Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
98.15% |
583 / 594 |
|
85.29% |
29 / 34 |
CRAP | |
0.00% |
0 / 1 |
| Linter | |
98.15% |
583 / 594 |
|
85.29% |
29 / 34 |
253 | |
0.00% |
0 / 1 |
| getTagsWithChangedMisnestingBehavior | |
100.00% |
19 / 19 |
|
100.00% |
1 / 1 |
7 | |||
| leftMostMisnestedDescendent | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
8 | |||
| getMatchingMisnestedNode | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| findEnclosingTemplateName | |
77.78% |
14 / 18 |
|
0.00% |
0 / 1 |
7.54 | |||
| findLintDSR | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
5 | |||
| hasIdenticalNestedTag | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
5 | |||
| hasMisnestableContent | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
10 | |||
| endTagOptional | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getHeadingAncestor | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
3 | |||
| matchedOpenTagPairExists | |
90.00% |
9 / 10 |
|
0.00% |
0 / 1 |
7.05 | |||
| lintTreeBuilderFixup | |
100.00% |
59 / 59 |
|
100.00% |
1 / 1 |
31 | |||
| lintFostered | |
100.00% |
28 / 28 |
|
100.00% |
1 / 1 |
11 | |||
| lintObsoleteTag | |
100.00% |
38 / 38 |
|
100.00% |
1 / 1 |
17 | |||
| lintBogusImageOptions | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
8 | |||
| lintDeletableTableTag | |
100.00% |
24 / 24 |
|
100.00% |
1 / 1 |
7 | |||
| findMatchingChild | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
3 | |||
| hasNoWrapCSS | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
3 | |||
| lintPWrapBugWorkaround | |
100.00% |
17 / 17 |
|
100.00% |
1 / 1 |
5 | |||
| lintMiscTidyReplacementIssues | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
8 | |||
| lintTidyWhitespaceBug | |
98.78% |
81 / 82 |
|
0.00% |
0 / 1 |
29 | |||
| lintMultipleUnclosedFormattingTags | |
100.00% |
22 / 22 |
|
100.00% |
1 / 1 |
9 | |||
| postProcessLints | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getWikitextListItemAncestor | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
6 | |||
| lintMultilineHtmlTableInList | |
100.00% |
16 / 16 |
|
100.00% |
1 / 1 |
5 | |||
| lintWikilinksInExtlink | |
100.00% |
21 / 21 |
|
100.00% |
1 / 1 |
11 | |||
| recordLargeTablesLint | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
1 | |||
| skipNonElementNodes | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
3 | |||
| lintLargeTables | |
100.00% |
24 / 24 |
|
100.00% |
1 / 1 |
8 | |||
| lintNightModeUnawareBackgroundColor | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
4 | |||
| lintMissingAltText | |
100.00% |
25 / 25 |
|
100.00% |
1 / 1 |
9 | |||
| lintDuplicateIds | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
4 | |||
| logWikitextFixups | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
1 | |||
| findLints | |
88.24% |
30 / 34 |
|
0.00% |
0 / 1 |
11.20 | |||
| run | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
2.01 | |||
| 1 | <?php |
| 2 | |
| 3 | declare( strict_types = 1 ); |
| 4 | |
| 5 | namespace Wikimedia\Parsoid\Wt2Html\DOM\Processors; |
| 6 | |
| 7 | use stdClass; |
| 8 | use Wikimedia\Assert\UnreachableException; |
| 9 | use Wikimedia\Parsoid\Config\Env; |
| 10 | use Wikimedia\Parsoid\Core\DomSourceRange; |
| 11 | use Wikimedia\Parsoid\DOM\Comment; |
| 12 | use Wikimedia\Parsoid\DOM\Element; |
| 13 | use Wikimedia\Parsoid\DOM\Node; |
| 14 | use Wikimedia\Parsoid\DOM\Text; |
| 15 | use Wikimedia\Parsoid\Ext\ParsoidExtensionAPI; |
| 16 | use Wikimedia\Parsoid\NodeData\DataParsoid; |
| 17 | use Wikimedia\Parsoid\NodeData\TempData; |
| 18 | use Wikimedia\Parsoid\NodeData\TemplateInfo; |
| 19 | use Wikimedia\Parsoid\Utils\DiffDOMUtils; |
| 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\Timing; |
| 25 | use Wikimedia\Parsoid\Utils\Utils; |
| 26 | use Wikimedia\Parsoid\Utils\WTUtils; |
| 27 | use Wikimedia\Parsoid\Wikitext\Consts; |
| 28 | use Wikimedia\Parsoid\Wt2Html\Wt2HtmlDOMProcessor; |
| 29 | |
| 30 | /** |
| 31 | * DOM pass that walks the DOM tree, detects specific wikitext patterns, |
| 32 | * and emits them as linter events. |
| 33 | */ |
| 34 | class Linter implements Wt2HtmlDOMProcessor { |
| 35 | private ?ParsoidExtensionAPI $extApi = null; |
| 36 | private ?string $obsoleteTagsRE = null; |
| 37 | private array $seenIds = []; |
| 38 | |
| 39 | /** @var array<string,bool>|null */ |
| 40 | private ?array $tagsWithChangedMisnestingBehavior = null; |
| 41 | |
| 42 | /** |
| 43 | * We are trying to find HTML5 tags that have different behavior compared to HTML4 |
| 44 | * in some misnesting scenarios around wikitext paragraphs. |
| 45 | * |
| 46 | * Ex: Input: <p><small>a</p><p>b</small></p> |
| 47 | * Tidy output: <p><small>a</small></p><p><small>b</small></p> |
| 48 | * HTML5 output: <p><small>a</small></p><p><small>b</small></p> |
| 49 | * |
| 50 | * So, all good here. |
| 51 | * But, see how output changes when we use <span> instead |
| 52 | * |
| 53 | * Ex: Input: <p><span>a</p><p>b</span></p> |
| 54 | * Tidy output: <p><span>a</span></p><p><span>b</span></p> |
| 55 | * HTML5 output: <p><span>a</span></p><p>b</p> |
| 56 | * |
| 57 | * The source wikitext is "<span>a\n\nb</span>". The difference persists even |
| 58 | * when you have "<span>a\n\n<div>b</div>" or "<span>a\n\n{|\n|x\n|}\nbar". |
| 59 | * |
| 60 | * This is because Tidy seems to be doing the equivalent of HTM5-treebuilder's |
| 61 | * active formatting element reconstruction step on all *inline* elements. |
| 62 | * However, HTML5 parsers only do that on formatting elements. So, we need |
| 63 | * to compute which HTML5 tags are subject to this differential behavior. |
| 64 | * |
| 65 | * We compute that by excluding the following tags from the list of all HTML5 tags |
| 66 | * - If our sanitizer doesn't allow them, they will be escaped => ignore them |
| 67 | * - HTML4 block tags are excluded (obviously) |
| 68 | * - Void tags don't matter since they cannot wrap anything (obviously) |
| 69 | * - Active formatting elements have special handling in the HTML5 tree building |
| 70 | * algorithm where they are reconstructed to wrap all originally intended content. |
| 71 | * (ex: <small> above) |
| 72 | * |
| 73 | * Here is the list of 22 HTML5 tags that are affected: |
| 74 | * ABBR, BDI, BDO, CITE, DATA, DEL, DFN, INS, KBD, MARK, |
| 75 | * Q, RB, RP, RT, RTC, RUBY, SAMP, SPAN, SUB, SUP, TIME, VAR |
| 76 | * |
| 77 | * https://phabricator.wikimedia.org/T176363#3628173 verifies that this list of |
| 78 | * tags all demonstrate this behavior. |
| 79 | * |
| 80 | * @return array |
| 81 | * @phan-return array<string,bool> |
| 82 | */ |
| 83 | private function getTagsWithChangedMisnestingBehavior(): array { |
| 84 | if ( $this->tagsWithChangedMisnestingBehavior === null ) { |
| 85 | // This set is frozen in time. It gets us down to the requisite |
| 86 | // 22 HTML5 tags above, but shouldn't be used for anything other |
| 87 | // than that. |
| 88 | $HTML4TidyBlockTags = PHPUtils::makeSet( [ |
| 89 | 'div', 'p', |
| 90 | # tables |
| 91 | 'table', 'tbody', 'thead', 'tfoot', 'caption', 'th', 'tr', 'td', |
| 92 | # lists |
| 93 | 'ul', 'ol', 'li', 'dl', 'dt', 'dd', |
| 94 | # HTML5 heading content |
| 95 | 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'hgroup', |
| 96 | # HTML5 sectioning content |
| 97 | 'article', 'aside', 'nav', 'section', 'footer', 'header', |
| 98 | 'figure', 'figcaption', 'fieldset', 'details', 'blockquote', |
| 99 | # other |
| 100 | 'hr', 'button', 'canvas', 'center', 'col', 'colgroup', 'embed', |
| 101 | 'map', 'object', 'pre', 'progress', 'video', |
| 102 | ] ); |
| 103 | $this->tagsWithChangedMisnestingBehavior = []; |
| 104 | foreach ( Consts::$HTML['HTML5Tags'] as $tag => $dummy ) { |
| 105 | if ( isset( Consts::$Sanitizer['AllowedLiteralTags'][$tag] ) && |
| 106 | !isset( $HTML4TidyBlockTags[$tag] ) && |
| 107 | !isset( Consts::$HTML['FormattingTags'][$tag] ) && |
| 108 | !isset( Consts::$HTML['VoidTags'][$tag] ) |
| 109 | ) { |
| 110 | $this->tagsWithChangedMisnestingBehavior[$tag] = true; |
