MessageWebImporter.php

<?php
declare( strict_types = 1 );
 
namespace MediaWiki\Extension\Translate\Synchronization;
 
use ContentHandler;
use DifferenceEngine;
use Html;
use InvalidArgumentException;
use Language;
use MediaWiki\Extension\Translate\MessageGroupProcessing\MessageGroups;
use MediaWiki\Extension\Translate\SystemUsers\FuzzyBot;
use MediaWiki\Extension\Translate\Utilities\Utilities;
use MediaWiki\MediaWikiServices;
use MediaWiki\Revision\SlotRecord;
use MessageGroup;
use MessageHandle;
use RequestContext;
use RuntimeException;
use Sanitizer;
use Title;
use User;
use Xml;
 
class MessageWebImporter {
    private Title $title;
    private User $user;
    private MessageGroup $group;
    private string $code;
    private $time;
    private const MAX_PROCESSING_TIME = 43;
 
    public function __construct( Title $title, User $user, $group = null, string $code = 'en' ) {
        $this->setTitle( $title );
        $this->setUser( $user );
        $this->setGroup( $group );
        $this->setCode( $code );
    }
 
    public function getTitle(): Title {
        return $this->title;
    }
 
    public function setTitle( Title $title ): void {
        $this->title = $title;
    }
 
    public function getUser(): User {
        return $this->user;
    }
 
    public function setUser( User $user ): void {
        $this->user = $user;
    }
 
    public function getGroup(): MessageGroup {
        return $this->group;
    }
 
    public function setGroup( $group ): void {
        if ( $group instanceof MessageGroup ) {
            $this->group = $group;
        } else {
            $this->group = MessageGroups::getGroup( $group );
        }
    }
 
    public function getCode(): string {
        return $this->code;
    }
 
    public function setCode( string $code = 'en' ): void {
        $this->code = $code;
    }
 
    protected function getAction(): string {
        return $this->getTitle()->getLocalURL();
    }
 
    protected function doHeader(): string {
        $formParams = [
            'method' => 'post',
            'action' => $this->getAction(),
            'class' => 'mw-translate-manage'
        ];
 
        $csrfTokenSet = RequestContext::getMain()->getCsrfTokenSet();
        return Xml::openElement( 'form', $formParams ) .
            Html::hidden( 'title', $this->getTitle()->getPrefixedText() ) .
            Html::hidden( 'token', $csrfTokenSet->getToken() ) .
            Html::hidden( 'process', 1 );
    }
 
    protected function doFooter(): string {
        return '</form>';
    }
 
    protected function allowProcess(): bool {
        $context = RequestContext::getMain();
        $request = $context->getRequest();
        $csrfTokenSet = $context->getCsrfTokenSet();
 
        return $request->wasPosted()
            && $request->getBool( 'process' )
            && $csrfTokenSet->matchTokenField( 'token' );
    }
 
    protected function getActions(): array {
        return [
            'import',
            $this->code === 'en' ? 'fuzzy' : 'conflict',
            'ignore',
        ];
    }
 
    public function execute( array $messages ): bool {
        $context = RequestContext::getMain();
        $output = $context->getOutput();
 
        // Set up diff engine
        $diff = new DifferenceEngine();
        $diff->showDiffStyle();
        $diff->setReducedLineNumbers();
 
        // Check whether we do processing
        $process = $this->allowProcess();
 
        // Initialise collection
        $group = $this->getGroup();
        $code = $this->getCode();
        $collection = $group->initCollection( $code );
        $collection->loadTranslations();
 
        $output->addHTML( $this->doHeader() );
 
        // Initialise variable to keep track whether all changes were imported
        // or not. If we're allowed to process, initially assume they were.
        $allDone = $process;
 
