*/
package org.codelibs.fess.crawler.processor;

import org.codelibs.fess.crawler.entity.ResponseData;

/**
 * @author shinsuke
 *
 */
public interface ResponseProcessor {

    void process(ResponseData responseData);

    public void setRuleId(final String ruleId) {
        this.ruleId = ruleId;
    }

    @Override
    public ResponseProcessor getResponseProcessor() {
        return responseProcessor;
    }

    public void setResponseProcessor(final ResponseProcessor responseProcessor) {
        this.responseProcessor = responseProcessor;
    }

		</postConstruct>
	</component>

	<component name="sitemapsRule" class="org.codelibs.fess.crawler.rule.impl.RegexRule" >
		<property name="ruleId">"sitemapsRule"</property>
		<property name="responseProcessor">
			<component class="org.codelibs.fess.crawler.processor.impl.SitemapsResponseProcessor">
			</component>
		</property>
		<postConstruct name="addRule">
			<arg>"url"</arg>

            responseData.setRuleId(rule.getRuleId());
            final ResponseProcessor responseProcessor = rule.getResponseProcessor();
            if (!(responseProcessor instanceof DefaultResponseProcessor)) {
                throw new CrawlingAccessException("The response processor is not DefaultResponseProcessor. responseProcessor: "
                        + responseProcessor + ", url: " + url);
            }

            } else {
                responseData.setRuleId(rule.getRuleId());
                final ResponseProcessor responseProcessor = rule.getResponseProcessor();
                if (responseProcessor instanceof DefaultResponseProcessor) {
                    final Transformer transformer = ((DefaultResponseProcessor) responseProcessor).getTransformer();
                    final ResultData resultData = transformer.transform(responseData);

        } else {
            responseData.setRuleId(rule.getRuleId());
            final ResponseProcessor responseProcessor = rule.getResponseProcessor();
            if (responseProcessor == null) {
                log(logHelper, LogType.NO_RESPONSE_PROCESSOR, crawlerContext, urlQueue, responseData, rule);
            } else {
                responseProcessor.process(responseData);
            }
        }

    }

 */
package org.codelibs.fess.crawler.processor.impl;

import org.codelibs.fess.crawler.entity.ResponseData;
import org.codelibs.fess.crawler.processor.ResponseProcessor;

/**
 * @author kuma
 *
 */
public class NullResponseProcessor implements ResponseProcessor {

    @Override
    public void process(final ResponseData responseData) {
        // do nothing
    }

import org.codelibs.fess.crawler.entity.ResponseData;
import org.codelibs.fess.crawler.processor.ResponseProcessor;

/**
 * @author shinsuke
 *
 */
public interface Rule extends Serializable {

    boolean match(ResponseData responseData);

    String getRuleId();

    ResponseProcessor getResponseProcessor();

			<arg>fileRule</arg>
		</postConstruct>
	</component>

	<component name="sitemapsRule" class="org.codelibs.fess.crawler.rule.impl.SitemapsRule">
		<property name="ruleId">"sitemapsRule"</property>
		<property name="responseProcessor">
			<component
				class="org.codelibs.fess.crawler.processor.impl.SitemapsResponseProcessor">
			</component>
		</property>
		<postConstruct name="addRule">
			<arg>"url"</arg>
			<arg>".*sitemap.*"</arg>

import org.codelibs.fess.crawler.entity.ResponseData;
import org.codelibs.fess.crawler.entity.ResultData;
import org.codelibs.fess.crawler.entity.UrlQueue;
import org.codelibs.fess.crawler.processor.ResponseProcessor;
import org.codelibs.fess.crawler.service.UrlQueueService;
import org.codelibs.fess.crawler.transformer.Transformer;
import org.codelibs.fess.crawler.util.CrawlingParameterUtil;
import org.slf4j.Logger;