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();

        htmlXpathExtractor = container.getComponent("htmlXpathExtractor");
        htmlXpathExtractor.init();
    }

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

        StandardCrawlerContainer container = new StandardCrawlerContainer().singleton("xmlExtractor", XmlExtractor.class);
        xmlExtractor = container.getComponent("xmlExtractor");
    }

    public void test_getXml_utf8() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/test_utf8.xml");
        final String content = xmlExtractor.getText(in, null).getContent();
        CloseableUtil.closeQuietly(in);