Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
15.86% |
23 / 145 |
|
0.00% |
0 / 7 |
CRAP | |
0.00% |
0 / 1 |
| MetaHandler | |
15.86% |
23 / 145 |
|
0.00% |
0 / 7 |
3608.48 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| handle | |
32.39% |
23 / 71 |
|
0.00% |
0 / 1 |
308.09 | |||
| needToWriteStartMeta | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
90 | |||
| needToWriteEndMeta | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
90 | |||
| needNewLineSepBeforeMeta | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
30 | |||
| before | |
0.00% |
0 / 22 |
|
0.00% |
0 / 1 |
156 | |||
| after | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
132 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Html2Wt\DOMHandlers; |
| 5 | |
| 6 | use Wikimedia\Parsoid\DOM\Element; |
| 7 | use Wikimedia\Parsoid\DOM\Node; |
| 8 | use Wikimedia\Parsoid\DOM\Text; |
| 9 | use Wikimedia\Parsoid\Html2Wt\DiffUtils; |
| 10 | use Wikimedia\Parsoid\Html2Wt\SerializerState; |
| 11 | use Wikimedia\Parsoid\Html2Wt\WTSUtils; |
| 12 | use Wikimedia\Parsoid\Utils\DiffDOMUtils; |
| 13 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 14 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 15 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 16 | use Wikimedia\Parsoid\Utils\WTUtils; |
| 17 | |
| 18 | class MetaHandler extends DOMHandler { |
| 19 | |
| 20 | public function __construct() { |
| 21 | parent::__construct( false ); |
| 22 | } |
| 23 | |
| 24 | /** @inheritDoc */ |
| 25 | public function handle( |
| 26 | Element $node, SerializerState $state, bool $wrapperUnmodified = false |
| 27 | ): ?Node { |
| 28 | $property = DOMCompat::getAttribute( $node, 'property' ) ?? ''; |
| 29 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 30 | $dmw = DOMDataUtils::getDataMw( $node ); |
| 31 | |
| 32 | if ( isset( $dp->src ) && |
| 33 | DOMUtils::matchTypeOf( $node, '#^mw:Placeholder(/|$)#' ) |
| 34 | ) { |
| 35 | $this->emitPlaceholderSrc( $node, $state ); |
| 36 | return $node->nextSibling; |
| 37 | } |
| 38 | |
| 39 | // Check for property before type so that page properties with |
| 40 | // templated attrs roundtrip properly. |
| 41 | // Ex: {{DEFAULTSORT:{{1x|foo}} }} |
| 42 | if ( $property ) { |
| 43 | preg_match( '#^mw\:PageProp/(.*)$#D', $property, $switchType ); |
| 44 | if ( $switchType ) { |
| 45 | $out = $switchType[1]; |
| 46 | $cat = preg_match( '/^(?:category)?(.*)/', $out, $catMatch ); |
| 47 | if ( $cat && ( |
| 48 | // Need this b/c support while RESTBase has Parsoid HTML |
| 49 | // in storage with meta tags for these. |
| 50 | // Can be removed as part of T335843 |
| 51 | $catMatch[1] === 'defaultsort' || $catMatch[1] === 'displaytitle' |
| 52 | ) ) { |
| 53 | $contentInfo = $state->serializer->serializedAttrVal( $node, 'content' ); |
| 54 | if ( WTUtils::hasExpandedAttrsType( $node ) ) { |
| 55 | $out = '{{' . $contentInfo['value'] . '}}'; |
| 56 | } elseif ( isset( $dp->src ) ) { |
| 57 | $colon = strpos( $dp->src, ':', 2 ); |
| 58 | $out = preg_replace( '/^([^:}]+).*$/D', "$1", $dp->src, 1 ); |
| 59 | if ( ( $colon === false ) && ( $contentInfo['value'] === '' ) ) { |
| 60 | $out .= '}}'; |
| 61 | } else { |
| 62 | $out .= ':' . $contentInfo['value'] . '}}'; |
| 63 | } |