| 111 | } |
| 112 | } |
| 113 | } |
| 114 | |
| 115 | return $this->tagsWithChangedMisnestingBehavior; |
| 116 | } |
| 117 | |
| 118 | /** |
| 119 | * Finds a matching node at the "start" of this node. |
| 120 | */ |
| 121 | private function leftMostMisnestedDescendent( ?Node $node, Element $match ): ?Element { |
| 122 | if ( !$node instanceof Element ) { |
| 123 | return null; |
| 124 | } |
| 125 | |
| 126 | if ( DOMUtils::isMarkerMeta( $node, 'mw:Placeholder/StrippedTag' ) ) { |
| 127 | $name = DOMDataUtils::getDataParsoid( $node )->name ?? null; |
| 128 | return $name === DOMCompat::nodeName( $match ) ? $node : null; |
| 129 | } |
| 130 | |
| 131 | if ( DOMCompat::nodeName( $node ) === DOMCompat::nodeName( $match ) ) { |
| 132 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 133 | if ( ( DOMDataUtils::getDataParsoid( $match )->stx ?? null ) === ( $dp->stx ?? null ) && |
| 134 | !empty( $dp->autoInsertedStart ) |
| 135 | ) { |
| 136 | if ( !empty( $dp->autoInsertedEnd ) ) { |
| 137 | return $this->getMatchingMisnestedNode( $node, $match ); |
| 138 | } else { |
| 139 | return $node; |
| 140 | } |
| 141 | } |
| 142 | } |
| 143 | |
| 144 | return $this->leftMostMisnestedDescendent( $node->firstChild, $match ); |
| 145 | } |
| 146 | |
| 147 | /** |
| 148 | * $node has an 'autoInsertedEnd' flag set on it. We are looking for |
| 149 | * its matching node that has an 'autoInsertedStart' flag set on it. |
| 150 | * This happens when the tree-builder fixes up misnested tags. |
| 151 | * This "adjacency" is wrt the HTML string. In a DOM, this can either |
| 152 | * be the next sibling OR, it might be the left-most-descendent of |
| 153 | * of $node's parent's sibling (and so on up the ancestor chain). |
| 154 | */ |
| 155 | private function getMatchingMisnestedNode( Node $node, Element $match ): ?Element { |
| 156 | if ( DOMUtils::atTheTop( $node ) ) { |
| 157 | return null; |
| 158 | } |
| 159 | |
| 160 | if ( DiffDOMUtils::nextNonSepSibling( $node ) ) { |
| 161 | return $this->leftMostMisnestedDescendent( DiffDOMUtils::nextNonSepSibling( $node ), $match ); |
| 162 | } |
| 163 | |
| 164 | return $this->getMatchingMisnestedNode( $node->parentNode, $match ); |
| 165 | } |
| 166 | |
| 167 | /** |
| 168 | * Given a tplInfo object, determine whether we are: |
| 169 | * - Not processing template content (could be extension or top level page) |
| 170 | * - Processing encapsulated content that is produced by a single template. |
| 171 | * If so, return the name of that template. |
| 172 | * - Processing encapsulated content that comes from multiple templates. |
| 173 | * If so, return a flag indicating this. |
| 174 | * |
| 175 | * FIXME: We might potentially be computing this information redundantly |
| 176 | * for every lint we find within this template's content. It could probably |
| 177 | * be cached in tplInfo after it is computed once. |
| 178 | */ |
| 179 | public static function findEnclosingTemplateName( Env $env, ?stdClass $tplInfo ): ?array { |
| 180 | if ( !$tplInfo ) { |
| 181 | return null; |
| 182 | } |
| 183 | |
| 184 | if ( !DOMUtils::hasTypeOf( $tplInfo->first, 'mw:Transclusion' ) ) { |
| 185 | return null; |
| 186 | } |
| 187 | $dmw = DOMDataUtils::getDataMw( $tplInfo->first ); |
| 188 | // This count check is conservative in that link suffixes and prefixes |
| 189 | // could artifically add an extra element to the parts array but we |
| 190 | // don't have a good way of distinguishing that right now. It will require |
| 191 | // a non-string representation for them and a change in spec along with |
| 192 | // a version bump and all that song and dance. If linting accuracy in these |
| 193 | // scenarios become a problem, we can revisit this. |
| 194 | if ( |
| 195 | !empty( $dmw->parts ) && |
| 196 | count( $dmw->parts ) === 1 |
| 197 | ) { |
| 198 | $p0 = $dmw->parts[0]; |
| 199 | if ( !( $p0 instanceof TemplateInfo ) ) { |
| 200 | throw new UnreachableException( |
| 201 | "a single part will always be a TemplateInfo not a string" |
| 202 | ); |
| 203 | } |
| 204 | $name = null; |
| 205 | if ( !empty( $p0->href ) ) { // Could be "function" |
| 206 | // PORT-FIXME: Should that be SiteConfig::relativeLinkPrefix() rather than './'? |
| 207 | $name = PHPUtils::stripPrefix( $p0->href, './' ); |
| 208 | } else { |
| 209 | // type === 'templatearg' or 'template' |
| 210 | $name = trim( $p0->targetWt ); |
| 211 | } |
| 212 | return [ 'name' => $name ]; |
| 213 | } else { |
| 214 | return [ 'multiPartTemplateBlock' => true ]; |
| 215 | } |
| 216 | } |
| 217 | |
| 218 | /** |
| 219 | * Compute the DSR information for the lint object. |
| 220 | * - In the common case, this is simply the DSR value of the node |
| 221 | * that generated the lint. But, occasionally, for some lints, |
| 222 | * we might have to post-process the node's DSR. |
| 223 | * - If the lint is found in template content, then the DSR spans |
| 224 | * the transclusion markup in the toplevel page source. |
| 225 | */ |
| 226 | public static function findLintDSR( |
| 227 | ?array $tplLintInfo, ?stdClass $tplInfo, ?DomSourceRange $nodeDSR, |
| 228 | ?callable $updateNodeDSR = null |
| 229 | ): ?DomSourceRange { |
| 230 | if ( $tplLintInfo !== null || ( $tplInfo && !Utils::isValidDSR( $nodeDSR ) ) ) { |
| 231 | return DOMDataUtils::getDataParsoid( $tplInfo->first )->dsr ?? null; |
| 232 | } else { |
| 233 | return $updateNodeDSR ? $updateNodeDSR( $nodeDSR ) : $nodeDSR; |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | /** |
| 238 | * Determine if a node has an identical nested tag (?) |
| 239 | */ |
| 240 | private function hasIdenticalNestedTag( Element $node, string $name ): bool { |
| 241 | $c = $node->firstChild; |
| 242 | while ( $c ) { |
| 243 | if ( $c instanceof Element ) { |
| 244 | if ( |
| 245 | DOMCompat::nodeName( $c ) === $name && |
| 246 | empty( DOMDataUtils::getDataParsoid( $c )->autoInsertedEnd ) |
| 247 | ) { |
| 248 | return true; |
| 249 | } |
| 250 | |
| 251 | return $this->hasIdenticalNestedTag( $c, $name ); |
| 252 | } |
| 253 | |
| 254 | $c = $c->nextSibling; |
| 255 | } |
| 256 | |
| 257 | return false; |
| 258 | } |
| 259 | |
| 260 | /** |
| 261 | * Determine if a node has misnestable content |
| 262 | */ |
| 263 | private function hasMisnestableContent( Node $node, string $name ): bool { |
| 264 | // For A, TD, TH, H* tags, Tidy doesn't seem to propagate |
| 265 | // the unclosed tag outside these tags. |
| 266 | // No need to check for tr/table since content cannot show up there |
| 267 | if ( DOMUtils::atTheTop( $node ) || preg_match( '/^(?:a|td|th|h\d)$/D', DOMCompat::nodeName( $node ) ) ) { |
| 268 | return false; |
| 269 | } |
| 270 | |
| 271 | $next = DiffDOMUtils::nextNonSepSibling( $node ); |
| 272 | if ( !$next ) { |
| 273 | return $this->hasMisnestableContent( $node->parentNode, $name ); |
| 274 | } |
| 275 | |
| 276 | $contentNode = null; |
| 277 | if ( DOMCompat::nodeName( $next ) === 'p' && !WTUtils::isLiteralHTMLNode( $next ) ) { |
| 278 | $contentNode = DiffDOMUtils::firstNonSepChild( $next ); |
| 279 | } else { |
| 280 | $contentNode = $next; |
| 281 | } |
| 282 | |
| 283 | // If the first "content" node we find is a matching |
| 284 | // stripped tag, we have nothing that can get misnested |
| 285 | return $contentNode && !( |
| 286 | $contentNode instanceof Element && |
| 287 | DOMUtils::isMarkerMeta( $contentNode, 'mw:Placeholder/StrippedTag' ) && |
| 288 | isset( DOMDataUtils::getDataParsoid( $contentNode )->name ) && |
| 289 | DOMDataUtils::getDataParsoid( $contentNode )->name === $name |
| 290 | ); |
| 291 | } |
| 292 | |
| 293 | /** |
| 294 | * Indicate whether an end tag is optional for this node |
| 295 | * |
| 296 | * See https://www.w3.org/TR/html5/syntax.html#optional-tags |
| 297 | * |
| 298 | * End tags for tr/td/th/li are entirely optional since they |
| 299 | * require a parent container and can only be followed by like |
| 300 | * kind. |
| 301 | * |
| 302 | * Caveat: <li>foo</li><ol>..</ol> and <li>foo<ol>..</ol> |
