import java.io.Serializable;

/**
 * The Sitemap interface represents a sitemap entity with location and last modification date.
 * 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();

<?xml version="1.0" encoding="UTF-8"?>
<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
   <sitemap>
      <loc>http://www.example.com/sitemap1.xml.gz</loc>
      <lastmod>2004-10-01T18:23:17+00:00</lastmod>
   </sitemap>
   <sitemap>
      <loc>http://www.example.com/sitemap2.xml.gz</loc>
      <lastmod>2005-01-01</lastmod>
   </sitemap>

        responseData.setResponseBody(file, false);
        return responseData;
    }

    private ResponseData getTestData3_OK() {
        final ResponseData responseData = new ResponseData();
        responseData.setUrl("http://example.com/sitemap.txt");
        File file = ResourceUtil.getResourceAsFile("sitemaps/sitemap1.txt");
        responseData.setResponseBody(file, false);
        return responseData;
    }

        assertEquals("http://www.example.com/sitemap2.xml.gz", sitemaps[1].getLoc());

    }

    public void test_parseXmlSitemapsIndexGz() {
        final InputStream in = ResourceUtil.getResourceAsStream("sitemaps/sitemap2.xml.gz");
        final SitemapSet sitemapSet = sitemapsHelper.parse(in);
        final Sitemap[] sitemaps = sitemapSet.getSitemaps();
        assertEquals(2, sitemaps.length);

    }

    /**
     * Adds a sitemap to this set.
     * @param sitemap the sitemap to add
     */
    public void addSitemap(final Sitemap sitemap) {
        sitemapList.add(sitemap);
    }

    /**
     * Removes a sitemap from this set.
     * @param sitemap the sitemap to remove
     */
    public void removeSitemap(final Sitemap sitemap) {
        sitemapList.remove(sitemap);
    }

    /**

import org.codelibs.core.lang.StringUtil;

/**
 * Represents a URL entry within a sitemap.
 *
 * <p>
 * This class encapsulates the properties of a URL as defined in the sitemap XML format,
 * including its location, last modification date, change frequency, and priority.
 * It implements the {@link Sitemap} interface.
 * </p>
 *
 * <p>

            final Set<RequestData> requestDataSet = new LinkedHashSet<>();
            for (final Sitemap sitemap : sitemapSet.getSitemaps()) {
                if (sitemap != null) {
                    requestDataSet.add(RequestDataBuilder.newRequestData().get().url(sitemap.getLoc()).build()); // TODO priority
                }
            }

package org.codelibs.fess.crawler.entity;

import org.codelibs.core.lang.StringUtil;

/**
 * Represents a Sitemap file entry, conforming to the Sitemap XML format.
 * This class holds information about a single Sitemap, including its location and last modification timestamp.
 * It implements the {@link Sitemap} interface.
 *
 * <p>
 * A Sitemap file provides search engines with a list of URLs available for crawling.

 * represents a valid sitemap. It uses a SitemapsHelper to validate the response body as an InputStream.
 * The rule checks if the URL matches the defined regex pattern and then validates the content as a sitemap.
 * If any exception occurs during the sitemap validation, it logs the error and returns false.
 *
 */
public class SitemapsRule extends RegexRule {
    /**

			<component class="org.codelibs.fess.crawler.processor.impl.SitemapsResponseProcessor">
			</component>
		</property>
		<postConstruct name="addRule">
			<arg>"url"</arg>
			<arg>"http[s]?:.*sitemap[^/]*\.xml.*|http[s]?:.*sitemap[^/]*\.gz.*|http[s]?:.*sitemap[^/]*\.txt.*"</arg>
		</postConstruct>
	</component>

	<component name="webHtmlRule" class="org.codelibs.fess.crawler.rule.impl.RegexRule" >
		<property name="ruleId">"webHtmlRule"</property>