| 64 | } else { |
| 65 | $magicWord = mb_strtoupper( $catMatch[1] ); |
| 66 | $out = '{{' . $magicWord . ':' . $contentInfo['value'] . '}}'; |
| 67 | } |
| 68 | } else { |
| 69 | $out = $state->getEnv()->getSiteConfig()->getMagicWordWT( |
| 70 | $switchType[1], $dp->magicSrc ?? '' ); |
| 71 | } |
| 72 | $state->emitChunk( $out, $node ); |
| 73 | } else { |
| 74 | ( new FallbackHTMLHandler )->handle( $node, $state ); |
| 75 | } |
| 76 | } elseif ( WTUtils::isAnnotationStartMarkerMeta( $node ) ) { |
| 77 | $annType = WTUtils::extractAnnotationType( $node ); |
| 78 | if ( $this->needToWriteStartMeta( $state, $node ) ) { |
| 79 | $datamw = DOMDataUtils::getDataMw( $node ); |
| 80 | $attrs = ""; |
| 81 | if ( isset( $datamw->attrs ) ) { |
| 82 | foreach ( get_object_vars( $datamw->attrs ) as $k => $v ) { |
| 83 | if ( $v === "" ) { |
| 84 | $attrs .= ' ' . $k; |
| 85 | } else { |
| 86 | $attrs .= ' ' . $k . '="' . $v . '"'; |
| 87 | } |
| 88 | } |
| 89 | } |
| 90 | // Follow-up on attributes sanitation to happen in T295168 |
| 91 | $state->emitChunk( '<' . $annType . $attrs . '>', $node ); |
| 92 | $state->openAnnotationRange( $annType, $datamw->extendedRange ?? false ); |
| 93 | } |
| 94 | } elseif ( WTUtils::isAnnotationEndMarkerMeta( $node ) ) { |
| 95 | if ( $this->needToWriteEndMeta( $state, $node ) ) { |
| 96 | $annType = WTUtils::extractAnnotationType( $node ); |
| 97 | $state->emitChunk( '</' . $annType . '>', $node ); |
| 98 | $state->closeAnnotationRange( $annType ); |
| 99 | } |
| 100 | } else { |
| 101 | switch ( DOMCompat::getAttribute( $node, 'typeof' ) ) { |
| 102 | case 'mw:Includes/IncludeOnly': |
| 103 | // Remove the dp.src when older revisions of HTML expire in RESTBase |
| 104 | $state->emitChunk( $dmw->src ?? $dp->src ?? '', $node ); |
| 105 | break; |
| 106 | case 'mw:Includes/IncludeOnly/End': |
| 107 | // Just ignore. |
| 108 | break; |
| 109 | case 'mw:Includes/NoInclude': |
| 110 | $state->emitChunk( $dp->src ?? '<noinclude>', $node ); |
| 111 | break; |
| 112 | case 'mw:Includes/NoInclude/End': |
| 113 | $state->emitChunk( $dp->src ?? '</noinclude>', $node ); |
| 114 | break; |
| 115 | case 'mw:Includes/OnlyInclude': |
| 116 | $state->emitChunk( $dp->src ?? '<onlyinclude>', $node ); |
| 117 | break; |
| 118 | case 'mw:Includes/OnlyInclude/End': |
| 119 | $state->emitChunk( $dp->src ?? '</onlyinclude>', $node ); |
| 120 | break; |
| 121 | case 'mw:DiffMarker/inserted': |
| 122 | case 'mw:DiffMarker/deleted': |
| 123 | case 'mw:DiffMarker/moved': |
| 124 | case 'mw:Separator': |
| 125 | // just ignore it |
| 126 | break; |
| 127 | default: |
| 128 | ( new FallbackHTMLHandler() )->handle( $node, $state ); |
| 129 | } |
| 130 | } |
| 131 | return $node->nextSibling; |
| 132 | } |
| 133 | |
| 134 | /** |
| 135 | * Decides if we need to write an annotation start meta at the place we encounter it |
| 136 | * @param SerializerState $state |
| 137 | * @param Element $node |
| 138 | * @return bool |
| 139 | */ |
| 140 | private function needToWriteStartMeta( SerializerState $state, Element $node ): bool { |
| 141 | if ( !$state->selserMode ) { |
| 142 | return true; |
| 143 | } |
| 144 | if ( WTUtils::isMovedMetaTag( $node ) ) { |
| 145 | $nextContentSibling = DOMCompat::getNextElementSibling( $node ); |
