/** Property name for proxy authentication scheme setting */
    public static final String PROXY_AUTH_SCHEME_PROPERTY = "proxyAuthScheme";

    /** Property name for proxy credentials setting */
    public static final String PROXY_CREDENTIALS_PROPERTY = "proxyCredentials";

    /** Property name for user agent setting */

     */
    protected <T> T get(final Class<T> clazz, final String sessionId, final String url) {
        final String id = getId(sessionId, url);
        final GetResponse response = getClient().get(c -> c.prepareGet().setIndex(index).setId(id).execute());
        if (response.isExists()) {
            final Map<String, Object> source = response.getSource();
            final T bean = BeanUtil.copyMapToNewBean(source, clazz, option -> {

        final InputStream in = ResourceUtil.getResourceAsStream("extractor/msoffice/test.doc");
        final ExtractData extractData = tikaExtractor.getText(in, null);
        final String content = extractData.getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);
        assertTrue(content.contains("テスト"));
        for (final String key : extractData.getKeySet()) {

            final DocumentBuilder builder = factory.newDocumentBuilder();

            final Document doc = builder.parse(is);

            final StringBuilder buf = new StringBuilder(1000);
            buf.append(getResultDataHeader());
            for (final Map.Entry<String, String> entry : fieldRuleMap.entrySet()) {
                final List<String> nodeStrList = new ArrayList<>();

    }

    public void test_execute_web() throws Exception {
        final CrawlerWebServer server = new CrawlerWebServer(7070);
        server.start();

        final String url = "http://localhost:7070/";
        try {
            final int maxCount = 50;
            final int numOfThread = 10;

            final File file = File.createTempFile("crawler-", "");
            file.delete();

     */
    protected ResponseData getResponseData(final String uri, final boolean includeContent) {
        final ResponseData responseData = new ResponseData();
        try {
            responseData.setMethod(includeContent ? Constants.GET_METHOD : Constants.HEAD_METHOD);
            final String filePath = preprocessUri(uri);
            responseData.setUrl(filePath);

     * as a String.
     */
    protected Class<?> dataClass = null;

    @Override
    protected void storeData(final ResponseData responseData, final ResultData resultData) {
        final DOMParser parser = getDomParser();
        try (final InputStream in = responseData.getResponseBody()) {
            final InputSource is = new InputSource(in);
            if (responseData.getCharSet() != null) {

     * Field name for ID.
     */
    public static final String ID = "id";

    /**
     * Field name for session ID.
     */
    public static final String SESSION_ID = "sessionId";

    /**
     * Field name for rule ID.
     */
    public static final String RULE_ID = "ruleId";

    /**
     * Field name for URL.
     */
    public static final String URL = "url";

    /**

    @Override
    public ExtractData getText(final InputStream in, final Map<String, String> params) {
        if (in == null) {
            throw new CrawlerSystemException("The inputstream is null.");
        }
        try {
            final BufferedInputStream bis = new BufferedInputStream(in);
            final String enc = getEncoding(bis);

    private final ThreadLocal<XPathAPI> xpathAPI = new ThreadLocal<>();

    /**
     * Creates a new HtmlExtractor instance.
     */
    public HtmlExtractor() {
        super();
    }

    @Override
    protected ExtractData createExtractData(final String content) {
        final DOMParser parser = getDomParser();
        try (final Reader reader = new StringReader(content)) {