From 9588603206c8c3bc6650416b71f0f635968e0fbb Mon Sep 17 00:00:00 2001
From: Reinhard Pointner <rednoah@users.noreply.github.com>
Date: Wed, 27 Nov 2013 17:09:19 +0000
Subject: [PATCH] * extra last-resort SxE pattern for space-less naming

@see
http://www.filebot.net/forums/viewtopic.php?f=4&t=1138
---
 .../filebot/media/MediaDetection.java            | 10 ++++++++--
 .../filebot/similarity/SeasonEpisodeMatcher.java | 16 +++++++++++++---
 .../similarity/SeasonEpisodeMatcherTest.java     |  7 +++++--
 3 files changed, 26 insertions(+), 7 deletions(-)

diff --git a/source/net/sourceforge/filebot/media/MediaDetection.java b/source/net/sourceforge/filebot/media/MediaDetection.java
index 070494ae..d8763177 100644
--- a/source/net/sourceforge/filebot/media/MediaDetection.java
+++ b/source/net/sourceforge/filebot/media/MediaDetection.java
@@ -248,11 +248,17 @@ public class MediaDetection {
 
 	public static Object getEpisodeIdentifier(CharSequence name, boolean strict) {
 		// check SxE first
-		Object match = new SeasonEpisodeMatcher(SeasonEpisodeMatcher.DEFAULT_SANITY, strict).match(name);
+		Object match = new SeasonEpisodeMatcher(SeasonEpisodeMatcher.DEFAULT_SANITY, true).match(name);
 
 		// then Date pattern
-		if (match == null)
+		if (match == null) {
 			match = new DateMatcher().match(name);
+		}
+
+		// check SxE non-strict
+		if (match == null && !strict) {
+			match = new SeasonEpisodeMatcher(SeasonEpisodeMatcher.DEFAULT_SANITY, false).match(name);
+		}
 
 		return match;
 	}
diff --git a/source/net/sourceforge/filebot/similarity/SeasonEpisodeMatcher.java b/source/net/sourceforge/filebot/similarity/SeasonEpisodeMatcher.java
index 1c193038..f17e11a8 100644
--- a/source/net/sourceforge/filebot/similarity/SeasonEpisodeMatcher.java
+++ b/source/net/sourceforge/filebot/similarity/SeasonEpisodeMatcher.java
@@ -19,13 +19,14 @@ import java.util.regex.Pattern;
 public class SeasonEpisodeMatcher {
 
 	public static final SeasonEpisodeFilter DEFAULT_SANITY = new SeasonEpisodeFilter(50, 50, 1000, 1970, 2100);
+	public static final SeasonEpisodeFilter STRICT_SANITY = new SeasonEpisodeFilter(10, 30, -1, -1, -1);
 
 	private SeasonEpisodeParser[] patterns;
 	private Pattern seasonPattern;
 
 	public SeasonEpisodeMatcher(SeasonEpisodeFilter sanity, boolean strict) {
 		// define variables
-		SeasonEpisodePattern Season_00_Episode_00, S00E00, SxE, Dot101, EP0, Num101;
+		SeasonEpisodePattern Season_00_Episode_00, S00E00, SxE, Dot101, EP0, Num101_TOKEN, Num101_SUBSTRING;
 
 		// match patterns like Season 01 Episode 02, ...
 		Season_00_Episode_00 = new SeasonEpisodePattern(null, "(?<!\\p{Alnum})(?i:season|series)[^\\p{Alnum}]{0,3}(\\d{1,4})[^\\p{Alnum}]{0,3}(?i:episode)[^\\p{Alnum}]{0,3}(\\d{1,4})[^\\p{Alnum}]{0,3}(?!\\p{Digit})");
@@ -83,7 +84,7 @@ public class SeasonEpisodeMatcher {
 		};
 
 		// match patterns like 01, 102, 1003, 10102 (enclosed in separators)
-		Num101 = new SeasonEpisodePattern(sanity, "(?<!\\p{Alnum})([0-2]?\\d?)(\\d{2})(\\d{2})?(?!\\p{Alnum})") {
+		Num101_TOKEN = new SeasonEpisodePattern(sanity, "(?<!\\p{Alnum})([0-2]?\\d?)(\\d{2})(\\d{2})?(?!\\p{Alnum})") {
 
 			@Override
 			protected Collection<SxE> process(MatchResult match) {
@@ -106,11 +107,20 @@ public class SeasonEpisodeMatcher {
 			}
 		};
 
+		// match patterns like 101, 102 (and greedily just grab the first)
+		Num101_SUBSTRING = new SeasonEpisodePattern(STRICT_SANITY, "(\\d{1})(\\d{2}).+") {
+
+			@Override
+			protected Collection<SxE> process(MatchResult match) {
+				return singleton(new SxE(match.group(1), match.group(2)));
+			}
+		};
+
 		// only use S00E00 and SxE pattern in strict mode
 		if (strict) {
 			patterns = new SeasonEpisodeParser[] { Season_00_Episode_00, S00E00, SxE, Dot101 };
 		} else {
-			patterns = new SeasonEpisodeParser[] { Season_00_Episode_00, S00E00, SxE, Dot101, new SeasonEpisodeUnion(EP0, Num101) };
+			patterns = new SeasonEpisodeParser[] { Season_00_Episode_00, S00E00, SxE, Dot101, new SeasonEpisodeUnion(EP0, Num101_TOKEN), Num101_SUBSTRING };
 		}
 
 		// season folder pattern for complementing partial sxe info from filename
diff --git a/test/net/sourceforge/filebot/similarity/SeasonEpisodeMatcherTest.java b/test/net/sourceforge/filebot/similarity/SeasonEpisodeMatcherTest.java
index 0eff1a11..e9581497 100644
--- a/test/net/sourceforge/filebot/similarity/SeasonEpisodeMatcherTest.java
+++ b/test/net/sourceforge/filebot/similarity/SeasonEpisodeMatcherTest.java
@@ -74,14 +74,17 @@ public class SeasonEpisodeMatcherTest {
 		// test season digits <= 19
 		assertEquals(new SxE(null, 16), matcher.match("E16").get(0));
 
-		// test look-behind
-		assertEquals(null, matcher.match("720p"));
+		// test look-ahead
+		assertEquals(asList(new SxE(7, 20)), matcher.match("720p"));
 
 		// test ambiguous match processing
 		assertEquals(asList(new SxE(1, 1), new SxE(UNDEFINED, 101)), matcher.match("Test.101"));
 
 		// test 4-digit
 		assertEquals(asList(new SxE(23, 21)), matcher.match("the.simpsons.2321.hdtv-lol"));
+
+		// test Num101_SUBSTRING
+		assertEquals(asList(new SxE(4, 07)), matcher.match("TWalkingDead4071080p"));
 	}
 
 	@Test