TranslatablePage.php

<?php
declare( strict_types = 1 );
 
namespace MediaWiki\Extension\Translate\PageTranslation;
 
use IDBAccessObject;
use LogicException;
use MediaWiki\Extension\Translate\MessageGroupProcessing\MessageGroups;
use MediaWiki\Extension\Translate\MessageGroupProcessing\RevTagStore;
use MediaWiki\Extension\Translate\MessageGroupProcessing\TranslatableBundle;
use MediaWiki\Extension\Translate\MessageLoading\MessageHandle;
use MediaWiki\Extension\Translate\Services;
use MediaWiki\Extension\Translate\Statistics\MessageGroupStats;
use MediaWiki\Extension\Translate\Utilities\Utilities;
use MediaWiki\Languages\LanguageNameUtils;
use MediaWiki\Linker\LinkTarget;
use MediaWiki\MediaWikiServices;
use MediaWiki\Page\PageIdentity;
use MediaWiki\Page\PageReference;
use MediaWiki\Revision\RevisionRecord;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\SpecialPage\SpecialPage;
use MediaWiki\Title\Title;
use RuntimeException;
use TextContent;
use Wikimedia\Rdbms\Database;
use Wikimedia\Rdbms\IResultWrapper;
use Wikimedia\Rdbms\SelectQueryBuilder;
use WikiPageMessageGroup;
 
class TranslatablePage extends TranslatableBundle {
    public const METADATA_KEYS = [
        'maxid',
        'priorityforce',
        'prioritylangs',
        'priorityreason',
        'transclusion',
        'version'
    ];
    public const DISPLAY_TITLE_UNIT_ID = 'Page display title';
 
    protected PageIdentity $title;
    protected RevTagStore $revTagStore;
    protected $text;
    protected $revision;
    protected $source;
    protected $pageDisplayTitle;
    private $targetLanguage;
 
    protected function __construct( PageIdentity $title ) {
        $this->title = $title;
        $this->revTagStore = Services::getInstance()->getRevTagStore();
    }
 
    public static function newFromText( Title $title, string $text ): self {
        $obj = new self( $title );
        $obj->text = $text;
        $obj->source = 'text';
 
        return $obj;
    }
 
    public static function newFromRevision( PageIdentity $title, int $revision ): self {
        $rev = MediaWikiServices::getInstance()
            ->getRevisionLookup()
            ->getRevisionByTitle( $title, $revision );
        if ( $rev === null ) {
            throw new RuntimeException( 'Revision is null' );
        }
 
        $obj = new self( $title );
        $obj->source = 'revision';
        $obj->revision = $revision;
 
        return $obj;
    }
 
    public static function newFromTitle( PageIdentity $title ): self {
        $obj = new self( $title );
        $obj->source = 'title';
 
        return $obj;
    }
 
    public function getTitle(): Title {
        return Title::castFromPageIdentity( $this->title );
    }
 
    public function getPageIdentity(): PageIdentity {
        return $this->title;
    }
 
    public function getText(): string {
        if ( $this->text !== null ) {
            return $this->text;
        }
 
        if ( $this->source === 'title' ) {
            $revision = $this->getMarkedTag();
            if ( !is_int( $revision ) ) {
                throw new LogicException(
                    "Trying to load a text for {$this->getPageIdentity()} which is not marked for translation"
                );
            }
            $this->revision = $revision;
        }
 
        $flags = Utilities::shouldReadFromPrimary()
            ? IDBAccessObject::READ_LATEST
            : IDBAccessObject::READ_NORMAL;
        $rev = MediaWikiServices::getInstance()
            ->getRevisionLookup()
            ->getRevisionByTitle( $this->getPageIdentity(), $this->revision, $flags );
        $content = $rev->getContent( SlotRecord::MAIN );
        $text = ( $content instanceof TextContent ) ? $content->getText() : null;
 
        if ( !is_string( $text ) ) {
            throw new RuntimeException( "Failed to load text for {$this->getPageIdentity()}" );
        }
 
