.singleton("tikaExtractor", TikaExtractor.class);
        textTransformer = container.getComponent("textTransformer");
        textTransformer.setName("textTransformer");
        ExtractorFactory extractorFactory = container.getComponent("extractorFactory");
        TikaExtractor tikaExtractor = container.getComponent("tikaExtractor");
        extractorFactory.addExtractor("text/plain", tikaExtractor);

        sitemap.setLoc("https://example.com/page1");

        SitemapSet sitemapSet = new SitemapSet();
        sitemapSet.addSitemap(sitemap);

        when(crawlerContainer.getComponent("sitemapsHelper")).thenReturn(sitemapsHelper);
        try {
            when(sitemapsHelper.parse(any(InputStream.class))).thenReturn(sitemapSet);
        } catch (Exception e) {

        final CrawlingInfoService crawlingInfoService = ComponentUtil.getComponent(CrawlingInfoService.class);
        final SearchLogService searchLogService = ComponentUtil.getComponent(SearchLogService.class);
        final JobLogService jobLogService = ComponentUtil.getComponent(JobLogService.class);
        final UserInfoService userInfoService = ComponentUtil.getComponent(UserInfoService.class);

                .<ExtractorFactory> singleton("extractorFactory", ExtractorFactory.class, factory -> {
                    TikaExtractor tikaExtractor = container.getComponent("tikaExtractor");
                    LhaExtractor lhaExtractor = container.getComponent("lhaExtractor");
                    factory.addExtractor("text/plain", tikaExtractor);
                    factory.addExtractor("text/html", tikaExtractor);

        final OpenSearchUrlFilterService urlFilterService = ComponentUtil.getComponent(OpenSearchUrlFilterService.class);
        final OpenSearchUrlQueueService urlQueueService = ComponentUtil.getComponent(OpenSearchUrlQueueService.class);
        final OpenSearchDataService dataService = ComponentUtil.getComponent(OpenSearchDataService.class);

        try {
            // clear url filter

        ruleManager = container.getComponent("ruleManager");

        SitemapsRule sitemapsRule = container.getComponent("sitemapsRule");
        sitemapsRule.setRuleId("sitemapsRule");
        sitemapsRule.addRule("url", ".*sitemap.*");
        ruleManager.addRule(sitemapsRule);

        RegexRule fileRule = container.getComponent("fileRule");
        fileRule.setRuleId("fileRule");

     * Creates a new instance of LastaCrawlerContainer.
     */
    public LastaCrawlerContainer() {
        initialize();
    }

    @Override
    public <T> T getComponent(final String name) {
        return SingletonLaContainerFactory.getContainer().getComponent(name);
    }

    @Override
    public boolean available() {
        return SingletonLaContainerFactory.hasContainer();
    }

    @Override

                .<ExtractorFactory> singleton("extractorFactory", ExtractorFactory.class, factory -> {
                    TikaExtractor tikaExtractor = container.getComponent("tikaExtractor");
                    PdfExtractor pdfExtractor = container.getComponent("pdfExtractor");
                    factory.addExtractor("text/plain", tikaExtractor);
                    factory.addExtractor("text/html", tikaExtractor);

                .<ExtractorFactory> singleton("extractorFactory", ExtractorFactory.class, factory -> {
                    TikaExtractor tikaExtractor = container.getComponent("tikaExtractor");
                    ZipExtractor zipExtractor = container.getComponent("zipExtractor");
                    factory.addExtractor("text/plain", tikaExtractor);
                    factory.addExtractor("text/html", tikaExtractor);

     * @throws ExtractException if extraction fails
     */
    public ExtractData extract() {
        final ExtractorFactory extractorFactory = crawlerContainer.getComponent("extractorFactory");

        DeferredFileOutputStream dfos = null;
        try (DeferredFileOutputStream out = new DeferredFileOutputStream(cacheFileSize, "fess-extractor-", ".out", null)) {
            dfos = out;