assertEquals("text/html", mimeMap.getMimeType("html"));
            assertEquals("text/html", mimeMap.getMimeType("htm"));
            assertEquals("image/jpeg", mimeMap.getMimeType("jpg"));
            assertEquals("image/jpeg", mimeMap.getMimeType("jpeg"));
            assertEquals("image/gif", mimeMap.getMimeType("gif"));
            assertEquals("image/png", mimeMap.getMimeType("png"));
        }

        try (final ResponseData responseData = storageClient.doGet("storage://fess/file1.txt")) {
            assertEquals("storage://fess/file1.txt", responseData.getUrl());
            assertEquals("text/plain", responseData.getMimeType());
            assertEquals("file1", new String(InputStreamUtil.getBytes(responseData.getResponseBody())));
            assertEquals(5, responseData.getContentLength());

     * @return the MIME type for the extension, or "application/octet-stream" if not found
     * @throws IOException if there is an error processing the MIME mappings
     */
    public String getMimeType(final String extension) throws IOException {
        return getMimeType(extension, "application/octet-stream");
    }

    /**
     * Returns the MIME type for the given file extension with a custom default.
     *

            assertEquals(5, responseData.getContentLength());
            assertEquals(baseUrl + "file1.txt", responseData.getUrl());
            assertEquals("text/plain", responseData.getMimeType());
            assertNull(responseData.getResponseBody());
        }
        try (final ResponseData responseData = smbClient.doHead(baseUrl + "dir1/file2.txt")) {
            assertEquals(0, responseData.getStatus());

        assertNotNull(responseData.getLastModified());
        assertEquals(Constants.GET_METHOD, responseData.getMethod());
        assertEquals("text/plain", responseData.getMimeType());
        assertTrue(responseData.getUrl().endsWith("test/text1.txt"));
        final String content = new String(InputStreamUtil.getBytes(responseData.getResponseBody()), "UTF-8");
        assertEquals("test1", content.trim());

        if (extractorFactory == null) {
            throw new FessSystemException("Could not find extractorFactory.");
        }
        final Extractor extractor = extractorFactory.getExtractor(responseData.getMimeType());
        if (logger.isDebugEnabled()) {
            logger.debug("url={}, extractor={}", responseData.getUrl(), extractor);
        }
        return extractor;
    }

        if (extractorFactory == null) {
            throw new FessSystemException("Could not find extractorFactory.");
        }
        Extractor extractor = extractorFactory.getExtractor(responseData.getMimeType());
        if (extractor == null) {
            extractor = ComponentUtil.getComponent("tikaExtractor");
            if (extractor == null) {
                throw new FessSystemException("Could not find tikaExtractor.");

                final ResultData resultData = transformer.transform(responseData);
                if (resultData == null) {
                    logger.warn("No data for ({}, {})", responseData.getUrl(), responseData.getMimeType());
                } else {
                    final UrlQueue<?> urlQueue = CrawlingParameterUtil.getUrlQueue();
                    processResult(urlQueue, responseData, resultData);
                }
            }

            }

            // Check MIME type condition
            String mimeTypeCondition = conditions.get("mimeType");
            if (mimeTypeCondition != null && responseData.getMimeType() != null) {
                if (!responseData.getMimeType().matches(mimeTypeCondition)) {
                    return false;
                }
            }

            // Check status code condition

            assertNotNull(responseData.getLastModified());
            assertEquals(Constants.GET_METHOD, responseData.getMethod());
            assertEquals("text/plain", responseData.getMimeType());
            assertTrue(responseData.getUrl().endsWith("text1.txt"));
            final String content = new String(InputStreamUtil.getBytes(responseData.getResponseBody()), "UTF-8");