* lots of fine-tuning and adjustments for fully-automated movie detection and episode matching

@see http://www.filebot.net/forums/viewtopic.php?f=4&t=832&p=5360#p5360

source/net/sourceforge/filebot/cli/CmdlineOperations.java

@@ -897,7 +897,7 @@ public class CmdlineOperations implements CmdlineInterface {
 		}
 		
 		// return first and only value
-		return probableMatches;
+		return probableMatches.size() <= 5 ? probableMatches : probableMatches.subList(0, 5);
 	}
 	
 	

source/net/sourceforge/filebot/media/MediaDetection.java

@@ -45,6 +45,7 @@ import net.sourceforge.filebot.similarity.DateMatcher;
 import net.sourceforge.filebot.similarity.DateMetric;
 import net.sourceforge.filebot.similarity.MetricAvg;
 import net.sourceforge.filebot.similarity.NameSimilarityMetric;
+import net.sourceforge.filebot.similarity.NumericSimilarityMetric;
 import net.sourceforge.filebot.similarity.SeasonEpisodeMatcher;
 import net.sourceforge.filebot.similarity.SeasonEpisodeMatcher.SeasonEpisodePattern;
 import net.sourceforge.filebot.similarity.SeasonEpisodeMatcher.SxE;
@@ -589,18 +590,35 @@ public class MediaDetection {
 	}
 	
 	
-	public static <T> List<T> sortBySimilarity(Collection<T> options, Collection<String> terms) throws IOException {
-		List<String> paragon = stripReleaseInfo(terms, true);
-		SimilarityMetric metric = new MetricAvg(new SequenceMatchSimilarity(), new NameSimilarityMetric(), new NameSimilarityMetric() {
+	private static SimilarityMetric getMovieMatchMetric() {
+		return new MetricAvg(new SequenceMatchSimilarity(), new NameSimilarityMetric(), new SequenceMatchSimilarity(0, true), new NumericSimilarityMetric() {
+			
+			private Pattern year = Pattern.compile("\\b\\d{4}\\b");
+			
 			
 			@Override
 			protected String normalize(Object object) {
-				return super.normalize(stripReleaseInfo(object.toString()).replaceAll("\\D+", " ")); // similarity of number patterns
+				Matcher ym = year.matcher(object.toString());
+				StringBuilder sb = new StringBuilder();
+				while (ym.find()) {
+					sb.append(ym.group()).append(' ');
+				}
+				return sb.toString().trim();
+			}
+			
+			
+			@Override
+			public float getSimilarity(Object o1, Object o2) {
+				return super.getSimilarity(o1, o2) * 2; // DOUBLE WEIGHT FOR YEAR MATCH
 			}
 		});
-		
+	}
+	
+	
+	public static <T> List<T> sortBySimilarity(Collection<T> options, Collection<String> terms) throws IOException {
+		List<String> paragon = stripReleaseInfo(terms, true);
 		List<T> sorted = new ArrayList<T>(options);
-		sort(sorted, new SimilarityComparator(metric, paragon.toArray()));
+		sort(sorted, new SimilarityComparator(getMovieMatchMetric(), paragon.toArray()));
 		return sorted;
 	}
 	
@@ -801,8 +819,8 @@ public class MediaDetection {
 		// DEBUG
 		// System.out.format("Query %s: %s%n", queryLookupService.getName(), querySet);
 		
-		final SimilarityMetric metric = new NameSimilarityMetric();
 		final Map<Movie, Float> probabilityMap = new LinkedHashMap<Movie, Float>();
+		final SimilarityMetric metric = getMovieMatchMetric();
 		for (String query : querySet) {
 			for (Movie movie : queryLookupService.searchMovie(query.toLowerCase(), locale)) {
 				probabilityMap.put(movie, metric.getSimilarity(query, movie));

source/net/sourceforge/filebot/similarity/EpisodeMatcher.java

@@ -57,12 +57,8 @@ public class EpisodeMatcher extends Matcher<File, Object> {
 			
 			if (uniqueFiles.equals(uniqueEpisodes)) {
 				Episode[] episodes = episodeSets.get(file).toArray(new Episode[0]);
-				Set<String> seriesNames = new HashSet<String>();
-				for (Episode ep : episodes) {
-					seriesNames.add(ep.getSeriesName());
-				}
 				
-				if (seriesNames.size() == 1) {
+				if (isMultiEpisode(episodes)) {
 					MultiEpisode episode = new MultiEpisode(episodes);
 					disjointMatchCollection.add(new Match<File, Object>(file, episode));
 					modified = true;
@@ -99,4 +95,27 @@ public class EpisodeMatcher extends Matcher<File, Object> {
 		return result;
 	}
 	
