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) {

        ;
        includeFilter = container.getComponent("includeFilter");
        includeFilter.setIncludeFilteringPattern("$1$2$3.*");
        excludeFilter = container.getComponent("excludeFilter");
        excludeFilter.setExcludeFilteringPattern("$1$2$3.*");
        domainFilter = container.getComponent("domainFilter");
        domainFilter.setIncludeFilteringPattern("http://$2/.*");

                .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);

        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

                .<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);

     * 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

        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");

                .<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);

                .<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);

container.singleton("extractorFactory", ExtractorFactory.class, factory -> {
    factory.addExtractor("text/html", container.getComponent("tikaExtractor"));
    factory.addExtractor("application/pdf", container.getComponent("tikaExtractor"));
});

Crawler crawler = container.getComponent("crawler");
crawler.addUrl("https://example.com");
crawler.crawlerContext.setMaxAccessCount(500);
String sessionId = crawler.execute();