import org.codelibs.fess.crawler.util.IgnoreCloseInputStream;

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

/**
 * Extractor implementation for LHA (LZH) archive files.
 * This extractor can extract text content from files within LHA archives
 * by using appropriate extractors for each contained file type.
 *
 * @author shinsuke
 */

                    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>

        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();

#### PDFs and Images
- PDF documents (text and metadata extraction)
- Images (JPEG, PNG, GIF, TIFF, BMP)
- Image metadata (EXIF, IPTC, XMP)

#### Archives and Compressed Files
- ZIP, TAR, GZ archives
- LHA compression format
- Nested archive extraction

#### Web and Markup
- HTML, XHTML with XPath support
- XML documents
- JSON and structured data

#### Media Files
- Audio formats (MP3, WAV, FLAC)

		<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