private final Map<String, String> params;

    private final CrawlerContainer crawlerContainer;

    private String mimeType;

    private String filename;

    private int cacheFileSize = 1_000_000;

    private String extractorName = "tikaExtractor";

    private long maxContentLength = -1;

    protected ExtractorBuilder(final CrawlerContainer crawlerContainer, final InputStream in, final Map<String, String> params) {

    private static final Logger logger = LoggerFactory.getLogger(HtmlTransformer.class);

    protected static final String LOCATION_HEADER = "Location";

    @Resource
    protected CrawlerContainer crawlerContainer;

    protected Map<String, String> featureMap = new HashMap<>();

    protected Map<String, String> propertyMap = new HashMap<>();

    protected Map<String, String> childUrlRuleMap = new LinkedHashMap<>();

public class ExtractorFactory {

    private static final Logger logger = LoggerFactory.getLogger(ExtractorFactory.class);

    @Resource
    protected CrawlerContainer crawlerContainer;

    protected Map<String, Extractor> extractorMap = new HashMap<>();

    public void addExtractor(final String key, final Extractor extractor) {
        if (StringUtil.isBlank(key)) {
            throw new CrawlerSystemException("The key is null.");
        }

    @Resource
    protected UrlQueueService<UrlQueue<?>> urlQueueService;

    @Resource
    protected DataService<AccessResult<?>> dataService;

    @Resource
    protected CrawlerContainer crawlerContainer;

    @Resource
    protected LogHelper logHelper;

    protected CrawlerClientFactory clientFactory;

    protected CrawlerContext crawlerContext;

    protected boolean noWaitOnFolder = false;

import org.slf4j.LoggerFactory;

import jakarta.annotation.PostConstruct;
import jakarta.annotation.PreDestroy;
import jakarta.annotation.Resource;

public class StandardCrawlerContainer implements CrawlerContainer {

    private final Logger logger = LoggerFactory.getLogger(StandardCrawlerContainer.class);

    private final Map<String, ComponentHolder<?>> singletonMap = new ConcurrentHashMap<>();

    private final Map<String, PDFParserConfig> pdfParserConfigMap = new ConcurrentHashMap<>();

    @PostConstruct
    public void init() {
        if (tikaConfig == null && crawlerContainer != null) {
            try {
                tikaConfig = crawlerContainer.getComponent("tikaConfig");
            } catch (final Exception e) {
                logger.debug("tikaConfig component is not found.", e);
            }
        }

                    }
                }

                if (file.canRead()) {
                    final MimeTypeHelper mimeTypeHelper = crawlerContainer.getComponent("mimeTypeHelper");
                    if (includeContent) {
                        if (logger.isDebugEnabled()) {
                            logger.debug("Parsing SmbFile Content: {}", filePath);

                            throw new CrawlingAccessException("Failed to retrieve: " + ftpInfo.toUrl());
                        }
                    }

                    final MimeTypeHelper mimeTypeHelper = crawlerContainer.getComponent("mimeTypeHelper");
                    try (InputStream is = new FileInputStream(tempFile)) {
                        responseData.setMimeType(mimeTypeHelper.getContentType(is, file.getName()));