[#336] fix disable OCR when --ocr=false

ICIJ · Feb 28, 2020 · e74f924 · e74f924
1 parent 48685a6
commit e74f924
Showing 1 changed file with 5 additions and 0 deletions.
diff --git a/datashare-app/src/main/java/org/icij/datashare/tasks/IndexTask.java b/datashare-app/src/main/java/org/icij/datashare/tasks/IndexTask.java
@@ -60,6 +60,11 @@ public IndexTask(final ElasticsearchSpewer spewer, final Publisher publisher, fi
         Extractor extractor = new Extractor(new DocumentFactory().withIdentifier(new DigestIdentifier(Entity.HASHER.toString(), Charset.defaultCharset())));
         extractor.setDigester(new UpdatableDigester(indexName, Entity.HASHER.toString()));
 
+        logger.info(String.valueOf(properties));
+        if ("false".equals(properties.getProperty("ocr"))) {
+            logger.info("disable OCR");
+            extractor.disableOcr();
+        }
         consumer = new DocumentConsumer(spewer, extractor, this.parallelism);
         if (propertiesProvider.getProperties().get(MAP_NAME_OPTION) != null) {
             logger.info("report map enabled with name set to {}", propertiesProvider.getProperties().get(MAP_NAME_OPTION));