CategoryViewer.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Category;
 
use InvalidArgumentException;
use MediaWiki\Collation\Collation;
use MediaWiki\Context\ContextSource;
use MediaWiki\Context\IContextSource;
use MediaWiki\Debug\DeprecationHelper;
use MediaWiki\Deferred\LinksUpdate\CategoryLinksTable;
use MediaWiki\Gallery\Exception\ImageGalleryClassNotFoundException;
use MediaWiki\Gallery\ImageGalleryBase;
use MediaWiki\HookContainer\ProtectedHookAccessorTrait;
use MediaWiki\Html\Html;
use MediaWiki\Language\ILanguageConverter;
use MediaWiki\Linker\LinkTarget;
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
use MediaWiki\Page\LinkCache;
use MediaWiki\Page\PageIdentity;
use MediaWiki\Page\PageReference;
use MediaWiki\Parser\ParserOutputFlags;
use MediaWiki\Title\Title;
use MediaWiki\Title\TitleValue;
use Wikimedia\HtmlArmor\HtmlArmor;
use Wikimedia\Rdbms\FakeResultWrapper;
use Wikimedia\Rdbms\SelectQueryBuilder;
 
class CategoryViewer extends ContextSource {
    use ProtectedHookAccessorTrait;
    use DeprecationHelper;
 
    public readonly int $limit;
    public array $articles = [];
    public array $articles_start_char = [];
    public array $children = [];
    public array $children_start_char = [];
    public bool $showGallery;
    public array $imgsNoGallery_start_char = [];
    public array $imgsNoGallery = [];
    public array $nextPage = [];
    protected array $prevPage = [];
    public array $flip = [];
 
    public readonly bool $sortByTimestamp;
 
    public readonly Collation $collation;
    public ImageGalleryBase $gallery;
 
    private readonly Category $cat;
 
    private readonly ILanguageConverter $languageConverter;
 
    public function __construct(
        protected PageIdentity $page,
        IContextSource $context,
        public readonly array $from = [],
        public readonly array $until = [],
        private array $query = [],
    ) {
        $this->deprecatePublicPropertyFallback(
            'title',
            '1.37',
            fn (): Title => Title::newFromPageIdentity( $this->page ),
            fn ( PageIdentity $page ) => $this->page = $page
        );
 
        $this->setContext( $context );
        $this->getOutput()->addModuleStyles( [
            'mediawiki.action.styles',
        ] );
        $this->limit = $context->getConfig()->get( MainConfigNames::CategoryPagingLimit );
        $this->sortByTimestamp = $context->getRequest()->getVal( 'cldsort' ) === 'timestamp';
        $this->cat = Category::newFromTitle( $page );
 
        $services = MediaWikiServices::getInstance();
        $this->collation = $services->getCollationFactory()->getCategoryCollation();
        $this->languageConverter = $services->getLanguageConverterFactory()->getLanguageConverter();
 
        unset( $this->query['title'] );
    }
 
    public function getHTML() {
        $this->showGallery = $this->getConfig()->get( MainConfigNames::CategoryMagicGallery )
            && !$this->getOutput()->getOutputFlag( ParserOutputFlags::NO_GALLERY );
 
        $this->clearCategoryState();
        $this->doCategoryQuery();
        $this->finaliseCategoryState();
 
        $html = $this->getSubcategorySection() .
            $this->getPagesSection() .
            $this->getImageSection();
 
        if ( $html === '' ) {
            $html = $this->msg( 'category-empty' )->parseAsBlock();
        }
 
        # put a div around the headings which are in the user language
        $lang = $this->getLanguage();
        return Html::rawElement( 'div', [
            'class' => 'mw-category-generated',
            'lang' => $lang->getHtmlCode(),
            'dir' => $lang->getDir()
        ], $html );
    }
 
    protected function clearCategoryState() {
        $this->articles = [];
        $this->articles_start_char = [];
        $this->children = [];
        $this->children_start_char = [];
        if ( $this->showGallery ) {
            // Note that null for mode is taken to mean use default.
            $mode = $this->getRequest()->getVal( 'gallerymode', null );
            try {
                $this->gallery = ImageGalleryBase::factory( $mode, $this->getContext() );
            } catch ( ImageGalleryClassNotFoundException ) {
                // User specified something invalid, fallback to default.
                $this->gallery = ImageGalleryBase::factory( false, $this->getContext() );
            }
 