| 303 | * generate different DOM trees, so explicit </li> tag |
| 304 | * is required to specify which of the two was intended. |
| 305 | * |
| 306 | * With that one caveat around nesting, the parse with/without |
| 307 | * the end tag is identical. For now, ignoring that caveat |
| 308 | * since they aren't like to show up in our corpus much. |
| 309 | * |
| 310 | * For the other tags in that w3c spec section, I haven't reasoned |
| 311 | * through when exactly they are optional. Not handling that complexity |
| 312 | * for now since those are likely uncommon use cases in our corpus. |
| 313 | */ |
| 314 | private function endTagOptional( Node $node ): bool { |
| 315 | static $tagNames = [ 'tr', 'td', 'th', 'li' ]; |
| 316 | return in_array( DOMCompat::nodeName( $node ), $tagNames, true ); |
| 317 | } |
| 318 | |
| 319 | /** |
| 320 | * Find the nearest ancestor heading tag |
| 321 | */ |
| 322 | private function getHeadingAncestor( Node $node ): ?Node { |
| 323 | while ( $node && !DOMUtils::isHeading( $node ) ) { |
| 324 | $node = $node->parentNode; |
| 325 | } |
| 326 | return $node; |
| 327 | } |
| 328 | |
| 329 | /** |
| 330 | * For formatting tags, Tidy seems to be doing this "smart" fixup of |
| 331 | * unclosed tags by looking for matching unclosed pairs of identical tags |
| 332 | * and if the content ends in non-whitespace text, it treats the second |
| 333 | * unclosed opening tag as a closing tag. But, a HTML5 parser won't do this. |
| 334 | * So, detect this pattern and flag for linter fixup. |
| 335 | */ |
| 336 | private function matchedOpenTagPairExists( Node $c, DataParsoid $dp ): bool { |
| 337 | $lc = $c->lastChild; |
| 338 | if ( !$lc instanceof Element || DOMCompat::nodeName( $lc ) !== DOMCompat::nodeName( $c ) ) { |
| 339 | return false; |
| 340 | } |
| 341 | |
| 342 | $lcDP = DOMDataUtils::getDataParsoid( $lc ); |
| 343 | if ( empty( $lcDP->autoInsertedEnd ) || ( $lcDP->stx ?? null ) !== ( $dp->stx ?? null ) ) { |
| 344 | return false; |
| 345 | } |
| 346 | |
| 347 | $prev = $lc->previousSibling; |
| 348 | // PORT-FIXME: Do we care about non-ASCII whitespace here? |
| 349 | if ( $prev instanceof Text && !preg_match( '/\s$/D', $prev->nodeValue ) ) { |
| 350 | return true; |
| 351 | } |
| 352 | |
| 353 | return false; |
| 354 | } |
| 355 | |
| 356 | /** |
| 357 | * Lint Treebuilder fixups marked by ProcessTreeBuilderFixups |
| 358 | * |
| 359 | * It handles the following scenarios: |
| 360 | * |
| 361 | * 1. Unclosed end tags (`missing-end-tag`, `missing-end-tag-in-heading`) |
| 362 | * 2. Invalid self-closed tags (`self-closed-tag`) |
| 363 | * 3. Stripped tags (`stripped-tag`) |
| 364 | * |
| 365 | * In addition, we have specialized categories for some patterns |
| 366 | * where we encounter unclosed end tags. |
| 367 | * |
| 368 | * 4. misnested-tag |
| 369 | * 5. html5-misnesting |
| 370 | * 6. multiple-unclosed-formatting-tags |
| 371 | * 7. unclosed-quotes-in-heading |
| 372 | */ |
| 373 | private function lintTreeBuilderFixup( |
| 374 | Env $env, Element $c, DataParsoid $dp, ?stdClass $tplInfo |
| 375 | ): void { |
| 376 | // This might have been processed as part of |
| 377 | // misnested-tag category identification. |
| 378 | if ( $dp->getTempFlag( TempData::LINTED ) ) { |
| 379 | return; |
| 380 | } |
| 381 | |
| 382 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 383 | // During DSR computation, stripped meta tags |
| 384 | // surrender their width to its previous sibling. |
| 385 | // We record the original DSR in the tmp attribute |
| 386 | // for that reason. |
| 387 | $dsr = self::findLintDSR( $tplLintInfo, $tplInfo, $dp->tmp->origDSR ?? $dp->dsr ?? null ); |
| 388 | $lintObj = null; |
| 389 | if ( DOMUtils::isMarkerMeta( $c, 'mw:Placeholder/StrippedTag' ) ) { |
| 390 | $lintObj = [ |
| 391 | 'dsr' => $dsr, |
| 392 | 'templateInfo' => $tplLintInfo, |
| 393 | 'params' => [ 'name' => $dp->name ?? null ], |
| 394 | ]; |
| 395 | $env->recordLint( 'stripped-tag', $lintObj ); |
| 396 | } |
| 397 | |
| 398 | // Dont bother linting for auto-inserted start/end or self-closing-tag if: |
| 399 | // 1. c is a void element |
| 400 | // Void elements won't have auto-inserted start/end tags |
| 401 | // and self-closing versions are valid for them. |
| 402 | // |
| 403 | // 2. c is tbody (FIXME: don't remember why we have this exception) |
| 404 | // |
| 405 | // 3. c is not an HTML element (unless they are i/b quotes or tables) |
| 406 | // |
| 407 | // 4. c doesn't have DSR info and doesn't come from a template either |
| 408 | $cNodeName = DOMCompat::nodeName( $c ); |
| 409 | $ancestor = null; |
| 410 | $isHtmlElement = WTUtils::hasLiteralHTMLMarker( $dp ); |
| 411 | if ( !Utils::isVoidElement( $cNodeName ) && |
| 412 | $cNodeName !== 'tbody' && |
| 413 | ( $isHtmlElement || DOMUtils::isQuoteElt( $c ) || $cNodeName === 'table' ) && |
| 414 | ( $tplInfo !== null || $dsr !== null ) |
| 415 | ) { |
| 416 | if ( !empty( $dp->selfClose ) && $cNodeName !== 'meta' ) { |
| 417 | $lintObj = [ |
| 418 | 'dsr' => $dsr, |
| 419 | 'templateInfo' => $tplLintInfo, |
| 420 | 'params' => [ 'name' => $cNodeName ], |
| 421 | ]; |
| 422 | $env->recordLint( 'self-closed-tag', $lintObj ); |
| 423 | // The other checks won't pass - no need to test them. |
| 424 | return; |
| 425 | } |
| 426 | |
| 427 | if ( |
| 428 | ( $dp->autoInsertedEnd ?? null ) === true && |
| 429 | ( $tplInfo || ( $dsr->openWidth ?? 0 ) > 0 ) |
| 430 | ) { |
| 431 | $lintObj = [ |
| 432 | 'dsr' => $dsr, |
| 433 | 'templateInfo' => $tplLintInfo, |
| 434 | 'params' => [ 'name' => $cNodeName ], |
| 435 | ]; |
| 436 | |
| 437 | // FIXME: This literal html marker check is strictly not required |
| 438 | // (a) we've already checked that above and know that isQuoteElt is |
| 439 | // not one of our tags. |
| 440 | // (b) none of the tags in the list have native wikitext syntax => |
| 441 | // they will show up as literal html tags. |
| 442 | // But, in the interest of long-term maintenance in the face of |
| 443 | // changes (to wikitext or html specs), let us make it explicit. |
| 444 | if ( $isHtmlElement && |
| 445 | isset( $this->getTagsWithChangedMisnestingBehavior()[DOMCompat::nodeName( $c )] ) && |
| 446 | $this->hasMisnestableContent( $c, DOMCompat::nodeName( $c ) ) && |
| 447 | // Tidy WTF moment here! |
| 448 | // I don't know why Tidy does something very different |
| 449 | // when there is an identical nested tag here. |
| 450 | // |
| 451 | // <p><span id='1'>a<span>X</span></p><p>b</span></p> |
| 452 | // vs. |
| 453 | // <p><span id='1'>a</p><p>b</span></p> OR |
| 454 | // <p><span id='1'>a<del>X</del></p><p>b</span></p> |
| 455 | // |
| 456 | // For the first snippet, Tidy only wraps "a" with the id='1' span |
| 457 | // For the second and third snippets, Tidy wraps "b" with the id='1' span as well. |
| 458 | // |
| 459 | // For the corresponding wikitext that generates the above token stream, |
| 460 | // Parsoid (and Remex) won't wrap 'b' with the id=1' span at all. |
| 461 | !$this->hasIdenticalNestedTag( $c, DOMCompat::nodeName( $c ) ) |
| 462 | ) { |
| 463 | $env->recordLint( 'html5-misnesting', $lintObj ); |
| 464 | } elseif ( |
| 465 | !$isHtmlElement && DOMUtils::isQuoteElt( $c ) && |
| 466 | // phpcs:ignore Generic.CodeAnalysis.AssignmentInCondition.Found |
| 467 | ( $ancestor = $this->getHeadingAncestor( $c->parentNode ) ) |
| 468 | ) { |
| 469 | $lintObj['params']['ancestorName'] = DOMCompat::nodeName( $ancestor ); |
| 470 | $env->recordLint( 'unclosed-quotes-in-heading', $lintObj ); |
| 471 | } else { |
| 472 | $adjNode = $this->getMatchingMisnestedNode( $c, $c ); |
| 473 | if ( $adjNode ) { |
| 474 | $adjDp = DOMDataUtils::getDataParsoid( $adjNode ); |
| 475 | $adjDp->setTempFlag( TempData::LINTED ); |
| 476 | $env->recordLint( 'misnested-tag', $lintObj ); |
| 477 | } elseif ( !$this->endTagOptional( $c ) && empty( $dp->autoInsertedStart ) ) { |
| 478 | $lintObj['params']['inTable'] = DOMUtils::hasNameOrHasAncestorOfName( $c, 'table' ); |