| 146 | // If the meta tag has been moved, it comes from its next element.... "almost". |
| 147 | // First exception is if we have several marker annotations in a row - then we need |
| 148 | // to pass them all. Second exception is if we have fostered content: then we're |
| 149 | // interested in what happens in the table, which happens _after_ the fostered content. |
| 150 | while ( $nextContentSibling !== null && |
| 151 | ( WTUtils::isMarkerAnnotation( $nextContentSibling ) || |
| 152 | !empty( DOMDataUtils::getDataParsoid( $nextContentSibling )->fostered ) |
| 153 | ) |
| 154 | ) { |
| 155 | $nextContentSibling = DOMCompat::getNextElementSibling( $nextContentSibling ); |
| 156 | } |
| 157 | |
| 158 | if ( $nextContentSibling !== null ) { |
| 159 | // When the content from which the meta tag comes gets |
| 160 | // deleted or modified, we emit _now_ so that we don't risk losing it. The range |
| 161 | // stays extended in the round-tripped version of the wikitext. |
| 162 | $nextdiffdata = DOMDataUtils::getDataParsoidDiff( $nextContentSibling ); |
| 163 | if ( DiffUtils::isDiffMarker( $nextContentSibling ) || |
| 164 | ( $nextdiffdata->diff ?? null ) ) { |
| 165 | return true; |
| 166 | } |
| 167 | |
| 168 | return !WTSUtils::origSrcValidInEditedContext( $state, $nextContentSibling ); |
| 169 | } |
| 170 | } |
| 171 | return true; |
| 172 | } |
| 173 | |
| 174 | /** |
| 175 | * Decides if we need to write an annotation end meta at the place we encounter it |
| 176 | * @param SerializerState $state |
| 177 | * @param Element $node |
| 178 | * @return bool |
| 179 | */ |
| 180 | private function needToWriteEndMeta( SerializerState $state, Element $node ): bool { |
| 181 | if ( !$state->selserMode ) { |
| 182 | return true; |
| 183 | } |
| 184 | if ( WTUtils::isMovedMetaTag( $node ) ) { |
| 185 | $prevElementSibling = DOMCompat::getPreviousElementSibling( $node ); |
| 186 | while ( $prevElementSibling !== null && |
| 187 | WTUtils::isMarkerAnnotation( $prevElementSibling ) |
| 188 | ) { |
| 189 | $prevElementSibling = DOMCompat::getPreviousElementSibling( $prevElementSibling ); |
| 190 | } |
| 191 | if ( $prevElementSibling ) { |
| 192 | $prevdiffdata = DOMDataUtils::getDataParsoidDiff( $prevElementSibling ); |
| 193 | |
| 194 | if ( |
| 195 | DiffUtils::isDiffMarker( $prevElementSibling ) || |
| 196 | $prevdiffdata !== null && $prevdiffdata->diff !== null |
| 197 | ) { |
| 198 | return true; |
| 199 | } |
| 200 | return !WTSUtils::origSrcValidInEditedContext( $state, $prevElementSibling ); |
| 201 | } |
| 202 | } |
| 203 | return true; |
| 204 | } |
| 205 | |
| 206 | /** |
| 207 | * We create a newline (or two) if: |
| 208 | * * the previous element is a block element |
| 209 | * * the previous element is text, AND we're not in an inline-text situation: this |
| 210 | * corresponds to text having been added in VE without creating a paragraph, which happens |
| 211 | * when inserting a new line before the <meta> tag in VE. The "we're not in an inline text" |
| 212 | * is a heuristic and doesn't work for the ends of line for instance, but it shouldn't add |
| 213 | * semantic whitespace either. |
| 214 | * @param Node $meta |
| 215 | * @param Node $otherNode |