            $this->gallery->setHideBadImages();
        } else {
            $this->imgsNoGallery = [];
            $this->imgsNoGallery_start_char = [];
        }
    }
 
    public function addSubcategoryObject( Category $cat, string $sortkey, int $pageLength ): void {
        $page = $cat->getPage();
        if ( !$page ) {
            return;
        }
 
        // Subcategory; strip the 'Category' namespace from the link text.
        $pageRecord = MediaWikiServices::getInstance()->getPageStore()
            ->getPageByReference( $page );
        if ( !$pageRecord ) {
            return;
        }
 
        $this->children[] = $this->generateLink(
            'subcat',
            $pageRecord,
            $pageRecord->isRedirect(),
            htmlspecialchars( str_replace( '_', ' ', $pageRecord->getDBkey() ) )
        );
 
        $this->children_start_char[] = $this->getStartKey( $sortkey );
    }
 
    private function getStartKey( string $sortkey ): string {
        if ( $this->sortByTimestamp ) {
            return $sortkey;
        }
        return $this->languageConverter->convert( $this->collation->getFirstLetter( $sortkey ) );
    }
 
    private function generateLink(
        string $type,
        PageReference $page,
        bool $isRedirect,
        ?string $html = null
    ): string {
        $link = null;
        $legacyTitle = MediaWikiServices::getInstance()->getTitleFactory()
            ->newFromPageReference( $page );
        // @phan-suppress-next-line PhanTypeMismatchArgument Type mismatch on pass-by-ref args
        $this->getHookRunner()->onCategoryViewer__generateLink( $type, $legacyTitle, $html, $link );
        if ( $link === null ) {
            $linkRenderer = MediaWikiServices::getInstance()->getLinkRenderer();
            if ( $html !== null ) {
                $html = new HtmlArmor( $html );
            }
            $link = $linkRenderer->makeLink( $page, $html );
        }
        if ( $isRedirect ) {
            $link = Html::rawElement(
                'span',
                [ 'class' => 'redirect-in-category' ],
                $link
            );
        }
 
        return $link;
    }
 
    public function addImage(
        PageReference $page,
        string $sortkey,
        int $pageLength,
        bool $isRedirect = false
    ): void {
        $title = MediaWikiServices::getInstance()->getTitleFactory()
            ->newFromPageReference( $page );
        if ( $this->showGallery ) {
            $flip = $this->flip['file'];
            if ( $flip ) {
                $this->gallery->insert( $title, '', '', '', [], ImageGalleryBase::LOADING_LAZY );
            } else {
                $this->gallery->add( $title, '', '', '', [], ImageGalleryBase::LOADING_LAZY );
            }
        } else {
            $this->imgsNoGallery[] = $this->generateLink( 'image', $page, $isRedirect );
 
            $this->imgsNoGallery_start_char[] = $this->getStartKey( $sortkey );
        }
    }
 
    public function addPage(
        PageReference $page,
        string $sortkey,
        int $pageLength,
        bool $isRedirect = false
    ): void {
        $this->articles[] = $this->generateLink( 'page', $page, $isRedirect );
 
        $this->articles_start_char[] = $this->getStartKey( $sortkey );
    }
 
    protected function finaliseCategoryState() {
        if ( $this->flip['subcat'] ) {
            $this->children = array_reverse( $this->children );
            $this->children_start_char = array_reverse( $this->children_start_char );
        }
        if ( $this->flip['page'] ) {
            $this->articles = array_reverse( $this->articles );
            $this->articles_start_char = array_reverse( $this->articles_start_char );
        }
        if ( !$this->showGallery && $this->flip['file'] ) {
            $this->imgsNoGallery = array_reverse( $this->imgsNoGallery );
            $this->imgsNoGallery_start_char = array_reverse( $this->imgsNoGallery_start_char );
        }
    }
 
    protected function doCategoryQuery() {
        $connProvider = MediaWikiServices::getInstance()->getConnectionProvider();
        $dbr = $connProvider->getReplicaDatabase();
        $categoryLinksDbr = $connProvider->getReplicaDatabase( CategoryLinksTable::VIRTUAL_DOMAIN );
 