| 479 | $category = $this->getHeadingAncestor( $c ) ? |
| 480 | 'missing-end-tag-in-heading' : 'missing-end-tag'; |
| 481 | $next = DiffDOMUtils::nextNonSepSibling( $c ); |
| 482 | if ( |
| 483 | // Skip if covered by deletable-table-tag |
| 484 | !( $cNodeName === 'table' && $next && |
| 485 | ( DOMCompat::nodeName( $c ) === 'table' ) ) |
| 486 | ) { |
| 487 | $env->recordLint( $category, $lintObj ); |
| 488 | } |
| 489 | if ( isset( Consts::$HTML['FormattingTags'][DOMCompat::nodeName( $c )] ) && |
| 490 | $this->matchedOpenTagPairExists( $c, $dp ) |
| 491 | ) { |
| 492 | $env->recordLint( 'multiple-unclosed-formatting-tags', $lintObj ); |
| 493 | } |
| 494 | } |
| 495 | } |
| 496 | } |
| 497 | } |
| 498 | } |
| 499 | |
| 500 | /** |
| 501 | * Lint fostered content marked by MarkFosteredContent. |
| 502 | * |
| 503 | * Lint category: `fostered`, `fostered-transparent` |
| 504 | * |
| 505 | * This will log cases like: |
| 506 | * |
| 507 | * {| |
| 508 | * foo |
| 509 | * |- |
| 510 | * | bar |
| 511 | * |} |
| 512 | * |
| 513 | * Here 'foo' gets fostered out. |
| 514 | */ |
| 515 | private function lintFostered( |
| 516 | Env $env, Element $node, DataParsoid $dp, ?stdClass $tplInfo |
| 517 | ): void { |
| 518 | if ( DOMCompat::nodeName( $node ) !== 'table' ) { |
| 519 | return; |
| 520 | } |
| 521 | |
| 522 | // The top-level nodes in the foster box are span/p wrapped |
| 523 | // and so, if we have fostered content, previous siblings to |
| 524 | // the table are expected to be elements. |
| 525 | $maybeFostered = $node->previousSibling; |
| 526 | |
| 527 | // Emit "fostered" or "fostered-transparent" depending on if the fostered |
| 528 | // content is entirely transparent or not. |
| 529 | // |
| 530 | // We're trying to find a balance between creating noise for wikignomes |
| 531 | // and avoiding dirty-diffs from DiscussionTools. DiscussionTools |
| 532 | // expects to know when pages have fostered content otherwise it can |
| 533 | // lead to corruption on edit. However, rendering transparent nodes |
| 534 | // often end up in fosterable positions, like category links from |
| 535 | // templates or include directives on template pages. |
| 536 | |
| 537 | $fosteredRenderingTransparent = false; |
| 538 | while ( |
| 539 | $maybeFostered instanceof Element && |
| 540 | !empty( DOMDataUtils::getDataParsoid( $maybeFostered )->fostered ) && |
| 541 | ( WTUtils::isRenderingTransparentNode( $maybeFostered ) || |
| 542 | // TODO: Section tags are rendering transparent but not sol transparent, |
| 543 | // and that method only considers WTUtils::isSolTransparentLink, though |
| 544 | // there is a FIXME to consider all link nodes. |
| 545 | ( DOMCompat::nodeName( $maybeFostered ) === 'link' && |
| 546 | DOMUtils::hasTypeOf( $maybeFostered, 'mw:Extension/section' ) ) ) |
| 547 | ) { |
| 548 | // Skip rendering-transparent nodes if they come from a template, |
| 549 | // since they'll roundtrip cleanly regardless |
| 550 | $fosteredRenderingTransparent = $fosteredRenderingTransparent || !$tplInfo; |
| 551 | |
| 552 | $maybeFostered = $maybeFostered->previousSibling; |
| 553 | } |
| 554 | |
| 555 | if ( |
| 556 | $maybeFostered instanceof Element && |
| 557 | !empty( DOMDataUtils::getDataParsoid( $maybeFostered )->fostered ) |
| 558 | ) { |
| 559 | $type = 'fostered'; |
| 560 | } elseif ( $fosteredRenderingTransparent ) { |
| 561 | $type = 'fostered-transparent'; |
| 562 | } else { |
| 563 | return; |
| 564 | } |
| 565 | |
| 566 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 567 | $lintObj = [ |
| 568 | 'dsr' => self::findLintDSR( |
| 569 | $tplLintInfo, $tplInfo, $dp->dsr ?? null |
| 570 | ), |
| 571 | 'templateInfo' => $tplLintInfo, |
| 572 | ]; |
| 573 | $env->recordLint( $type, $lintObj ); |
| 574 | } |
| 575 | |
| 576 | /** |
| 577 | * Lint obsolete HTML tags. |
| 578 | * |
| 579 | * Lint category: `obsolete-tag`, `tidy-font-bug` |
| 580 | */ |
| 581 | private function lintObsoleteTag( |
| 582 | Env $env, Element $c, DataParsoid $dp, ?stdClass $tplInfo |
| 583 | ): void { |
| 584 | if ( !$this->obsoleteTagsRE ) { |
| 585 | $elts = []; |
| 586 | foreach ( Consts::$HTML['OlderHTMLTags'] as $tag => $dummy ) { |
| 587 | // Looks like all existing editors let editors add the <big> tag. |
| 588 | // VE has a button to add <big>, it seems so does the WikiEditor |
| 589 | // and JS wikitext editor. So, don't flag BIG as an obsolete tag. |
| 590 | if ( $tag !== 'big' ) { |
| 591 | $elts[] = preg_quote( $tag, '/' ); |
| 592 | } |
| 593 | } |
| 594 | $this->obsoleteTagsRE = '/^(?:' . implode( '|', $elts ) . ')$/D'; |
| 595 | } |
| 596 | |
| 597 | $tplLintInfo = null; |
| 598 | if ( ( empty( $dp->autoInsertedStart ) || empty( $dp->autoInsertedEnd ) ) && |
| 599 | preg_match( $this->obsoleteTagsRE, DOMCompat::nodeName( $c ) ) |
| 600 | ) { |
| 601 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 602 | $lintObj = [ |
| 603 | 'dsr' => self::findLintDSR( $tplLintInfo, $tplInfo, $dp->dsr ?? null ), |
| 604 | 'templateInfo' => $tplLintInfo, |
| 605 | 'params' => [ 'name' => DOMCompat::nodeName( $c ) ], |
| 606 | ]; |
| 607 | $env->recordLint( 'obsolete-tag', $lintObj ); |
| 608 | } |
| 609 | |
| 610 | if ( DOMCompat::nodeName( $c ) === 'font' && $c->hasAttribute( 'color' ) ) { |
| 611 | /* ---------------------------------------------------------- |
| 612 | * Tidy migrates <font> into the link in these cases |
| 613 | * <font>[[Foo]]</font> |
| 614 | * <font>[[Foo]]l</font> (link-trail) |
| 615 | * <font><!--boo-->[[Foo]]</font> |
| 616 | * <font>__NOTOC__[[Foo]]</font> |
| 617 | * <font>[[Category:Foo]][[Foo]]</font> |
| 618 | * <font>{{1x|[[Foo]]}}</font> |
| 619 | * |
| 620 | * Tidy does not migrate <font> into the link in these cases |
| 621 | * <font> [[Foo]]</font> |
| 622 | * <font>[[Foo]] </font> |
| 623 | * <font>[[Foo]]L</font> (not a link-trail) |
| 624 | * <font>[[Foo]][[Bar]]</font> |
| 625 | * <font>[[Foo]][[Bar]]</font> |
| 626 | * |
| 627 | * <font> is special. |
| 628 | * This behavior is not seen with other formatting tags. |
| 629 | * |
| 630 | * Remex/parsoid won't do any of this. |
| 631 | * This difference in behavior only matters when the font tag |
| 632 | * specifies a link colour because the link no longer renders |
| 633 | * as blue/red but in the font-specified colour. |
| 634 | * ---------------------------------------------------------- */ |
| 635 | $tidyFontBug = $c->firstChild !== null; |
| 636 | $haveLink = false; |
| 637 | for ( $n = $c->firstChild; $n; $n = $n->nextSibling ) { |
| 638 | $nodeName = DOMCompat::nodeName( $n ); |
| 639 | if ( $nodeName !== 'a' && |
| 640 | !WTUtils::isRenderingTransparentNode( $n ) && |
| 641 | !WTUtils::isTplMarkerMeta( $n ) |
| 642 | ) { |
| 643 | $tidyFontBug = false; |
| 644 | break; |
| 645 | } |
| 646 | |
| 647 | if ( $nodeName === 'a' || $nodeName === 'figure' ) { |
| 648 | if ( !$haveLink ) { |
| 649 | $haveLink = true; |
| 650 | } else { |
| 651 | $tidyFontBug = false; |
| 652 | break; |
| 653 | } |
| 654 | } |
| 655 | } |
| 656 | |
| 657 | if ( $tidyFontBug ) { |
| 658 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 659 | $env->recordLint( 'tidy-font-bug', [ |
| 660 | 'dsr' => self::findLintDSR( $tplLintInfo, $tplInfo, $dp->dsr ?? null ), |
| 661 | 'templateInfo' => $tplLintInfo, |
| 662 | 'params' => [ 'name' => 'font' ] |
| 663 | ] ); |
| 664 | } |
| 665 | } |
| 666 | } |
| 667 | |
| 668 | /** |
| 669 | * Log bogus (=unrecognized) media options. |
| 670 | * |
| 671 | * See - https://www.mediawiki.org/wiki/Help:Images#Syntax |
| 672 | * |
| 673 | * Lint category: `bogus-image-options` |
| 674 | */ |
| 675 | private function lintBogusImageOptions( |
| 676 | Env $env, Node $c, DataParsoid $dp, ?stdClass $tplInfo |
| 677 | ): void { |
| 678 | // Despite the lint category name, this checks all media, not just images |
| 679 | if ( WTUtils::isGeneratedFigure( $c ) && !empty( $dp->optList ) ) { |
| 680 | $items = []; |
| 681 | $bogusPx = $dp->getTempFlag( TempData::BOGUS_PX ); |
