*
 * <p>Example usage:
 * <pre>
 *   Crawler crawler = new Crawler();
 *   crawler.addUrl("http://example.com/");
 *   crawler.execute();
 *   crawler.close();
 * </pre>
 */
public class Crawler implements Runnable, AutoCloseable {

    private static final Logger logger = LogManager.getLogger(Crawler.class);

    /**
     * Service for managing URL queues during crawling.
     */

 * This job launches a separate crawler process that can crawl web sites, file systems,
 * and data sources based on the configured crawling settings.
 *
 * <p>The job supports selective crawling by specifying configuration IDs for different
 * types of crawlers (web, file, data). It manages the crawler process lifecycle,
 * handles timeout scenarios, and ensures proper cleanup of resources.</p>
 *

import org.codelibs.fess.Constants;
import org.codelibs.fess.crawler.Crawler;
import org.codelibs.fess.crawler.entity.AccessResult;
import org.codelibs.fess.crawler.entity.AccessResultData;
import org.codelibs.fess.crawler.entity.OpenSearchAccessResult;
import org.codelibs.fess.crawler.entity.OpenSearchUrlQueue;
import org.codelibs.fess.crawler.service.DataService;
import org.codelibs.fess.crawler.service.UrlFilterService;

        // Default constructor
    }

    /**
     * Enable or disable incremental crawling.
     * When enabled, only new or modified documents are crawled.
     */
    @Size(max = 10)
    public String incrementalCrawling;

    /**
     * Number of days to keep crawled documents before cleanup.
     * Set to -1 to disable automatic cleanup.
     */
    @Required
    @Min(-1)
    @Max(1000)
    @ValidateTypeFailure

import org.codelibs.fess.Constants;
import org.codelibs.fess.crawler.builder.RequestDataBuilder;
import org.codelibs.fess.crawler.client.CrawlerClient;
import org.codelibs.fess.crawler.client.CrawlerClientFactory;
import org.codelibs.fess.crawler.entity.ResponseData;
import org.codelibs.fess.crawler.entity.ResultData;
import org.codelibs.fess.crawler.exception.ChildUrlsException;
import org.codelibs.fess.crawler.exception.CrawlerSystemException;

 * governing permissions and limitations under the License.
 */
package org.codelibs.fess.crawler.util;

import org.codelibs.fess.crawler.CrawlerContext;
import org.codelibs.fess.crawler.entity.AccessResult;
import org.codelibs.fess.crawler.entity.UrlQueue;
import org.codelibs.fess.crawler.service.DataService;
import org.codelibs.fess.crawler.service.UrlQueueService;

/**

    /**
     * Interval in milliseconds between crawler thread executions.
     * Used to control the rate at which new crawler threads are started
     * and the frequency of status checks.
     */
    protected long crawlingExecutionInterval = Constants.DEFAULT_CRAWLING_EXECUTION_INTERVAL;

    /**
     * Thread priority for crawler threads.
     * Defaults to normal thread priority.
     */

import org.codelibs.fess.app.service.FailureUrlService;
import org.codelibs.fess.crawler.builder.RequestDataBuilder;
import org.codelibs.fess.crawler.client.CrawlerClient;
import org.codelibs.fess.crawler.entity.RequestData;
import org.codelibs.fess.crawler.entity.ResponseData;
import org.codelibs.fess.crawler.entity.UrlQueue;
import org.codelibs.fess.crawler.log.LogType;
import org.codelibs.fess.exception.ContainerNotAvailableException;

 * This class defines tag patterns that match HTML elements based on tag name, CSS class, ID, or custom attributes.
 * It is used to identify and remove unwanted HTML elements from crawled documents.
 */
public class PrunedTag {
    /** The HTML tag name to match (e.g., "div", "span", "p") */
    private final String tag;
    /** The ID attribute value to match */
    private String id;

import org.codelibs.core.lang.StringUtil;
import org.codelibs.fess.Constants;
import org.codelibs.fess.crawler.entity.AccessResult;
import org.codelibs.fess.crawler.entity.AccessResultData;
import org.codelibs.fess.crawler.entity.UrlQueue;
import org.codelibs.fess.crawler.util.CrawlingParameterUtil;
import org.codelibs.fess.crawler.util.FieldConfigs;
import org.codelibs.fess.mylasta.direction.FessConfig;