        $this->nextPage = [
            'page' => null,
            'subcat' => null,
            'file' => null,
        ];
        $this->prevPage = [
            'page' => null,
            'subcat' => null,
            'file' => null,
        ];
 
        $this->flip = [ 'page' => false, 'subcat' => false, 'file' => false ];
 
        // The DB column driving ordering and pagination for this view: either the
        // collation-based cl_sortkey (default), or cl_timestamp when the caller
        // asked for members ordered by when they were added to the category.
        $sortField = $this->sortByTimestamp ? 'cl_timestamp' : 'cl_sortkey';
 
        foreach ( [ 'page', 'subcat', 'file' ] as $type ) {
            # Get the sort values for start/end, if applicable. Note that if
            # the collation in the database differs from the one
            # set in $wgCategoryCollation, pagination might go totally haywire.
            $extraConds = [ 'cl_type' => $type ];
            if ( isset( $this->from[$type] ) ) {
                $extraConds[] = $categoryLinksDbr->expr(
                    $sortField,
                    '>=',
                    $this->sortByTimestamp
                        // @phan-suppress-next-line PhanTypeMismatchArgumentNullable
                        ? $categoryLinksDbr->timestamp( $this->from[$type] )
                        // @phan-suppress-next-line PhanTypeMismatchArgumentNullable
                        : $this->collation->getSortKey( $this->from[$type] )
                );
            } elseif ( isset( $this->until[$type] ) ) {
                $extraConds[] = $categoryLinksDbr->expr(
                    $sortField,
                    '<',
                    $this->sortByTimestamp
                        // @phan-suppress-next-line PhanTypeMismatchArgumentNullable
                        ? $categoryLinksDbr->timestamp( $this->until[$type] )
                        // @phan-suppress-next-line PhanTypeMismatchArgumentNullable
                        : $this->collation->getSortKey( $this->until[$type] )
                );
                $this->flip[$type] = true;
            }
 
            $queryBuilder = $categoryLinksDbr->newSelectQueryBuilder();
            $queryBuilder->select( array_merge(
                LinkCache::getSelectFields(),
                [
                    'cl_sortkey',
                    'cl_sortkey_prefix',
                    'cl_timestamp',
                    'collation_name',
                ]
            ) )
                ->from( 'page' )
                ->andWhere( $extraConds );
 
            if ( $this->flip[$type] ) {
                $queryBuilder->orderBy( $sortField, SelectQueryBuilder::SORT_DESC );
            } else {
                $queryBuilder->orderBy( $sortField );
            }
 
            $queryBuilder
                ->join( 'categorylinks', null, [ 'cl_from = page_id' ] )
                ->join( 'linktarget', null, 'cl_target_id = lt_id' )
                ->straightJoin( 'collation', null, 'cl_collation_id = collation_id' )
                ->where( [ 'lt_title' => $this->page->getDBkey(), 'lt_namespace' => NS_CATEGORY ] )
                ->useIndex( [ 'categorylinks' => $this->sortByTimestamp ? 'cl_timestamp_id' : 'cl_sortkey_id' ] )
                ->limit( $this->limit + 1 );
 
            $res = $queryBuilder->caller( __METHOD__ )->fetchResultSet();
 
            $categoryTitles = [];
            $pageRows = [];
 
            foreach ( $res as $row ) {
                $pageRows[] = $row;
                if ( (int)$row->page_namespace === NS_CATEGORY ) {
                    $categoryTitles[] = $row->page_title;
                }
            }
 
            $categoryFields = [ 'cat_id', 'cat_title', 'cat_subcats', 'cat_pages', 'cat_files' ];
 
