Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
7.69% |
12 / 156 |
|
0.00% |
0 / 11 |
CRAP | |
0.00% |
0 / 1 |
| ContentUtils | |
7.69% |
12 / 156 |
|
0.00% |
0 / 11 |
2262.35 | |
0.00% |
0 / 1 |
| toXML | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| ppToXML | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| createAndLoadDocument | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
6 | |||
| createAndLoadDocumentFragment | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| stripUnnecessaryWrappersAndSyntheticNodes | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
72 | |||
| processAttributeEmbeddedDom | |
52.17% |
12 / 23 |
|
0.00% |
0 / 1 |
20.94 | |||
| extApiWrapper | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| shiftDSR | |
0.00% |
0 / 33 |
|
0.00% |
0 / 1 |
132 | |||
| convertOffsets | |
0.00% |
0 / 54 |
|
0.00% |
0 / 1 |
156 | |||
| dumpNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| dumpDOM | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Utils; |
| 5 | |
| 6 | use Wikimedia\Assert\UnreachableException; |
| 7 | use Wikimedia\Parsoid\Config\Env; |
| 8 | use Wikimedia\Parsoid\Config\SiteConfig; |
| 9 | use Wikimedia\Parsoid\Core\DOMCompat; |
| 10 | use Wikimedia\Parsoid\Core\DomSourceRange; |
| 11 | use Wikimedia\Parsoid\DOM\Document; |
| 12 | use Wikimedia\Parsoid\DOM\DocumentFragment; |
| 13 | use Wikimedia\Parsoid\DOM\Element; |
| 14 | use Wikimedia\Parsoid\DOM\Node; |
| 15 | use Wikimedia\Parsoid\Ext\ParsoidExtensionAPI; |
| 16 | use Wikimedia\Parsoid\Mocks\MockEnv; |
| 17 | use Wikimedia\Parsoid\Mocks\MockSiteConfig; |
| 18 | use Wikimedia\Parsoid\Wt2Html\XHtmlSerializer; |
| 19 | |
| 20 | /** |
| 21 | * These utilities are for processing content that's generated |
| 22 | * by parsing source input (ex: wikitext) |
| 23 | */ |
| 24 | class ContentUtils { |
| 25 | |
| 26 | /** |
| 27 | * XML Serializer. |
| 28 | * |
| 29 | * @param Node $node |
| 30 | * @param array $options XHtmlSerializer options. |
| 31 | * @return string |
| 32 | */ |
| 33 | public static function toXML( Node $node, array $options = [] ): string { |
| 34 | return XHtmlSerializer::serialize( $node, $options )['html']; |
| 35 | } |
| 36 | |
| 37 | /** |
| 38 | * dataobject aware XML serializer, to be used in the DOM post-processing phase. |
| 39 | * |
| 40 | * @param Node $node |
| 41 | * @param array $options |
| 42 | * Data attribute options, see DOMDataUtils::storeDataAttribs() for |
| 43 | * details. In addition, setting `$options['fragment']` to true |
| 44 | * should be used when serializing a DocumentFragment unconnected to |
| 45 | * the parent document; this ensures that we don't mistakenly mark |
| 46 | * the top level document as "unloaded" if we were just serializing |
| 47 | * a fragment. |
| 48 | * |
| 49 | * Eventually most places which serialize using the `fragment` option |
| 50 | * should be converted to store the DocumentFragment natively, instead |
| 51 | * of as a string (T348161). |
| 52 | * |
| 53 | * @return string |
| 54 | */ |
| 55 | public static function ppToXML( Node $node, array $options = [] ): string { |
| 56 | $doc = $node->ownerDocument ?? $node; |
| 57 | DOMDataUtils::visitAndStoreDataAttribs( $node, $options ); |
| 58 | if ( !( $options['fragment'] ?? false ) ) { |
| 59 | DOMDataUtils::getBag( $doc )->loaded = false; |
| 60 | } |
| 61 | return self::toXML( $node, $options ); |
| 62 | } |
| 63 | |
| 64 | /** |
| 65 | * Create a new prepared document with the given HTML and load the |
| 66 | * data attributes. |
| 67 | * |
| 68 | * Don't use this inside of the parser pipeline: it shouldn't be necessary |
| 69 | * to create new documents when parsing or serializing. A document lives |
