* set foundation for possible future fine-tuning of SxE pattern matching combined with release info knowledge

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

@@ -111,16 +111,20 @@ public class MediaDetection {
 		return releaseInfo.getLanguageSuffix(getName(file));
 	}
 
+	public static SeasonEpisodeMatcher getSeasonEpisodeMatcher(boolean strict) {
+		return new SeasonEpisodeMatcherWithFilter(strict);
+	}
+
 	public static boolean isEpisode(String name, boolean strict) {
 		return parseEpisodeNumber(name, strict) != null || parseDate(name) != null;
 	}
 
 	public static List<SxE> parseEpisodeNumber(String string, boolean strict) {
-		return new SeasonEpisodeMatcher(SeasonEpisodeMatcher.DEFAULT_SANITY, strict).match(string);
+		return getSeasonEpisodeMatcher(strict).match(string);
 	}
 
 	public static List<SxE> parseEpisodeNumber(File file, boolean strict) {
-		return new SeasonEpisodeMatcher(SeasonEpisodeMatcher.DEFAULT_SANITY, strict).match(file);
+		return getSeasonEpisodeMatcher(strict).match(file);
 	}
 
 	public static Date parseDate(Object object) {
@@ -248,7 +252,7 @@ public class MediaDetection {
 
 	public static Object getEpisodeIdentifier(CharSequence name, boolean strict) {
 		// check SxE first
-		Object match = new SeasonEpisodeMatcher(SeasonEpisodeMatcher.DEFAULT_SANITY, true).match(name);
+		Object match = getSeasonEpisodeMatcher(true).match(name);
 
 		// then Date pattern
 		if (match == null) {
@@ -257,7 +261,7 @@ public class MediaDetection {
 
 		// check SxE non-strict
 		if (match == null && !strict) {
-			match = new SeasonEpisodeMatcher(SeasonEpisodeMatcher.DEFAULT_SANITY, false).match(name);
+			match = getSeasonEpisodeMatcher(false).match(name);
 		}
 
 		return match;
@@ -1040,6 +1044,33 @@ public class MediaDetection {
 		}
 	}
 
+	private static class SeasonEpisodeMatcherWithFilter extends SeasonEpisodeMatcher {
+
+		private final Pattern ignorePattern = MediaDetection.releaseInfo.getVideoFormatPattern(false);
+
+		public SeasonEpisodeMatcherWithFilter(boolean strict) {
+			super(DEFAULT_SANITY, strict);
+		}
+
+		protected String clean(CharSequence name) {
+			return ignorePattern.matcher(name).replaceAll("");
+		}
+
+		@Override
+		public List<SxE> match(CharSequence name) {
+			return super.match(clean(name));
+		}
+
+		@Override
+		protected List<String> tokenizeTail(File file) {
+			List<String> tail = super.tokenizeTail(file);
+			for (int i = 0; i < tail.size(); i++) {
+				tail.set(i, clean(tail.get(i)));
+			}
+			return tail;
+		}
+	}
+
 	public static void storeMetaInfo(File file, Object model) {
 		// only for Episode / Movie objects
 		if ((model instanceof Episode || model instanceof Movie) && file.exists()) {

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

@@ -119,7 +119,7 @@ public class ReleaseInfo {
 				Pattern languageSuffix = getLanguageSuffixPattern(languages, strict);
 				Pattern languageTag = getLanguageTagPattern(languages);
 				Pattern videoSource = getVideoSourcePattern();
-				Pattern videoFormat = getVideoFormatPattern();
+				Pattern videoFormat = getVideoFormatPattern(strict);
 				Pattern resolution = getResolutionPattern();
 				Pattern queryBlacklist = getBlacklistPattern();
 
@@ -226,10 +226,10 @@ public class ReleaseInfo {
 		return compile("(?<!\\p{Alnum})(\\d{4}|[6-9]\\d{2})x(\\d{4}|[4-9]\\d{2})(?!\\p{Alnum})");
 	}
 
-	public Pattern getVideoFormatPattern() {
+	public Pattern getVideoFormatPattern(boolean strict) {
 		// pattern matching any video source name
 		String pattern = getBundle(getClass().getName()).getString("pattern.video.format");
-		return compile("(?<!\\p{Alnum})(" + pattern + ")(?!\\p{Alnum})", CASE_INSENSITIVE);
+		return strict ? compile("(?<!\\p{Alnum})(" + pattern + ")(?!\\p{Alnum})", CASE_INSENSITIVE) : compile(pattern, CASE_INSENSITIVE);
 	}
 
 	public Pattern getVideoSourcePattern() {

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

@@ -107,7 +107,7 @@ public class SeasonEpisodeMatcher {
 			}
 		};
 
-		// match patterns like 101, 102 (and greedily just grab the first)
+		// (last-resort) match patterns like 101, 102 (and greedily just grab the first)
 		Num101_SUBSTRING = new SeasonEpisodePattern(STRICT_SANITY, "(\\d{1})(\\d{2}).+") {
 
 			@Override
@@ -148,17 +148,17 @@ public class SeasonEpisodeMatcher {
 
 	public List<SxE> match(File file) {
 		// take folder name into consideration as much as file name but put priority on file name
-		List<File> pathTail = listPathTail(file, 2, true);
+		List<String> tail = tokenizeTail(file);
 
 		for (SeasonEpisodeParser pattern : patterns) {
-			for (File tail : pathTail) {
-				List<SxE> match = pattern.match(tail.getName());
+			for (int t = 0; t < tail.size(); t++) {
+				List<SxE> match = pattern.match(tail.get(t));
 
 				if (!match.isEmpty()) {
 					// current pattern did match
 					for (int i = 0; i < match.size(); i++) {
-						if (match.get(i).season < 0) {
-							Matcher sm = seasonPattern.matcher(file.getPath());
+						if (match.get(i).season < 0 && t < tail.size() - 1) {
+							Matcher sm = seasonPattern.matcher(tail.get(t + 1));
 							if (sm.find()) {
 								match.set(i, new SxE(Integer.parseInt(sm.group(1)), match.get(i).episode));
 							}
@@ -171,6 +171,14 @@ public class SeasonEpisodeMatcher {
 		return null;
 	}
 
+	protected List<String> tokenizeTail(File file) {
+		List<String> tail = new ArrayList<String>(2);
+		for (File f : listPathTail(file, 2, true)) {
+			tail.add(getName(f));
+		}
+		return tail;
+	}
+
 	public int find(CharSequence name, int fromIndex) {
 		for (SeasonEpisodeParser pattern : patterns) {
 			int index = pattern.find(name, fromIndex);