            $categoryData = [];
            if ( $categoryTitles !== [] ) {
                $categoryRes = $dbr->newSelectQueryBuilder()
                    ->select( $categoryFields )
                    ->from( 'category' )
                    ->where( [ 'cat_title' => $categoryTitles ] )
                    ->caller( __METHOD__ )
                    ->fetchResultSet();
 
                foreach ( $categoryRes as $catRow ) {
                    $categoryData[$catRow->cat_title] = $catRow;
                }
            }
 
            foreach ( $pageRows as $row ) {
                if ( (int)$row->page_namespace === NS_CATEGORY ) {
                    $catRow = $categoryData[$row->page_title] ?? null;
                    foreach ( $categoryFields as $field ) {
                        $row->$field = $catRow->$field ?? null;
                    }
                }
            }
 
            // Convert modified pageRows back to result wrapper for hook
            $res = new FakeResultWrapper( $pageRows );
            $this->getHookRunner()->onCategoryViewer__doCategoryQuery( $type, $res );
            $linkCache = MediaWikiServices::getInstance()->getLinkCache();
 
            $count = 0;
            foreach ( $pageRows as $row ) {
                $title = Title::newFromRow( $row );
                $linkCache->addGoodLinkObjFromRow( $title, $row );
 
                if ( $this->sortByTimestamp ) {
                    // Raw DB value, used verbatim for from/until pagination.
                    $pageCursor = $row->cl_timestamp;
                    // Group entries by the day they were added to the category.
                    $groupLabel = $this->getLanguage()->userDate( $row->cl_timestamp, $this->getUser() );
                } else {
                    $pageCursor = $title->getCategorySortkey( $row->cl_sortkey_prefix );
                    $groupLabel = $pageCursor;
                }
 
                if ( ++$count > $this->limit ) {
                    # We've reached the one extra which shows that there
                    # are additional pages to be had. Stop here...
                    $this->nextPage[$type] = $pageCursor;
                    break;
                }
                if ( $count == $this->limit ) {
                    $this->prevPage[$type] = $pageCursor;
                }
 
                if ( $title->getNamespace() === NS_CATEGORY ) {
                    $cat = Category::newFromRow( $row, $title );
                    $this->addSubcategoryObject( $cat, $groupLabel, $row->page_len );
                } elseif ( $title->getNamespace() === NS_FILE ) {
                    $this->addImage( $title, $groupLabel, $row->page_len, $row->page_is_redirect );
                } else {
                    $this->addPage( $title, $groupLabel, $row->page_len, $row->page_is_redirect );
                }
            }
        }
    }
 
    protected function getSubcategorySection() {
        # Don't show subcategories section if there are none.
        $html = '';
        $localCount = count( $this->children );
        $databaseCount = $this->cat->getSubcatCount();
        // This function should be called even if the result isn't used, it has side-effects
        $countMessage = $this->getCountMessage( $localCount, $databaseCount, 'subcat' );
 
        if ( $localCount > 0 ) {
            $html .= Html::openElement( 'div', [ 'id' => 'mw-subcategories' ] ) . "\n";
            $html .= Html::rawElement( 'h2', [], $this->msg( 'subcategories' )->parse() ) . "\n";
            $html .= $countMessage;
            $html .= $this->getSectionPagingLinks( 'subcat' );
            $html .= $this->formatList( $this->children, $this->children_start_char );
            $html .= $this->getSectionPagingLinks( 'subcat' );
            $html .= "\n" . Html::closeElement( 'div' );
        }
        return $html;
    }
 
    protected function getPagesSection() {
        $name = $this->getOutput()->getUnprefixedDisplayTitle();
        # Don't show articles section if there are none.
        $html = '';
 
        # @todo FIXME: Here and in the other two sections: we don't need to bother
        # with this rigmarole if the entire category contents fit on one page
        # and have already been retrieved.  We can just use $rescnt in that
        # case and save a query and some logic.
        $databaseCount = $this->cat->getPageCount( Category::COUNT_CONTENT_PAGES );
        $localCount = count( $this->articles );
        // This function should be called even if the result isn't used, it has side-effects
        $countMessage = $this->getCountMessage( $localCount, $databaseCount, 'page' );
 
