Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
9.92% |
13 / 131 |
|
0.00% |
0 / 7 |
CRAP | |
0.00% |
0 / 1 |
| MetaHandler | |
9.92% |
13 / 131 |
|
0.00% |
0 / 7 |
3651.20 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| handle | |
22.81% |
13 / 57 |
|
0.00% |
0 / 1 |
244.63 | |||
| needToWriteStartMeta | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
110 | |||
| 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\Core\DOMCompat; |
| 7 | use Wikimedia\Parsoid\DOM\Element; |
| 8 | use Wikimedia\Parsoid\DOM\Node; |
| 9 | use Wikimedia\Parsoid\DOM\Text; |
| 10 | use Wikimedia\Parsoid\Html2Wt\DiffUtils; |
| 11 | use Wikimedia\Parsoid\Html2Wt\SerializerState; |
| 12 | use Wikimedia\Parsoid\Html2Wt\WTSUtils; |
| 13 | use Wikimedia\Parsoid\Utils\DiffDOMUtils; |
| 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 = $state->getEnv()->getSiteConfig()->getMagicWordWT( |
| 46 | $switchType[1], $dp->magicSrc ?? '' |
| 47 | ); |
| 48 | $state->emitChunk( $out, $node ); |
| 49 | } else { |
| 50 | ( new FallbackHTMLHandler )->handle( $node, $state ); |
| 51 | } |
| 52 | } elseif ( WTUtils::isAnnotationStartMarkerMeta( $node ) ) { |
| 53 | $annType = WTUtils::extractAnnotationType( $node ); |
| 54 | if ( $this->needToWriteStartMeta( $state, $node ) ) { |
| 55 | $datamw = DOMDataUtils::getDataMw( $node ); |
| 56 | $attrs = ""; |
| 57 | foreach ( ( $datamw->getExtAttribs() ?? [] ) as $k => $v ) { |
| 58 | // numeric $k will get converted to int by PHP when |
| 59 | // they are used as array keys |
| 60 | $k = (string)$k; |
| 61 | if ( $v === "" ) { |
| 62 | $attrs .= ' ' . $k; |
| 63 | } else { |
| 64 | $attrs .= ' ' . $k . '="' . $v . '"'; |
| 65 | } |
| 66 | } |
| 67 | // Follow-up on attributes sanitation to happen in T295168 |
| 68 | $state->emitChunk( '<' . $annType . $attrs . '>', $node ); |
| 69 | $state->openAnnotationRange( $annType, $datamw->extendedRange ?? false ); |
| 70 | } |
| 71 | } elseif ( WTUtils::isAnnotationEndMarkerMeta( $node ) ) { |
| 72 | if ( $this->needToWriteEndMeta( $state, $node ) ) { |
| 73 | $annType = WTUtils::extractAnnotationType( $node ); |
| 74 | $state->emitChunk( '</' . $annType . '>', $node ); |
| 75 | $state->closeAnnotationRange( $annType ); |
| 76 | } |
| 77 | } else { |
| 78 | switch ( DOMCompat::getAttribute( $node, 'typeof' ) ) { |
| 79 | case 'mw:Includes/IncludeOnly': |
| 80 | // Remove the dp.src when older revisions of HTML expire in RESTBase |
| 81 | $state->emitChunk( $dmw->src ?? $dp->src ?? '', $node ); |
| 82 | break; |
| 83 | case 'mw:Includes/IncludeOnly/End': |
| 84 | // Just ignore. |
| 85 | break; |
| 86 | case 'mw:Includes/NoInclude': |
| 87 | $state->emitChunk( $dp->src ?? '<noinclude>', $node ); |
| 88 | break; |
| 89 | case 'mw:Includes/NoInclude/End': |
| 90 | $state->emitChunk( $dp->src ?? '</noinclude>', $node ); |
| 91 | break; |
| 92 | case 'mw:Includes/OnlyInclude': |
| 93 | $state->emitChunk( $dp->src ?? '<onlyinclude>', $node ); |
| 94 | break; |
| 95 | case 'mw:Includes/OnlyInclude/End': |
| 96 | $state->emitChunk( $dp->src ?? '</onlyinclude>', $node ); |
| 97 | break; |
| 98 | case 'mw:DiffMarker/inserted': |
| 99 | case 'mw:DiffMarker/deleted': |
| 100 | case 'mw:DiffMarker/moved': |
| 101 | case 'mw:Separator': |
| 102 | // just ignore it |
| 103 | break; |
| 104 | default: |
| 105 | ( new FallbackHTMLHandler() )->handle( $node, $state ); |
| 106 | } |
| 107 | } |
| 108 | return $node->nextSibling; |
| 109 | } |
| 110 | |
| 111 | /** |