        // Determine changes for each message.
        $changed = [];
 
        foreach ( $messages as $key => $value ) {
            $old = null;
            $isExistingMessageFuzzy = false;
 
            if ( isset( $collection[$key] ) ) {
                // This returns null if no existing translation is found
                $old = $collection[$key]->translation();
                $isExistingMessageFuzzy = $collection[$key]->hasTag( 'fuzzy' );
            }
 
            if ( $old === null ) {
                // We found a new translation for this message of the
                // current group: import it.
                if ( $process ) {
                    $action = 'import';
                    $this->doAction(
                        $action,
                        $group,
                        $key,
                        $value
                    );
                }
                // Show the user that we imported the new translation
                $para = '<code class="mw-tmi-new">' . htmlspecialchars( $key ) . '</code>';
                $name = $context->msg( 'translate-manage-import-new' )->rawParams( $para )
                    ->escaped();
                $text = Utilities::convertWhiteSpaceToHTML( $value );
                $changed[] = self::makeSectionElement( $name, 'new', $text );
            } else {
                // No changes at all, ignore
                if ( $old === (string)$value ) {
                    continue;
                }
 
                // Check if the message is already fuzzy in the system, and then determine if there are changes
                $oldTextForDiff = $old;
                if ( $isExistingMessageFuzzy ) {
                    if ( MessageHandle::makeFuzzyString( $old ) === (string)$value ) {
                        continue;
                    }
 
                    // Normalize the display of FUZZY message diffs so that if an old message has
                    // a fuzzy tag, then that is added to the text used in the diff.
                    $oldTextForDiff = MessageHandle::makeFuzzyString( $old );
                }
 
                $oldContent = ContentHandler::makeContent( $oldTextForDiff, $diff->getTitle() );
                $newContent = ContentHandler::makeContent( $value, $diff->getTitle() );
                $diff->setContent( $oldContent, $newContent );
                $text = $diff->getDiff( '', '' );
 
                // This is a changed translation. Note it for the next steps.
                $type = 'changed';
 
                // Get the user instructions for the current message,
                // submitted together with the form
                $action = $context->getRequest()
                    ->getVal( self::escapeNameForPHP( "action-$type-$key" ) );
 
                if ( $process ) {
                    if ( $changed === [] ) {
                        // Initialise the HTML list showing the changes performed
                        $changed[] = '<ul>';
                    }
 
                    if ( $action === null ) {
                        // We have been told to process the messages, but not
                        // what to do with this one. Tell the user.
                        $message = $context->msg(
                            'translate-manage-inconsistent',
                            wfEscapeWikiText( "action-$type-$key" )
                        )->parse();
                        $changed[] = "<li>$message</li></ul>";
 
                        // Also stop any further processing for the other messages.
                        $process = false;
                    } else {
                        // Check processing time
                        if ( !isset( $this->time ) ) {
                            $this->time = (int)wfTimestamp();
                        }
 
                        // We have all the necessary information on this changed
                        // translation: actually process the message
                        $messageKeyAndParams = $this->doAction(
                            $action,
                            $group,
                            $key,
                            $value
                        );
 
                        // Show what we just did, adding to the list of changes
                        $msgKey = array_shift( $messageKeyAndParams );
                        $params = $messageKeyAndParams;
                        $message = $context->msg( $msgKey, $params )->parse();
                        $changed[] = "<li>$message</li>";
 
                        // Stop processing further messages if too much time
                        // has been spent.
                        if ( $this->checkProcessTime() ) {
                            $process = false;
                            $message = $context->msg( 'translate-manage-toolong' )
                                ->numParams( self::MAX_PROCESSING_TIME )->parse();
                            $changed[] = "<li>$message</li></ul>";
                        }
 
                        continue;
                    }
                }
 
                // We are not processing messages, or no longer, or this was an
                // un-actionable translation. We will eventually return false
                $allDone = false;
 
                // Prepare to ask the user what to do with this message
                $actions = $this->getActions();
                $defaultAction = $action ?: 'import';
 
                $act = [];
 
                // Give grep a chance to find the usages:
                // translate-manage-action-import, translate-manage-action-conflict,
                // translate-manage-action-ignore, translate-manage-action-fuzzy
                foreach ( $actions as $action ) {
                    $label = $context->msg( "translate-manage-action-$action" )->text();
                    $name = self::escapeNameForPHP( "action-$type-$key" );
                    $id = Sanitizer::escapeIdForAttribute( "action-$key-$action" );
                    $act[] = Xml::radioLabel( $label, $name, $action, $id, $action === $defaultAction );
                }
 
                $param = '<code class="mw-tmi-diff">' . htmlspecialchars( $key ) . '</code>';
                $name = $context->msg( 'translate-manage-import-diff' )
                    ->rawParams( $param, implode( ' ', $act ) )
                    ->escaped();
 