+	
+	private boolean isMultiEpisode(Episode[] episodes) {
+		// check episode sequence integrity
+		Integer seqIndex = null;
+		for (Episode ep : episodes) {
+			if (seqIndex != null && !ep.getEpisode().equals(seqIndex + 1))
+				return false;
+			
+			seqIndex = ep.getEpisode();
+		}
+		
+		// check drill-down integrity
+		String seriesName = null;
+		for (Episode ep : episodes) {
+			if (seriesName != null && !seriesName.equals(ep.getSeriesName()))
+				return false;
+			
+			seriesName = ep.getSeriesName();
+		}
+		
+		return true;
+	}
+	
 }

source/net/sourceforge/filebot/similarity/EpisodeMetrics.java

@@ -503,7 +503,7 @@ public enum EpisodeMetrics implements SimilarityMetric {
 						if (seriesInfo != null) {
 							if (seriesInfo.getRatingCount() > 0) {
 								float rating = max(0, seriesInfo.getRating().floatValue());
-								return seriesInfo.getRatingCount() >= 15 ? rating : rating / 2; // PENALIZE SHOWS WITH FEW RATINGS
+								return seriesInfo.getRatingCount() >= 15 ? rating : 0; // PENALIZE SHOWS WITH FEW RATINGS
 							} else {
 								return -1; // BIG PENALTY FOR SHOWS WITH 0 RATINGS
 							}

source/net/sourceforge/filebot/similarity/SequenceMatchSimilarity.java

@@ -11,7 +11,17 @@ import java.util.Locale;
 
 public class SequenceMatchSimilarity implements SimilarityMetric {
 	
-	private final CommonSequenceMatcher commonSequenceMatcher = new CommonSequenceMatcher(getLenientCollator(Locale.ROOT), 10, false);
+	private final CommonSequenceMatcher commonSequenceMatcher;
+	
+	
+	public SequenceMatchSimilarity() {
+		this(10, false);
+	}
+	
+	
+	public SequenceMatchSimilarity(int commonSequenceMaxStartIndex, boolean returnFirstMatch) {
+		this.commonSequenceMatcher = new CommonSequenceMatcher(getLenientCollator(Locale.ROOT), commonSequenceMaxStartIndex, returnFirstMatch);
+	}
 	
 	
 	@Override

website/data/query-blacklist.txt

@@ -92,6 +92,7 @@ A.Release.Lounge
 ABC
 AC3D
 Addic7ed.com
+AMC-TEST
 Anaglyph
 Anime[s]?
 Arte
@@ -142,6 +143,7 @@ Extended.Version
 ExtraScene
 ExtraTorrent
 Fantasy
+FileBot
 Final.Cut
 FIXED
 Fra
@@ -244,6 +246,7 @@ tt\d{2,8}
 TVOON.DE
 tvp
 tvrecorder
+UE
 ultimate.edition
 UNCUT
 unrated

website/data/release-groups.txt

@@ -543,6 +543,7 @@ DeTvaVe
 DEViSE
 DEWSTRR
 DFA
+DFE
 DFQ
 DFTU
 DGAS

website/scripts/amc.groovy

@@ -145,10 +145,10 @@ def groups = input.groupBy{ f ->
 		def mn = norm(mov.name)
 		
 		// S00E00 | 2012.07.21 | One Piece 217 | Firefly - Serenity | [Taken 1, Taken 2, Taken 3, Taken 4, ..., Taken 10]
-		if ((parseEpisodeNumber(fn, true) || parseDate(fn) || ([dn, fn].find{ it =~ sn && matchMovie(it, true) == null } && (parseEpisodeNumber(fn.after(sn), false) || fn.after(sn) =~ /\d{1,2}\D+\d{1,2}/) && matchMovie(fn, true) == null) || (fn.after(sn) ==~ /.{0,3} - .+/ && matchMovie(fn, true) == null) || f.dir.listFiles{ it.isVideo() && norm(it.name) =~ sn && it.name =~ /\b\d{1,3}\b/}.size() >= 10) && !tryQuietly{ def m = detectMovie(f, true); m.year >= 1950 && f.listPath().reverse().take(3).find{ it.name =~ m.year } }) {
+		if ((parseEpisodeNumber(fn, true) || parseDate(fn) || ([dn, fn].find{ it =~ sn && matchMovie(it, true) == null } && (parseEpisodeNumber(fn.after(sn), false) || fn.after(sn) =~ /\d{1,2}\D+\d{1,2}/) && matchMovie(fn, true) == null) || (fn.after(sn) ==~ /.{0,3} - .+/ && matchMovie(fn, true) == null) || f.dir.listFiles{ it.isVideo() && norm(it.name) =~ sn && it.name =~ /\b\d{1,3}\b/}.size() >= 10) && !(mov.year >= 1950 && f.listPath().reverse().take(3).find{ it.name =~ mov.year }) || mov.year < 1900) {
 			_log.fine("Exclude Movie: $mov")
 			mov = null
-		} else if (mn ==~ fn || (detectMovie(f, true) && [dn, fn].find{ it =~ /(19|20)\d{2}/ }) || [dn, fn].find{ it =~ mn && !(it.after(mn) =~ /\b\d{1,3}\b/) && !(it.before(mn).contains(sn)) }) {
+		} else if (mn ==~ fn || [dn, fn].find{ it =~ /\b/+mov.year+/\b/ } || [dn, fn].find{ it =~ mn && !(it.after(mn) =~ /\b\d{1,3}\b/) && !(it.before(mn).contains(sn)) } || (detectMovie(f, true) && [dn, fn].find{ it =~ /(19|20)\d{2}/ })) {
 			_log.fine("Exclude Series: $tvs")
 			tvs = null
 		}