MessageCollection.php

Go to the documentation of this file.

<?php
use MediaWiki\Extension\Translate\MessageGroupProcessing\RevTagStore;
use MediaWiki\Extension\Translate\SystemUsers\FuzzyBot;
use MediaWiki\MediaWikiServices;
use MediaWiki\Revision\RevisionRecord;
use MediaWiki\Revision\SlotRecord;
 
class MessageCollection implements ArrayAccess, Iterator, Countable {
    private const MAX_ITEMS_PER_QUERY = 2000;
 
    public $code;
    private $definitions = null;
    private $infile = [];
    // Keys and messages.
 
    protected $keys = [];
    protected $messages = [];
    private $reverseMap;
    // Database resources
 
    private $dbInfo;
    private $dbData;
    private $dbReviewData;
    protected $tags = [];
    private $properties = [];
    private $authors = [];
 
    public function __construct( $code ) {
        $this->code = $code;
    }
 
    public static function newFromDefinitions( MessageDefinitions $definitions, $code ) {
        $collection = new self( $code );
        $collection->definitions = $definitions;
        $collection->resetForNewLanguage( $code );
 
        return $collection;
    }
 
    public function getLanguage() {
        return $this->code;
    }
 
    // Data setters
 
    public function setInFile( array $messages ) {
        $this->infile = $messages;
    }
 
    public function setTags( $type, array $keys ) {
        $this->tags[$type] = $keys;
    }
 
    public function keys() {
        return $this->keys;
    }
 
    public function getTitles() {
        return array_values( $this->keys );
    }
 
    public function getMessageKeys() {
        return array_keys( $this->keys );
    }
 
    public function getTags( $type ) {
        return $this->tags[$type] ?? [];
    }
 
    public function getAuthors() {
        $this->loadTranslations();
 
        $authors = array_flip( $this->authors );
 
        foreach ( $this->messages as $m ) {
            // Check if there are authors
            $author = $m->getProperty( 'last-translator-text' );
 
            if ( $author === null ) {
                continue;
            }
 
            if ( !isset( $authors[$author] ) ) {
                $authors[$author] = 1;
            } else {
                $authors[$author]++;
            }
        }
 
        # arsort( $authors, SORT_NUMERIC );
        ksort( $authors );
        $fuzzyBot = FuzzyBot::getName();
        $filteredAuthors = [];
        foreach ( $authors as $author => $edits ) {
            if ( $author !== $fuzzyBot ) {
                $filteredAuthors[] = $author;
            }
        }
 
        return $filteredAuthors;
    }
 
    public function addCollectionAuthors( $authors, $mode = 'append' ) {
        switch ( $mode ) {
            case 'append':
                $authors = array_merge( $this->authors, $authors );
                break;
            case 'set':
                break;
            default:
                throw new MWException( "Invalid mode $mode" );
        }
 
        $this->authors = array_unique( $authors );
    }
 
    // Data modifiers
 
    public function loadTranslations() {
        // Performance optimization: Instead of building conditions based on key in every
        // method, build them once and pass it on to each of them.
        $dbr = TranslateUtils::getSafeReadDB();
        $titleConds = $this->getTitleConds( $dbr );
 
        $this->loadData( $this->keys, $titleConds );
        $this->loadInfo( $this->keys, $titleConds );
        $this->loadReviewInfo( $this->keys, $titleConds );
        $this->initMessages();
    }
 
    public function resetForNewLanguage( $code ) {
        $this->code = $code;
        $this->keys = $this->fixKeys();
        $this->dbInfo = [];
        $this->dbData = [];
        $this->dbReviewData = [];
        $this->messages = null;
        $this->infile = [];
        $this->authors = [];
 
        unset( $this->tags['fuzzy'] );
        $this->reverseMap = null;
    }
 
    public function slice( $offset, $limit ) {
        $indexes = array_keys( $this->keys );
 
        if ( $offset === '' ) {
            $offset = 0;
        }
 
        // Handle string offsets
        if ( !ctype_digit( (string)$offset ) ) {
            $pos = array_search( $offset, array_keys( $this->keys ), true );
            // Now offset is always an integer, suitable for array_slice
            $offset = $pos !== false ? $pos : count( $this->keys );
        }
 