| 682 | foreach ( $dp->optList as $item ) { |
| 683 | if ( |
| 684 | $item['ck'] === 'bogus' || |
| 685 | ( $bogusPx && $item['ck'] === 'width' ) |
| 686 | ) { |
| 687 | $items[] = $item['ak']; |
| 688 | } |
| 689 | } |
| 690 | if ( $items ) { |
| 691 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 692 | $env->recordLint( 'bogus-image-options', [ |
| 693 | 'dsr' => self::findLintDSR( $tplLintInfo, $tplInfo, $dp->dsr ?? null ), |
| 694 | 'templateInfo' => $tplLintInfo, |
| 695 | 'params' => [ 'items' => $items ] |
| 696 | ] ); |
| 697 | } |
| 698 | } |
| 699 | } |
| 700 | |
| 701 | /** |
| 702 | * Lint tables Tidy deletes. |
| 703 | * |
| 704 | * Lint category: `deletable-table-tag` |
| 705 | * |
| 706 | * In this example below, the second table is in a fosterable position |
| 707 | * (inside a <tr>). The tree builder closes the first table at that point |
| 708 | * and starts a new table there. We are detecting this pattern because |
| 709 | * Tidy does something very different here. It strips the inner table |
| 710 | * and retains the outer table. So, for preserving rendering of pages |
| 711 | * that are tailored for Tidy, editors have to fix up this wikitext |
| 712 | * to strip the inner table (to mimic what Tidy does). |
| 713 | * |
| 714 | * {| style='border:1px solid red;' |
| 715 | * |a |
| 716 | * |- |
| 717 | * {| style='border:1px solid blue;' |
| 718 | * |b |
| 719 | * |c |
| 720 | * |} |
| 721 | * |} |
| 722 | */ |
| 723 | private function lintDeletableTableTag( |
| 724 | Env $env, Node $c, DataParsoid $dp, ?stdClass $tplInfo |
| 725 | ): void { |
| 726 | if ( DOMCompat::nodeName( $c ) === 'table' ) { |
| 727 | $prev = DiffDOMUtils::previousNonSepSibling( $c ); |
| 728 | if ( $prev instanceof Element && DOMCompat::nodeName( $prev ) === 'table' && |
| 729 | !empty( DOMDataUtils::getDataParsoid( $prev )->autoInsertedEnd ) |
| 730 | ) { |
| 731 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 732 | $dsr = self::findLintDSR( |
| 733 | $tplLintInfo, |
| 734 | $tplInfo, |
| 735 | $dp->dsr ?? null, |
| 736 | static function ( ?DomSourceRange $nodeDSR ): ?DomSourceRange { |
| 737 | // Identify the dsr-span of the opening tag |
| 738 | // of the table that needs to be deleted |
| 739 | $x = $nodeDSR === null ? null : ( clone $nodeDSR ); |
| 740 | if ( !empty( $x->openWidth ) ) { |
| 741 | $x->end = $x->innerStart(); |
| 742 | $x->openWidth = 0; |
| 743 | $x->closeWidth = 0; |
| 744 | } |
| 745 | return $x; |
| 746 | } |
| 747 | ); |
| 748 | $lintObj = [ |
| 749 | 'dsr' => $dsr, |
| 750 | 'templateInfo' => $tplLintInfo, |
| 751 | 'params' => [ 'name' => 'table' ], |
| 752 | ]; |
| 753 | $env->recordLint( 'deletable-table-tag', $lintObj ); |
| 754 | } |
| 755 | } |
| 756 | } |
| 757 | |
| 758 | /** |
| 759 | * Find the first child passing the filter. |
| 760 | */ |
| 761 | private function findMatchingChild( Node $node, callable $filter ): ?Node { |
| 762 | $c = $node->firstChild; |
| 763 | while ( $c && !$filter( $c ) ) { |
| 764 | $c = $c->nextSibling; |
| 765 | } |
| 766 | |
| 767 | return $c; |
| 768 | } |
| 769 | |
| 770 | /** |
| 771 | * Test if the node has a 'nowrap' CSS rule |
| 772 | * |
| 773 | * In the general case, this CSS can come from a class, |
| 774 | * or from a <style> tag or a stylesheet or even from JS code. |
| 775 | * But, for now, we are restricting this inspection to inline CSS |
| 776 | * since the intent is to aid editors in fixing patterns that |
| 777 | * can be automatically detected. |
| 778 | * |
| 779 | * Special case for enwiki that has Template:nowrap which |
| 780 | * assigns class='nowrap' with CSS white-space:nowrap in |
| 781 | * MediaWiki:Common.css |
| 782 | */ |
| 783 | private function hasNoWrapCSS( Node $node ): bool { |
| 784 | return $node instanceof Element && ( |
| 785 | str_contains( DOMCompat::getAttribute( $node, 'style' ) ?? '', 'nowrap' ) || |
| 786 | DOMUtils::hasClass( $node, 'nowrap' ) |
| 787 | ); |
| 788 | } |
| 789 | |
| 790 | /** |
| 791 | * Lint bad P wrapping. |
| 792 | * |
| 793 | * Lint category: `pwrap-bug-workaround` |
| 794 | */ |
| 795 | private function lintPWrapBugWorkaround( |
| 796 | Env $env, Element $node, DataParsoid $dp, ?stdClass $tplInfo |
| 797 | ): void { |
| 798 | if ( |
| 799 | !DOMUtils::isWikitextBlockNode( $node ) && |
| 800 | DOMUtils::isWikitextBlockNode( $node->parentNode ) && |
| 801 | $this->hasNoWrapCSS( $node ) |
| 802 | ) { |
| 803 | $p = $this->findMatchingChild( $node, static function ( $e ) { |
| 804 | return DOMCompat::nodeName( $e ) === 'p'; |
| 805 | } ); |
| 806 | if ( $p ) { |
| 807 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 808 | $lintObj = [ |
| 809 | 'dsr' => self::findLintDSR( $tplLintInfo, $tplInfo, $dp->dsr ?? null ), |
| 810 | 'templateInfo' => $tplLintInfo, |
| 811 | 'params' => [ |
| 812 | 'root' => DOMCompat::nodeName( $node->parentNode ), |
| 813 | 'child' => DOMCompat::nodeName( $node ), |
| 814 | ] |
| 815 | ]; |
| 816 | $env->recordLint( 'pwrap-bug-workaround', $lintObj ); |
| 817 | } |
| 818 | } |
| 819 | } |
| 820 | |
| 821 | /** |
| 822 | * Lint Tidy div span flip. |
| 823 | * |
| 824 | * Lint category: `misc-tidy-replacement-issues` |
| 825 | */ |
| 826 | private function lintMiscTidyReplacementIssues( |
| 827 | Env $env, Element $node, DataParsoid $dp, ?stdClass $tplInfo |
| 828 | ): void { |
| 829 | if ( DOMCompat::nodeName( $node ) !== 'span' ) { |
| 830 | return; |
| 831 | } |
| 832 | |
| 833 | $fc = DiffDOMUtils::firstNonSepChild( $node ); |
| 834 | if ( !$fc instanceof Element || DOMCompat::nodeName( $fc ) !== 'div' ) { |
| 835 | return; |
| 836 | } |
| 837 | |
| 838 | // No style/class attributes -- so, this won't affect rendering |
| 839 | if ( !$node->hasAttribute( 'class' ) && !$node->hasAttribute( 'style' ) && |
| 840 | !$fc->hasAttribute( 'class' ) && !$fc->hasAttribute( 'style' ) |
| 841 | ) { |
| 842 | return; |
| 843 | } |
| 844 | |
| 845 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 846 | $lintObj = [ |
| 847 | 'dsr' => self::findLintDSR( $tplLintInfo, $tplInfo, $dp->dsr ?? null ), |
| 848 | 'templateInfo' => $tplLintInfo, |
| 849 | 'params' => [ 'subtype' => 'div-span-flip' ] |
| 850 | ]; |
| 851 | $env->recordLint( 'misc-tidy-replacement-issues', $lintObj ); |
| 852 | } |
| 853 | |
| 854 | /** |
| 855 | * Lint tidy whitespace bug. |
| 856 | * |
| 857 | * Lint category: `tidy-whitespace-bug` |
| 858 | */ |
| 859 | private function lintTidyWhitespaceBug( |
| 860 | Env $env, Node $node, DataParsoid $dp, ?stdClass $tplInfo |
| 861 | ): void { |
| 862 | // We handle a run of nodes in one shot. |
| 863 | // No need to reprocess repeatedly. |
| 864 | if ( $dp->getTempFlag( TempData::PROCESSED_TIDY_WS_BUG ) ) { |
| 865 | return; |
| 866 | } |
| 867 | |
| 868 | // Find the longest run of nodes that are affected by white-space:nowrap CSS |
| 869 | // in a way that leads to unsightly rendering in HTML5 compliant browsers. |
| 870 | // |
| 871 | // Check if Tidy does buggy whitespace hoisting there to provide the browser |
| 872 | // opportunities to split the content in short segments. |
| 873 | // |
| 874 | // If so, editors would need to edit this run of nodes to introduce |
| 875 | // whitespace breaks as necessary so that HTML5 browsers get that |
| 876 | // same opportunity when Tidy is removed. |
| 877 | $s = null; |
| 878 | $nowrapNodes = []; |
| 879 | '@phan-var array<array{node:Node,tidybug:bool,hasLeadingWS:bool}> $nowrapNodes'; |
| 880 | $startNode = $node; |
| 881 | $haveTidyBug = false; |
| 882 | $runLength = 0; |
| 883 | |
| 884 | // <br>, <wbr>, <hr> break a line |
| 885 | while ( $node && !DOMUtils::isRemexBlockNode( $node ) && |
| 886 | !in_array( DOMCompat::nodeName( $node ), [ 'hr', 'br', 'wbr' ], true ) |
| 887 | ) { |
| 888 | if ( $node instanceof Text || !$this->hasNoWrapCSS( $node ) ) { |
| 889 | // No CSS property that affects whitespace. |
| 890 | $s = $node->textContent; |
| 891 | if ( preg_match( '/^(\S*)\s/', $s, $m ) ) { // PORT-FIXME: non-ASCII whitespace? |
