* trust OpenSubtitles hash matches, but only the ones with high SeenCount, and scrap any hash match that doesn't unambiguously match a single movie, i.e. scrap hash collisions

source/net/sourceforge/filebot/cli/ScriptShell.lib.groovy

@@ -205,7 +205,7 @@ def detectSeriesName(files, locale = Locale.ENGLISH) {
 	return names == null || names.isEmpty() ? null : names.toList()[0]
 }
 
-def detectMovie(movieFile, strict = true, queryLookupService = TheMovieDB, hashLookupService = null, locale = Locale.ENGLISH) {
+def detectMovie(movieFile, strict = true, queryLookupService = TheMovieDB, hashLookupService = OpenSubtitles, locale = Locale.ENGLISH) {
 	def movies = MediaDetection.detectMovie(movieFile, hashLookupService, queryLookupService, locale, strict)
 	return movies == null || movies.isEmpty() ? null : movies.toList()[0]
 }

source/net/sourceforge/filebot/web/OpenSubtitlesClient.java

@@ -281,7 +281,8 @@ public class OpenSubtitlesClient implements SubtitleProvider, VideoHashSubtitleS
 			for (int bn = 0; bn < ceil((float) hashes.size() / batchSize); bn++) {
 				List<String> batch = hashes.subList(bn * batchSize, min((bn * batchSize) + batchSize, hashes.size()));
 				
-				for (Entry<String, Movie> it : xmlrpc.checkMovieHash(batch).entrySet()) {
+				int minSeenCount = 20; // make sure we don't get mismatches by making sure the hash has not been confirmed numerous times
+				for (Entry<String, Movie> it : xmlrpc.checkMovieHash(batch, minSeenCount).entrySet()) {
 					result.put(hashMap.get(it.getKey()), it.getValue());
 					getCache().putData("getMovieDescriptor", it.getKey(), locale, it.getValue());
 				}

source/net/sourceforge/filebot/web/OpenSubtitlesXmlRpc.java

@@ -223,23 +223,45 @@ public class OpenSubtitlesXmlRpc {
 	
 	
 	@SuppressWarnings("unchecked")
-	public Map<String, Movie> checkMovieHash(Collection<String> hashes) throws XmlRpcFault {
+	public Map<String, Movie> checkMovieHash(Collection<String> hashes, int minSeenCount) throws XmlRpcFault {
-		Map<?, ?> response = invoke("CheckMovieHash", token, hashes);
-		
-		Map<String, ?> movieHashData = (Map<String, ?>) response.get("data");
 		Map<String, Movie> movieHashMap = new HashMap<String, Movie>();
 		
-		for (Entry<String, ?> entry : movieHashData.entrySet()) {
+		Map<?, ?> response = invoke("CheckMovieHash2", token, hashes);
-			// empty associative arrays are deserialized as array, not as map
+		Object payload = response.get("data");
-			if (entry.getValue() instanceof Map) {
-				Map<String, String> info = (Map<String, String>) entry.getValue();
 		
-				String hash = info.get("MovieHash");
+		if (payload instanceof Map) {
-				String name = info.get("MovieName");
+			Map<String, ?> movieHashData = (Map<String, ?>) payload;
-				int year = Integer.parseInt(info.get("MovieYear"));
+			for (Entry<String, ?> entry : movieHashData.entrySet()) {
-				int imdb = Integer.parseInt(info.get("MovieImdbID"));
+				// empty associative arrays are deserialized as array, not as map
+				if (entry.getValue() instanceof List) {
+					String hash = entry.getKey();
+					List<Movie> matches = new ArrayList<Movie>();
 					
-				movieHashMap.put(hash, new Movie(name, year, imdb, -1));
+					List<?> hashMatches = (List<?>) entry.getValue();
+					for (Object match : hashMatches) {
+						if (match instanceof Map) {
+							Map<String, String> info = (Map<String, String>) match;
+							int seenCount = Integer.parseInt(info.get("SeenCount"));
+							
+							// require minimum SeenCount before this hash match is considered trusted
+							if (seenCount >= minSeenCount) {
+								String name = info.get("MovieName");
+								int year = Integer.parseInt(info.get("MovieYear"));
+								int imdb = Integer.parseInt(info.get("MovieImdbID"));
+								
+								matches.add(new Movie(name, year, imdb, -1));
+							}
+						}
+					}
+					
+					if (matches.size() == 1) {
+						// perfect unambiguous match
+						movieHashMap.put(hash, matches.get(0));
+					} else if (matches.size() > 1) {
+						// multiple hash matches => ignore all
+						Logger.getLogger(getClass().getName()).log(Level.WARNING, "Ignore hash match due to hash collision: " + matches);
+					}
+				}
 			}
 		}
 		

test/net/sourceforge/filebot/web/OpenSubtitlesXmlRpcTest.java

@@ -10,15 +10,15 @@ import java.nio.ByteBuffer;
 import java.util.List;
 import java.util.Map;
 
-import org.junit.AfterClass;
-import org.junit.BeforeClass;
-import org.junit.Test;
-
 import net.sourceforge.filebot.web.OpenSubtitlesSubtitleDescriptor.Property;
 import net.sourceforge.filebot.web.OpenSubtitlesXmlRpc.Query;
 import net.sourceforge.filebot.web.OpenSubtitlesXmlRpc.SubFile;
 import net.sourceforge.filebot.web.OpenSubtitlesXmlRpc.TryUploadResponse;
 
+import org.junit.AfterClass;
+import org.junit.BeforeClass;
+import org.junit.Test;
+
 
 public class OpenSubtitlesXmlRpcTest {
 	
@@ -35,7 +35,7 @@ public class OpenSubtitlesXmlRpcTest {
 	@Test
 	public void search() throws Exception {
 		List<Movie> list = xmlrpc.searchMoviesOnIMDB("babylon 5");
-		Movie sample = (Movie) list.get(0);
+		Movie sample = list.get(0);
 		
 		// check sample entry
 		assertEquals("Babylon 5", sample.getName());
@@ -131,7 +131,7 @@ public class OpenSubtitlesXmlRpcTest {
 	
 	@Test
 	public void checkMovieHash() throws Exception {
-		Map<String, Movie> results = xmlrpc.checkMovieHash(singleton("d7aa0275cace4410"));
+		Map<String, Movie> results = xmlrpc.checkMovieHash(singleton("d7aa0275cace4410"), 0);
 		Movie movie = results.get("d7aa0275cace4410");
 		
 		assertEquals("Iron Man", movie.getName());
@@ -142,7 +142,7 @@ public class OpenSubtitlesXmlRpcTest {
 	
 	@Test
 	public void checkMovieHashInvalid() throws Exception {
-		Map<String, Movie> results = xmlrpc.checkMovieHash(singleton("0123456789abcdef"));
+		Map<String, Movie> results = xmlrpc.checkMovieHash(singleton("0123456789abcdef"), 0);
 		
 		// no movie info
 		assertTrue(results.isEmpty());