set default properties for tika

Author: lrog

src/main/java/uk/ac/kcl/itemProcessors/TikaDocumentItemProcessor.java

@@ -53,25 +53,26 @@ public class TikaDocumentItemProcessor extends TLItemProcessor implements ItemPr
 
  private static final org.slf4j.Logger LOG = LoggerFactory.getLogger(TikaDocumentItemProcessor.class);
 
- // whether shall we parse all the document data into XHTML format
- @Value("${tika.keepTags}")
- private boolean keepTags;
-
  // the name of the field in the resulting JSON file that will hold the content of the parsed document
  @Value("${tika.tikaFieldName}")
  String tikaFieldName;
 
- // configuration of Tika module plus individual parsers
- private TikaConfig tikaConfig;
 
- @Value("${tika.tesseract.timeout:#{null}}")
+ // whether shall we parse all the document data into XHTML format
+ @Value("${tika.keepTags:false}")
+ private boolean keepTags;
+
+ @Value("${tika.tesseract.timeout:120}")
  Integer tesseractTimeout;
  private TesseractOCRConfig tesseractConfig;
 
- @Value("${tika.convert.timeout:#{null}}")
+ @Value("${tika.convert.timeout:120}")
  Integer convertTimeout;
- private ImageMagickConfig imgConfig;
 
+
+ // configuration of Tika module plus individual parsers
+ private TikaConfig tikaConfig;
+ private ImageMagickConfig imgConfig;
  private AutoDetectParser parser;
 
 
@@ -88,13 +89,13 @@ public void init() throws IOException, SAXException, TikaException{
 
  // load tesseract ocr configuration
  tesseractConfig = new TesseractOCRConfig();
- if (tesseractTimeout != null && tesseractTimeout > 0) {
+ if (tesseractTimeout > 0) {
  tesseractConfig.setTimeout(tesseractTimeout);
  }
 
  // load image magick configuration -- used for tiff conversion
  imgConfig = new ImageMagickConfig();
- if (convertTimeout != null && convertTimeout > 0) {
+ if (convertTimeout > 0) {
  imgConfig.setTimeout(convertTimeout);
  }