOMUtils::migrateChildren( $child, $td, $child ); |
| 217 | $child->parentNode->removeChild( $child ); |
| 218 | $child = $next; |
| 219 | } else { |
| 220 | $child = $child->nextSibling; |
| 221 | } |
| 222 | } else { |
| 223 | $child = $child->nextSibling; |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | // $dtState->tplInfo can be null when information is hoisted |
| 228 | // from children to $td because DOMTraverser hasn't seen the |
| 229 | // children yet! |
| 230 | if ( !$dtState->tplInfo ) { |
| 231 | $dtState->tplInfo = (object)[ |
| 232 | 'first' => $td, |
| 233 | 'last' => $td, |
| 234 | 'clear' => false |
| 235 | ]; |
| 236 | } |
| 237 | } |
| 238 | |
| 239 | /** |
| 240 | * Collect potential attribute content. |
| 241 | * |
| 242 | * We expect this to be text nodes without a pipe character followed by one or |
| 243 | * more nowiki spans, followed by a template encapsulation with pure-text and |
| 244 | * nowiki content. Collection stops when encountering a pipe character. |
| 245 | * |
| 246 | * @param Env $env |
| 247 | * @param Element $cell known to be <td> / <th> |
| 248 | * @param ?Element $templateWrapper |
| 249 | * @return ?array |
| 250 | */ |
| 251 | public function collectAttributishContent( |
| 252 | Env $env, Element $cell, ?Element $templateWrapper |
| 253 | ): ?array { |
| 254 | $buf = []; |
| 255 | $nowikis = []; |
| 256 | $transclusions = $templateWrapper ? [ $templateWrapper ] : []; |
| 257 | |
| 258 | // Some of this logic could be replaced by DSR-based recovery of |
| 259 | // wikitext that is outside templates. But since we have to walk over |
| 260 | // templated content in this fashion anyway, we might as well use the |
| 261 | // same logic uniformly. |
| 262 | |
| 263 | $traverse = static function ( ?Node $child ) use ( |
| 264 | &$traverse, &$buf, &$nowikis, &$transclusions |
| 265 | ): bool { |
| 266 | while ( $child ) { |
| 267 | if ( $child instanceof Comment ) { |
| 268 | // Legacy parser strips comments during parsing => drop them. |
| 269 | } elseif ( $child instanceof Text ) { |
| 270 | $text = $child->nodeValue; |
| 271 | $buf[] = $text; |
| 272 | |
| 273 | // Are we done accumulating? |
| 274 | if ( preg_match( '/(?:^|[^|])\|(?:[^|]|$)/D', $text ) ) { |
| 275 | return true; |
| 276 | } |
| 277 | } else { |
| 278 | '@phan-var Element $child'; /** @var Element $child */ |
| 279 | if ( DOMUtils::hasTypeOf( $child, 'mw:Transclusion' ) ) { |
| 280 | $transclusions[] = $child; |
| 281 | } |
| 282 | |
| 283 | if ( WTUtils::isFirstExtensionWrapperNode( $child ) ) { |
| 284 | // "|" chars in extension content don't trigger table-cell parsing |
| 285 | // since they have higher precedence in tokenization. The extension |
| 286 | // content will simply be dropped (but any side effects it had will |
| 287 | // continue to apply. Ex: <ref> tags might leave an orphaned ref in |
| 288 | // the <references> section). |
| 289 | $child = WTUtils::skipOverEncapsulatedContent( $child ); |
| 290 | continue; |
| 291 | } elseif ( DOMUtils::hasTypeOf( $child, 'mw:Entity' ) ) { |
| 292 | // Get entity's wikitext source, not rendered content. |
| 293 | // " " is "\n" which breaks attribute parsing! |
| 294 | $buf[] = DOMDataUtils::getDataParsoid( $child )->src ?? $child->textContent; |
