From 811f945daf9ee431faf6423b7528e9b4c7329d69 Mon Sep 17 00:00:00 2001
From: Reinhard Pointner <rednoah@users.noreply.github.com>
Date: Wed, 4 Jul 2012 10:20:52 +0000
Subject: [PATCH] * improved series detection for certain cases

---
 .../filebot/media/MediaDetection.java         | 20 +++++++++++-------
 .../filebot/media/ReleaseInfo.java            |  5 ++++-
 website/data/query-blacklist.txt              |  4 +++-
 website/data/release-groups.txt               | 21 +++++++++++--------
 4 files changed, 32 insertions(+), 18 deletions(-)

diff --git a/source/net/sourceforge/filebot/media/MediaDetection.java b/source/net/sourceforge/filebot/media/MediaDetection.java
index e174bc1b..329bd828 100644
--- a/source/net/sourceforge/filebot/media/MediaDetection.java
+++ b/source/net/sourceforge/filebot/media/MediaDetection.java
@@ -214,14 +214,20 @@ public class MediaDetection {
 			}
 			
 			// check foldernames first
-			List<String> matches = matchSeriesByName(folders);
+			List<String> matches = matchSeriesByName(folders, 0);
 			
 			// check all filenames if necessary
 			if (matches.isEmpty()) {
-				matches = matchSeriesByName(filenames);
+				matches = matchSeriesByName(filenames, 0);
 			}
 			
-			names.addAll(matches);
+			// use lenient sub sequence matching only as fallback
+			if (matches.size() > 0) {
+				names.addAll(matches);
+			} else {
+				names.addAll(matchSeriesByName(folders, 3));
+				names.addAll(matchSeriesByName(filenames, 3));
+			}
 		} catch (Exception e) {
 			Logger.getLogger(MediaDetection.class.getClass().getName()).log(Level.WARNING, "Failed to match folder structure: " + e.getMessage(), e);
 		}
@@ -244,8 +250,8 @@ public class MediaDetection {
 	}
 	
 	
-	public static List<String> matchSeriesByName(Collection<String> names) throws Exception {
-		HighPerformanceMatcher nameMatcher = new HighPerformanceMatcher(0);
+	public static List<String> matchSeriesByName(Collection<String> names, int maxStartIndex) throws Exception {
+		HighPerformanceMatcher nameMatcher = new HighPerformanceMatcher(maxStartIndex);
 		List<String> matches = new ArrayList<String>();
 		
 		String[] seriesIndex = releaseInfo.getSeriesList();
@@ -602,8 +608,8 @@ public class MediaDetection {
 		private static final Map<String, CollationKey[]> transformCache = synchronizedMap(new WeakHashMap<String, CollationKey[]>(65536));
 		
 		
-		public HighPerformanceMatcher(int commonWordSequenceMaxStartIndex) {
-			super(collator, commonWordSequenceMaxStartIndex);
+		public HighPerformanceMatcher(int maxStartIndex) {
+			super(collator, maxStartIndex);
 		}
 		
 		
diff --git a/source/net/sourceforge/filebot/media/ReleaseInfo.java b/source/net/sourceforge/filebot/media/ReleaseInfo.java
index ce576e91..bb1d8256 100644
--- a/source/net/sourceforge/filebot/media/ReleaseInfo.java
+++ b/source/net/sourceforge/filebot/media/ReleaseInfo.java
@@ -146,7 +146,10 @@ public class ReleaseInfo {
 		for (Pattern it : stopwords) {
 			Matcher matcher = it.matcher(item);
 			if (matcher.find()) {
-				item = item.substring(0, matcher.start()); // use substring before the matched stopword
+				String substring = item.substring(0, matcher.start()); // use substring before the matched stopword
+				if (normalizePunctuation(substring).length() >= 3) {
+					item = substring; // make sure that the substring has enough data
+				}
 			}
 		}
 		return item;
diff --git a/website/data/query-blacklist.txt b/website/data/query-blacklist.txt
index a4daf721..05e5ea8b 100644
--- a/website/data/query-blacklist.txt
+++ b/website/data/query-blacklist.txt
@@ -14,6 +14,7 @@
 [1-9].?of.?[1-9]
 ^(TV.)?(Show|Serie)[s]?
 ^[0-9]{1,2}[.]
+^[lp][^\p{Alnum}]
 ^AUDIO_TS$
 ^BDMV$
 ^Cover
@@ -21,7 +22,6 @@
 ^Film[s]?
 ^HVDVD_TS$
 ^Info
-^l[^\p{Alnum}]
 ^Movie[s]?
 ^New$
 ^Other$
@@ -34,6 +34,7 @@
 ^VIDEO_TS$
 A.Release.Lounge
 Anime[s]?
+BBC
 By.Cool.Release
 CD[0]?[1-3]
 CN
@@ -121,6 +122,7 @@ unrated
 unrated.edition
 UsaBit.com
 Video[s]?
+www.speed.cd
 www.torentz.3xforum.ro
 www.Torrenting.com
 www[.]
diff --git a/website/data/release-groups.txt b/website/data/release-groups.txt
index 4daf1a77..a19f4efc 100644
--- a/website/data/release-groups.txt
+++ b/website/data/release-groups.txt
@@ -89,16 +89,16 @@ BTSD
 BTT
 BugZ
 BULLDOZER
-Bunny
+BUNNY
 BWB
 C4TV
 CAMELOT
 CBGB
 CDD
 CDDHD
-Chakra
+CHAKRA
 chaostheory
-Chara
+CHARA
 charliebartlett
 CHD
 CHDBits
@@ -114,6 +114,7 @@ CiNEFiLE
 CiNEFOX
 CLASSiC
 CLDD
+CLUE
 cntc
 COALiTiON
 Cocksure
@@ -139,7 +140,6 @@ danger2u
 danirl
 DARM
 DASH
-DATA
 DAW
 DCA
 DDC
@@ -178,7 +178,7 @@ DNR
 DON
 DoNE
 DOT
-doubt
+DOUBT
 DOWN
 DRHD
 DUPLI
@@ -235,7 +235,7 @@ FilmHD
 FiNaLe
 fjall
 FLAiTE
-Flaket
+FLAKET
 fLAMEhd
 FLAWL3SS
 Flomp-Rumbel
@@ -251,7 +251,7 @@ FQM
 FRAGMENT
 FraMeSToR
 FRIGGHD
-Frost
+FROST
 FSiHD
 Ft4U
 FTVDT
@@ -259,7 +259,7 @@ FTW-FM
 FTW-HD
 fty
 FUCT
-Funner
+FUNNER
 FXG
 FxM
 G3N3
@@ -268,7 +268,6 @@ Gazdi
 GB
 GECKOS
 GEHENNA
-george.c
 GFW
 GFY
 GiNJi
@@ -557,7 +556,9 @@ RTA
 RUBY
 RuDE
 RUDOS
+RUNNER
 RUSTLE
+RWD
 Ryugan
 S26
 SAiMORNY
@@ -660,6 +661,7 @@ TRUEFRENCH
 tRuEHD
 TsH
 tsn
+TURKiSO
 TUSAHD
 TVA
 TW
@@ -700,6 +702,7 @@ w4f
 WAF
 WANKAZ
 WASTE
+WAT
 WAVEY
 WHATELSE
 WHiiZz