        if ( $localCount > 0 ) {
            $html .= Html::openElement( 'div', [ 'id' => 'mw-pages' ] ) . "\n";
            $html .= Html::rawElement(
                'h2',
                [],
                $this->msg( 'category_header' )->rawParams( $name )->parse()
            ) . "\n";
            $html .= $countMessage;
            $html .= $this->getSectionPagingLinks( 'page' );
            $html .= $this->formatList( $this->articles, $this->articles_start_char );
            $html .= $this->getSectionPagingLinks( 'page' );
            $html .= "\n" . Html::closeElement( 'div' );
        }
        return $html;
    }
 
    protected function getImageSection() {
        $name = $this->getOutput()->getUnprefixedDisplayTitle();
        $html = '';
        $localCount = $this->showGallery ?
            $this->gallery->count() :
            count( $this->imgsNoGallery );
        $databaseCount = $this->cat->getFileCount();
        // This function should be called even if the result isn't used, it has side-effects
        $countMessage = $this->getCountMessage( $localCount, $databaseCount, 'file' );
 
        if ( $localCount > 0 ) {
            $html .= Html::openElement( 'div', [ 'id' => 'mw-category-media' ] ) . "\n";
            $html .= Html::rawElement(
                'h2',
                [],
                $this->msg( 'category-media-header' )->rawParams( $name )->parse()
            ) . "\n";
            $html .= $countMessage;
            $html .= $this->getSectionPagingLinks( 'file' );
            if ( $this->showGallery ) {
                $html .= $this->gallery->toHTML();
            } else {
                $html .= $this->formatList( $this->imgsNoGallery, $this->imgsNoGallery_start_char );
            }
            $html .= $this->getSectionPagingLinks( 'file' );
            $html .= "\n" . Html::closeElement( 'div' );
        }
        return $html;
    }
 
    private function getSectionPagingLinks( string $type ): string {
        if ( isset( $this->until[$type] ) ) {
            // The new value for the until parameter should be pointing to the first
            // result displayed on the page which is the second last result retrieved
            // from the database.The next link should have a from parameter pointing
            // to the until parameter of the current page.
            if ( $this->nextPage[$type] !== null ) {
                return $this->pagingLinks(
                    $this->prevPage[$type] ?? '',
                    // @phan-suppress-next-line PhanTypeMismatchArgumentNullable
                    $this->until[$type],
                    $type
                );
            }
 
            // If the nextPage variable is null, it means that we have reached the first page
            // and therefore the previous link should be disabled.
            return $this->pagingLinks(
                '',
                // @phan-suppress-next-line PhanTypeMismatchArgumentNullable
                $this->until[$type],
                $type
            );
        } elseif ( $this->nextPage[$type] !== null || isset( $this->from[$type] ) ) {
            return $this->pagingLinks(
                $this->from[$type] ?? '',
                $this->nextPage[$type] ?? '',
                $type
            );
        }
 
        return '';
    }
 
    private function formatList( array $articles, array $articles_start_char, int $cutoff = 6 ): string {
        $list = '';
        if ( count( $articles ) > $cutoff ) {
            $list = self::columnList( $articles, $articles_start_char );
        } elseif ( count( $articles ) > 0 ) {
            // for short lists of articles in categories.
            $list = self::shortList( $articles, $articles_start_char );
        }
 
        $pageLang = MediaWikiServices::getInstance()->getTitleFactory()
            ->newFromPageIdentity( $this->page )
            ->getPageLanguage();
        $attribs = [ 'lang' => $pageLang->getHtmlCode(), 'dir' => $pageLang->getDir(),
            'class' => 'mw-content-' . $pageLang->getDir() ];
        $list = Html::rawElement( 'div', $attribs, $list );
 
        return $list;
    }
 
    public static function columnList(
        $articles,
        $articles_start_char,
        $cssClasses = 'mw-category mw-category-columns'
    ) {
        $columns = array_combine( $articles, $articles_start_char );
 
        $ret = Html::openElement( 'div', [ 'class' => $cssClasses ] );
 
        $colContents = [];
 
        # Kind of like array_flip() here, but we keep duplicates in an
        # array instead of dropping them.
        foreach ( $columns as $article => $char ) {
            $colContents[$char][] = $article;
        }
 
        foreach ( $colContents as $char => $articles ) {
            # Change space to non-breaking space to keep headers aligned
            $h3char = $char === ' ' ? "\u{00A0}" : htmlspecialchars( $char );
 