        $this->text = $text;
 
        return $this->text;
    }
 
    public function getRevision(): ?int {
        return $this->revision;
    }
 
    public function getSourceLanguageCode(): string {
        return $this->getTitle()->getPageLanguage()->getCode();
    }
 
    public function getMessageGroupId(): string {
        return self::getMessageGroupIdFromTitle( $this->getPageIdentity() );
    }
 
    public static function getMessageGroupIdFromTitle( PageReference $page ): string {
        return 'page-' . MediaWikiServices::getInstance()->getTitleFormatter()->getPrefixedText( $page );
    }
 
    public function getMessageGroup(): ?WikiPageMessageGroup {
        $groupId = $this->getMessageGroupId();
        $group = MessageGroups::getGroup( $groupId );
        if ( !$group || $group instanceof WikiPageMessageGroup ) {
            return $group;
        }
 
        throw new RuntimeException(
            "Expected $groupId to be of type WikiPageMessageGroup; got " .
            get_class( $group )
        );
    }
 
    public function hasPageDisplayTitle(): bool {
        // Cached value
        if ( $this->pageDisplayTitle !== null ) {
            return $this->pageDisplayTitle;
        }
 
        // Check if title section exists in list of sections
        $factory = Services::getInstance()->getTranslationUnitStoreFactory();
        $store = $factory->getReader( $this->getPageIdentity() );
        $this->pageDisplayTitle = in_array( self::DISPLAY_TITLE_UNIT_ID, $store->getNames() );
 
        return $this->pageDisplayTitle;
    }
 
    public function getPageDisplayTitle( string $languageCode ): ?string {
        // Return null if title not marked for translation
        if ( !$this->hasPageDisplayTitle() ) {
            return null;
        }
 
        // Display title from DB
        $section = str_replace( ' ', '_', self::DISPLAY_TITLE_UNIT_ID );
        $page = MediaWikiServices::getInstance()->getTitleFormatter()->getPrefixedDBkey( $this->getPageIdentity() );
 
        try {
            $group = $this->getMessageGroup();
        } catch ( RuntimeException $e ) {
            return null;
        }
 
        // Sanity check, seems to happen during moves
        if ( !$group ) {
            return null;
        }
 
        return $group->getMessage( "$page/$section", $languageCode, IDBAccessObject::READ_NORMAL );
    }
 
    public function getStrippedSourcePageText(): string {
        $parser = Services::getInstance()->getTranslatablePageParser();
        $text = $parser->cleanupTags( $this->getText() );
        $text = preg_replace( '~<languages\s*/>\n?~s', '', $text );
 
        return $text;
    }
 
    public static function getTranslationPageFromTitle( Title $title ): ?TranslationPage {
        $self = self::isTranslationPage( $title );
        return $self ? $self->getTranslationPage( $self->targetLanguage ) : null;
    }
 
    public function getTranslationPage( string $targetLanguage ): TranslationPage {
        $mwServices = MediaWikiServices::getInstance();
        $config = $mwServices->getMainConfig();
        $services = Services::getInstance();
        $parser = $services->getTranslatablePageParser();
        $parserOutput = $parser->parse( $this->getText() );
        $pageVersion = (int)$services->getMessageGroupMetadata()
            ->get( $this->getMessageGroupId(), 'version' );
        $wrapUntranslated = $pageVersion >= 2;
        $languageFactory = $mwServices->getLanguageFactory();
 
        return new TranslationPage(
            $parserOutput,
            $this->getMessageGroup(),
            $languageFactory->getLanguage( $targetLanguage ),
            $languageFactory->getLanguage( $this->getSourceLanguageCode() ),
            $config->get( 'TranslateKeepOutdatedTranslations' ),
            $wrapUntranslated,
            $this->getTitle()
        );
    }
 
    public function addMarkedTag( int $revision, array $value = null ) {
        $this->revTagStore->replaceTag( $this->getPageIdentity(), RevTagStore::TP_MARK_TAG, $revision, $value );
        self::clearSourcePageCache();
    }
 
    public function addReadyTag( int $revision ): void {
        $this->revTagStore->replaceTag( $this->getPageIdentity(), RevTagStore::TP_READY_TAG, $revision );
        if ( !self::isSourcePage( $this->getPageIdentity() ) ) {
            self::clearSourcePageCache();
        }
    }
 
    public function getMarkedTag(): ?int {
        return $this->revTagStore->getLatestRevisionWithTag( $this->getPageIdentity(), RevTagStore::TP_MARK_TAG );
    }
 
    public function getReadyTag(): ?int {
        return $this->revTagStore->getLatestRevisionWithTag( $this->getPageIdentity(), RevTagStore::TP_READY_TAG );
    }
 
    public function getTranslationUrl( $code = false ): string {
        $params = [
            'group' => $this->getMessageGroupId(),
            'action' => 'page',
            'filter' => '',
            'language' => $code,
        ];
 