| 216 | * @return bool |
| 217 | */ |
| 218 | private function needNewLineSepBeforeMeta( Node $meta, Node $otherNode ) { |
| 219 | return ( $otherNode !== $meta->parentNode |
| 220 | && ( $otherNode instanceof Element && |
| 221 | DOMUtils::isWikitextBlockNode( $otherNode ) || |
| 222 | ( $otherNode instanceof Text && |
| 223 | DOMUtils::isWikitextBlockNode( DiffDOMUtils::nextNonSepSibling( $meta ) ) |
| 224 | ) |
| 225 | ) ); |
| 226 | } |
| 227 | |
| 228 | /** @inheritDoc */ |
| 229 | public function before( Element $node, Node $otherNode, SerializerState $state ): array { |
| 230 | if ( WTUtils::isAnnotationStartMarkerMeta( $node ) ) { |
| 231 | if ( $this->needNewLineSepBeforeMeta( $node, $otherNode ) ) { |
| 232 | return [ 'min' => 2 ]; |
| 233 | } else { |
| 234 | return []; |
| 235 | } |
| 236 | } |
| 237 | if ( WTUtils::isAnnotationEndMarkerMeta( $node ) ) { |
| 238 | if ( $this->needNewLineSepBeforeMeta( $node, $otherNode ) ) { |
| 239 | return [ |
| 240 | 'min' => 1 |
| 241 | ]; |
| 242 | } else { |
| 243 | return []; |
| 244 | } |
| 245 | } |
| 246 | |
| 247 | $type = DOMCompat::getAttribute( $node, 'typeof' ) ?? |
| 248 | DOMCompat::getAttribute( $node, 'property' ); |
| 249 | if ( $type && str_contains( $type, 'mw:PageProp/categorydefaultsort' ) ) { |
| 250 | if ( $otherNode instanceof Element |
| 251 | && DOMCompat::nodeName( $otherNode ) === 'p' |
| 252 | && ( DOMDataUtils::getDataParsoid( $otherNode )->stx ?? null ) !== 'html' |
| 253 | ) { |
| 254 | // Since defaultsort is outside the p-tag, we need 2 newlines |
| 255 | // to ensure that it go back into the p-tag when parsed. |
| 256 | return [ 'min' => 2 ]; |
| 257 | } else { |
| 258 | return [ 'min' => 1 ]; |
| 259 | } |
| 260 | } elseif ( WTUtils::isNewElt( $node ) && |
| 261 | // Placeholder and annotation metas or <*include*> tags don't need to be serialized on |
| 262 | // their own line |
| 263 | !DOMUtils::matchTypeOf( $node, '#^mw:(Placeholder|Includes|Annotation)(/|$)#' ) |
| 264 | ) { |
| 265 | return [ 'min' => 1 ]; |
| 266 | } else { |
| 267 | return []; |
| 268 | } |
| 269 | } |
| 270 | |
| 271 | /** @inheritDoc */ |
| 272 | public function after( Element $node, Node $otherNode, SerializerState $state ): array { |
| 273 | if ( WTUtils::isAnnotationEndMarkerMeta( $node ) ) { |
| 274 | if ( $otherNode !== $node->parentNode && $otherNode instanceof Element && |
| 275 | DOMUtils::isWikitextBlockNode( $otherNode ) ) { |
| 276 | return [ 'min' => 2 ]; |
| 277 | } else { |
| 278 | return []; |
| 279 | } |
| 280 | } |
| 281 | if ( WTUtils::isAnnotationStartMarkerMeta( $node ) ) { |
| 282 | if ( $otherNode !== $node->parentNode && $otherNode instanceof Element && |
| 283 | DOMUtils::isWikitextBlockNode( $otherNode ) ) { |
| 284 | return [ 'min' => 1 ]; |
| 285 | } else { |
| 286 | return []; |
| 287 | } |
| 288 | } |
| 289 | |
| 290 | // No diffs |
| 291 | if ( WTUtils::isNewElt( $node ) && |
| 292 | // Placeholder and annotation metas or <*include*> tags don't need to be serialized on |
| 293 | // their own line |
| 294 | !DOMUtils::matchTypeOf( $node, '#^mw:(Placeholder|Includes|Annotation)(/|$)#' ) |
| 295 | ) { |
| 296 | return [ 'min' => 1 ]; |
| 297 | } else { |
| 298 | return []; |
| 299 | } |
| 300 | } |
| 301 | } |