| 295 | } elseif ( DOMUtils::hasTypeOf( $child, 'mw:Nowiki' ) ) { |
| 296 | // Nowiki span were added to protect otherwise |
| 297 | // meaningful wikitext chars used in attributes. |
| 298 | // Save the content and add in a marker to splice out later. |
| 299 | $nowikis[] = $child->textContent; |
| 300 | $buf[] = '<nowiki-marker>'; |
| 301 | } elseif ( DOMUtils::hasRel( $child, 'mw:WikiLink' ) || |
| 302 | WTUtils::isGeneratedFigure( $child ) |
| 303 | ) { |
| 304 | // Wikilinks/images abort attribute parsing |
| 305 | return true; |
| 306 | } else { |
| 307 | if ( $traverse( $child->firstChild ) ) { |
| 308 | return true; |
| 309 | } |
| 310 | } |
| 311 | } |
| 312 | |
| 313 | $child = $child->nextSibling; |
| 314 | } |
| 315 | |
| 316 | return false; |
| 317 | }; |
| 318 | |
| 319 | if ( $traverse( $cell->firstChild ) ) { |
| 320 | return [ |
| 321 | 'txt' => implode( '', $buf ), |
| 322 | 'nowikis' => $nowikis, |
| 323 | 'transclusions' => $transclusions, |
| 324 | ]; |
| 325 | } else { |
| 326 | return null; |
| 327 | } |
| 328 | } |
| 329 | |
| 330 | /** |
| 331 | * T46498, second part of T52603 |
| 332 | * |
| 333 | * Handle wikitext like |
| 334 | * ``` |
| 335 | * {| |
| 336 | * |{{nom|Bar}} |
| 337 | * |} |
| 338 | * ``` |
| 339 | * where nom expands to `style="foo" class="bar"|Bar`. The attributes are |
| 340 | * tokenized and stripped from the table contents. |
| 341 | * |
| 342 | * This method works well for the templates documented in |
| 343 | * https://en.wikipedia.org/wiki/Template:Table_cell_templates/doc |
| 344 | * |
| 345 | * Nevertheless, there are some limitations: |
| 346 | * - We assume that attributes don't contain wiki markup (apart from <nowiki>) |
| 347 | * and end up in text or nowiki nodes. |
| 348 | * - Only a single table cell is produced / opened by the template that |
| 349 | * contains the attributes. This limitation could be lifted with more |
| 350 | * aggressive re-parsing if really needed in practice. |
| 351 | * - There is only a single transclusion in the table cell content. This |
| 352 | * limitation can be lifted with more advanced data-mw construction. |
| 353 | * |
| 354 | * $cell known to be <td> / <th> |
| 355 | */ |
| 356 | public function reparseTemplatedAttributes( |
| 357 | Frame $frame, Element $cell, ?Element $templateWrapper, DTSTate $dtState |
| 358 | ): void { |
| 359 | $env = $frame->getEnv(); |
| 360 | // Collect attribute content and examine it |
| 361 | $attributishContent = $this->collectAttributishContent( $env, $cell, $templateWrapper ); |
| 362 | if ( !$attributishContent ) { |
| 363 | return; |
| 364 | } |
| 365 | |
| 366 | /** |
| 367 | * FIXME: These checks are insufficient. |
| 368 | * Previous rounds of table fixups might have created this cell without |
| 369 | * any templated content (the while loop in handleTableCellTemplates). |
| 370 | * Till we figure out a reliable test for this, we'll reparse attributes always. |
| 371 | * |
| 372 | * // This DOM pass is trying to bridge broken parses across |
| 373 | * // template boundaries. So, if templates aren't involved, |
| 374 | * // no reason to reparse. |
| 375 | * if ( count( $attributishContent['transclusions'] ) === 0 && |