                $changed[] = self::makeSectionElement( $name, $type, $text );
            }
        }
 
        if ( !$process ) {
            $collection->filter( 'hastranslation', false );
            $keys = $collection->getMessageKeys();
 
            $diff = array_diff( $keys, array_keys( $messages ) );
 
            foreach ( $diff as $s ) {
                $para = '<code class="mw-tmi-deleted">' . htmlspecialchars( $s ) . '</code>';
                $name = $context->msg( 'translate-manage-import-deleted' )->rawParams( $para )->escaped();
                $text = Utilities::convertWhiteSpaceToHTML( $collection[$s]->translation() );
                $changed[] = self::makeSectionElement( $name, 'deleted', $text );
            }
        }
 
        if ( $process || ( $changed === [] && $code !== 'en' ) ) {
            if ( $changed === [] ) {
                $output->addWikiMsg( 'translate-manage-nochanges-other' );
            }
 
            if ( $changed === [] || !str_starts_with( end( $changed ), '<li>' ) ) {
                $changed[] = '<ul>';
            }
 
            $changed[] = '</ul>';
 
            $languageName = Utilities::getLanguageName( $code, $context->getLanguage()->getCode() );
            $message = $context
                ->msg( 'translate-manage-import-done', $group->getId(), $group->getLabel(), $languageName )
                ->parse();
            $changed[] = Html::successBox( $message );
            $output->addHTML( implode( "\n", $changed ) );
        } else {
            // END
            if ( $changed !== [] ) {
                if ( $code === 'en' ) {
                    $output->addWikiMsg( 'translate-manage-intro-en' );
                } else {
                    $lang = Utilities::getLanguageName(
                        $code,
                        $context->getLanguage()->getCode()
                    );
                    $output->addWikiMsg( 'translate-manage-intro-other', $lang );
                }
                $output->addHTML( Html::hidden( 'language', $code ) );
                $output->addHTML( implode( "\n", $changed ) );
                $output->addHTML( Xml::submitButton( $context->msg( 'translate-manage-submit' )->text() ) );
            } else {
                $output->addWikiMsg( 'translate-manage-nochanges' );
            }
        }
 
        $output->addHTML( $this->doFooter() );
 
        return $allDone;
    }
 
    private function doAction(
        string $action,
        MessageGroup $group,
        string $key,
        string $message
    ): array {
        global $wgTranslateDocumentationLanguageCode;
 
        $comment = '';
        $code = $this->getCode();
        $title = $this->makeTranslationTitle( $group, $key, $code );
 
        if ( $action === 'import' || $action === 'conflict' ) {
            if ( $action === 'import' ) {
                $comment = wfMessage( 'translate-manage-import-summary' )->inContentLanguage()->plain();
            } else {
                $comment = wfMessage( 'translate-manage-conflict-summary' )->inContentLanguage()->plain();
                $message = MessageHandle::makeFuzzyString( $message );
            }
 
            return self::doImport( $title, $message, $comment, $this->getUser() );
        } elseif ( $action === 'ignore' ) {
            return [ 'translate-manage-import-ignore', $key ];
        } elseif ( $action === 'fuzzy' && $code !== 'en' &&
            $code !== $wgTranslateDocumentationLanguageCode
        ) {
            $message = MessageHandle::makeFuzzyString( $message );
 
            return self::doImport( $title, $message, $comment, $this->getUser() );
        } elseif ( $action === 'fuzzy' && $code === 'en' ) {
            return self::doFuzzy( $title, $message, $comment, $this->getUser() );
        } else {
            throw new InvalidArgumentException( "Unhandled action $action" );
        }
    }
 
    protected function checkProcessTime() {
        return (int)wfTimestamp() - $this->time >= self::MAX_PROCESSING_TIME;
    }
 
    private static function doImport(
        Title $title,
        string $message,
        string $summary,
        User $user
    ): array {
        $wikiPage = MediaWikiServices::getInstance()->getWikiPageFactory()->newFromTitle( $title );
        $content = ContentHandler::makeContent( $message, $title );
        $status = $wikiPage->doUserEditContent(
            $content,
            $user,
            $summary
        );
        $success = $status->isOK();
 
        if ( $success ) {
            return [ 'translate-manage-import-ok',
                wfEscapeWikiText( $title->getPrefixedText() )
            ];
        }
 