| 70 | * on the environment which can be used to create fragments. The bag added |
| 71 | * as a dynamic property to the PHP wrapper around the libxml doc |
| 72 | * is at risk of being GC-ed. |
| 73 | * |
| 74 | * @param string $html |
| 75 | * @param array $options |
| 76 | * @param ?SiteConfig $siteConfig |
| 77 | * @return Document |
| 78 | */ |
| 79 | public static function createAndLoadDocument( |
| 80 | string $html, array $options = [], ?SiteConfig $siteConfig = null |
| 81 | ): Document { |
| 82 | $doc = DOMUtils::parseHTML( $html, validateXMLNames: true ); |
| 83 | $siteConfig ??= $options['siteConfig'] ?? null; |
| 84 | if ( $siteConfig === null ) { |
| 85 | PHPUtils::deprecated( __METHOD__ . ' without SiteConfig', '0.24' ); |
| 86 | $siteConfig = new MockSiteConfig( [] ); |
| 87 | } |
| 88 | DOMDataUtils::prepareAndLoadDoc( $doc, $siteConfig, $options ); |
| 89 | return $doc; |
| 90 | } |
| 91 | |
| 92 | /** |
| 93 | * @param Document $doc |
| 94 | * @param string $html |
| 95 | * @param ?array $options Not used |
| 96 | * @return DocumentFragment |
| 97 | */ |
| 98 | public static function createAndLoadDocumentFragment( |
| 99 | Document $doc, string $html, ?array $options = null |
| 100 | ): DocumentFragment { |
| 101 | if ( $options !== null ) { |
| 102 | // $options are deprecated and ignored |
| 103 | PHPUtils::deprecated( __METHOD__ . ' with $options', '0.23' ); |
| 104 | } |
| 105 | $domFragment = $doc->createDocumentFragment(); |
| 106 | DOMUtils::setFragmentInnerHTML( $domFragment, $html ); |
| 107 | // This non-lazy loading is primarily to ensure that $doc's counters |
| 108 | // are updated based on $domFragment. This is needed since $domFragment |
| 109 | // could be a newly-constructed fragment that didn't originally exist |
| 110 | // in $doc, the parent DOM, and we want global unique ids within it. |
| 111 | DOMDataUtils::visitAndLoadDataAttribs( $domFragment ); |
| 112 | return $domFragment; |
| 113 | } |
| 114 | |
| 115 | /** |
| 116 | * Strip Parsoid-inserted section wrappers, annotation wrappers, and synthetic nodes |
| 117 | * (fallback id spans with HTML4 ids for headings, auto-generated TOC metas |
| 118 | * and possibly other such in the future) from the DOM. |
| 119 | * |
| 120 | * @param Element $node |
| 121 | */ |
| 122 | public static function stripUnnecessaryWrappersAndSyntheticNodes( Element $node ): void { |
| 123 | $n = $node->firstChild; |
| 124 | while ( $n ) { |
| 125 | $next = $n->nextSibling; |
| 126 | if ( $n instanceof Element ) { |
| 127 | if ( DOMUtils::nodeName( $n ) === 'meta' && |
| 128 | ( DOMDataUtils::getDataMw( $n )->autoGenerated ?? false ) |
| 129 | ) { |
| 130 | // Strip auto-generated synthetic meta tags |
| 131 | $n->parentNode->removeChild( $n ); |
| 132 | } elseif ( WTUtils::isFallbackIdSpan( $n ) ) { |
| 133 | // Strip <span typeof='mw:FallbackId' ...></span> |
| 134 | $n->parentNode->removeChild( $n ); |
| 135 | } else { |
| 136 | // Recurse into subtree before stripping this |
| 137 | self::stripUnnecessaryWrappersAndSyntheticNodes( $n ); |
| 138 | |
| 139 | // Strip <section> tags and synthetic extended-annotation-region wrappers |
| 140 | if ( WTUtils::isParsoidSectionTag( $n ) || |
| 141 | DOMUtils::hasTypeOf( $n, 'mw:ExtendedAnnRange' ) ) { |
| 142 | DOMUtils::migrateChildren( $n, $n->parentNode, $n ); |
| 143 | $n->parentNode->removeChild( $n ); |
| 144 | } |
| 145 | } |
| 146 | } |
| 147 | $n = $next; |
| 148 | } |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * Extensions might be interested in examining their content embedded |
| 153 | * in data-mw attributes that don't otherwise show up in the DOM. |