| 376 | * !WTUtils::fromEncapsulatedContent( $cell ) |
| 377 | * ) { |
| 378 | * return; |
| 379 | * } |
| 380 | */ |
| 381 | |
| 382 | $attrText = $attributishContent['txt']; |
| 383 | if ( !preg_match( '/(^[^|]+\|)([^|]|$)/D', $attrText, $matches ) ) { |
| 384 | return; |
| 385 | } |
| 386 | $attributishPrefix = $matches[1]; |
| 387 | |
| 388 | // Splice in nowiki content. We added in <nowiki> markers to prevent the |
| 389 | // above regexps from matching on nowiki-protected chars. |
| 390 | if ( str_contains( $attributishPrefix, '<nowiki-marker>' ) ) { |
| 391 | $attributishPrefix = preg_replace_callback( |
| 392 | '/<nowiki-marker>/', |
| 393 | static function ( $unused ) use ( &$attributishContent ) { |
| 394 | // This is a little tricky. We want to use the content from the |
| 395 | // nowikis to reparse the string to key/val pairs but the rule, |
| 396 | // single_cell_table_args, will invariably get tripped up on |
| 397 | // newlines which, to this point, were shuttled through in the |
| 398 | // nowiki. Core sanitizer will do this replacement in attr vals |
| 399 | // so it's a safe normalization to do here. |
| 400 | return preg_replace( '/\s+/', ' ', array_shift( $attributishContent['nowikis'] ) ); |
| 401 | }, |
| 402 | $attributishPrefix |
| 403 | ); |
| 404 | } |
| 405 | |
| 406 | // re-parse the attributish prefix |
| 407 | $attributeTokens = $this->tokenizer->tokenizeTableCellAttributes( $attributishPrefix, false ); |
| 408 | |
| 409 | // No attributes => nothing more to do! |
| 410 | if ( !$attributeTokens ) { |
| 411 | return; |
| 412 | } |
| 413 | |
| 414 | // Note that `row_syntax_table_args` (the rule used for tokenizing above) |
| 415 | // returns an array consisting of [table_attributes, spaces, pipe] |
| 416 | $attrs = $attributeTokens[0]; |
| 417 | |
| 418 | // Sanitize attrs and transfer them to the td node |
| 419 | Sanitizer::applySanitizedArgs( $env->getSiteConfig(), $cell, $attrs ); |
| 420 | $cellDp = DOMDataUtils::getDataParsoid( $cell ); |
| 421 | // Reparsed cells start off as non-mergeable-table cells |
| 422 | // and preserve that property after reparsing |
| 423 | $cellDp->setTempFlag( TempData::NON_MERGEABLE_TABLE_CELL ); |
| 424 | $cellDp->setTempFlag( TempData::NO_ATTRS, false ); |
| 425 | |
| 426 | // If the transclusion node was embedded within the td node, |
| 427 | // lift up the about group to the td node. |
| 428 | $transclusions = $attributishContent['transclusions']; |
| 429 | if ( $transclusions && ( $cell !== $transclusions[0] || count( $transclusions ) > 1 ) ) { |
| 430 | $this->hoistTransclusionInfo( $frame, $transclusions, $cell, $dtState ); |
| 431 | } |
| 432 | |
| 433 | // Drop content that has been consumed by the reparsed attribute content. |
| 434 | // NOTE: We serialize and reparse data-object-id attributes as well which |
| 435 | // ensures stashed data-* attributes continue to be usable. |
| 436 | // FIXME: This is too naive. What about all the care we showed in `collectAttributishContent`? |
| 437 | DOMCompat::setInnerHTML( $cell, |
| 438 | preg_replace( '/^[^|]*\|/', '', DOMCompat::getInnerHTML( $cell ) ) ); |
| 439 | } |
| 440 | |
| 441 | /** |
| 442 | * Possibilities: |