            $ret .= Html::openElement( 'div', [ 'class' => 'mw-category-group' ] );
            $ret .= Html::rawElement( 'h3', [], $h3char ) . "\n";
            $ret .= Html::openElement( 'ul' );
            $ret .= implode(
                "\n",
                array_map(
                    static fn ( $article ) => Html::rawElement( 'li', [], $article ),
                    $articles
                )
            );
            $ret .= Html::closeElement( 'ul' ) . Html::closeElement( 'div' );
 
        }
 
        $ret .= Html::closeElement( 'div' );
        return $ret;
    }
 
    public static function shortList( $articles, $articles_start_char ) {
        return self::columnList( $articles, $articles_start_char, 'mw-category' );
    }
 
    private function pagingLinks( string $first, string $last, string $type = '' ): string {
        $prevLink = $this->msg( 'prev-page' )->escaped();
 
        $linkRenderer = MediaWikiServices::getInstance()->getLinkRenderer();
        if ( $first != '' ) {
            $prevQuery = $this->query;
            $prevQuery["{$type}until"] = $first;
            unset( $prevQuery["{$type}from"] );
            $prevLink = $linkRenderer->makeKnownLink(
                $this->addFragmentToTitle( $this->page, $type ),
                new HtmlArmor( $prevLink ),
                [],
                $prevQuery
            );
        }
 
        $nextLink = $this->msg( 'next-page' )->escaped();
 
        if ( $last != '' ) {
            $lastQuery = $this->query;
            $lastQuery["{$type}from"] = $last;
            unset( $lastQuery["{$type}until"] );
            $nextLink = $linkRenderer->makeKnownLink(
                $this->addFragmentToTitle( $this->page, $type ),
                new HtmlArmor( $nextLink ),
                [],
                $lastQuery
            );
        }
 
        return $this->msg( 'categoryviewer-pagedlinks' )->rawParams( $prevLink, $nextLink )->escaped();
    }
 
    private function addFragmentToTitle( PageReference $page, string $section ): LinkTarget {
        $fragment = match ( $section ) {
            'page' => 'mw-pages',
            'subcat' => 'mw-subcategories',
            'file' => 'mw-category-media',
            default => throw new InvalidArgumentException( __METHOD__ . " Invalid section $section." )
        };
        return new TitleValue( $page->getNamespace(), $page->getDBkey(), $fragment );
    }
 
    private function getCountMessage( int $localCount, int $databaseCount, string $type ): string {
        // There are three cases:
        //   1) The category table figure seems good.  It might be wrong, but
        //      we can't do anything about it if we don't recalculate it on ev-
        //      ery category view.
        //   2) The category table figure isn't good, like it's smaller than the
        //      number of actual results, *but* the number of results is less
        //      than $this->limit and there's no offset.  In this case we still
        //      know the right figure.
        //   3) We have no idea.
 
        // Check if there's a "from" or "until" for anything
 
        // This is a little ugly, but we seem to use different names
        // for the paging types then for the messages.
        $msgType = $type === 'page' ? 'article' : $type;
 
        $fromOrUntil = false;
        if ( isset( $this->from[$type] ) || isset( $this->until[$type] ) ) {
            $fromOrUntil = true;
        }
 
        if ( $databaseCount == $localCount ||
            ( ( $localCount == $this->limit || $fromOrUntil ) && $databaseCount > $localCount )
        ) {
            // Case 1: seems good.
            $totalCount = $databaseCount;
        } elseif ( $localCount < $this->limit && !$fromOrUntil ) {
            // Case 2: not good, but salvageable.  Use the number of results.
            $totalCount = $localCount;
        } else {
            // Case 3: hopeless.  Don't give a total count at all.
            // Messages: category-subcat-count-limited, category-article-count-limited,
            // category-file-count-limited
            return $this->msg( "category-$msgType-count-limited" )->numParams( $localCount )->parseAsBlock();
        }
        // Messages: category-subcat-count, category-article-count, category-file-count
        return $this->msg( "category-$msgType-count" )->numParams( $localCount, $totalCount )->parseAsBlock();
    }
}