urlFilter.init(sessionId);

        urlFilter.addExclude(".*\\.(jpg|png|gif)$");

        assertTrue(urlFilter.match("https://example.com/page.html"));
        assertTrue(urlFilter.match("https://example.com/document.pdf"));
        assertFalse(urlFilter.match("https://example.com/image.jpg"));
        assertFalse(urlFilter.match("https://example.com/photo.png"));

});

// Configure URL filtering
container.singleton("urlFilter", UrlFilterImpl.class, filter -> {
    filter.addInclude("https://example.com/.*");
    filter.addExclude(".*\\.(css|js|png|jpg|gif)$");
});

// Configure content extraction
container.singleton("tikaExtractor", TikaExtractor.class);
container.singleton("extractorFactory", ExtractorFactory.class, factory -> {

        logger.info(content);
        assertTrue(content.contains("Test"));
    }

    public void test_getTika_jpg() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/image/test.jpg");
        final ExtractData extractData = tikaExtractor.getText(in, null);
        final String content = extractData.getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);