| 443 | * - $cell and $prev are both <td>s (or both <th>s) |
| 444 | * - Common case |
| 445 | * - Ex: "|align=left {{tpl returning | foobar}}" |
| 446 | * So, "|align=left |foobar" is the combined string |
| 447 | * - Combined cell is a <td> (will be <th> if both were <th>s) |
| 448 | * - We assign new attributes to $cell and drop $prev |
| 449 | * - $cell is <td> and $prev is <th> |
| 450 | * - Ex: "!align=left {{tpl returning | foobar}}" |
| 451 | * So, "!align=left |foobar" is the combined string |
| 452 | * - The combined cell will be a <th> with attributes "align=left" |
| 453 | * and content "foobar" |
| 454 | * - $cell is <th> and $prev is <td> |
| 455 | * - Ex: "|align=left {{tpl returning !scope=row | foobar}}" |
| 456 | * So "|align=left !scope=row | foobar" is the combined string |
| 457 | * and we need to drop the th-attributes entirely after combining |
| 458 | * - The combined cell will be a <td> |
| 459 | * - $cell's attribute is dropped |
| 460 | * - $prev's content is dropped |
| 461 | * |
| 462 | * FIXME: There are a number of other merge possibilities that end up |
| 463 | * in this function that aren't accounted for yet! Couple of them are |
| 464 | * in the unsupported scenario 1/2 buckets below. |
| 465 | * |
| 466 | * @param Frame $frame |
| 467 | * @param Element $cell |
| 468 | * @return bool |
| 469 | */ |
| 470 | private function combineAttrsWithPreviousCell( Frame $frame, Element $cell ): bool { |
| 471 | // UNSUPPORTED SCENARIO 1: |
| 472 | // In this cell-combining scenario, $prev can have attributes only if it |
| 473 | // also had content. See example below: |
| 474 | // Ex: |class="foo"|bar{{1x|1={{!}}foo}} |
| 475 | // should parse as <td class="foo">bar|foo</td> |
| 476 | // In this case, the attributes/content of $cell would end up as the |
| 477 | // content of the combined cell. |
| 478 | // |
| 479 | // UNSUPPORTED SCENARIO 2: |
| 480 | // The template produced attributes as well as maybe a new cell. |
| 481 | // Ex: |class="foo"{{1x| foo}} and |class="foo"{{1x| foo}} |
| 482 | // We let the more general 'reparseTemplatedAttributes' code handle |
| 483 | // this scenario for now. |
| 484 | $prev = $cell->previousSibling; |
| 485 | DOMUtils::assertElt( $prev ); |
| 486 | $prevDp = DOMDataUtils::getDataParsoid( $prev ); |
| 487 | |
| 488 | // Build the attribute string |
| 489 | $prevCellSrc = PHPUtils::safeSubstr( |
| 490 | $frame->getSrcText(), $prevDp->dsr->start, $prevDp->dsr->length() ); |
| 491 | // "|" or "!", but doesn't matter since we discard that anyway |
| 492 | $reparseSrc = substr( $prevCellSrc, $prevDp->dsr->openWidth ) . "|"; |
| 493 | |
| 494 | // Reparse the attributish prefix |
| 495 | $env = $frame->getEnv(); |
| 496 | $attributeTokens = $this->tokenizer->tokenizeTableCellAttributes( $reparseSrc, false ); |
| 497 | if ( !is_array( $attributeTokens ) ) { |
| 498 | $env->log( "error/wt2html", |
| 499 | "TableFixups: Failed to successfully reparse $reparseSrc as table cell attributes" ); |
| 500 | return false; |
| 501 | } |
| 502 | |
| 503 | // Update data-mw, DSR if $cell is an encapsulation wrapper |
| 504 | $cellDp = DOMDataUtils::getDataParsoid( $cell ); |
| 505 | if ( DOMUtils::hasTypeOf( $cell, 'mw:Transclusion' ) ) { |
