Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 83 |
|
0.00% |
0 / 3 |
CRAP | |
0.00% |
0 / 1 |
| LiFixups | |
0.00% |
0 / 83 |
|
0.00% |
0 / 3 |
1260 | |
0.00% |
0 / 1 |
| getMigrationInfo | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
42 | |||
| findLastMigratableNode | |
0.00% |
0 / 20 |
|
0.00% |
0 / 1 |
90 | |||
| migrateTrailingCategories | |
0.00% |
0 / 54 |
|
0.00% |
0 / 1 |
420 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Wt2Html\DOM\Handlers; |
| 5 | |
| 6 | use Wikimedia\Assert\Assert; |
| 7 | use Wikimedia\Parsoid\DOM\Comment; |
| 8 | use Wikimedia\Parsoid\DOM\Element; |
| 9 | use Wikimedia\Parsoid\DOM\Node; |
| 10 | use Wikimedia\Parsoid\DOM\Text; |
| 11 | use Wikimedia\Parsoid\Utils\DiffDOMUtils; |
| 12 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 13 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 14 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 15 | use Wikimedia\Parsoid\Utils\DTState; |
| 16 | use Wikimedia\Parsoid\Utils\WTUtils; |
| 17 | |
| 18 | class LiFixups { |
| 19 | |
| 20 | private static function getMigrationInfo( Node $c ): array { |
| 21 | $tplRoot = WTUtils::findFirstEncapsulationWrapperNode( $c ); |
| 22 | if ( $tplRoot !== null ) { |
| 23 | // Check if everything between tplRoot and c is migratable. |
| 24 | $prev = $tplRoot->previousSibling; |
| 25 | while ( $c !== $prev ) { |
| 26 | if ( !WTUtils::isCategoryLink( $c ) && |
| 27 | !( DOMCompat::nodeName( $c ) === 'span' && preg_match( '/^\s*$/D', $c->textContent ) ) |
| 28 | ) { |
| 29 | return [ 'tplRoot' => $tplRoot, 'migratable' => false ]; |
| 30 | } |
| 31 | |
| 32 | $c = $c->previousSibling; |
| 33 | } |
| 34 | } |
| 35 | |
| 36 | return [ 'tplRoot' => $tplRoot, 'migratable' => true ]; |
| 37 | } |
| 38 | |
| 39 | private static function findLastMigratableNode( Node $li ): ?Node { |
| 40 | $sentinel = null; |
| 41 | $c = DiffDOMUtils::lastNonSepChild( $li ); |
| 42 | // c is known to be a category link. |
| 43 | // fail fast in parser tests if something changes. |
| 44 | Assert::invariant( WTUtils::isCategoryLink( $c ), 'c is known to be a category link' ); |
| 45 | while ( $c ) { |
| 46 | // Handle template units first |
| 47 | $info = self::getMigrationInfo( $c ); |
| 48 | if ( !$info['migratable'] ) { |
| 49 | break; |
| 50 | } elseif ( $info['tplRoot'] !== null ) { |
| 51 | $c = $info['tplRoot']; |
| 52 | } |
| 53 | |
| 54 | if ( $c instanceof Text ) { |
| 55 | // Update sentinel if we hit a newline. |
| 56 | // We want to migrate these newlines and |
| 57 | // everything following them out of 'li'. |
| 58 | if ( preg_match( '/\n\s*$/D', $c->nodeValue ) ) { |
| 59 | $sentinel = $c; |
| 60 | } |
| 61 | |
| 62 | // If we didn't hit pure whitespace, we are done! |
| 63 | if ( !preg_match( '/^\s*$/D', $c->nodeValue ) ) { |
| 64 | break; |
| 65 | } |
| 66 | } elseif ( $c instanceof Comment ) { |
| 67 | $sentinel = $c; |
| 68 | } elseif ( !WTUtils::isCategoryLink( $c ) ) { |
| 69 | // We are done if we hit anything but text |
| 70 | // or category links. |
| 71 | break; |
| 72 | } |
| 73 | |
| 74 | $c = $c->previousSibling; |
| 75 | } |
| 76 | |
| 77 | return $sentinel; |
| 78 | } |
| 79 | |
| 80 | /** |
| 81 | * Earlier in the parsing pipeline, we suppress all newlines |
| 82 | * and other whitespace before categories which causes category |
| 83 | * links to be swallowed into preceding paragraphs and list items. |
| 84 | * |
| 85 | * However, with wikitext like this: `*a\n\n[[Category:Foo]]`, this |
| 86 | * could prevent proper roundtripping (because we suppress newlines |
| 87 | * when serializing list items). This needs addressing because |
| 88 | * this pattern is extremely common (some list at the end of the page |
| 89 | * followed by a list of categories for the page). |
| 90 | * @param Element $li |
| 91 | * @param DTState $state |