| 112 | * Decides if we need to write an annotation start meta at the place we encounter it |
| 113 | * @param SerializerState $state |
| 114 | * @param Element $node |
| 115 | * @return bool |
| 116 | */ |
| 117 | private function needToWriteStartMeta( SerializerState $state, Element $node ): bool { |
| 118 | if ( !$state->selserMode ) { |
| 119 | return true; |
| 120 | } |
| 121 | if ( WTUtils::isMovedMetaTag( $node ) ) { |
| 122 | $nextContentSibling = DOMCompat::getNextElementSibling( $node ); |
| 123 | // If the meta tag has been moved, it comes from its next element.... "almost". |
| 124 | // First exception is if we have several marker annotations in a row - then we need |
| 125 | // to pass them all. Second exception is if we have fostered content: then we're |
| 126 | // interested in what happens in the table, which happens _after_ the fostered content. |
| 127 | while ( $nextContentSibling !== null && |
| 128 | ( WTUtils::isMarkerAnnotation( $nextContentSibling ) || |
| 129 | !empty( DOMDataUtils::getDataParsoid( $nextContentSibling )->fostered ) |
| 130 | ) |
| 131 | ) { |
| 132 | $nextContentSibling = DOMCompat::getNextElementSibling( $nextContentSibling ); |
| 133 | } |
| 134 | |
| 135 | if ( $nextContentSibling !== null ) { |
| 136 | // When the content from which the meta tag comes gets |
| 137 | // deleted or modified, we emit _now_ so that we don't risk losing it. The range |
| 138 | // stays extended in the round-tripped version of the wikitext. |
| 139 | $nextdiffdata = DOMDataUtils::getDataParsoidDiff( $nextContentSibling ); |
| 140 | if ( |
| 141 | DiffUtils::isDiffMarker( $nextContentSibling ) || |
| 142 | ( $nextdiffdata && !$nextdiffdata->isEmpty() ) |
| 143 | ) { |
| 144 | return true; |
| 145 | } |
| 146 | |
| 147 | return !WTSUtils::origSrcValidInEditedContext( $state, $nextContentSibling ); |
| 148 | } |
| 149 | } |
| 150 | return true; |
| 151 | } |
| 152 | |
| 153 | /** |
| 154 | * Decides if we need to write an annotation end meta at the place we encounter it |
| 155 | * @param SerializerState $state |
| 156 | * @param Element $node |
| 157 | * @return bool |
| 158 | */ |
| 159 | private function needToWriteEndMeta( SerializerState $state, Element $node ): bool { |
| 160 | if ( !$state->selserMode ) { |
| 161 | return true; |
| 162 | } |
| 163 | if ( WTUtils::isMovedMetaTag( $node ) ) { |
| 164 | $prevElementSibling = DOMCompat::getPreviousElementSibling( $node ); |
| 165 | while ( $prevElementSibling !== null && |
| 166 | WTUtils::isMarkerAnnotation( $prevElementSibling ) |
| 167 | ) { |
| 168 | $prevElementSibling = DOMCompat::getPreviousElementSibling( $prevElementSibling ); |
| 169 | } |
| 170 | if ( $prevElementSibling ) { |
| 171 | $prevdiffdata = DOMDataUtils::getDataParsoidDiff( $prevElementSibling ); |
| 172 | |
| 173 | if ( |
| 174 | DiffUtils::isDiffMarker( $prevElementSibling ) || |
| 175 | ( $prevdiffdata && !$prevdiffdata->isEmpty() ) |
| 176 | ) { |
| 177 | return true; |
| 178 | } |
| 179 | return !WTSUtils::origSrcValidInEditedContext( $state, $prevElementSibling ); |
| 180 | } |
| 181 | } |
| 182 | return true; |
| 183 | } |
| 184 | |
| 185 | /** |
| 186 | * We create a newline (or two) if: |
| 187 | * * the previous element is a block element |
| 188 | * * the previous element is text, AND we're not in an inline-text situation: this |
| 189 | * corresponds to text having been added in VE without creating a paragraph, which happens |
| 190 | * when inserting a new line before the <meta> tag in VE. The "we're not in an inline text" |