| 506 | $dataMW = DOMDataUtils::getDataMw( $cell ); |
| 507 | array_unshift( $dataMW->parts, $prevCellSrc ); |
| 508 | $cellDSR = $cellDp->dsr ?? null; |
| 509 | if ( $cellDSR && $cellDSR->start ) { |
| 510 | $cellDSR->start -= strlen( $prevCellSrc ); |
| 511 | } |
| 512 | } |
| 513 | |
| 514 | $parent = $cell->parentNode; |
| 515 | if ( DOMCompat::nodeName( $cell ) === DOMCompat::nodeName( $prev ) ) { |
| 516 | // Matching cell types |
| 517 | $combinedCell = $cell; |
| 518 | $combinedCellDp = $cellDp; |
| 519 | $parent->removeChild( $prev ); |
| 520 | } else { |
| 521 | // Different cell types |
| 522 | $combinedCell = $prev; |
| 523 | |
| 524 | // Remove all content on $prev which will |
| 525 | // become the new combined cell |
| 526 | while ( $prev->firstChild ) { |
| 527 | $prev->removeChild( $prev->firstChild ); |
| 528 | } |
| 529 | // Note that this implicitly migrates data-mw and data-parsoid |
| 530 | foreach ( DOMUtils::attributes( $cell ) as $k => $v ) { |
| 531 | $combinedCell->setAttribute( $k, $v ); |
| 532 | } |
| 533 | DOMUtils::migrateChildren( $cell, $combinedCell ); |
| 534 | $parent->removeChild( $cell ); |
| 535 | |
| 536 | $combinedCellDp = DOMDataUtils::getDataParsoid( $combinedCell ); |
| 537 | } |
| 538 | |
| 539 | // Note that `row_syntax_table_args` (the rule used for tokenizing above) |
| 540 | // returns an array consisting of [table_attributes, spaces, pipe] |
| 541 | $attrs = $attributeTokens[0]; |
| 542 | Sanitizer::applySanitizedArgs( $env->getSiteConfig(), $combinedCell, $attrs ); |
| 543 | // Combined cells don't merge further |
| 544 | $combinedCellDp->setTempFlag( TempData::NON_MERGEABLE_TABLE_CELL ); |
| 545 | $combinedCellDp->setTempFlag( TempData::NO_ATTRS, false ); |
| 546 | |
| 547 | return true; |
| 548 | } |
| 549 | |
| 550 | private const NO_REPARSING = 0; |
| 551 | private const COMBINE_WITH_PREV_CELL = 1; |
| 552 | private const OTHER_REPARSE = 2; |
| 553 | |
| 554 | /** |
| 555 | * $cell is known to be <td>/<th> |
| 556 | */ |
| 557 | private function getReparseType( Element $cell, DTState $dtState ): int { |
| 558 | $inTplContent = $dtState->tplInfo !== null; |
| 559 | $dp = DOMDataUtils::getDataParsoid( $cell ); |
| 560 | if ( !$dp->getTempFlag( TempData::NON_MERGEABLE_TABLE_CELL ) && |
| 561 | !$dp->getTempFlag( TempData::FAILED_REPARSE ) && |
| 562 | // This is a good proxy for what we need: "Is $cell a template wrapper?". |
| 563 | // That info won't be available for nested templates unless we want |
| 564 | // to use the more expensive hacky check as used in "stripDoubleTDs" |
| 565 | // above. "inTplContent" is sufficient because we won't have mergeable |
| 566 | // cells for wikitext that doesn't get any part of its content from |
| 567 | // templates because NON_MERGEABLE_TABLE_CELL prevents such merges. |
| 568 | $inTplContent |
| 569 | ) { |
| 570 | // Look for opportunities where table cells could combine. This requires |
| 571 | // $cell to be a templated cell. But, we don't support combining |
| 572 | // templated cells with other templated cells. So, previous sibling |
| 573 | // cannot be templated. |
| 574 | // |
| 575 | // So, bail out of scenarios where prevDp comes from a template (the checks |
