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

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

        url = "http://test.com/hoge1.pdf";
        resourceName = null;
        params.put(ExtractData.URL, url);
        params.put(ExtractData.RESOURCE_NAME_KEY, resourceName);
        assertNull(pdfExtractor.getPassword(params));

        url = "http://test.com/hoge1.pdf";
        resourceName = "hoge2.pdf";
        params.put(ExtractData.URL, url);

        final Map<String, String> params = new HashMap<>();
        params.put(ExtractData.RESOURCE_NAME_KEY, "test-document.pdf");

        final ExtractData result = filenameExtractor.getText(in, params);

        assertNotNull(result);
        assertEquals("test-document.pdf", result.getContent());
    }

    /**
     * Test extraction with null parameters map.
     */

        void testCaseInsensitiveExtensions() throws IOException {
            assertEquals("application/pdf", mimeMap.getMimeType("PDF"));
            assertEquals("application/pdf", mimeMap.getMimeType("Pdf"));
            assertEquals("application/pdf", mimeMap.getMimeType("pDf"));
            assertEquals("text/html", mimeMap.getMimeType("HTML"));
            assertEquals("text/html", mimeMap.getMimeType("HtMl"));

		<property name="commandList">
			["${path}/generate-thumbnail",
			"pdf",
			"${url}",
			"${outputFile}"]
		</property>
		<property name="generatorList">
			["${path}/generate-thumbnail"]
		</property>
		<postConstruct name="addCondition">
			<arg>"mimetype"</arg>
			<arg>"application/pdf"
			</arg>
		</postConstruct>
		<postConstruct name="register"></postConstruct>
	</component>

		</postConstruct>
		<postConstruct name="addRule">
			<arg>"mimeType"</arg>
			<!-- Supported MIME type -->
			<arg>
  "(application/xml"
+ "|application/xhtml+xml"
+ "|application/rdf+xml"
+ "|application/pdf"
+ "|application/x-freemind"
+ "|text/xml"
+ "|text/xml-external-parsed-entity)"
			</arg>
		</postConstruct>
	</component>

	<component name="fsFileRule" class="org.codelibs.fess.crawler.rule.impl.RegexRule" >

        }
    }

    /**
     * Adds or updates a MIME type to file type mapping.
     *
     * @param mimetype the MIME type to map (e.g., "application/pdf")
     * @param filetype the file type classification (e.g., "pdf")
     */
    public void add(final String mimetype, final String filetype) {
        mimetypeMap.put(mimetype, filetype);
    }

    /**

application/msword             doc              # Microsoft Word
application/octet-stream       bin exe ani      # Binary File
application/oda                oda
application/pagemaker          pm5 pt5 pm       # PageMaker
application/pdf                pdf              # Adobe Acrobat
application/postscript         ai eps ps        # Postscript File
application/rtf                rtf              # Rich Text File
application/toolbook           tbk              # Toolbook

  if [[ -z "${im_cmd}" ]] ; then
    echo "ImageMagick (convert or magick) does not exist."
    exit 1
  fi
  check_command pdftoppm
  check_command unoconv
  tmp_pdf_file=/tmp/thumbnail.$$.pdf
  unoconv -e PageRange=1-1 -o ${tmp_pdf_file} -f pdf "${target_file}"
  if [[ ! -f ${tmp_pdf_file} ]] ; then
    echo "unoconv does not work."
    exit 1
  fi
  tmp_png_prefix=/tmp/thumbnail.png.$$