* Stops the crawling process.
     * Sets the crawler status to DONE and interrupts all crawler threads.
     */
    public void stop() {
        if (logger.isInfoEnabled()) {
            logger.info("Stopping crawler: sessionId={}, status={}", crawlerContext.sessionId, crawlerContext.getStatus());
        }
        crawlerContext.setStatus(CrawlerStatus.DONE);
        try {
            if (crawlerThreadGroup != null) {

                    crawler.awaitTermination(crawlingExecutionInterval);
                    if (crawler.getCrawlerContext().getStatus() == CrawlerStatus.DONE && !Constants.DONE.equals(crawlerStatusList.get(i))) {
                        crawlerStatusList.set(i, Constants.DONE);
                        final String sid = crawler.getCrawlerContext().getSessionId();
                        indexUpdater.addFinishedSessionId(sid);

        assertTrue(content.contains("Email"));
        assertTrue(content.contains("Age"));
        assertTrue(content.contains("Department"));

        // Verify data extraction
        assertTrue(content.contains("John Doe"));
        assertTrue(content.contains("******@****.***"));
        assertTrue(content.contains("Engineering"));
    }

    public void test_columnMetadata() {

    }

    public void test_putValue() {
        // Test putValue method
        ExtractData data = new ExtractData();

        data.putValue("title", "Test Title");
        data.putValue("author", "John Doe");

        String[] titleValues = data.getValues("title");
        assertNotNull(titleValues);
        assertEquals(1, titleValues.length);
        assertEquals("Test Title", titleValues[0]);

Name,Email,Age,Department
John Doe,******@****.***,30,Engineering
Jane Smith,******@****.***,28,Marketing
Bob Johnson,******@****.***,35,Sales

                } finally {
                    done.set(true);
                }
            }, Thread.currentThread().getName() + "-pdf");
            task.setDaemon(isDaemonThread);
            task.start();
            task.join(timeout);
            if (!done.get()) {
                for (int i = 0; i < 100 && !done.get(); i++) {
                    task.interrupt();

{
  "title": "Sample Document",
  "author": "John Doe",
  "version": "1.0",
  "published": "2025-01-15",
  "tags": ["crawler", "extractor", "json"],
  "content": {
    "summary": "This is a sample JSON document for testing",
    "body": "The extractor should handle nested objects and arrays properly"
  },
  "metadata": {
    "created_at": "2025-01-01T00:00:00Z",
    "updated_at": "2025-01-15T12:00:00Z"
  }

        assertTrue(content.contains("author"));
        assertTrue(content.contains("John Doe"));

        // Verify metadata extraction
        assertNotNull(extractData.getValues("title"));
        assertEquals("Sample Document", extractData.getValues("title")[0]);
        assertNotNull(extractData.getValues("author"));
        assertEquals("John Doe", extractData.getValues("author")[0]);
    }

    public void test_getText_nested() {

        crawlerThread.run();

        verify(urlQueueService, times(1)).poll(anyString());
    }

    /**
     * Test run method with crawler status DONE.
     */
    public void test_run_statusDone() throws Exception {
        crawlerContext.setStatus(CrawlerStatus.DONE);

        crawlerThread.run();

        verify(urlQueueService, times(0)).poll(anyString());
    }

    /**
     * Test setNoWaitOnFolder.

                if (!dataCrawlingThreadList.get(i).isRunning() && Constants.RUNNING.equals(dataCrawlingThreadStatusList.get(i))) {
                    dataCrawlingThreadStatusList.set(i, Constants.DONE);
                }
                if (!Constants.DONE.equals(dataCrawlingThreadStatusList.get(i))) {
                    finishedAll = false;
                }
            }
        }
        dataCrawlingThreadList.clear();