| 576 | // for isValidDSR( $prevDp-> dsr ) and valid opening tag width catch this. |
| 577 | $prev = $cell->previousSibling; |
| 578 | $prevDp = $prev instanceof Element ? DOMDataUtils::getDataParsoid( $prev ) : null; |
| 579 | if ( $prevDp && |
| 580 | !WTUtils::hasLiteralHTMLMarker( $prevDp ) && |
| 581 | Utils::isValidDSR( $prevDp->dsr ?? null, true ) && |
| 582 | !DOMUtils::hasTypeOf( $prev, 'mw:Transclusion' ) && |
| 583 | !str_contains( DOMCompat::getInnerHTML( $prev ), "\n" ) |
| 584 | ) { |
| 585 | return self::COMBINE_WITH_PREV_CELL; |
| 586 | } |
| 587 | } |
| 588 | |
| 589 | $cellIsTd = DOMCompat::nodeName( $cell ) === 'td'; |
| 590 | $testRE = $cellIsTd ? '/[|]/' : '/[!|]/'; |
| 591 | $child = $cell->firstChild; |
| 592 | while ( $child ) { |
| 593 | if ( !$inTplContent && DOMUtils::hasTypeOf( $child, 'mw:Transclusion' ) ) { |
| 594 | $inTplContent = true; |
| 595 | } |
| 596 | |
| 597 | if ( $inTplContent && |
| 598 | $child instanceof Text && |
| 599 | preg_match( $testRE, $child->textContent ) |
| 600 | ) { |
| 601 | return self::OTHER_REPARSE; |
| 602 | } |
| 603 | |
| 604 | if ( $child instanceof Element ) { |
| 605 | if ( WTUtils::isFirstExtensionWrapperNode( $child ) ) { |
| 606 | // "|" chars in extension/language variant content don't trigger |
| 607 | // table-cell parsing since they have higher precedence in tokenization |
| 608 | $child = WTUtils::skipOverEncapsulatedContent( $child ); |
| 609 | } else { |
| 610 | if ( DOMUtils::hasRel( $child, 'mw:WikiLink' ) || |
| 611 | WTUtils::isGeneratedFigure( $child ) |
| 612 | ) { |
| 613 | // Wikilinks/images abort attribute parsing |
| 614 | return self::NO_REPARSING; |
| 615 | } |
| 616 | // FIXME: Ugly for now |
| 617 | $outerHTML = DOMCompat::getOuterHTML( $child ); |
| 618 | if ( preg_match( $testRE, $outerHTML ) && |
| 619 | ( $inTplContent || preg_match( '/"mw:Transclusion"/', $outerHTML ) ) |
| 620 | ) { |
| 621 | // A "|" char in the HTML will trigger table cell tokenization. |
| 622 | // Ex: "| foobar <div> x | y </div>" will split the <div> |
| 623 | // in table-cell tokenization context. |
| 624 | return self::OTHER_REPARSE; |
| 625 | } |
| 626 | $child = $child->nextSibling; |
| 627 | } |
| 628 | } else { |
| 629 | $child = $child->nextSibling; |
| 630 | } |
| 631 | } |
| 632 | |
| 633 | return self::NO_REPARSING; |
| 634 | } |
| 635 | |
| 636 | /** |
| 637 | * In a wikitext-syntax-table-parsing context, the meaning of |
| 638 | * "|", "||", "!", "!!" is context-sensitive. Additionally, the |
| 639 | * complete syntactical construct for a table cell (including leading |
| 640 | * pipes, attributes, and content-separating pipe char) might straddle |
| 641 | * a template boundary - with some content coming from the top-level and |
| 642 | * some from a template. |
| 643 | * |
| 644 | * This impacts parsing of tables when some cells are templated since |
| 645 | * Parsoid parses template content independent of top-level content |
| 646 | * (without any preceding context). This means that Parsoid's table-cell |
| 647 | * parsing in templated contexts might be incorrect |
| 648 | * |
| 649 | * To deal with this, Parsoid implements this table-fixups pass that |
| 650 | * has to deal with cell-merging and cell-reparsing scenarios. |