        $translate = SpecialPage::getTitleFor( 'Translate' );
 
        return $translate->getLocalURL( $params );
    }
 
    public function getMarkedRevs(): IResultWrapper {
        $db = Utilities::getSafeReadDB();
 
        return $db->newSelectQueryBuilder()
            ->select( [ 'rt_revision', 'rt_value' ] )
            ->from( 'revtag' )
            ->where( [
                'rt_page' => $this->getPageIdentity()->getId(),
                'rt_type' => RevTagStore::TP_MARK_TAG,
            ] )
            ->orderBy( 'rt_revision', SelectQueryBuilder::SORT_DESC )
            ->caller( __METHOD__ )
            ->fetchResultSet();
    }
 
    public function getTranslationPages(): array {
        $mwServices = MediaWikiServices::getInstance();
 
        $messageGroup = $this->getMessageGroup();
        $knownLanguageCodes = $messageGroup ? $messageGroup->getTranslatableLanguages() : null;
        $knownLanguageCodes ??= Utilities::getLanguageNames( LanguageNameUtils::AUTONYMS );
 
        $prefixedDbTitleKey = $this->getPageIdentity()->getDBkey() . '/';
        $baseNamespace = $this->getPageIdentity()->getNamespace();
 
        // Build a link batch query for all translation pages
        $linkBatch = $mwServices->getLinkBatchFactory()->newLinkBatch();
        foreach ( array_keys( $knownLanguageCodes ) as $code ) {
            $linkBatch->add( $baseNamespace, $prefixedDbTitleKey . $code );
        }
 
        $translationPages = [];
        foreach ( $linkBatch->getPageIdentities() as $pageIdentity ) {
            if ( $pageIdentity->exists() ) {
                $translationPages[] = Title::castFromPageIdentity( $pageIdentity );
            }
        }
 
        return $translationPages;
    }
 
    public function getTranslationUnitPages( ?string $code = null ): array {
        return $this->getTranslationUnitPagesByTitle( $this->title, $code );
    }
 
    public function getTranslationPercentages(): array {
        // Calculate percentages for the available translations
        try {
            $group = $this->getMessageGroup();
        } catch ( RuntimeException $e ) {
            return [];
        }
 
        if ( !$group ) {
            return [];
        }
 
        $titles = $this->getTranslationPages();
        $temp = MessageGroupStats::forGroup( $this->getMessageGroupId() );
        $stats = [];
 
        foreach ( $titles as $t ) {
            $handle = new MessageHandle( $t );
            $code = $handle->getCode();
 
            // Sometimes we want to display 0.00 for pages for which translation
            // hasn't started yet.
            $stats[$code] = 0.00;
            if ( ( $temp[$code][MessageGroupStats::TOTAL] ?? 0 ) > 0 ) {
                $total = $temp[$code][MessageGroupStats::TOTAL];
                $translated = $temp[$code][MessageGroupStats::TRANSLATED];
                $percentage = $translated / $total;
                $stats[$code] = sprintf( '%.2f', $percentage );
            }
        }
 
        // Content language is always up-to-date
        $stats[$this->getSourceLanguageCode()] = 1.00;
 
        return $stats;
    }
 
    public function getTransRev( string $suffix ) {
        $title = Title::makeTitle( NS_TRANSLATIONS, $suffix );
 
        $db = Utilities::getSafeReadDB();
 
        return $db->newSelectQueryBuilder()
            ->select( 'rt_value' )
            ->from( 'revtag' )
            ->where( [
                'rt_page' => $title->getArticleID(),
                'rt_type' => RevTagStore::TRANSVER_PROP,
            ] )
            ->orderBy( 'rt_revision', SelectQueryBuilder::SORT_DESC )
            ->caller( __METHOD__ )
            ->fetchField();
    }
 
    public function supportsTransclusion(): ?bool {
        $transclusion = Services::getInstance()
            ->getMessageGroupMetadata()
            ->get( $this->getMessageGroupId(), 'transclusion' );
        if ( $transclusion === false ) {
            return null;
        }
 
        return $transclusion === '1';
    }
 
    public function getRevisionRecordWithFallback(): ?RevisionRecord {
        $title = $this->getTitle();
        $store = MediaWikiServices::getInstance()->getRevisionStore();
        $revRecord = $store->getRevisionByTitle( $title->getSubpage( $this->targetLanguage ) );
        if ( $revRecord ) {
            return $revRecord;
        }
 