        // False means that cannot go back or forward
        $backwardsOffset = $forwardsOffset = false;
        // Backwards paging uses numerical indexes, see below
 
        // Can only skip this if no offset has been provided or the
        // offset is zero. (offset - limit ) > 1 does not work, because
        // users can end in offest=2, limit=5 and can't see the first
        // two messages. That's also why it is capped into zero with
        // max(). And finally make the offsets to be strings even if
        // they are numbers in this case.
        if ( $offset > 0 ) {
            $backwardsOffset = (string)( max( 0, $offset - $limit ) );
        }
 
        // Forwards paging uses keys. If user opens view Untranslated,
        // translates some messages and then clicks next, the first
        // message visible in the page is the first message not shown
        // in the previous page (unless someone else translated it at
        // the same time). If we used integer offsets, we would skip
        // same number of messages that were translated, because they
        // are no longer in the list. For backwards paging this is not
        // such a big issue, so it still uses integer offsets, because
        // we would need to also implement "direction" to have it work
        // correctly.
        if ( isset( $indexes[$offset + $limit] ) ) {
            $forwardsOffset = $indexes[$offset + $limit];
        }
 
        $this->keys = array_slice( $this->keys, $offset, $limit, true );
 
        return [ $backwardsOffset, $forwardsOffset, $offset ];
    }
 
    public function filter( $type, $condition = true, $value = null ) {
        if ( !in_array( $type, self::getAvailableFilters(), true ) ) {
            throw new MWException( "Unknown filter $type" );
        }
        $this->applyFilter( $type, $condition, $value );
    }
 
    public static function getAvailableFilters() {
        return [
            'fuzzy',
            'optional',
            'ignored',
            'hastranslation',
            'changed',
            'translated',
            'reviewer',
            'last-translator',
        ];
    }
 
    protected function applyFilter( $filter, $condition, $value ) {
        $keys = $this->keys;
        if ( $filter === 'fuzzy' ) {
            $keys = $this->filterFuzzy( $keys, $condition );
        } elseif ( $filter === 'hastranslation' ) {
            $keys = $this->filterHastranslation( $keys, $condition );
        } elseif ( $filter === 'translated' ) {
            $fuzzy = $this->filterFuzzy( $keys, false );
            $hastranslation = $this->filterHastranslation( $keys, false );
            // Fuzzy messages are not counted as translated messages
            $translated = $this->filterOnCondition( $hastranslation, $fuzzy );
            $keys = $this->filterOnCondition( $keys, $translated, $condition );
        } elseif ( $filter === 'changed' ) {
            $keys = $this->filterChanged( $keys, $condition );
        } elseif ( $filter === 'reviewer' ) {
            $keys = $this->filterReviewer( $keys, $condition, $value );
        } elseif ( $filter === 'last-translator' ) {
            $keys = $this->filterLastTranslator( $keys, $condition, $value );
        } else {
            // Filter based on tags.
            if ( !isset( $this->tags[$filter] ) ) {
                if ( $filter !== 'optional' && $filter !== 'ignored' ) {
                    throw new MWException( "No tagged messages for custom filter $filter" );
                }
                $keys = $this->filterOnCondition( $keys, [], $condition );
            } else {
                $taggedKeys = array_flip( $this->tags[$filter] );
                $keys = $this->filterOnCondition( $keys, $taggedKeys, $condition );
            }
        }
 