| 651 | * |
| 652 | * HTML-syntax cells and non-templated cells without any templated content |
| 653 | * are not subject to this transformation and can be skipped right away. |
| 654 | * |
| 655 | * FIXME: This pass can benefit from a customized procsssor rather than |
| 656 | * piggyback on top of DOMTraverser since the DOM can be significantly |
| 657 | * mutated in these handlers. |
| 658 | * |
| 659 | * @param Element $cell $cell is known to be <td>/<th> |
| 660 | * @param Frame $frame |
| 661 | * @param DTState $dtState |
| 662 | * @return mixed |
| 663 | */ |
| 664 | public function handleTableCellTemplates( Element $cell, Frame $frame, DTState $dtState ) { |
| 665 | if ( WTUtils::isLiteralHTMLNode( $cell ) ) { |
| 666 | return true; |
| 667 | } |
| 668 | |
| 669 | // Deal with <th> special case where "!! foo" is parsed as <th>! foo</th> |
| 670 | // but should have been parsed as <th>foo</th> when not the first child |
| 671 | if ( DOMCompat::nodeName( $cell ) === 'th' && |
| 672 | DOMUtils::hasTypeOf( $cell, 'mw:Transclusion' ) && |
| 673 | // This is checking that previous sibling is not "\n" which would |
| 674 | // signal that this <th> is on a fresh line and the "!" shouldn't be stripped. |
| 675 | // If this weren't template output, we would check for "stx" === 'row'. |
| 676 | // FIXME: Note that ths check is fragile and doesn't work always, but this is |
| 677 | // the price we pay for Parsoid's independent template parsing! |
| 678 | $cell->previousSibling instanceof Element |
| 679 | ) { |
| 680 | $fc = DiffDOMUtils::firstNonSepChild( $cell ); |
| 681 | if ( $fc instanceof Text ) { |
| 682 | $leadingText = $fc->nodeValue; |
| 683 | if ( str_starts_with( $leadingText, "!" ) ) { |
| 684 | $fc->nodeValue = substr( $leadingText, 1 ); |
| 685 | } |
| 686 | } |
| 687 | } |
| 688 | |
| 689 | $reparseType = $this->getReparseType( $cell, $dtState ); |
| 690 | if ( $reparseType === self::NO_REPARSING ) { |
| 691 | return true; |
| 692 | } |
| 693 | |
| 694 | $cellDp = DOMDataUtils::getDataParsoid( $cell ); |
| 695 | if ( $reparseType === self::COMBINE_WITH_PREV_CELL ) { |
| 696 | if ( $this->combineAttrsWithPreviousCell( $frame, $cell ) ) { |
| 697 | return true; |
| 698 | } else { |
| 699 | // Clear property and retry $cell for other reparses |
| 700 | // The DOMTraverser will resume the handler on the |
| 701 | // returned $cell. |
| 702 | $cellDp->setTempFlag( TempData::FAILED_REPARSE ); |
| 703 | return $cell; |
| 704 | } |
| 705 | } |
| 706 | |
| 707 | // If the cell didn't have attrs, extract and reparse templated attrs |
| 708 | if ( $cellDp->getTempFlag( TempData::NO_ATTRS ) ) { |
| 709 | $templateWrapper = DOMUtils::hasTypeOf( $cell, 'mw:Transclusion' ) ? $cell : null; |
| 710 | $this->reparseTemplatedAttributes( $frame, $cell, $templateWrapper, $dtState ); |
| 711 | } |
| 712 | |
| 713 | // Now, examine the <td> to see if it hides additional <td>s |
| 714 | // and split it up if required. |
| 715 | // |
| 716 | // DOMTraverser will process the new cell and invoke |
| 717 | // handleTableCellTemplates on it which ensures that |
| 718 | // if any addition attribute fixup or splits are required, |
| 719 | // they will get done. |
| 720 | $newCell = null; |
