final List<Map<String, Object>> docs = new ArrayList<>(docPerReq);
            try (final DocumentReader documentReader = reader.get()) {
                Map<String, Object> doc = documentReader.read();
                while (doc != null) {
                    if (Thread.currentThread().isInterrupted()) {
                        break;
                    }
                    docs.add(doc);

import org.codelibs.fess.crawler.exception.RobotsTxtException;

/**
 * Robots.txt Specifications:
 * <ul>
 * <li><a href=
 * "https://developers.google.com/webmasters/control-crawl-index/docs/robots_txt"
 * >https://developers.google.com/webmasters/control-crawl-index/docs/robots_txt
 * </a></li>
 * </ul>
 *
 * @author bowez
 * @author shinsuke
 *
 */
public class RobotsTxtHelper {

    /** Pattern for parsing user-agent records. */

    <root-XML localName="map" namespaceURI="http://docs.oasis-open.org/namespace"/>
    <glob pattern="*.ditamap"/>
  </mime-type>
  <mime-type type="application/dita+xml;format=topic">
    <sub-class-of type="application/dita+xml"/>
    <_comment>DITA Topic</_comment>
    <root-XML localName="topic"/>
    <root-XML localName="topic" namespaceURI="http://docs.oasis-open.org/namespace"/>

    }

    public void test_getText_mswordx() {
        InputStream in = ResourceUtil.getResourceAsStream("extractor/msoffice/test.docx");
        Map<String, String> params = new HashMap<String, String>();
        params.put("resourceName", "test.docx");
        ExtractData extractData = jodExtractor.getText(in, params);
        String content = extractData.getContent();
        CloseableUtil.closeQuietly(in);

Shinsuke Sugaya <******@****.***> 1444529815 +0900

Shinsuke Sugaya <******@****.***> 1444529815 +0900

        assertContentType("application/vnd.openxmlformats-officedocument.wordprocessingml.document", "test/text1.txt", "hoge.docx");
        assertContentType("application/vnd.openxmlformats-officedocument.wordprocessingml.document", "extractor/msoffice/test.docx",
                "hoge.docx");
        assertContentType("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "test/text1.txt", "hoge.xlsx");

Shinsuke Sugaya <******@****.***> 1444529815 +0900

Shinsuke Sugaya <******@****.***> 1444529815 +0900

        private final Node doc;

        public DefaultNamespaceContext(final Node doc) {
            this.doc = doc;
        }

        @Override
        public String getNamespaceURI(final String prefix) {
            return getNamespaceForPrefix(prefix, doc);
        }