sitemapsHelper = container.getComponent("sitemapsHelper");
    }

    public void test_parseXmlSitemaps() {
        final InputStream in = ResourceUtil.getResourceAsStream("sitemaps/sitemap1.xml");
        final SitemapSet sitemapSet = sitemapsHelper.parse(in);
        final Sitemap[] sitemaps = sitemapSet.getSitemaps();
        assertEquals(5, sitemaps.length);

        final InputStream in = ResourceUtil.getResourceAsStream("extractor/test_utf8.html");
        final String content = htmlXpathExtractor.getText(in, null).getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);
        assertTrue(content.contains("テスト"));
    }

    public void test_getHtml_sjis() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/test_sjis.html");

        htmlExtractor = container.getComponent("htmlExtractor");
        htmlExtractor.addMetadata("title", "//TITLE");
    }

    public void test_getHtml_utf8() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/test_utf8.html");
        final ExtractData data = htmlExtractor.getText(in, null);
        final String content = data.getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);

                });
        emlExtractor = container.getComponent("emlExtractor");
    }

    public void test_getText() throws IOException {
        try (final InputStream in = ResourceUtil.getResourceAsStream("extractor/eml/sample1.eml")) {
            ExtractData data = emlExtractor.getText(in, null);
            final String content = data.getContent();
            logger.info(content);

                })//
        ;

        tarExtractor = container.getComponent("tarExtractor");
    }

    public void test_getText() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/tar/test.tar");
        final String content = tarExtractor.getText(in, null).getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);
        assertTrue(content.contains("テスト"));

    @Override
    protected void tearDown() throws Exception {
        jodExtractor.destroy();
        super.tearDown();
    }

    public void test_getText_ooow() {
        InputStream in = ResourceUtil.getResourceAsStream("extractor/ooo/test.odt");
        Map<String, String> params = new HashMap<String, String>();
        params.put("resourceName", "test.odt");
        ExtractData extractData = jodExtractor.getText(in, params);

        final InputStream in = ResourceUtil.getResourceAsStream("extractor/test_sjis.xml");
        final String content = xmlExtractor.getText(in, null).getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);
        assertTrue(content.contains("テスト"));
    }

    public void test_getXml_entity() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/test_entity.xml");

        msWordExtractor = container.getComponent("msWordExtractor");
    }

    public void test_getText() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/msoffice/test.doc");
        final String content = msWordExtractor.getText(in, null).getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);

                })//
        ;

        zipExtractor = container.getComponent("zipExtractor");
    }

    public void test_getText() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/zip/test.zip");
        final String content = zipExtractor.getText(in, null).getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);
        assertTrue(content.contains("テスト"));

        msExcelExtractor = container.getComponent("msExcelExtractor");
    }

    public void test_getText() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/msoffice/test.xls");
        final String content = msExcelExtractor.getText(in, null).getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);