switch inst.MemBytes {
		case 1:
			prefix = "byte "
		case 2:
			prefix = "word "
		case 4:
			prefix = "dword "
		case 8:
			prefix = "qword "
		case 16:
			prefix = "xmmword "
		case 32:
			prefix = "ymmword "
		}
		switch inst.Op {
		case INVLPG:
			prefix = "byte "
		case STOSB, MOVSB, CMPSB, LODSB, SCASB:
			prefix = "byte "
		case STOSW, MOVSW, CMPSW, LODSW, SCASW:
			prefix = "word "

        LhaExtractor lhaExtractor = container.getComponent("lhaExtractor");
        PasswordBasedExtractor pdfExtractor = container.getComponent("pdfExtractor");
        extractorFactory.addExtractor("application/msword", tikaExtractor);
        extractorFactory.addExtractor("application/vnd.ms-excel", tikaExtractor);
        extractorFactory.addExtractor("application/vnd.ms-powerpoint", tikaExtractor);

        assertContentType("text/html", "html/test1.html", "hoge.htm");
        assertContentType("text/html", "html/test1.shtml", "hoge.shtml");

        assertContentType("application/msword", "test/text1.txt", "hoge.doc");
        assertContentType("application/msword", "extractor/msoffice/test.doc", "hoge.doc");
        assertContentType("application/vnd.ms-excel", "test/text1.txt", "hoge.xls");

			<arg>"mimetype"</arg>
			<arg>"application/vnd.openxmlformats-officedocument.presentationml.presentation"</arg>
		</postConstruct>
		<postConstruct name="addCondition">
			<arg>"mimetype"</arg>
			<arg>"application/msword"</arg>
		</postConstruct>
		<postConstruct name="addCondition">
			<arg>"mimetype"</arg>
			<arg>"application/vnd.ms-excel"</arg>
		</postConstruct>
		<postConstruct name="addCondition">
			<arg>"mimetype"</arg>

application/lotus-123          wks              # Lotus 123
application/mac-binhex40       hqx              # Macintosh binhexed archives
application/mspowerpoint       ppt              # Microsoft Powerpoint
application/msword             doc              # Microsoft Word
application/octet-stream       bin exe ani      # Binary File
application/oda                oda
application/pagemaker          pm5 pt5 pm       # PageMaker

# cache
crawler.document.cache.enabled=true
crawler.document.cache.max.size=2621440
crawler.document.cache.supported.mimetypes=text/html

        assertEquals("ぶんるい", extractData.getValues("cp:category")[0]);
        assertEquals("1", extractData.getValues("meta:page-count")[0]);
        assertEquals("application/msword", extractData.getValues("Content-Type")[0]);
    }

    public void test_getTika_mswordx() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/msoffice/test.docx");

      createConstant(APPLICATION_TYPE, "vnd.ms-powerpoint");

  /** <a href="http://goo.gl/XDQ1h2">Microsoft Word</a> documents. */
  public static final MediaType MICROSOFT_WORD = createConstant(APPLICATION_TYPE, "msword");

  /**
   * Media type for <a
   * href="https://en.wikipedia.org/wiki/Dynamic_Adaptive_Streaming_over_HTTP">Dynamic Adaptive
   * Streaming over HTTP (DASH)</a>. This is <a

      createConstant(APPLICATION_TYPE, "vnd.ms-powerpoint");

  /** <a href="http://goo.gl/XDQ1h2">Microsoft Word</a> documents. */
  public static final MediaType MICROSOFT_WORD = createConstant(APPLICATION_TYPE, "msword");

  /**
   * Media type for <a
   * href="https://en.wikipedia.org/wiki/Dynamic_Adaptive_Streaming_over_HTTP">Dynamic Adaptive
   * Streaming over HTTP (DASH)</a>. This is <a

				"application/mosskey-data",
				"application/mosskey-request",
				"application/mp4",
				"application/mpeg4-generic",
				"application/mpeg4-iod",
				"application/mpeg4-iod-xmt",
				"application/msword",
				"application/msword2",
				"application/msword5",
				"application/mxf",
				"application/nasdata",
				"application/news-checkgroups",
				"application/news-groupinfo",