| 191 | * is a heuristic and doesn't work for the ends of line for instance, but it shouldn't add |
| 192 | * semantic whitespace either. |
| 193 | * @param Node $meta |
| 194 | * @param Node $otherNode |
| 195 | * @return bool |
| 196 | */ |
| 197 | private function needNewLineSepBeforeMeta( Node $meta, Node $otherNode ) { |
| 198 | return ( $otherNode !== $meta->parentNode |
| 199 | && ( |
| 200 | ( $otherNode instanceof Element && DOMUtils::isWikitextBlockNode( $otherNode ) ) || |
| 201 | ( $otherNode instanceof Text && |
| 202 | DOMUtils::isWikitextBlockNode( DiffDOMUtils::nextNonSepSibling( $meta ) ) |
| 203 | ) |
| 204 | ) ); |
| 205 | } |
| 206 | |
| 207 | /** @inheritDoc */ |
| 208 | public function before( Element $node, Node $otherNode, SerializerState $state ): array { |
| 209 | if ( WTUtils::isAnnotationStartMarkerMeta( $node ) ) { |
| 210 | if ( $this->needNewLineSepBeforeMeta( $node, $otherNode ) ) { |
| 211 | return [ 'min' => 2 ]; |
| 212 | } else { |
| 213 | return []; |
| 214 | } |
| 215 | } |
| 216 | if ( WTUtils::isAnnotationEndMarkerMeta( $node ) ) { |
| 217 | if ( $this->needNewLineSepBeforeMeta( $node, $otherNode ) ) { |
| 218 | return [ |
| 219 | 'min' => 1 |
| 220 | ]; |
| 221 | } else { |
| 222 | return []; |
| 223 | } |
| 224 | } |
| 225 | |
| 226 | $type = DOMCompat::getAttribute( $node, 'typeof' ) ?? |
| 227 | DOMCompat::getAttribute( $node, 'property' ); |
| 228 | if ( $type && str_contains( $type, 'mw:PageProp/categorydefaultsort' ) ) { |
| 229 | if ( $otherNode instanceof Element |
| 230 | && DOMUtils::nodeName( $otherNode ) === 'p' |
| 231 | && ( DOMDataUtils::getDataParsoid( $otherNode )->stx ?? null ) !== 'html' |
| 232 | ) { |
| 233 | // Since defaultsort is outside the p-tag, we need 2 newlines |
| 234 | // to ensure that it go back into the p-tag when parsed. |
| 235 | return [ 'min' => 2 ]; |
| 236 | } else { |
| 237 | return [ 'min' => 1 ]; |
| 238 | } |
| 239 | } elseif ( WTUtils::isNewElt( $node ) && |
| 240 | // Placeholder and annotation metas or <*include*> tags don't need to be serialized on |
| 241 | // their own line |
| 242 | !DOMUtils::matchTypeOf( $node, '#^mw:(Placeholder|Includes|Annotation)(/|$)#' ) |
| 243 | ) { |
| 244 | return [ 'min' => 1 ]; |
| 245 | } else { |
| 246 | return []; |
| 247 | } |
| 248 | } |
| 249 | |
| 250 | /** @inheritDoc */ |
| 251 | public function after( Element $node, Node $otherNode, SerializerState $state ): array { |
| 252 | if ( WTUtils::isAnnotationEndMarkerMeta( $node ) ) { |
| 253 | if ( $otherNode !== $node->parentNode && $otherNode instanceof Element && |
| 254 | DOMUtils::isWikitextBlockNode( $otherNode ) ) { |
| 255 | return [ 'min' => 2 ]; |
| 256 | } else { |
| 257 | return []; |
| 258 | } |
| 259 | } |
| 260 | if ( WTUtils::isAnnotationStartMarkerMeta( $node ) ) { |
| 261 | if ( $otherNode !== $node->parentNode && $otherNode instanceof Element && |
| 262 | DOMUtils::isWikitextBlockNode( $otherNode ) ) { |
| 263 | return [ 'min' => 1 ]; |
| 264 | } else { |
| 265 | return []; |
| 266 | } |
| 267 | } |
| 268 | |
| 269 | // No diffs |
| 270 | if ( WTUtils::isNewElt( $node ) && |
| 271 | // Placeholder and annotation metas or <*include*> tags don't need to be serialized on |
| 272 | // their own line |
| 273 | !DOMUtils::matchTypeOf( $node, '#^mw:(Placeholder|Includes|Annotation)(/|$)#' ) |
| 274 | ) { |
| 275 | return [ 'min' => 1 ]; |
| 276 | } else { |
| 277 | return []; |
| 278 | } |
| 279 | } |
| 280 | } |