| 721 | $isTd = DOMCompat::nodeName( $cell ) === 'td'; |
| 722 | $ownerDoc = $cell->ownerDocument; |
| 723 | $child = $cell->firstChild; |
| 724 | while ( $child ) { |
| 725 | $next = $child->nextSibling; |
| 726 | |
| 727 | if ( $newCell ) { |
| 728 | $newCell->appendChild( $child ); |
| 729 | } elseif ( $child instanceof Text || $this->isSimpleTemplatedSpan( $child ) ) { |
| 730 | // FIXME: This skips over scenarios like <div>foo||bar</div>. |
| 731 | $cellName = DOMCompat::nodeName( $cell ); |
| 732 | $hasSpanWrapper = !( $child instanceof Text ); |
| 733 | $match = $match1 = $match2 = null; |
| 734 | |
| 735 | // Find the first match of || |
| 736 | preg_match( '/^((?:[^|]*(?:\|[^|])?)*)\|\|([^|].*)?$/D', $child->textContent, $match1 ); |
| 737 | if ( $isTd ) { |
| 738 | $match = $match1; |
| 739 | } else { |
| 740 | // Find the first match !! |
| 741 | preg_match( '/^((?:[^!]*(?:\![^!])?)*)\!\!([^!].*)?$/D', $child->textContent, $match2 ); |
| 742 | |
| 743 | // Pick the shortest match |
| 744 | if ( $match1 && $match2 ) { |
| 745 | $match = strlen( $match1[1] ?? '' ) < strlen( $match2[1] ?? '' ) |
| 746 | ? $match1 |
| 747 | : $match2; |
| 748 | } else { |
| 749 | $match = $match1 ?: $match2; |
| 750 | } |
| 751 | } |
| 752 | |
| 753 | if ( $match ) { |
| 754 | $child->textContent = $match[1] ?? ''; |
| 755 | |
| 756 | $newCell = $ownerDoc->createElement( $cellName ); |
| 757 | if ( $hasSpanWrapper ) { |
| 758 | /** |
| 759 | * $hasSpanWrapper above ensures $child is a span. |
| 760 | * |
| 761 | * @var Element $child |
| 762 | */ |
| 763 | '@phan-var Element $child'; |
| 764 | // Fix up transclusion wrapping |
| 765 | $about = DOMCompat::getAttribute( $child, 'about' ); |
| 766 | $this->hoistTransclusionInfo( $frame, [ $child ], $cell, $dtState ); |
| 767 | } else { |
| 768 | // Refetch the about attribute since 'reparseTemplatedAttributes' |
| 769 | // might have added one to it. |
| 770 | $about = DOMCompat::getAttribute( $cell, 'about' ); |
| 771 | } |
| 772 | |
| 773 | // about may not be present if the cell was inside |
| 774 | // wrapped template content rather than being part |
| 775 | // of the outermost wrapper. |
| 776 | if ( $about !== null ) { |
| 777 | $newCell->setAttribute( 'about', $about ); |
| 778 | if ( $dtState->tplInfo && $dtState->tplInfo->last === $cell ) { |
| 779 | $dtState->tplInfo->last = $newCell; |
| 780 | } |
| 781 | } |
| 782 | $newCell->appendChild( $ownerDoc->createTextNode( $match[2] ?? '' ) ); |
| 783 | $cell->parentNode->insertBefore( $newCell, $cell->nextSibling ); |
| 784 | |
| 785 | // Set data-parsoid noAttrs flag |
| 786 | $newCellDp = DOMDataUtils::getDataParsoid( $newCell ); |
| 787 | // This new cell has 'row' stx (would be set if the tokenizer had parsed it) |
| 788 | $newCellDp->stx = 'row'; |
| 789 | $newCellDp->setTempFlag( TempData::NO_ATTRS ); |
| 790 | // It is important to set this so that when $newCell is processed by this pass, |
| 791 | // it won't accidentally recombine again with the previous cell! |
| 792 | $newCellDp->setTempFlag( TempData::NON_MERGEABLE_TABLE_CELL ); |
| 793 | } |
| 794 | } |
| 795 | |
| 796 | $child = $next; |
| 797 | } |
| 798 | |
| 799 | return true; |
| 800 | } |
| 801 | } |