| 154 | * |
| 155 | * Ex: inline media captions that aren't rendered, language variant markup, |
| 156 | * attributes that are transcluded. More scenarios might be added later. |
| 157 | * |
| 158 | * @param SiteConfig $siteConfig |
| 159 | * @param Element $elt The node whose data attributes need to be examined |
| 160 | * @param callable(DocumentFragment):bool $proc |
| 161 | * The processor that will process the embedded HTML. |
| 162 | * This processor will be provided a DocumentFragment |
| 163 | * and is expected to return true if that fragment was modified. |
| 164 | */ |
| 165 | public static function processAttributeEmbeddedDom( |
| 166 | SiteConfig $siteConfig, Element $elt, callable $proc |
| 167 | ): void { |
| 168 | // Expanded attributes and media captions |
| 169 | if ( DOMUtils::matchTypeOf( $elt, '/^mw:ExpandedAttrs$/' ) || |
| 170 | WTUtils::isInlineMedia( $elt ) ) { |
| 171 | $dmw = DOMDataUtils::getDataMwIfExists( $elt ); |
| 172 | foreach ( $dmw?->embeddedDocumentFragments() ?? [] as $df ) { |
| 173 | $proc( $df ); |
| 174 | } |
| 175 | } |
| 176 | |
| 177 | // Language variant markup |
| 178 | if ( DOMUtils::matchTypeOf( $elt, '/^mw:LanguageVariant$/' ) ) { |
| 179 | $dmwv = DOMDataUtils::getDataMwVariant( $elt ); |
| 180 | foreach ( $dmwv?->embeddedDocumentFragments() ?? [] as $df ) { |
| 181 | $proc( $df ); |
| 182 | } |
| 183 | } |
| 184 | |
| 185 | // Process extension-specific embedded DocumentFragments |
| 186 | $extTagName = WTUtils::getExtTagName( $elt ); |
| 187 | if ( $extTagName ) { |
| 188 | $extConfig = $siteConfig->getExtTagConfig( $extTagName ); |
| 189 | if ( $extConfig['options']['wt2html']['embedsDomInAttributes'] ?? false ) { |
| 190 | $tagHandler = $siteConfig->getExtTagImpl( $extTagName ); |
| 191 | $extAPI = self::extApiWrapper( $siteConfig, $elt->ownerDocument ); |
| 192 | $tagHandler->processAttributeEmbeddedDom( $extAPI, $elt, $proc ); |
| 193 | } |
| 194 | } |
| 195 | $key = WTUtils::getPFragmentHandlerKey( $elt ); |
| 196 | if ( $key ) { |
| 197 | $config = $siteConfig->getPFragmentHandlerConfig( $key ); |
| 198 | if ( $config['options']['embedsDomInAttributes'] ?? false ) { |
| 199 | $handler = $siteConfig->getPFragmentHandlerImpl( $key ); |
| 200 | $extAPI = self::extApiWrapper( $siteConfig, $elt->ownerDocument ); |
| 201 | $handler->processAttributeEmbeddedDom( $extAPI, $elt, $proc ); |
| 202 | } |
| 203 | } |
| 204 | } |
| 205 | |
| 206 | /** |
| 207 | * Temporary backward-compatibility thunk to wrap a SiteConfig as |
| 208 | * a ParsoidExtensionAPI. |
| 209 | */ |
| 210 | private static function extApiWrapper( |
| 211 | SiteConfig $siteConfig, Document $topLevelDoc |
| 212 | ): ParsoidExtensionAPI { |
| 213 | // This is a backward-compatibility hack! |
| 214 | return new ParsoidExtensionAPI( new MockEnv( [ |
| 215 | 'siteConfig' => $siteConfig, |
| 216 | 'topLevelDoc' => $topLevelDoc, |
| 217 | ] ) ); |
| 218 | } |
| 219 | |
| 220 | /** |
| 221 | * Shift the DOM Source Range (DSR) of a DOM fragment. |
| 222 | * @param Env $env |
| 223 | * @param Node $rootNode |
| 224 | * @param callable $dsrFunc |
| 225 | */ |
| 226 | public static function shiftDSR( |
| 227 | Env $env, Node $rootNode, callable $dsrFunc |
| 228 | ): void { |
| 229 | $siteConfig = $env->getSiteConfig(); |
| 230 | $convertNode = static function ( Node $node ) use ( |
| 231 | $siteConfig, $dsrFunc, &$convertNode |
| 232 | ): void { |
| 233 | if ( !( $node instanceof Element ) ) { |
| 234 | return; |
| 235 | } |
| 236 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 237 | if ( isset( $dp->dsr ) ) { |
| 238 | $dp->dsr = $dsrFunc( clone $dp->dsr ); |
