bugfixes

- fixed missing bounds, by simply omitting those scenes
- fixed duplicate S2 scenes, by also comparing filenames

gms-metadatacrawler/src/main/java/de/potsdam/gfz/gms/metadatacrawler/MetadatacrawlerModule.java

@@ -85,6 +85,16 @@ public class MetadatacrawlerModule {
 		}
 	}
 
+	private static class DBSceneSets {
+		private Set<String>	entityids	= new HashSet<>();
+		private Set<String>	filenames	= new HashSet<>();
+
+		public void clear() {
+			entityids.clear();
+			filenames.clear();
+		}
+	}
+
 	/**
 	 * Returns a set of scene entity ids of the existing scene data in the gms database
 	 * 
@@ -92,11 +102,11 @@ public class MetadatacrawlerModule {
 	 * @return
 	 * @throws Exception
 	 */
-	private static Set<String> getExistingScenes(DatasetMetadata dataset, Timestamp latestAcquisitionDate) throws Exception {
+	private static DBSceneSets getExistingScenes(DatasetMetadata dataset, Timestamp latestAcquisitionDate) throws Exception {
 
 		GeoMultiSensMetadataCrawler gmsCrawler = new GeoMultiSensMetadataCrawler();
 		BlockingQueue<SceneMetadata> queue = new LinkedBlockingQueue<>();
-		Set<String> entityIdSet = new HashSet<>();
+		DBSceneSets dbScenes = new DBSceneSets();
 
 		gmsCrawler.startCrawling(null, null, queue, dataset);
 
@@ -106,7 +116,11 @@ public class MetadatacrawlerModule {
 				continue;
 			}
 
-			entityIdSet.add(scene.entityid);
+			dbScenes.entityids.add(scene.entityid);
+			if (scene.filename != null && !scene.filename.isEmpty()) {
+				// ignore file extension
+				dbScenes.filenames.add(scene.filename.substring(0, scene.filename.lastIndexOf(".")));
+			}
 
 			// check date
 			if (latestAcquisitionDate.before(scene.acquisitiondate)) {
@@ -114,7 +128,7 @@ public class MetadatacrawlerModule {
 			}
 		}
 
-		return entityIdSet;
+		return dbScenes;
 
 	}
 
@@ -129,9 +143,9 @@ public class MetadatacrawlerModule {
 		System.out.print("getting existing scenes for dataset: ");
 
 		Timestamp latestDateInDB = new Timestamp(Timestamp.MIN);
-		Set<String> dbScenes = getExistingScenes(dataset, latestDateInDB);
+		DBSceneSets dbScenes = getExistingScenes(dataset, latestDateInDB);
 
-		System.out.println("done (number of scenes in DB:" + dbScenes.size() + ")");
+		System.out.println("done (number of scenes in DB:" + dbScenes.entityids.size() + ")");
 		System.out.println("latest scene timestamp in DB: " + latestDateInDB.toISO8601String());
 
 		BlockingQueue<SceneMetadata> queue = new LinkedBlockingQueue<>();
@@ -161,12 +175,25 @@ public class MetadatacrawlerModule {
 				continue;
 			}
 
-			if (dbScenes.contains(scene.entityid)) {
+			if (dbScenes.entityids.contains(scene.entityid)) {
 				// we already have this scene information - skip
 				continue;
 			}
 
-			if (scene.entityid == null || scene.acquisitiondate == null) {
+			if (dataset.id == 249) {
+				// the controller will alter the entityid for downloaded scenes - so we need to check the corresponding filename
+
+				if (scene.filename != null && !scene.filename.isEmpty()) {
+					// check file name (ignore file extension)
+
+					if (dbScenes.filenames.contains(scene.filename.substring(0, scene.filename.lastIndexOf(".")))) {
+						// we already have this scene information - skip
+						continue;
+					}
+				}
+			}
+
+			if (scene.entityid == null || scene.acquisitiondate == null || scene.bounds == null) {
 				// skip invalid scenemetadata
 				continue;
 			}