| 892 | $runLength += strlen( $m[1] ); |
| 893 | $nowrapNodes[] = [ |
| 894 | 'node' => $node, |
| 895 | 'tidybug' => false, |
| 896 | 'hasLeadingWS' => ( preg_match( '/^\s/', $s ) === 1 ), // PORT-FIXME: non-ASCII whitespace? |
| 897 | ]; |
| 898 | break; |
| 899 | } else { |
| 900 | $nowrapNodes[] = [ 'node' => $node, 'tidybug' => false ]; |
| 901 | $runLength += strlen( $s ); |
| 902 | } |
| 903 | } else { |
| 904 | // Find last non-comment child of node |
| 905 | $last = $node->lastChild; |
| 906 | while ( $last instanceof Comment ) { |
| 907 | $last = $last->previousSibling; |
| 908 | } |
| 909 | |
| 910 | $bug = false; |
| 911 | if ( $last instanceof Text && |
| 912 | preg_match( '/\s$/D', $last->nodeValue ) // PORT-FIXME: non-ASCII whitespace? |
| 913 | ) { |
| 914 | // In this scenario, when Tidy hoists the whitespace to |
| 915 | // after the node, that whitespace is not subject to the |
| 916 | // nowrap CSS => browsers can break content there. |
| 917 | // |
| 918 | // But, non-Tidy libraries won't hoist the whitespace. |
| 919 | // So, browsers don't have a place to break content. |
| 920 | $bug = true; |
| 921 | $haveTidyBug = true; |
| 922 | } |
| 923 | |
| 924 | $nowrapNodes[] = [ 'node' => $node, 'tidybug' => $bug ]; |
| 925 | $runLength += strlen( $node->textContent ); |
| 926 | } |
| 927 | |
| 928 | // Don't cross template boundaries at the top-level |
| 929 | if ( $tplInfo && $tplInfo->last === $node ) { |
| 930 | // Exiting a top-level template |
| 931 | break; |
| 932 | } elseif ( !$tplInfo && WTUtils::findFirstEncapsulationWrapperNode( $node ) ) { |
| 933 | // Entering a top-level template |
| 934 | break; |
| 935 | } |
| 936 | |
| 937 | // Move to the next non-comment sibling |
| 938 | $node = $node->nextSibling; |
| 939 | while ( $node instanceof Comment ) { |
| 940 | $node = $node->nextSibling; |
| 941 | } |
| 942 | } |
| 943 | |
| 944 | $markProcessedNodes = static function () use ( &$nowrapNodes ) { // Helper |
| 945 | foreach ( $nowrapNodes as $o ) { |
| 946 | // Phan fails at applying the instanceof type restriction to the array member when analyzing the |
| 947 | // following call, but is fine when it's copied to a local variable. |
| 948 | $node = $o['node']; |
| 949 | if ( $node instanceof Element ) { |
| 950 | DOMDataUtils::getDataParsoid( $node )->setTempFlag( TempData::PROCESSED_TIDY_WS_BUG ); |
| 951 | } |
| 952 | } |
| 953 | }; |
| 954 | |
| 955 | if ( !$haveTidyBug ) { |
| 956 | // Mark processed nodes and bail |
| 957 | $markProcessedNodes(); |
| 958 | return; |
| 959 | } |
| 960 | |
| 961 | // Find run before startNode that doesn't have a whitespace break |
| 962 | $prev = $startNode->previousSibling; |
| 963 | while ( $prev && !DOMUtils::isRemexBlockNode( $prev ) ) { |
| 964 | if ( !( $prev instanceof Comment ) ) { |
| 965 | $s = $prev->textContent; |
| 966 | // Find the last \s in the string |
| 967 | if ( preg_match( '/\s(\S*)$/D', $s, $m ) ) { // PORT-FIXME: non-ASCII whitespace here? |
| 968 | $runLength += strlen( $m[1] ); |
| 969 | break; |
| 970 | } else { |
| 971 | $runLength += strlen( $s ); |
| 972 | } |
| 973 | } |
| 974 | $prev = $prev->previousSibling; |
| 975 | } |
| 976 | |
| 977 | $lintConfig = $env->getLinterConfig(); |
| 978 | $tidyWhitespaceBugMaxLength = $lintConfig['tidyWhitespaceBugMaxLength'] ?? 100; |
| 979 | |
| 980 | if ( $runLength < $tidyWhitespaceBugMaxLength ) { |
| 981 | // Mark processed nodes and bail |
| 982 | $markProcessedNodes(); |
| 983 | return; |
| 984 | } |
| 985 | |
| 986 | // For every node where Tidy hoists whitespace, |
| 987 | // emit an event to flag a whitespace fixup opportunity. |
| 988 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 989 | $n = count( $nowrapNodes ) - 1; |
| 990 | foreach ( $nowrapNodes as $i => $o ) { |
| 991 | if ( $o['tidybug'] && $i < $n && empty( $nowrapNodes[$i + 1]['hasLeadingWS'] ) ) { |
| 992 | $nowrapNode = $o['node']; // (see above) |
| 993 | $lintObj = [ |
| 994 | 'dsr' => self::findLintDSR( |
| 995 | $tplLintInfo, |
| 996 | $tplInfo, |
| 997 | $nowrapNode instanceof Element |
| 998 | ? DOMDataUtils::getDataParsoid( $nowrapNode )->dsr ?? null |
| 999 | : null |
| 1000 | ), |
| 1001 | 'templateInfo' => $tplLintInfo, |
| 1002 | 'params' => [ |
| 1003 | 'node' => DOMCompat::nodeName( $o['node'] ), |
| 1004 | 'sibling' => DOMCompat::nodeName( $o['node']->nextSibling ) |
| 1005 | ] |
| 1006 | ]; |
| 1007 | |
| 1008 | $env->recordLint( 'tidy-whitespace-bug', $lintObj ); |
| 1009 | } |
| 1010 | } |
| 1011 | |
| 1012 | $markProcessedNodes(); |
| 1013 | } |
| 1014 | |
| 1015 | /** |
| 1016 | * Detect multiple-unclosed-formatting-tags errors. |
| 1017 | * |
| 1018 | * Since unclosed <small> and <big> tags accumulate their effects |
| 1019 | * in HTML5 parsers (unlike in Tidy where it seems to suppress |
| 1020 | * multiple unclosed elements of the same name), such pages will |
| 1021 | * break pretty spectacularly with Remex. |
| 1022 | * |
| 1023 | * Ex: https://it.wikipedia.org/wiki/Hubert_H._Humphrey_Metrodome?oldid=93017491#Note |
| 1024 | * |
| 1025 | * Lint category: `multiple-unclosed-formatting-tags` |
| 1026 | */ |
| 1027 | private function lintMultipleUnclosedFormattingTags( array $lints, Env $env ): void { |
| 1028 | $firstUnclosedTag = [ |
| 1029 | 'small' => null, |
| 1030 | 'big' => null |
| 1031 | ]; |
| 1032 | $multiUnclosedTagName = null; |
| 1033 | foreach ( $lints as $item ) { |
| 1034 | // Unclosed tags in tables don't leak out of the table |
| 1035 | if ( $item['type'] === 'missing-end-tag' && !$item['params']['inTable'] ) { |
| 1036 | if ( $item['params']['name'] === 'small' || $item['params']['name'] === 'big' ) { |
| 1037 | $tagName = $item['params']['name']; |
| 1038 | // @phan-suppress-next-line PhanPossiblyUndeclaredVariable |
| 1039 | if ( !$firstUnclosedTag[$tagName] ) { |
| 1040 | $firstUnclosedTag[$tagName] = $item; |
| 1041 | } else { |
| 1042 | $multiUnclosedTagName = $tagName; |
| 1043 | break; |
| 1044 | } |
| 1045 | } |
| 1046 | } |
| 1047 | } |
| 1048 | |
| 1049 | if ( $multiUnclosedTagName ) { |
| 1050 | $item = $firstUnclosedTag[$multiUnclosedTagName]; |
| 1051 | if ( isset( $item['dsr'] ) ) { |
| 1052 | $item['dsr'] = DomSourceRange::newFromJsonArray( $item['dsr'] ); |
| 1053 | } |
| 1054 | $env->recordLint( 'multiple-unclosed-formatting-tags', [ |
| 1055 | 'params' => $item['params'], |
| 1056 | 'dsr' => $item['dsr'], |
| 1057 | 'templateInfo' => $item['templateInfo'], |
| 1058 | ] ); |
| 1059 | } |
| 1060 | } |
| 1061 | |
| 1062 | /** |
| 1063 | * Post-process an array of lints |
| 1064 | */ |
| 1065 | private function postProcessLints( array $lints, Env $env ): void { |
| 1066 | $this->lintMultipleUnclosedFormattingTags( $lints, $env ); |
| 1067 | } |
| 1068 | |
| 1069 | /** |
| 1070 | * Get wikitext list item ancestor |
| 1071 | */ |
| 1072 | private function getWikitextListItemAncestor( ?Node $node ): ?Node { |
| 1073 | while ( $node && !DOMUtils::isListItem( $node ) ) { |
| 1074 | $node = $node->parentNode; |
| 1075 | } |
| 1076 | |
| 1077 | if ( $node && !WTUtils::isLiteralHTMLNode( $node ) && |
| 1078 | !WTUtils::fromExtensionContent( $node, 'references' ) |
| 1079 | ) { |
| 1080 | return $node; |
| 1081 | } |
| 1082 | |
| 1083 | return null; |
| 1084 | } |
| 1085 | |
| 1086 | /** |
| 1087 | * Lint a PHP parser bug. |
| 1088 | * |
| 1089 | * When an HTML table is nested inside a list, if any part of the table |
| 1090 | * is on a new line, the PHP parser misnests the list and the table. |
| 1091 | * Tidy fixes the misnesting one way (puts table inside/outside the list) |
| 1092 | * HTML5 parser fixes it another way (list expands to rest of the page!) |
| 1093 | * |
| 1094 | * Lint category: `multiline-html-table-in-list` |
| 1095 | */ |
| 1096 | private function lintMultilineHtmlTableInList( |
| 1097 | Env $env, Element $node, DataParsoid $dp, ?stdClass $tplInfo |
| 1098 | ): void { |
| 1099 | $li = null; |
| 1100 | if ( !WTUtils::isLiteralHTMLNode( $node ) || |
| 1101 | DOMCompat::nodeName( $node ) !== 'table' || |