        $text = "Failed to import new version of page {$title->getPrefixedText()}\n";
        $text .= $status->getWikiText();
        throw new RuntimeException( $text );
    }
 
    public static function doFuzzy(
        Title $title,
        string $message,
        string $comment,
        ?User $user
    ): array {
        $context = RequestContext::getMain();
        $services = MediaWikiServices::getInstance();
 
        if ( !$context->getUser()->isAllowed( 'translate-manage' ) ) {
            return [ 'badaccess-group0' ];
        }
 
        // Edit with fuzzybot if there is no user.
        if ( !$user ) {
            $user = FuzzyBot::getUser();
        }
 
        // Work on all subpages of base title.
        $handle = new MessageHandle( $title );
        $titleText = $handle->getKey();
 
        $revStore = $services->getRevisionStore();
        $queryInfo = $revStore->getQueryInfo( [ 'page' ] );
        $dbw = $services->getDBLoadBalancer()->getConnection( DB_PRIMARY );
        $rows = $dbw->select(
            $queryInfo['tables'],
            $queryInfo['fields'],
            [
                'page_namespace' => $title->getNamespace(),
                'page_latest=rev_id',
                'page_title' . $dbw->buildLike( "$titleText/", $dbw->anyString() ),
            ],
            __METHOD__,
            [],
            $queryInfo['joins']
        );
 
        $changed = [];
        $slots = $revStore->getContentBlobsForBatch( $rows, [ SlotRecord::MAIN ] )->getValue();
 
        foreach ( $rows as $row ) {
            global $wgTranslateDocumentationLanguageCode;
 
            $translationTitle = Title::makeTitle( (int)$row->page_namespace, $row->page_title );
 
            // No fuzzy for English original or documentation language code.
            if ( $translationTitle->getSubpageText() === 'en' ||
                $translationTitle->getSubpageText() === $wgTranslateDocumentationLanguageCode
            ) {
                // Use imported text, not database text.
                $text = $message;
            } elseif ( isset( $slots[$row->rev_id] ) ) {
                $slot = $slots[$row->rev_id][SlotRecord::MAIN];
                $text = MessageHandle::makeFuzzyString( $slot->blob_data );
            } else {
                $text = MessageHandle::makeFuzzyString(
                    Utilities::getTextFromTextContent(
                        $revStore->newRevisionFromRow( $row )->getContent( SlotRecord::MAIN )
                    )
                );
            }
 
            // Do actual import
            $changed[] = self::doImport(
                $translationTitle,
                $text,
                $comment,
                $user
            );
        }
 
        // Format return text
        $text = '';
        foreach ( $changed as $c ) {
            $key = array_shift( $c );
            $text .= '* ' . $context->msg( $key, $c )->plain() . "\n";
        }
 
        return [ 'translate-manage-import-fuzzy', "\n" . $text ];
    }
 
    private function makeTranslationTitle( MessageGroup $group, string $key, string $code ): Title {
        $ns = $group->getNamespace();
 
        return Title::makeTitleSafe( $ns, "$key/$code" );
    }
 
    public static function makeSectionElement(
        string $legend,
        string $type,
        string $content,
        Language $lang = null
    ): string {
        $containerParams = [ 'class' => "mw-tpt-sp-section mw-tpt-sp-section-type-{$type}" ];
        $legendParams = [ 'class' => 'mw-tpt-sp-legend' ];
        $contentParams = [ 'class' => 'mw-tpt-sp-content' ];
        if ( $lang ) {
            $contentParams['dir'] = $lang->getDir();
            $contentParams['lang'] = $lang->getCode();
        }
 
        return Html::rawElement( 'div', $containerParams,
            Html::rawElement( 'div', $legendParams, $legend ) .
                Html::rawElement( 'div', $contentParams, $content )
        );
    }
 
    private static function escapeNameForPHP( string $name ): string {
        $replacements = [
            '(' => '(OP)',
            ' ' => '(SP)',
            "\t" => '(TAB)',
            '.' => '(DOT)',
            "'" => '(SQ)',
            "\"" => '(DQ)',
            '%' => '(PC)',
            '&' => '(AMP)',
        ];
 
        return strtr( $name, $replacements );
    }
}