        $this->keys = $keys;
    }
 
    public function filterUntranslatedOptional(): void {
        $optionalKeys = array_flip( $this->tags['optional'] ?? [] );
        // Convert plain message keys to array<string,TitleValue>
        $optional = $this->filterOnCondition( $this->keys, $optionalKeys, false );
        // Then get reduce that list to those which have no translation. Ensure we don't
        // accidentally populate the info cache with too few keys.
        $this->loadInfo( $this->keys );
        $untranslatedOptional = $this->filterHastranslation( $optional, true );
        // Now remove that list from the full list
        $this->keys = $this->filterOnCondition( $this->keys, $untranslatedOptional );
    }
 
    protected function filterOnCondition( array $keys, array $condKeys, $condition = true ) {
        if ( $condition === true ) {
            // Delete $condKeys from $keys
            foreach ( array_keys( $condKeys ) as $key ) {
                unset( $keys[$key] );
            }
        } else {
            // Keep the keys which are in $condKeys
            foreach ( array_keys( $keys ) as $key ) {
                if ( !isset( $condKeys[$key] ) ) {
                    unset( $keys[$key] );
                }
            }
        }
 
        return $keys;
    }
 
    protected function filterFuzzy( array $keys, $condition ) {
        $this->loadInfo( $keys );
 
        $origKeys = [];
        if ( $condition === false ) {
            $origKeys = $keys;
        }
 
        foreach ( $this->dbInfo as $row ) {
            if ( $row->rt_type !== null ) {
                unset( $keys[$this->rowToKey( $row )] );
            }
        }
 
        if ( $condition === false ) {
            $keys = array_diff( $origKeys, $keys );
        }
 
        return $keys;
    }
 
    protected function filterHastranslation( array $keys, $condition ) {
        $this->loadInfo( $keys );
 
        $origKeys = [];
        if ( $condition === false ) {
            $origKeys = $keys;
        }
 
        foreach ( $this->dbInfo as $row ) {
            unset( $keys[$this->rowToKey( $row )] );
        }
 
        // Check also if there is something in the file that is not yet in the database
        foreach ( array_keys( $this->infile ) as $inf ) {
            unset( $keys[$inf] );
        }
 
        // Remove the messages which do not have a translation from the list
        if ( $condition === false ) {
            $keys = array_diff( $origKeys, $keys );
        }
 
        return $keys;
    }
 
    protected function filterChanged( array $keys, $condition ) {
        $this->loadData( $keys );
 
        $origKeys = [];
        if ( $condition === false ) {
            $origKeys = $keys;
        }
 
        $revStore = MediaWikiServices::getInstance()->getRevisionStore();
        $infileRows = [];
        foreach ( $this->dbData as $row ) {
            $mkey = $this->rowToKey( $row );
            if ( isset( $this->infile[$mkey] ) ) {
                $infileRows[] = $row;
            }
        }
 
        $revisions = $revStore->newRevisionsFromBatch( $infileRows, [
            'slots' => [ SlotRecord::MAIN ],
            'content' => true
        ] )->getValue();
        foreach ( $infileRows as $row ) {
            $rev = $revisions[$row->rev_id];
            if ( $rev ) {
                $content = $rev->getContent( SlotRecord::MAIN );
                if ( $content ) {
                    $mkey = $this->rowToKey( $row );
                    if ( $this->infile[$mkey] === $content->getText() ) {
                        // Remove unchanged messages from the list
                        unset( $keys[$mkey] );
                    }
                }
            }
        }
 
        // Remove the messages which have changed from the original list
        if ( $condition === false ) {
            $keys = $this->filterOnCondition( $origKeys, $keys );
        }
 
        return $keys;
    }
 
    protected function filterReviewer( array $keys, $condition, $user ) {
        $this->loadReviewInfo( $keys );
        $origKeys = $keys;
 
        /* This removes messages from the list which have certain
         * reviewer (among others) */
        $userId = (int)$user;
        foreach ( $this->dbReviewData as $row ) {
            if ( $user === null || (int)$row->trr_user === $userId ) {
                unset( $keys[$this->rowToKey( $row )] );
            }
        }
 
        if ( $condition === false ) {
            $keys = array_diff( $origKeys, $keys );
        }
 
        return $keys;
    }
 
    protected function filterLastTranslator( array $keys, $condition, $user ) {
        $this->loadData( $keys );
        $origKeys = $keys;
 