        // Fetch the source fallback
        return $store->getRevisionByTitle( $title->getSubpage( $this->getSourceLanguageCode() ) );
    }
 
    public function isMoveable(): bool {
        return $this->getMarkedTag() !== null;
    }
 
    public function isDeletable(): bool {
        return $this->getMarkedTag() !== null;
    }
 
    public static function isTranslationPage( Title $title ) {
        $handle = new MessageHandle( $title );
        if ( !Utilities::isTranslationPage( $handle ) ) {
            return false;
        }
 
        $languageCode = $handle->getCode();
        $newTitle = $handle->getTitleForBase();
 
        $page = self::newFromTitle( $newTitle );
 
        if ( $page->getMarkedTag() === null ) {
            return false;
        }
 
        $page->targetLanguage = $languageCode;
 
        return $page;
    }
 
    public static function parseTranslationUnit( LinkTarget $translationUnit ): array {
        // Format is Translations:SourcePageNamespace:SourcePageName/SectionName/LanguageCode.
        // We will drop the namespace immediately here.
        $parts = explode( '/', $translationUnit->getText() );
 
        // LanguageCode and SectionName are guaranteed to not have '/'.
        $language = array_pop( $parts );
        $section = array_pop( $parts );
        $sourcepage = implode( '/', $parts );
 
        return [
            'sourcepage' => $sourcepage,
            'section' => $section,
            'language' => $language
        ];
    }
 
    public static function isSourcePage( PageIdentity $page ): bool {
        if ( !$page->exists() ) {
            // No point in loading all translatable pages if the page
            // doesn’t exist. This also avoids PreconditionExceptions
            // if $page is a Title pointing to a non-proper page like
            // a special page.
            return false;
        }
 
        $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
        $cacheKey = $cache->makeKey( 'pagetranslation', 'sourcepages' );
 
        $translatablePageIds = $cache->getWithSetCallback(
            $cacheKey,
            $cache::TTL_HOUR * 2,
            static function ( $oldValue, &$ttl, array &$setOpts ) {
                $dbr = MediaWikiServices::getInstance()->getDBLoadBalancer()->getConnection( DB_REPLICA );
                $setOpts += Database::getCacheSetOptions( $dbr );
 
                return RevTagStore::getTranslatableBundleIds(
                    RevTagStore::TP_MARK_TAG, RevTagStore::TP_READY_TAG
                );
            },
            [
                'checkKeys' => [ $cacheKey ],
                'pcTTL' => $cache::TTL_PROC_SHORT,
                'pcGroup' => __CLASS__ . ':1',
                'version' => 2,
            ]
        );
 
        return isset( $translatablePageIds[$page->getId()] );
    }
 
    public static function clearSourcePageCache(): void {
        $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
        $cache->touchCheckKey( $cache->makeKey( 'pagetranslation', 'sourcepages' ) );
    }
 
    public static function determineStatus(
        ?int $readyRevisionId,
        ?int $markRevisionId,
        int $latestRevisionId
    ): ?TranslatablePageStatus {
        $status = null;
        if ( $markRevisionId === null ) {
            // Never marked, check that the latest version is ready
            if ( $readyRevisionId === $latestRevisionId ) {
                $status = TranslatablePageStatus::PROPOSED;
            } else {
                // Otherwise, ignore such pages
                return null;
            }
        } elseif ( $readyRevisionId === $latestRevisionId ) {
            if ( $markRevisionId === $readyRevisionId ) {
                // Marked and latest version is fine
                $status = TranslatablePageStatus::ACTIVE;
            } else {
                $status = TranslatablePageStatus::OUTDATED;
            }
        } else {
            // Marked but latest version is not fine
            $status = TranslatablePageStatus::BROKEN;
        }
 
        return new TranslatablePageStatus( $status );
    }
}
 
class_alias( TranslatablePage::class, 'TranslatablePage' );