factory.addExtractor("application/x-lha", lhaExtractor);

                })//
        ;

        lhaExtractor = container.getComponent("lhaExtractor");

    }

    public void test_getText() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/lha/test.lzh");
        final String content = lhaExtractor.getText(in, null).getContent();

			<!-- Supported MIME type -->
			<arg>
  "(application/xml"
+ "|application/xhtml+xml"
+ "|application/rdf+xml"
+ "|application/pdf"
+ "|application/x-freemind"
+ "|application/lha"
+ "|application/x-lha"
+ "|application/x-lha-compressed"
+ "|text/xml"
+ "|text/xml-external-parsed-entity"
+ "|text/html)"
			</arg>
		</postConstruct>
	</component>

import org.codelibs.fess.crawler.util.IgnoreCloseInputStream;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import jp.gr.java_conf.dangan.util.lha.LhaFile;
import jp.gr.java_conf.dangan.util.lha.LhaHeader;

/**
 * Extractor implementation for LHA.
 *
 * @author shinsuke
 *
 */
public class LhaExtractor extends AbstractExtractor {

        extractorFactory.addExtractor("application/vnd.visio", tikaExtractor);
        extractorFactory.addExtractor("application/pdf", pdfExtractor);
        extractorFactory.addExtractor("application/x-lha", lhaExtractor);
        extractorFactory.addExtractor("application/x-lharc", lhaExtractor);

    }

    public void test_addExtractor() {
        final ExtractorFactory extractorFactory = new ExtractorFactory();

        assertContentType("application/gzip", "extractor/gz/test.tar.gz", "hoge.tar.gz");
        assertContentType("application/zip", "extractor/zip/test.zip", "hoge.zip");
        assertContentType("application/x-lharc", "extractor/lha/test.lzh", "hoge.lzh"); // TODO is it correct?

        assertContentType("application/xml", "extractor/test.mm", "hoge.mm");

        assertContentType("message/rfc822", "extractor/eml/sample1.eml", "sample1.eml");

		<postConstruct name="addExtractor">
			<arg>[
				"application/pdf"
				]</arg>
			<arg>pdfExtractor</arg>
		</postConstruct>
		<postConstruct name="addExtractor">
			<arg>[
				"application/x-lha",
				"application/x-lharc"
				]</arg>
			<arg>lhaExtractor</arg>
		</postConstruct>
		<postConstruct name="addExtractor">
			<arg>[
				"message/rfc822"
				]</arg>
			<arg>emlExtractor</arg>

      <match value="\377\037" type="string" offset="0"/>
      <match value="0145405" type="host16" offset="0"/>
    </magic>
    <glob pattern="*.bin"/>
    <glob pattern="*.dms"/>
    <glob pattern="*.lha"/>
    <glob pattern="*.lrf"/>
    <glob pattern="*.lzh"/>
    <glob pattern="*.so"/>
    <glob pattern="*.dist"/>
    <glob pattern="*.distz"/>
    <glob pattern="*.pkg"/>
    <glob pattern="*.bpk"/>

0513          ; valid                                  # 5.0  CYRILLIC SMALL LETTER EL WITH HOOK
0514          ; mapped                 ; 0515          # 5.1  CYRILLIC CAPITAL LETTER LHA
0515          ; valid                                  # 5.1  CYRILLIC SMALL LETTER LHA
0516          ; mapped                 ; 0517          # 5.1  CYRILLIC CAPITAL LETTER RHA
0517          ; valid                                  # 5.1  CYRILLIC SMALL LETTER RHA