        $user = (int)$user;
        foreach ( $this->dbData as $row ) {
            if ( (int)$row->rev_user === $user ) {
                unset( $keys[$this->rowToKey( $row )] );
            }
        }
 
        if ( $condition === false ) {
            $keys = array_diff( $origKeys, $keys );
        }
 
        return $keys;
    }
 
    protected function fixKeys() {
        $newkeys = [];
 
        $pages = $this->definitions->getPages();
        foreach ( $pages as $key => $baseTitle ) {
            $newkeys[$key] = new TitleValue(
                $baseTitle->getNamespace(),
                $baseTitle->getDBkey() . '/' . $this->code
            );
        }
 
        return $newkeys;
    }
 
    protected function loadInfo( array $keys, ?array $titleConds = null ) {
        if ( $this->dbInfo !== [] ) {
            return;
        }
 
        if ( !count( $keys ) ) {
            $this->dbInfo = new EmptyIterator();
            return;
        }
 
        $dbr = TranslateUtils::getSafeReadDB();
        $tables = [ 'page', 'revtag' ];
        $fields = [ 'page_namespace', 'page_title', 'rt_type' ];
        $joins = [ 'revtag' =>
        [
            'LEFT JOIN',
            [ 'page_id=rt_page', 'page_latest=rt_revision', 'rt_type' => RevTagStore::FUZZY_TAG ]
        ]
        ];
 
        $titleConds = $titleConds ?? $this->getTitleConds( $dbr );
        $iterator = new AppendIterator();
        foreach ( $titleConds as $conds ) {
            $iterator->append( $dbr->select( $tables, $fields, $conds, __METHOD__, [], $joins ) );
        }
 
        $this->dbInfo = $iterator;
 
        // Populate and cache reverse map now, since if call to initMesages is delayed (e.g. a
        // filter that calls loadData() is used, or ::slice is used) the reverse map will not
        // contain all the entries that are present in our $iterator and will throw notices.
        $this->getReverseMap();
    }
 
    protected function loadReviewInfo( array $keys, ?array $titleConds = null ) {
        if ( $this->dbReviewData !== [] ) {
            return;
        }
 
        if ( !count( $keys ) ) {
            $this->dbReviewData = new EmptyIterator();
            return;
        }
 
        $dbr = TranslateUtils::getSafeReadDB();
        $tables = [ 'page', 'translate_reviews' ];
        $fields = [ 'page_namespace', 'page_title', 'trr_user' ];
        $joins = [ 'translate_reviews' =>
            [
                'JOIN',
                [ 'page_id=trr_page', 'page_latest=trr_revision' ]
            ]
        ];
 
        $titleConds = $titleConds ?? $this->getTitleConds( $dbr );
        $iterator = new AppendIterator();
        foreach ( $titleConds as $conds ) {
            $iterator->append( $dbr->select( $tables, $fields, $conds, __METHOD__, [], $joins ) );
        }
 
        $this->dbReviewData = $iterator;
 
        // Populate and cache reverse map now, since if call to initMesages is delayed (e.g. a
        // filter that calls loadData() is used, or ::slice is used) the reverse map will not
        // contain all the entries that are present in our $iterator and will throw notices.
        $this->getReverseMap();
    }
 
    protected function loadData( array $keys, ?array $titleConds = null ) {
        if ( $this->dbData !== [] ) {
            return;
        }
 
        if ( !count( $keys ) ) {
            $this->dbData = new EmptyIterator();
            return;
        }
 
        $dbr = TranslateUtils::getSafeReadDB();
        $revisionStore = MediaWikiServices::getInstance()->getRevisionStore();
        $revQuery = $revisionStore->getQueryInfo( [ 'page' ] );
        $tables = $revQuery['tables'];
        $fields = $revQuery['fields'];
        $joins = $revQuery['joins'];
 
