import org.codelibs.core.io.CloseableUtil;
import org.codelibs.core.io.ResourceUtil;
import org.codelibs.fess.crawler.container.StandardCrawlerContainer;
import org.codelibs.fess.crawler.exception.CrawlerSystemException;
import org.dbflute.utflute.core.PlainTestCase;

/**
 * @author shinsuke
 *
 */
public class TextExtractorTest extends PlainTestCase {

import org.codelibs.fess.crawler.entity.AccessResultDataImpl;
import org.codelibs.fess.crawler.entity.ResponseData;
import org.codelibs.fess.crawler.entity.ResultData;
import org.codelibs.fess.crawler.exception.CrawlerSystemException;
import org.dbflute.utflute.core.PlainTestCase;

/**
 * @author shinsuke
 *
 */
public class BinaryTransformerTest extends PlainTestCase {
    public BinaryTransformer binaryTransformer;

        try {
            server.start();
        } catch (final Exception e) {
            throw new CrawlerSystemException(e);
        }
    }

    public void stop() {
        try {
            server.stop();
            server.join();
        } catch (final Exception e) {
            throw new CrawlerSystemException(e);
        } finally {
            if (tempDocRoot) {
                docRoot.delete();
            }

import org.codelibs.fess.crawler.entity.AccessResultData;
import org.codelibs.fess.crawler.entity.ResponseData;
import org.codelibs.fess.crawler.entity.ResultData;
import org.codelibs.fess.crawler.exception.CrawlerSystemException;
import org.codelibs.fess.crawler.exception.CrawlingAccessException;

/**
 * The BinaryTransformer class is responsible for transforming binary data from a ResponseData object

import org.codelibs.fess.crawler.entity.AccessResultDataImpl;
import org.codelibs.fess.crawler.entity.ResponseData;
import org.codelibs.fess.crawler.entity.ResultData;
import org.codelibs.fess.crawler.exception.CrawlerSystemException;
import org.codelibs.fess.crawler.extractor.ExtractorFactory;
import org.codelibs.fess.crawler.extractor.impl.TikaExtractor;
import org.dbflute.utflute.core.PlainTestCase;

/**
 * @author shinsuke
 *
 */

import org.codelibs.fess.crawler.entity.ExtractData;
import org.codelibs.fess.crawler.entity.ResponseData;
import org.codelibs.fess.crawler.entity.ResultData;
import org.codelibs.fess.crawler.exception.CrawlerSystemException;
import org.codelibs.fess.crawler.exception.CrawlingAccessException;
import org.codelibs.fess.crawler.extractor.Extractor;
import org.codelibs.fess.crawler.extractor.ExtractorFactory;

import java.util.regex.Pattern;

import javax.xml.xpath.XPathNodes;

import org.codelibs.core.lang.StringUtil;
import org.codelibs.fess.crawler.entity.ExtractData;
import org.codelibs.fess.crawler.exception.CrawlerSystemException;
import org.codelibs.fess.crawler.exception.ExtractException;
import org.codelibs.fess.crawler.util.XPathAPI;
import org.codelibs.nekohtml.parsers.DOMParser;
import org.w3c.dom.Document;
import org.w3c.dom.Node;

/**
 * RobotsTxtException is an exception class that represents an error related to robots.txt processing during web crawling.
 * It extends CrawlerSystemException and provides constructors to create instances with a message and/or a cause.
 *
 */
public class RobotsTxtException extends CrawlerSystemException {

    private static final long serialVersionUID = 1L;

    /**

import org.codelibs.core.io.CloseableUtil;
import org.codelibs.core.io.ResourceUtil;
import org.codelibs.fess.crawler.container.StandardCrawlerContainer;
import org.codelibs.fess.crawler.exception.CrawlerSystemException;
import org.codelibs.fess.crawler.exception.MaxLengthExceededException;
import org.codelibs.fess.crawler.extractor.ExtractorFactory;
import org.codelibs.fess.crawler.helper.impl.MimeTypeHelperImpl;

/**
 * {@link ChildUrlsException} is thrown when child URLs are found during crawling.
 * It extends {@link CrawlerSystemException} and holds a set of {@link RequestData}
 * representing the child URLs that caused the exception.
 *
 */
public class ChildUrlsException extends CrawlerSystemException {

    private static final long serialVersionUID = 1L;

    /**
     * The list of child URLs.
     */