* It uses a ConcurrentMap to store the last access time for each host.
 * The delayBeforeProcessing method is overridden to introduce a delay before processing a URL,
 * ensuring that requests to the same host are not made too frequently.
 * The delay is calculated based on the configured delayMillisBeforeProcessing parameter.
 * If the time since the last request to the host is less than the configured delay,

        this.loc = loc;
    }

    /**
     * Returns the last modification date of this sitemap entry.
     * @return the last modification date
     */
    @Override
    public String getLastmod() {
        return lastmod;
    }

    /**
     * Sets the last modification date of this sitemap entry.
     * @param lastmod the last modification date to set
     */

 * It extends the Serializable interface to allow sitemap objects to be serialized.
 */
public interface Sitemap extends Serializable {

    /**
     * Retrieves the location (URL) of the sitemap.
     *
     * @return the location of the sitemap as a String.
     */
    String getLoc();

    /**
     * Retrieves the last modification date of the sitemap.
     *

    /**
     * Returns the last modified time of the accessed resource.
     *
     * @return the last modified time
     */
    Long getLastModified();

    /**
     * Sets the last modified time of the accessed resource.
     *
     * @param lastModified the last modified time
     */
    /**
     * Sets the last modified time of the accessed resource.
     *

     * @see org.codelibs.fess.crawler.entity.Sitemap#getLastmod()
     */
    @Override
    public String getLastmod() {
        return lastmod;
    }

    /**
     * Sets the last modification time of the sitemap.
     * @param lastmod the last modification time in W3C Datetime format
     */
    public void setLastmod(final String lastmod) {
        this.lastmod = lastmod;
    }

    /**

    void setDepth(Integer depth);

    /**
     * Returns the last modified timestamp of the URL in the queue.
     *
     * @return the last modified timestamp as a Long value.
     */
    Long getLastModified();

    /**
     * Sets the last modified timestamp for the URL queue entry.
     *
     * @param lastModified the timestamp of the last modification in milliseconds since epoch.
     */

        } finally {
            FileUtil.deleteInBackground(tempFile);
        }
    }

    /**
     * Marks the current position in this input stream. A subsequent call to the reset method repositions this stream at the last marked position so that subsequent reads re-read the same bytes.
     * This method delegates to {@link FileInputStream#mark(int)}.
     *

package org.codelibs.fess.crawler.entity;

/**
 * Implementation of the {@link UrlQueue} interface.
 * This class represents a URL to be crawled, storing its ID, session ID,
 * HTTP method, URL, metadata, encoding, parent URL, depth, last modified time,
 * creation time, and weight.
 *
 * @param <IDTYPE> the type of the identifier for the URL queue entry
 */
public class UrlQueueImpl<IDTYPE> implements UrlQueue<IDTYPE> {

     */
    public static final String PARENT_URL = "parentUrl";

    /**
     * Field name for depth.
     */
    public static final String DEPTH = "depth";

    /**
     * Field name for last modified timestamp.
     */
    public static final String LAST_MODIFIED = "lastModified";

    /**
     * Field name for creation time.
     */
    public static final String CREATE_TIME = "createTime";

    /** Indicates the cleanup phase of crawling. */
    CLEANUP_CRAWLING,
    /** Indicates an unsupported URL was encountered when crawling started. */
    UNSUPPORTED_URL_AT_CRAWLING_STARTED,
    /** Indicates checking the last modified date of a resource. */
    CHECK_LAST_MODIFIED,
    /** Indicates the resource has not been modified. */
    NOT_MODIFIED,
    /** Indicates getting content from a resource. */
    GET_CONTENT,