        $titleConds = $titleConds ?? $this->getTitleConds( $dbr );
        $iterator = new AppendIterator();
        foreach ( $titleConds as $conds ) {
            $conds = [ 'page_latest = rev_id', $conds ];
            $iterator->append( $dbr->select( $tables, $fields, $conds, __METHOD__, [], $joins ) );
        }
 
        $this->dbData = $iterator;
 
        // Populate and cache reverse map now, since if call to initMesages is delayed (e.g. a
        // filter that calls loadData() is used, or ::slice is used) the reverse map will not
        // contain all the entries that are present in our $iterator and will throw notices.
        $this->getReverseMap();
    }
 
    protected function getTitleConds( $db ) {
        $titles = $this->getTitles();
        $chunks = array_chunk( $titles, self::MAX_ITEMS_PER_QUERY );
        $results = [];
 
        foreach ( $chunks as $titles ) {
            // Array of array( namespace, pagename )
            $byNamespace = [];
            foreach ( $titles as $title ) {
                $namespace = $title->getNamespace();
                $pagename = $title->getDBkey();
                $byNamespace[$namespace][] = $pagename;
            }
 
            $conds = [];
            foreach ( $byNamespace as $namespaces => $pagenames ) {
                $cond = [
                    'page_namespace' => $namespaces,
                    'page_title' => $pagenames,
                ];
 
                $conds[] = $db->makeList( $cond, LIST_AND );
            }
 
            $results[] = $db->makeList( $conds, LIST_OR );
        }
 
        return $results;
    }
 
    protected function rowToKey( $row ) {
        $map = $this->getReverseMap();
        if ( isset( $map[$row->page_namespace][$row->page_title] ) ) {
            return $map[$row->page_namespace][$row->page_title];
        } else {
            wfWarn( "Got unknown title from the database: {$row->page_namespace}:{$row->page_title}" );
 
            return null;
        }
    }
 
    public function getReverseMap() {
        if ( isset( $this->reverseMap ) ) {
            return $this->reverseMap;
        }
 
        $map = [];
        foreach ( $this->keys as $mkey => $title ) {
            $map[$title->getNamespace()][$title->getDBkey()] = $mkey;
        }
 
        $this->reverseMap = $map;
        return $this->reverseMap;
    }
 
    public function initMessages() {
        if ( $this->messages !== null ) {
            return;
        }
 
        $messages = [];
        $definitions = $this->definitions->getDefinitions();
        $revStore = MediaWikiServices::getInstance()->getRevisionStore();
        $queryFlags = TranslateUtils::shouldReadFromPrimary() ? $revStore::READ_LATEST : 0;
        foreach ( array_keys( $this->keys ) as $mkey ) {
            $messages[$mkey] = new ThinMessage( $mkey, $definitions[$mkey] );
        }
 
        if ( $this->dbData !== null ) {
            $slotRows = $revStore->getContentBlobsForBatch(
                $this->dbData, [ SlotRecord::MAIN ], $queryFlags
            )->getValue();
 
            foreach ( $this->dbData as $row ) {
                $mkey = $this->rowToKey( $row );
                if ( !isset( $messages[$mkey] ) ) {
                    continue;
                }
                $messages[$mkey]->setRow( $row );
                $messages[$mkey]->setProperty( 'revision', $row->page_latest );
 
                if ( isset( $slotRows[$row->rev_id][SlotRecord::MAIN] ) ) {
                    $slot = $slotRows[$row->rev_id][SlotRecord::MAIN];
                    $messages[$mkey]->setTranslation( $slot->blob_data );
                }
            }
        }
 
        if ( $this->dbInfo !== null ) {
            $fuzzy = [];
            foreach ( $this->dbInfo as $row ) {
                if ( $row->rt_type !== null ) {
                    $fuzzy[] = $this->rowToKey( $row );
                }
            }
 
