;
        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);

        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");

        clientFactory.addClient("smb1:.*", container.getComponent("smb1Client"));
        clientFactory.addClient("ftp:.*", container.getComponent("ftpClient"));
        clientFactory.addClient("storage:.*", container.getComponent("storageClient"));
    }

    public void test_getClient() {
        String url;
        CrawlerClient client;

        url = "http://hoge.com/";
        client = clientFactory.getClient(url);

        super.setUp();
        StandardCrawlerContainer container = new StandardCrawlerContainer().singleton("contentLengthHelper", ContentLengthHelper.class);
        contentLengthHelper = container.getComponent("contentLengthHelper");
    }

    public void test_getMaxLength() {
        final String mimeType = "text/plain";

        assertEquals(DEFAULT_MAX_LENGTH, contentLengthHelper.getMaxLength(mimeType));

                .<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);

                    rule.setResponseProcessor(container.getComponent("defaultResponseProcessor"));
                });

        crawler = container.getComponent("crawler");
        dataService = container.getComponent("dataService");
        urlQueueService = container.getComponent("urlQueueService");
        fileTransformer = container.getComponent("fileTransformer");

    }

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();

                .singleton("extractorFactory", ExtractorFactory.class);
        extractorFactory = container.getComponent("extractorFactory");
        TikaExtractor tikaExtractor = container.getComponent("tikaExtractor");
        LhaExtractor lhaExtractor = container.getComponent("lhaExtractor");
        PasswordBasedExtractor pdfExtractor = container.getComponent("pdfExtractor");
        extractorFactory.addExtractor("application/msword", tikaExtractor);

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