| 1102 | // phpcs:ignore Generic.CodeAnalysis.AssignmentInCondition.Found |
| 1103 | !( $li = $this->getWikitextListItemAncestor( $node ) ) || |
| 1104 | !str_contains( DOMCompat::getOuterHTML( $node ), "\n" ) |
| 1105 | ) { |
| 1106 | return; |
| 1107 | } |
| 1108 | |
| 1109 | // We have an HTML table nested inside a list |
| 1110 | // that has a newline break in its outer HTML |
| 1111 | // => we are in trouble with the PHP Parser + Remex combo |
| 1112 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 1113 | $lintObj = [ |
| 1114 | 'dsr' => self::findLintDSR( $tplLintInfo, $tplInfo, $dp->dsr ?? null ), |
| 1115 | 'templateInfo' => $tplLintInfo, |
| 1116 | 'params' => [ |
| 1117 | 'name' => 'table', |
| 1118 | 'ancestorName' => DOMCompat::nodeName( $li ), |
| 1119 | ], |
| 1120 | ]; |
| 1121 | $env->recordLint( 'multiline-html-table-in-list', $lintObj ); |
| 1122 | } |
| 1123 | |
| 1124 | /** |
| 1125 | * Log wikilinks or media in external links. |
| 1126 | * |
| 1127 | * HTML tags can be nested but this is not the case for <a> tags |
| 1128 | * which when nested outputs the <a> tags adjacent to each other |
| 1129 | * In the example below, [[Google]] is a wikilink that is nested |
| 1130 | * in the outer external link |
| 1131 | * [http://google.com This is [[Google]]'s search page] |
| 1132 | * |
| 1133 | * Linter category: `wikilink-in-extlink` |
| 1134 | */ |
| 1135 | private function lintWikilinksInExtlink( |
| 1136 | Env $env, Element $node, DataParsoid $dp, ?stdClass $tplInfo |
| 1137 | ): void { |
| 1138 | if ( |
| 1139 | DOMCompat::nodeName( $node ) === 'a' && |
| 1140 | DOMUtils::hasRel( $node, "mw:ExtLink" ) && |
| 1141 | // Images in extlinks will end up with broken up extlinks inside the |
| 1142 | // <figure> DOM. Those have 'misnested' flag set on them. Ignore those. |
| 1143 | empty( $dp->misnested ) |
| 1144 | ) { |
| 1145 | $next = $node->nextSibling; |
| 1146 | $lintError = $next instanceof Element && |
| 1147 | !empty( DOMDataUtils::getDataParsoid( $next )->misnested ) && |
| 1148 | // This check may not be necessary but ensures that we are |
| 1149 | // really in a link-in-link misnested scenario. |
| 1150 | DOMUtils::treeHasElement( $next, 'a', true ); |
| 1151 | |
| 1152 | // Media as opposed to most instances of img (barring the link= trick), don't result |
| 1153 | // in misnesting according the html5 spec since we're actively suppressing links in |
| 1154 | // their structure. However, since timed media is inherently clickable, being nested |
| 1155 | // in an extlink could surprise a user clicking on it by navigating away from the page. |
| 1156 | if ( !$lintError ) { |
| 1157 | DOMUtils::visitDOM( $node, static function ( $element ) use ( &$lintError ) { |
| 1158 | if ( $element instanceof Element && |
| 1159 | ( DOMCompat::nodeName( $element ) === 'audio' || |
| 1160 | DOMCompat::nodeName( $element ) === 'video' ) |
| 1161 | ) { |
| 1162 | $lintError = true; |
| 1163 | } |
| 1164 | } ); |
| 1165 | } |
| 1166 | if ( $lintError ) { |
| 1167 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 1168 | $lintObj = [ |
| 1169 | 'dsr' => self::findLintDSR( $tplLintInfo, $tplInfo, $dp->dsr ?? null ), |
| 1170 | 'templateInfo' => $tplLintInfo, |
| 1171 | ]; |
| 1172 | $env->recordLint( 'wikilink-in-extlink', $lintObj ); |
| 1173 | } |
| 1174 | } |
| 1175 | } |
| 1176 | |
| 1177 | private function recordLargeTablesLint( |
| 1178 | Env $env, ?stdClass $tplInfo, Element $node, int $numColumns, int $columnsMax |
| 1179 | ): void { |
| 1180 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 1181 | $lintObj = [ |
| 1182 | 'dsr' => self::findLintDSR( |
| 1183 | $tplLintInfo, $tplInfo, DOMDataUtils::getDataParsoid( $node )->dsr ?? null |
| 1184 | ), |
| 1185 | 'templateInfo' => $tplLintInfo, |
| 1186 | 'params' => [ |
| 1187 | 'name' => 'table', |
| 1188 | 'columns' => $numColumns, |
| 1189 | 'columnsMax' => $columnsMax, |
| 1190 | ], |
| 1191 | ]; |
| 1192 | $env->recordLint( 'large-tables', $lintObj ); |
| 1193 | } |
| 1194 | |
| 1195 | /** |
| 1196 | * TODO: In the future, this may merit being moved to DOMUtils |
| 1197 | * along with its "previous" variant. |
| 1198 | */ |
| 1199 | private function skipNonElementNodes( ?Node $n ): ?Element { |
| 1200 | while ( $n && !( $n instanceof Element ) ) { |
| 1201 | $n = $n->nextSibling; |
| 1202 | } |
| 1203 | return $n; |
| 1204 | } |
| 1205 | |
| 1206 | /** |
| 1207 | * Lint large tables. |
| 1208 | * |
| 1209 | * Identify articles having overly-large tables |
| 1210 | * to help editors optimize their articles. |
| 1211 | * |
| 1212 | * Linter category: `large-tables` |
| 1213 | */ |
| 1214 | private function lintLargeTables( |
| 1215 | Env $env, Element $node, DataParsoid $dp, ?stdClass $tplInfo |
| 1216 | ): void { |
| 1217 | if ( DOMCompat::nodeName( $node ) !== 'table' ) { |
| 1218 | return; |
| 1219 | } |
| 1220 | |
| 1221 | // Skip tables that have nested tables in them as they are likely |
| 1222 | // to be used for layout and not for data representation. |
| 1223 | // We may check nested tables in the next iteration of this lint. |
| 1224 | $nestedTables = $node->getElementsByTagName( 'table' ); |
| 1225 | if ( $nestedTables->length > 0 ) { |
| 1226 | return; |
| 1227 | } |
| 1228 | |
| 1229 | $lintConfig = $env->getLinterConfig(); |
| 1230 | $maxColumns = $lintConfig['maxTableColumnHeuristic'] ?? 5; |
| 1231 | $maxRowsToCheck = $lintConfig['maxTableRowsToCheck'] ?? 10; |
| 1232 | |
| 1233 | $trCount = 0; |
| 1234 | $tbody = DOMCompat::querySelector( $node, 'tbody' ); |
| 1235 | // empty table |
| 1236 | if ( !$tbody ) { |
| 1237 | return; |
| 1238 | } |
| 1239 | $tr = self::skipNonElementNodes( $tbody->firstChild ); |
| 1240 | while ( $tr && $trCount < $maxRowsToCheck ) { |
| 1241 | $numTh = $tr->getElementsByTagName( 'th' )->length; |
| 1242 | if ( $numTh > $maxColumns ) { |
| 1243 | $this->recordLargeTablesLint( $env, $tplInfo, $node, $numTh, $maxColumns ); |
| 1244 | return; |
| 1245 | } |
| 1246 | |
| 1247 | $numTd = $tr->getElementsByTagName( 'td' )->length; |
| 1248 | if ( $numTd > $maxColumns ) { |
| 1249 | $this->recordLargeTablesLint( $env, $tplInfo, $node, $numTd, $maxColumns ); |
| 1250 | return; |
| 1251 | } |
| 1252 | |
| 1253 | $tr = self::skipNonElementNodes( $tr->nextSibling ); |
| 1254 | $trCount++; |
| 1255 | } |
| 1256 | } |
| 1257 | |
| 1258 | /** |
| 1259 | * Log inline background color style rules without a color style rule. |
| 1260 | * |
| 1261 | * This function identifies elements with inline style attributes |
| 1262 | * that have background color set but don't have a color style rule. |
| 1263 | * It records linter events for such elements to help editors make |
| 1264 | * their articles comply with WCAG color contrast rules. |
| 1265 | * |
| 1266 | * Linter category: `night-mode-unaware-background-color` |
| 1267 | */ |
| 1268 | private function lintNightModeUnawareBackgroundColor( |
| 1269 | Env $env, Element $node, DataParsoid $dp, ?stdClass $tplInfo |
| 1270 | ): void { |
| 1271 | // Get inline style attribute value |
| 1272 | $styleAttrValue = DOMCompat::getAttribute( $node, 'style' ); |
| 1273 | |
| 1274 | // Check if background color is set but font color is not |
| 1275 | if ( |
| 1276 | ( $styleAttrValue !== null ) && |
| 1277 | preg_match( '/(^|;)\s*background(-color)?\s*:/i', $styleAttrValue ) && |
| 1278 | !preg_match( '/(^|;)\s*color\s*:/i', $styleAttrValue ) |
| 1279 | ) { |
| 1280 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 1281 | $lintObj = [ |
| 1282 | 'dsr' => self::findLintDSR( $tplLintInfo, $tplInfo, $dp->dsr ?? null ), |
| 1283 | 'templateInfo' => $tplLintInfo, |
| 1284 | ]; |
| 1285 | $env->recordLint( 'night-mode-unaware-background-color', $lintObj ); |
| 1286 | } |
| 1287 | } |
| 1288 | |
| 1289 | /** |
| 1290 | * Lint for missing image alt text |
| 1291 | * |
| 1292 | * Linter category: `missing-image-alt-text` |
| 1293 | */ |
| 1294 | private function lintMissingAltText( |