            $this->setTags( 'fuzzy', $fuzzy );
        }
 
        // Copy tags if any.
        foreach ( $this->tags as $type => $keys ) {
            foreach ( $keys as $mkey ) {
                if ( isset( $messages[$mkey] ) ) {
                    $messages[$mkey]->addTag( $type );
                }
            }
        }
 
        // Copy properties if any.
        foreach ( $this->properties as $type => $keys ) {
            foreach ( $keys as $mkey => $value ) {
                if ( isset( $messages[$mkey] ) ) {
                    $messages[$mkey]->setProperty( $type, $value );
                }
            }
        }
 
        // Copy infile if any.
        foreach ( $this->infile as $mkey => $value ) {
            if ( isset( $messages[$mkey] ) ) {
                $messages[$mkey]->setInfile( $value );
            }
        }
 
        foreach ( $this->dbReviewData as $row ) {
            $mkey = $this->rowToKey( $row );
            if ( !isset( $messages[$mkey] ) ) {
                continue;
            }
            $messages[$mkey]->appendProperty( 'reviewers', $row->trr_user );
        }
 
        // Set the status property
        foreach ( $messages as $obj ) {
            if ( $obj->hasTag( 'fuzzy' ) ) {
                $obj->setProperty( 'status', 'fuzzy' );
            } elseif ( is_array( $obj->getProperty( 'reviewers' ) ) ) {
                $obj->setProperty( 'status', 'proofread' );
            } elseif ( $obj->translation() !== null ) {
                $obj->setProperty( 'status', 'translated' );
            } else {
                $obj->setProperty( 'status', 'untranslated' );
            }
        }
 
        $this->messages = $messages;
    }
 
    public function offsetExists( $offset ): bool {
        return isset( $this->keys[$offset] );
    }
 
    public function offsetGet( $offset ): ?TMessage {
        return $this->messages[$offset] ?? null;
    }
 
    public function offsetSet( $offset, $value ): void {
        $this->messages[$offset] = $value;
    }
 
    public function offsetUnset( $offset ): void {
        unset( $this->keys[$offset] );
    }
 
    public function __get( $name ) {
        throw new MWException( __METHOD__ . ": Trying to access unknown property $name" );
    }
 
    public function __set( $name, $value ) {
        throw new MWException( __METHOD__ . ": Trying to modify unknown property $name" );
    }
 
    public function rewind(): void {
        reset( $this->keys );
    }
 
    #[\ReturnTypeWillChange]
    public function current() {
        if ( !count( $this->keys ) ) {
            return false;
        }
 
        // @phan-suppress-next-line PhanTypeArraySuspiciousNullable
        return $this->messages[key( $this->keys )];
    }
 
    public function key(): ?string {
        return key( $this->keys );
    }
 
    public function next(): void {
        next( $this->keys );
    }
 
    public function valid(): bool {
        return isset( $this->messages[key( $this->keys )] );
    }
 
    public function count(): int {
        return count( $this->keys() );
    }
 
}
 
class MessageDefinitions {
    private $namespace;
    private $messages;
    private $pages;
 
    public function __construct( array $messages, $namespace = false ) {
        $this->namespace = $namespace;
        $this->messages = $messages;
    }
 
    public function getDefinitions() {
        return $this->messages;
    }
 
    public function getPages() {
        $namespace = $this->namespace;
        if ( $this->pages !== null ) {
            return $this->pages;
        }
 
        $pages = [];
        foreach ( array_keys( $this->messages ) as $key ) {
            if ( $namespace === false ) {
                // pages are in format ex. "8:jan"
                [ $tns, $tkey ] = explode( ':', $key, 2 );
                $title = Title::makeTitleSafe( $tns, $tkey );
            } else {
                $title = Title::makeTitleSafe( $namespace, $key );
            }
 
            if ( !$title ) {
                wfWarn( "Invalid title ($namespace:)$key" );
                continue;
            }
 
            $pages[$key] = $title;
        }
 
        $this->pages = $pages;
 
        return $this->pages;
    }
}