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

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

    }

    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);

        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");

        // Normal extensions should be allowed
        assertEquals("\\share\\file.txt", validator.validatePath("\\share\\file.txt"));
        assertEquals("\\share\\document.docx", validator.validatePath("\\share\\document.docx"));
        assertEquals("\\share\\archive.tar.gz", validator.validatePath("\\share\\archive.tar.gz"));
    }

    @Test
    public void testInternationalCharacters() throws Exception {

    }

    @ParameterizedTest
    @DisplayName("Test decode with various file names")
    @CsvSource({ "a.txt", "very_long_filename_with_many_characters_to_test_buffer_handling.docx", "file with spaces.pdf", "文件.txt", // Unicode filename
            "file-with-special-chars!@#$%^&().bin" })
    void testDecodeWithVariousFileNames(String fileName) throws SMBProtocolDecodingException {

        extensionMap.put("odt", "txt");
        extensionMap.put("ott", "txt");
        extensionMap.put("sxw", "txt");
        extensionMap.put("rtf", "txt");
        extensionMap.put("doc", "txt");
        extensionMap.put("docx", "txt");
        extensionMap.put("wpd", "txt");
        extensionMap.put("txt", "txt");
        extensionMap.put("html", "txt");
        // Spreadsheet Formats
        extensionMap.put("ods", "tsv");

    void testGetFilename() throws SMBProtocolDecodingException {
        // Prepare test data
        String expectedFilename = "longfilename.docx";
        byte[] buffer = createValidBuffer(expectedFilename, "LONGFI~1.DOC", true);

        // Decode
        fileBothDirectoryInfo.decode(buffer, 0, buffer.length);

        // Verify

        assertEquals("application/msword", extractData.getValues("Content-Type")[0]);
    }

    public void test_getTika_mswordx() {
        final InputStream in = ResourceUtil.getResourceAsStream("extractor/msoffice/test.docx");
        final ExtractData extractData = tikaExtractor.getText(in, null);
        final String content = extractData.getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);

        Map<String, Object> doc1 = new HashMap<>();
        doc1.put("title", "Doc1");
        Map<String, Object> doc2 = new HashMap<>();
        doc2.put("title", "Doc2");
        Map<String, Object> doc3 = new HashMap<>();
        doc3.put("title", "Doc3");
        documentList.add(doc1);
        documentList.add(doc2);
        documentList.add(doc3);