| 239 | // We don't need to setDataParsoid because dp is not a copy |
| 240 | |
| 241 | // This is a bit of a hack, but we use this function to |
| 242 | // clear DSR properties as well. See below as well. |
| 243 | if ( $dp->dsr === null ) { |
| 244 | unset( $dp->dsr ); |
| 245 | } |
| 246 | } |
| 247 | $tmp = $dp->getTemp(); |
| 248 | if ( isset( $tmp->origDSR ) ) { |
| 249 | // Even though tmp shouldn't escape Parsoid, go ahead and |
| 250 | // convert to enable hybrid testing. |
| 251 | $tmp->origDSR = $dsrFunc( clone $tmp->origDSR ); |
| 252 | if ( $tmp->origDSR === null ) { |
| 253 | unset( $tmp->origDSR ); |
| 254 | } |
| 255 | } |
| 256 | if ( isset( $dp->extTagOffsets ) ) { |
| 257 | $dp->extTagOffsets = $dsrFunc( clone $dp->extTagOffsets ); |
| 258 | if ( $dp->extTagOffsets === null ) { |
| 259 | unset( $dp->extTagOffsets ); |
| 260 | } |
| 261 | } |
| 262 | |
| 263 | // Handle embedded HTML in attributes |
| 264 | self::processAttributeEmbeddedDom( |
| 265 | $siteConfig, $node, |
| 266 | static function ( DocumentFragment $df ) use ( $convertNode ): bool { |
| 267 | DOMPostOrder::traverse( $df, $convertNode ); |
| 268 | return true; |
| 269 | } ); |
| 270 | |
| 271 | // DOMFragments will have already been unpacked when DSR shifting is run |
| 272 | if ( DOMUtils::hasTypeOf( $node, 'mw:DOMFragment' ) ) { |
| 273 | throw new UnreachableException( "Shouldn't encounter these nodes here." ); |
| 274 | } |
| 275 | |
| 276 | // However, extensions can choose to handle sealed fragments whenever |
| 277 | // they want and so may be returned in subpipelines which could |
| 278 | // subsequently be shifted |
| 279 | if ( DOMUtils::matchTypeOf( $node, '#^mw:DOMFragment/sealed/\w+$#D' ) ) { |
| 280 | if ( $dp->html ?? null ) { |
| 281 | DOMPostOrder::traverse( $dp->html, $convertNode ); |
| 282 | } |
| 283 | } |
| 284 | }; |
| 285 | DOMPostOrder::traverse( $rootNode, $convertNode ); |
| 286 | } |
| 287 | |
| 288 | /** |
| 289 | * Convert DSR offsets in a Document between utf-8/ucs2/codepoint |
| 290 | * indices. |
| 291 | * |
| 292 | * Offset types are: |
| 293 | * - 'byte': Bytes (UTF-8 encoding), e.g. PHP `substr()` or `strlen()`. |
| 294 | * - 'char': Unicode code points (encoding irrelevant), e.g. PHP `mb_substr()` or `mb_strlen()`. |
| 295 | * - 'ucs2': 16-bit code units (UTF-16 encoding), e.g. JavaScript `.substring()` or `.length`. |
| 296 | * |
| 297 | * @see TokenUtils::convertTokenOffsets for a related function on tokens. |
| 298 | * |
| 299 | * @param Env $env |
| 300 | * @param Document $doc The document to convert |
| 301 | * @param string $from Offset type to convert from. |
| 302 | * @param string $to Offset type to convert to. |
| 303 | */ |
| 304 | public static function convertOffsets( |
| 305 | Env $env, |
| 306 | Document $doc, |
| 307 | string $from, |
| 308 | string $to |
| 309 | ): void { |
| 310 | $env->setCurrentOffsetType( $to ); |
| 311 | if ( $from === $to ) { |
| 312 | return; // Hey, that was easy! |
| 313 | } |
| 314 | $source = $env->topFrame->getSource(); |
| 315 | $offsetMap = []; |
| 316 | $offsets = []; |
| 317 | $collect = static function ( int $n ) use ( &$offsetMap, &$offsets ): void { |
| 318 | if ( !array_key_exists( $n, $offsetMap ) ) { |
| 319 | $box = (object)[ 'value' => $n ]; |
| 320 | $offsetMap[$n] = $box; |
| 321 | $offsets[] =& $box->value; |
| 322 | } |
| 323 | }; |
| 324 | // Collect DSR offsets throughout the document |
| 325 | $collectDSR = static function ( DomSourceRange $dsr ) use ( $collect, $source, $env ): DomSourceRange { |
| 326 | // Validate DSR source |
| 327 | // FIXME T405759: $dsr->source shouldn't be null but we haven't |
| 328 | // fixed all our code yet. Also, technically we |