| 1295 | Env $env, Element $c, DataParsoid $dp, ?stdClass $tplInfo |
| 1296 | ): void { |
| 1297 | if ( !WTUtils::isGeneratedFigure( $c ) ) { |
| 1298 | return; |
| 1299 | } |
| 1300 | |
| 1301 | // Extract the media element in its standard place |
| 1302 | $media = $c->firstChild->firstChild ?? null; |
| 1303 | if ( !( $media instanceof Element ) || DOMCompat::nodeName( $media ) !== 'img' ) { |
| 1304 | // Videos and such are handled differently; check only |
| 1305 | // simple image output for alt text. |
| 1306 | return; |
| 1307 | } |
| 1308 | |
| 1309 | if ( $media->hasAttribute( 'alt' ) ) { |
| 1310 | // Present and accounted for, either via explicit markup |
| 1311 | // or filling in from an inline caption or other future |
| 1312 | // source. |
| 1313 | // |
| 1314 | // Note that an explicit empty alt text will be counted |
| 1315 | // as present, as this may be done deliberately for |
| 1316 | // spacer images or similar. |
| 1317 | return; |
| 1318 | } |
| 1319 | |
| 1320 | // Follow the parent tree looking for aria-hidden=true or equivalent roles |
| 1321 | for ( $node = $media; $node->parentNode; $node = $node->parentNode ) { |
| 1322 | $hidden = strtolower( DOMCompat::getAttribute( $node, 'aria-hidden' ) ?? '' ); |
| 1323 | $role = strtolower( DOMCompat::getAttribute( $node, 'role' ) ?? '' ); |
| 1324 | if ( $hidden === 'true' |
| 1325 | || $role === 'presentation' |
| 1326 | || $role === 'none' ) { |
| 1327 | // This entire subtree is excluded from the accessibility tree. |
| 1328 | return; |
| 1329 | } |
| 1330 | } |
| 1331 | |
| 1332 | $resource = DOMCompat::getAttribute( $media, 'resource' ) ?? ''; |
| 1333 | $file = basename( urldecode( $resource ) ); |
| 1334 | |
| 1335 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 1336 | $lintObj = [ |
| 1337 | 'dsr' => self::findLintDSR( $tplLintInfo, $tplInfo, $dp->dsr ?? null ), |
| 1338 | 'templateInfo' => $tplLintInfo, |
| 1339 | 'params' => [ |
| 1340 | 'file' => $file, |
| 1341 | ] |
| 1342 | ]; |
| 1343 | $env->recordLint( 'missing-image-alt-text', $lintObj ); |
| 1344 | } |
| 1345 | |
| 1346 | /** |
| 1347 | * Lint duplicate ids in the page |
| 1348 | * |
| 1349 | * Linter category: `duplicate-ids` |
| 1350 | */ |
| 1351 | private function lintDuplicateIds( |
| 1352 | Env $env, Element $node, DataParsoid $dp, ?stdClass $tplInfo |
| 1353 | ) { |
| 1354 | $id = DOMCompat::getAttribute( $node, 'id' ); |
| 1355 | if ( $id === null || $id === '' ) { |
| 1356 | return; |
| 1357 | } |
| 1358 | if ( !isset( $this->seenIds[$id] ) ) { |
| 1359 | $this->seenIds[$id] = 1; |
| 1360 | return; |
| 1361 | } |
| 1362 | $tplLintInfo = self::findEnclosingTemplateName( $env, $tplInfo ); |
| 1363 | $lintObj = [ |
| 1364 | 'dsr' => self::findLintDSR( |
| 1365 | $tplLintInfo, $tplInfo, $dp->dsr ?? null |
| 1366 | ), |
| 1367 | 'templateInfo' => $tplLintInfo, |
| 1368 | 'params' => [ 'id' => $id ], |
| 1369 | ]; |
| 1370 | $env->recordLint( 'duplicate-ids', $lintObj ); |
| 1371 | } |
| 1372 | |
| 1373 | /** |
| 1374 | * Log wikitext fixups |
| 1375 | */ |
| 1376 | private function logWikitextFixups( |
| 1377 | Element $node, Env $env, ?stdClass $tplInfo |
| 1378 | ): void { |
| 1379 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 1380 | $this->lintTreeBuilderFixup( $env, $node, $dp, $tplInfo ); |
| 1381 | $this->lintDeletableTableTag( $env, $node, $dp, $tplInfo ); // For T161341 |
| 1382 | $this->lintPWrapBugWorkaround( $env, $node, $dp, $tplInfo ); // For T161306 |
| 1383 | $this->lintObsoleteTag( $env, $node, $dp, $tplInfo ); |
| 1384 | $this->lintBogusImageOptions( $env, $node, $dp, $tplInfo ); |
| 1385 | $this->lintTidyWhitespaceBug( $env, $node, $dp, $tplInfo ); |
| 1386 | $this->lintMiscTidyReplacementIssues( $env, $node, $dp, $tplInfo ); |
| 1387 | $this->lintMultilineHtmlTableInList( $env, $node, $dp, $tplInfo ); |
| 1388 | $this->lintWikilinksInExtlink( $env, $node, $dp, $tplInfo ); |
| 1389 | $this->lintLargeTables( $env, $node, $dp, $tplInfo ); |
| 1390 | $this->lintNightModeUnawareBackgroundColor( $env, $node, $dp, $tplInfo ); |
| 1391 | $this->lintFostered( $env, $node, $dp, $tplInfo ); |
| 1392 | $this->lintMissingAltText( $env, $node, $dp, $tplInfo ); |
| 1393 | $this->lintDuplicateIds( $env, $node, $dp, $tplInfo ); |
| 1394 | } |
| 1395 | |
| 1396 | /** |
| 1397 | * Walk the DOM and compute lints for the entire tree. |
| 1398 | * - When we enter encapsulated content (templates or extensions), |
| 1399 | * compute "tplInfo" (misnamed given that it can be an extension) |
| 1400 | * so that lints from the templates' content can be mapped back |
| 1401 | * to the transclusion that generated them. |
| 1402 | * - When we process extensions, if we have a lint handler for the |
| 1403 | * extension, let the extension's lint handler compute lints. |
| 1404 | */ |
| 1405 | private function findLints( |
| 1406 | Node $root, Env $env, ?stdClass $tplInfo = null |
| 1407 | ): void { |
| 1408 | $node = $root->firstChild; |
| 1409 | while ( $node !== null ) { |
| 1410 | if ( !$node instanceof Element ) { |
| 1411 | $node = $node->nextSibling; |
| 1412 | continue; |
| 1413 | } |
| 1414 | |
| 1415 | // !tplInfo check is to protect against templated content in |
| 1416 | // extensions which might in turn be nested in templated content. |
| 1417 | if ( !$tplInfo && WTUtils::isFirstEncapsulationWrapperNode( $node ) ) { |
| 1418 | $aboutSibs = WTUtils::getAboutSiblings( $node, DOMCompat::getAttribute( $node, 'about' ) ); |
| 1419 | $tplInfo = (object)[ |
| 1420 | 'first' => $node, |
| 1421 | 'last' => end( $aboutSibs ), |
| 1422 | 'dsr' => DOMDataUtils::getDataParsoid( $node )->dsr ?? null, |
| 1423 | // FIXME: This is not being used. Instead the code is recomputing |
| 1424 | // this info in findEnclosingTemplateName. |
| 1425 | 'isTemplated' => DOMUtils::hasTypeOf( $node, 'mw:Transclusion' ), |
| 1426 | ]; |
| 1427 | } |
| 1428 | |
| 1429 | $handled = false; |
| 1430 | |
| 1431 | // Let native extensions lint their content |
| 1432 | $nativeExt = WTUtils::getNativeExt( $env, $node ); |
| 1433 | if ( $nativeExt ) { |
| 1434 | if ( !$this->extApi ) { |
| 1435 | $this->extApi = new ParsoidExtensionAPI( $env ); |
| 1436 | } |
| 1437 | $handled = $nativeExt->lintHandler( |
| 1438 | $this->extApi, |
| 1439 | $node, |
| 1440 | function ( $extRootNode ) use ( $env, $tplInfo ) { |
| 1441 | $this->findLints( |
| 1442 | $extRootNode, $env, |
| 1443 | empty( $tplInfo->isTemplated ) ? null : $tplInfo |
| 1444 | ); |
| 1445 | } |
| 1446 | ); |
| 1447 | // NOTE: See the note in WrapSectionsState::shouldOmitFromTOC() |
| 1448 | // but we've assumed extension content is contained in a single |
| 1449 | // wrapper node and it's safe to move to $node->nextSibling. |
| 1450 | } |
| 1451 | |
| 1452 | // Default node handler |
| 1453 | if ( $handled === false ) { |
| 1454 | // Lint this node |
| 1455 | $this->logWikitextFixups( $node, $env, $tplInfo ); |
| 1456 | |
| 1457 | // Lint subtree |
| 1458 | $this->findLints( $node, $env, $tplInfo ); |
| 1459 | } |
| 1460 | |
| 1461 | if ( $tplInfo && $tplInfo->last === $node ) { |
| 1462 | $tplInfo = null; |
| 1463 | } |
| 1464 | |
| 1465 | $node = $node->nextSibling; |
| 1466 | } |
| 1467 | } |
| 1468 | |
| 1469 | /** |
| 1470 | * This is only invoked on the top-level document |
| 1471 | * @inheritDoc |
| 1472 | */ |
| 1473 | public function run( |
| 1474 | Env $env, Node $root, array $options = [], bool $atTopLevel = false |
| 1475 | ): void { |
| 1476 | if ( !$env->linting() ) { |
| 1477 | return; |
| 1478 | } |
| 1479 | |
| 1480 | // Track time spent linting so we can evaluate benefits |
| 1481 | // of migrating this code off the critical path to its own |
| 1482 | // post processor. |
| 1483 | $siteConfig = $env->getSiteConfig(); |
| 1484 | $timer = Timing::start( $siteConfig ); |
| 1485 | |
| 1486 | $this->findLints( $root, $env ); |
| 1487 | $this->postProcessLints( $env->getLints(), $env ); |
| 1488 | |
| 1489 | $timer->end( "linting", "linting", [] ); |
| 1490 | } |
| 1491 | |
| 1492 | } |