| 92 | * @return bool |
| 93 | */ |
| 94 | public static function migrateTrailingCategories( Element $li, DTState $state ): bool { |
| 95 | // * Don't bother fixing up template content when processing the full page |
| 96 | if ( $state->tplInfo ?? null ) { |
| 97 | return true; |
| 98 | } |
| 99 | |
| 100 | // If there is migratable content inside a list item |
| 101 | // (categories preceded by newlines), |
| 102 | // * migrate it out of the outermost list |
| 103 | // * and fix up the DSR of list items and list along the rightmost path. |
| 104 | if ( $li->nextSibling === null && DOMUtils::isList( $li->parentNode ) && |
| 105 | WTUtils::isCategoryLink( DiffDOMUtils::lastNonSepChild( $li ) ) |
| 106 | ) { |
| 107 | |
| 108 | // Find the outermost list -- content will be moved after it |
| 109 | $outerList = $li->parentNode; |
| 110 | while ( DOMUtils::isListItem( $outerList->parentNode ) ) { |
| 111 | $p = $outerList->parentNode; |
| 112 | // Bail if we find ourself on a path that is not the rightmost path. |
| 113 | if ( $p->nextSibling !== null ) { |
| 114 | return true; |
| 115 | } |
| 116 | $outerList = $p->parentNode; |
| 117 | } |
| 118 | |
| 119 | // Find last migratable node |
| 120 | $sentinel = self::findLastMigratableNode( $li ); |
| 121 | if ( !$sentinel ) { |
| 122 | return true; |
| 123 | } |
| 124 | |
| 125 | // Migrate (and update DSR) |
| 126 | $c = $li->lastChild; |
| 127 | $liDsr = DOMDataUtils::getDataParsoid( $li )->dsr ?? null; |
| 128 | $newEndDsr = -1; // dummy to eliminate useless null checks |
| 129 | while ( true ) { |
| 130 | if ( $c instanceof Element ) { |
| 131 | $dsr = DOMDataUtils::getDataParsoid( $c )->dsr ?? null; |
| 132 | $newEndDsr = $dsr->start ?? -1; |
| 133 | $outerList->parentNode->insertBefore( $c, $outerList->nextSibling ); |
| 134 | } elseif ( $c instanceof Text ) { |
| 135 | if ( preg_match( '/^\s*$/D', $c->nodeValue ) ) { |
| 136 | $newEndDsr -= strlen( $c->nodeValue ); |
| 137 | $outerList->parentNode->insertBefore( $c, $outerList->nextSibling ); |
| 138 | } else { |
| 139 | // Split off the newlines into its own node and migrate it |
| 140 | $nls = $c->nodeValue; |
| 141 | $c->nodeValue = preg_replace( '/\s+$/D', '', $c->nodeValue, 1 ); |
| 142 | $nls = substr( $nls, strlen( $c->nodeValue ) ); |
| 143 | $nlNode = $c->ownerDocument->createTextNode( $nls ); |
| 144 | $outerList->parentNode->insertBefore( $nlNode, $outerList->nextSibling ); |
| 145 | $newEndDsr -= strlen( $nls ); |
| 146 | } |
| 147 | } elseif ( $c instanceof Comment ) { |
| 148 | $newEndDsr -= WTUtils::decodedCommentLength( $c ); |
| 149 | $outerList->parentNode->insertBefore( $c, $outerList->nextSibling ); |
| 150 | } |
| 151 | |
| 152 | if ( $c === $sentinel ) { |
| 153 | break; |
| 154 | } |
| 155 | |
| 156 | $c = $li->lastChild; |
| 157 | } |
| 158 | |
| 159 | // Update DSR of all listitem & list nodes till |
| 160 | // we hit the outermost list we started with. |
| 161 | $delta = null; |
| 162 | if ( $liDsr && $newEndDsr >= 0 ) { |
| 163 | $delta = $liDsr->end - $newEndDsr; |
| 164 | } |
| 165 | |
| 166 | // If there is no delta to adjust dsr by, we are done |
| 167 | if ( !$delta ) { |
| 168 | return true; |
| 169 | } |
| 170 | |
| 171 | // Fix DSR along the rightmost path to outerList |
| 172 | $list = null; |
| 173 | while ( $outerList !== $list ) { |
| 174 | $list = $li->parentNode; |
| 175 | DOMUtils::assertElt( $list ); |
| 176 | |
| 177 | $liDp = DOMDataUtils::getDataParsoid( $li ); |
| 178 | if ( !empty( $liDp->dsr ) ) { |
| 179 | $liDp->dsr->end -= $delta; |
| 180 | } |
| 181 | |
| 182 | $listDp = DOMDataUtils::getDataParsoid( $list ); |
| 183 | if ( !empty( $listDp->dsr ) ) { |
| 184 | $listDp->dsr->end -= $delta; |
| 185 | } |
| 186 | $li = $list->parentNode; |
| 187 | } |
| 188 | } |
| 189 | |
| 190 | return true; |
| 191 | } |
| 192 | } |