| 329 | // could/should collect a list of all the different $dsr |
| 330 | // sources and then run this multiple times, once for each |
| 331 | // source text. |
| 332 | if ( !( |
| 333 | $dsr->source === null || $dsr->source === $source |
| 334 | ) ) { |
| 335 | // T409345 |
| 336 | $env->log( |
| 337 | 'error/wt2html', |
| 338 | "Bad source in ::convertOffsets (T409345): ", |
| 339 | $env->getContextTitle()->getFullText(), |
| 340 | mb_substr( $dsr->source->getSrcText(), 0, 100 ) |
| 341 | ); |
| 342 | // Don't collect (or mutate) this, we don't know where |
| 343 | // it came from. |
| 344 | return $dsr; |
| 345 | } |
| 346 | if ( $dsr->start !== null ) { |
| 347 | $collect( $dsr->start ); |
| 348 | $collect( $dsr->innerStart() ); |
| 349 | } |
| 350 | if ( $dsr->end !== null ) { |
| 351 | $collect( $dsr->innerEnd() ); |
| 352 | $collect( $dsr->end ); |
| 353 | } |
| 354 | return $dsr; |
| 355 | }; |
| 356 | $body = DOMCompat::getBody( $doc ); |
| 357 | self::shiftDSR( $env, $body, $collectDSR ); |
| 358 | if ( count( $offsets ) === 0 ) { |
| 359 | return; /* nothing to do (shouldn't really happen) */ |
| 360 | } |
| 361 | // Now convert these offsets |
| 362 | TokenUtils::convertOffsets( |
| 363 | $source->getSrcText(), $from, $to, $offsets |
| 364 | ); |
| 365 | // Apply converted offsets |
| 366 | $applyDSR = static function ( DomSourceRange $dsr ) use ( $offsetMap, $source ): DomSourceRange { |
| 367 | if ( $dsr->source !== null && $dsr->source !== $source ) { |
| 368 | return $dsr; |
| 369 | } |
| 370 | $start = $dsr->start; |
| 371 | $openWidth = $dsr->openWidth; |
| 372 | if ( $start !== null ) { |
| 373 | $start = $offsetMap[$start]->value; |
| 374 | $openWidth = $offsetMap[$dsr->innerStart()]->value - $start; |
| 375 | } |
| 376 | $end = $dsr->end; |
| 377 | $closeWidth = $dsr->closeWidth; |
| 378 | if ( $end !== null ) { |
| 379 | $end = $offsetMap[$end]->value; |
| 380 | $closeWidth = $end - $offsetMap[$dsr->innerEnd()]->value; |
| 381 | } |
| 382 | return new DomSourceRange( |
| 383 | $start, $end, $openWidth, $closeWidth, source: $source |
| 384 | ); |
| 385 | }; |
| 386 | self::shiftDSR( $env, $body, $applyDSR ); |
| 387 | } |
| 388 | |
| 389 | /** |
| 390 | * @param Node $node |
| 391 | * @param array $options |
| 392 | * @return string |
| 393 | */ |
| 394 | private static function dumpNode( Node $node, array $options ): string { |
| 395 | return self::toXML( $node, $options + [ 'noSideEffects' => true ] ); |
| 396 | } |
| 397 | |
| 398 | /** |
| 399 | * Dump the DOM with attributes. |
| 400 | * |
| 401 | * @param Node $rootNode |
| 402 | * @param string $title |
| 403 | * @param array $options Associative array of options: |
| 404 | * - quiet: Suppress separators |
| 405 | * |
| 406 | * storeDataAttribs options: |
| 407 | * - discardDataParsoid |
| 408 | * - keepTmp |
| 409 | * - storeInPageBundle |
| 410 | * - storeDiffMark |
| 411 | * - env |
| 412 | * - idIndex |
| 413 | * |
| 414 | * XHtmlSerializer options: |
| 415 | * - smartQuote |
| 416 | * - innerXML |
| 417 | * - captureOffsets |
| 418 | * - addDoctype |
| 419 | * @return string The dump result |
| 420 | */ |
| 421 | public static function dumpDOM( |
| 422 | Node $rootNode, string $title = '', array $options = [] |
| 423 | ): string { |
| 424 | $buf = ''; |
| 425 | if ( empty( $options['quiet'] ) ) { |
| 426 | $buf .= "----- {$title} -----\n"; |
| 427 | } |
| 428 | $buf .= self::dumpNode( $rootNode, $options ) . "\n"; |
| 429 | |
| 430 | if ( empty( $options['quiet'] ) ) { |
| 431 | $buf .= str_repeat( '-', mb_strlen( $title ) + 12 ) . "\n"; |
| 432 | } |
| 433 | return $buf; |
| 434 | } |
| 435 | |
| 436 | } |