exception = new SmbOperationException(SmbOperationException.ErrorCode.FILE_EXISTS, "document.pdf already exists");
        assertTrue(exception.getMessage().contains("FILE_EXISTS"));
        assertTrue(exception.getMessage().contains("File already exists"));
        assertTrue(exception.getMessage().contains("document.pdf already exists"));

        // Without custom message

        pdfResult.setMethod("GET");
        pdfResult.setMimeType("application/pdf");
        pdfResult.setParentUrl("http://www.parent.com/");
        pdfResult.setRuleId("pdfRule");
        pdfResult.setSessionId(sessionId);
        pdfResult.setStatus(200);
        pdfResult.setUrl("http://www.example.com/document.pdf");
        dataService.store(pdfResult);

        // Filter by MIME type

						<option value="html" <c:if test="${as.filetype.contains('html')}">selected</c:if>><la:message
								key="labels.advance_search_filetype_html"
							/></option>
						<option value="pdf" <c:if test="${as.filetype.contains('pdf')}">selected</c:if>><la:message
								key="labels.advance_search_filetype_pdf"
							/></option>
						<option value="word" <c:if test="${as.filetype.contains('word')}">selected</c:if>><la:message

        DataStoreParams paramMap = new DataStoreParams();
        paramMap.put("url_exclude_pattern", ".*\\.pdf$");

        boolean result1 = indexUpdateCallback.isUrlCrawlable(paramMap, "http://example.com/test.html");
        boolean result2 = indexUpdateCallback.isUrlCrawlable(paramMap, "http://example.com/test.pdf");

        assertTrue(result1);
        assertFalse(result2);

    }

    @Test
    public void test_parseExtraQueries_sameGroupOrJoin() {
        setupViewHelperWithFacetGroups(
                createFacetQueryView("filetype", "HTML", "filetype:HTML", "WORD", "filetype:WORD", "PDF", "filetype:PDF"));
        final MockHttpServletRequest request = new MockHttpServletRequest();
        request.setParameterValues("ex_q", new String[] { "filetype:HTML", "filetype:WORD" });

        when(locator.getCanonicalURL()).thenReturn("smb://server/share/");

        // Mock directory listing
        SmbFile file1 = mock(SmbFile.class);
        when(file1.getName()).thenReturn("document.pdf");
        when(file1.isDirectory()).thenReturn(false);
        when(file1.length()).thenReturn(2048L);
        when(file1.lastModified()).thenReturn(System.currentTimeMillis());

   * href="ftp://www.3gpp.org/tsg_sa/TSG_SA/TSGS_23/Docs/PDF/SP-040065.pdf#page=10">3GPP TS
   * 26.244</a>.
   *
   * @since 20.0
   */
  public static final MediaType THREE_GPP_VIDEO = createConstant(VIDEO_TYPE, "3gpp");

  /**
   * The 3G2 multimedia container format. For more information, see <a
   * href="http://www.3gpp2.org/Public_html/specs/C.S0050-B_v1.0_070521.pdf#page=16">3GPP2
   * C.S0050-B</a>.
   *

### Protocols

HTTP/HTTPS, File, FTP/FTPS, SMB/CIFS (SMB1/SMB2+), Storage (MinIO via `storage://`), S3 (`s3://`), GCS (`gcs://`)

### Content Formats

Office (Word, Excel, PowerPoint), PDF, Archives (ZIP, TAR, GZ, LHA), HTML, XML, JSON, Markdown, Media metadata, Images (EXIF/IPTC/XMP), Email (EML)

---

## Architecture

### Module Structure

```
fess-crawler-parent/

            }
            return buf.toString();
        }
        return url;
    }

    /**
     * Appends search parameters to PDF URLs for highlighting.
     *
     * @param document the document data map
     * @param url the PDF URL
     * @return the URL with search parameters
     */
    protected String appendPDFSearchWord(final Map<String, Object> document, final String url) {

        assertEquals(128, result.length());
    }

    @Test
    public void test_generateId_urlWithUnicode() {
        String result = crawlingInfoHelper.generateId("file:///home/ユーザー/ドキュメント.pdf");
        assertNotNull(result);
        assertEquals(128, result.length());
    }

    @Test
    public void test_generateId_deterministic() {