void testKnownExtensions() throws IOException {
            assertEquals("application/pdf", mimeMap.getMimeType("pdf"));
            assertEquals("application/msword", mimeMap.getMimeType("doc"));
            assertEquals("application/vnd.ms-excel", mimeMap.getMimeType("xls"));
            assertEquals("text/html", mimeMap.getMimeType("html"));

        assertEquals("text", types[2]);
    }

    public void test_getTypes_withDuplicates() {
        fileTypeHelper.add("application/pdf", "document");
        fileTypeHelper.add("application/msword", "document");
        fileTypeHelper.add("text/plain", "text");

        String[] types = fileTypeHelper.getTypes();
        assertEquals(2, types.length);

        Arrays.sort(types);

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

        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");

# Seed value for popular word suggestion.
suggest.popular.word.seed=0
# Tags for popular word suggestion.
suggest.popular.word.tags=
# Fields for popular word suggestion.
suggest.popular.word.fields=
# Excluded words for popular word suggestion.
suggest.popular.word.excludes=
# Number of popular words to suggest.
suggest.popular.word.size=10
# Window size for popular word suggestion.

  /**
   * <a
   * href="https://learn.microsoft.com/en-us/archive/blogs/vsofficedeveloper/office-2007-file-format-mime-types-for-http-content-streaming-2">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

  /**
   * <a
   * href="https://learn.microsoft.com/en-us/archive/blogs/vsofficedeveloper/office-2007-file-format-mime-types-for-http-content-streaming-2">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

    String INDEX_INDICES_TIMEOUT = "index.indices.timeout";

    /** The key of the configuration. e.g. text/html=html<br>
     * application/msword=word<br>
     * application/vnd.openxmlformats-officedocument.wordprocessingml.document=word<br>
     * application/vnd.ms-excel=excel<br>
     * application/vnd.ms-excel.sheet.2=excel<br>
     * application/vnd.ms-excel.sheet.3=excel<br>

        ProtwordsItem item1 = new ProtwordsItem(1, "  word  ");
        assertEquals("  word  ", item1.getInput());
        assertEquals("  word  ", item1.toLineString());

        ProtwordsItem item2 = new ProtwordsItem(2, "\tword\n");
        assertEquals("\tword\n", item2.getInput());
        assertEquals("\tword\n", item2.toLineString());

        ProtwordsItem item3 = new ProtwordsItem(3, " ");

        String word = "test word";
        float boost = 1.5f;
        List<String> readings = Arrays.asList("reading1");
        List<String> fields = Arrays.asList("field1");
        List<String> tags = Arrays.asList("tag1");
        List<String> roles = null;

        elevateWord = new ElevateWord(word, boost, readings, fields, tags, roles);

        assertEquals(word, elevateWord.getElevateWord());