assertEquals(2, activeStatuses.size());
        assertTrue(activeStatuses.contains(CrawlerStatus.INITIALIZING));
        assertTrue(activeStatuses.contains(CrawlerStatus.RUNNING));
        assertFalse(activeStatuses.contains(CrawlerStatus.DONE));

        // Create empty EnumSet
        EnumSet<CrawlerStatus> emptySet = EnumSet.noneOf(CrawlerStatus.class);
        assertEquals(0, emptySet.size());

 * Provides common functionality and properties for crawler rules.
 *
 * <p>
 * This class handles the registration of rules with the {@link RuleManager}
 * and provides getter and setter methods for common properties such as
 * {@code ruleId} and {@code responseProcessor}.
 * </p>
 *
 * <p>
 * Subclasses should extend this class and implement the abstract methods

    /** The property name for maximum cached content size. */
    public static final String MAX_CACHED_CONTENT_SIZE = "maxCachedContentSize";

    /** The crawler container. */
    @Resource
    protected CrawlerContainer crawlerContainer;

    /** The initialization parameter map. */
    protected Map<String, Object> initParamMap;

    /** The maximum cached content size in bytes. Default is 1MB. */

 * This class provides methods to handle authentication details such as
 * authentication scope, credentials, and authentication scheme.
 *
 * <p>
 * It includes constructors to initialize these details and getter and setter
 * methods to access and modify them.
 * </p>
 *
 * <p>
 * Example usage:
 * </p>
 * <pre>
 * {@code
 * AuthScope authScope = new AuthScope("example.com", 80);

        this.maxContentLength = maxContentLength;
        return this;
    }

    /**
     * Sets the cache file size threshold in bytes.
     *
     * @param cacheFileSize the cache file size to set
     * @return this builder instance for method chaining
     */
    public ExtractorBuilder cacheFileSize(final int cacheFileSize) {
        this.cacheFileSize = cacheFileSize;

            final List<SuggestItem> items) {
        this.index = index;
        this.tookMs = tookMs;
        this.words = words;
        num = words.size();
        this.total = total;
        this.items = items;
    }

    /**
     * Returns the index name.
     * @return The index name.
     */
    public String getIndex() {
        return index;
    }

    public static final int FORBIDDEN_STATUS_CODE = 403;

    /**
     * The HTTP status code for Not Found.
     */
    public static final int NOT_FOUND_STATUS_CODE = 404;

    /**
     * The HTTP status code for Server Error.
     */
    public static final int SERVER_ERROR_STATUS_CODE = 500;

    /**
     * A constant for no transformer.
     */
    public static final String NO_TRANSFORMER = "NONE";

    /**

        auth.setPort(0);
        assertTrue(auth.matches("ftp://hostname:21/test/aaa.html"));
        assertTrue(auth.matches("ftp://hostname/test/aaa.html"));

        // Test with different server
        auth.setServer("otherhost");
        assertFalse(auth.matches("ftp://hostname:21/test/aaa.html"));
        assertTrue(auth.matches("ftp://otherhost:21/test/aaa.html"));

        // Test with IP address

 * The class also supports configuration options such as:
 * </p>
 * <ul>
 *   <li>Output encoding</li>
 *   <li>Maximum compression ratio and uncompression size</li>
 *   <li>Initial buffer size</li>
 *   <li>Memory size for temporary file storage</li>
 *   <li>Maximum term sizes for alphanumeric and symbolic terms</li>
 *   <li>Custom Tika configuration</li>
 *   <li>Tesseract OCR configuration for image-based documents</li>

        final ResultData resultData = htmlTransformer.transform(responseData);
        assertEquals(content, new String(resultData.getData()));
        assertEquals(1, resultData.getChildUrlSet().size());
        assertEquals("http://hoge/test2.html", resultData.getChildUrlSet().iterator().next().getUrl());
    }

    public void test_transform_urllink() {