From b2fbba3a2dc16d60814a54f35371f42e68d6bd2a Mon Sep 17 00:00:00 2001
From: Reinhard Pointner <rednoah@users.noreply.github.com>
Date: Thu, 9 Feb 2012 13:50:14 +0000
Subject: [PATCH] * improved auto-detection for date-based episodes

---
 .../filebot/cli/CmdlineOperations.java        |  2 +-
 .../filebot/similarity/DateMatcher.java       | 98 +++++++++++++++++++
 .../filebot/similarity/DateMetric.java        | 62 ++----------
 .../similarity/SeasonEpisodeMatcher.java      |  2 +-
 .../filebot/similarity/SeriesNameMatcher.java | 33 +++++--
 .../filebot/ui/rename/EpisodeListMatcher.java |  2 +-
 .../similarity/SeriesNameMatcherTest.java     |  4 +-
 7 files changed, 137 insertions(+), 66 deletions(-)
 create mode 100644 source/net/sourceforge/filebot/similarity/DateMatcher.java

diff --git a/source/net/sourceforge/filebot/cli/CmdlineOperations.java b/source/net/sourceforge/filebot/cli/CmdlineOperations.java
index e4021de6..fa043c2e 100644
--- a/source/net/sourceforge/filebot/cli/CmdlineOperations.java
+++ b/source/net/sourceforge/filebot/cli/CmdlineOperations.java
@@ -106,7 +106,7 @@ public class CmdlineOperations implements CmdlineInterface {
 		
 		for (File f : mediaFiles) {
 			// count SxE matches
-			if (nameMatcher.matchBySeasonEpisodePattern(f.getName()) != null) {
+			if (nameMatcher.matchByEpisodeIdentifier(f.getName()) != null) {
 				sxe++;
 			}
 			
diff --git a/source/net/sourceforge/filebot/similarity/DateMatcher.java b/source/net/sourceforge/filebot/similarity/DateMatcher.java
new file mode 100644
index 00000000..3f215018
--- /dev/null
+++ b/source/net/sourceforge/filebot/similarity/DateMatcher.java
@@ -0,0 +1,98 @@
+
+package net.sourceforge.filebot.similarity;
+
+
+import java.util.regex.MatchResult;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+import net.sourceforge.filebot.web.Date;
+
+
+public class DateMatcher {
+	
+	private final DatePattern[] patterns;
+	
+	
+	public DateMatcher() {
+		patterns = new DatePattern[2];
+		
+		// match yyyy-mm-dd patterns like 2010-10-24, 2009/6/1, etc.
+		patterns[0] = new DatePattern("(?<!\\p{Alnum})(\\d{4})[^\\p{Alnum}](\\d{1,2})[^\\p{Alnum}](\\d{1,2})(?!\\p{Alnum})", new int[] { 1, 2, 3 });
+		
+		// match dd-mm-yyyy patterns like 1.1.2010, 01/06/2010, etc.
+		patterns[1] = new DatePattern("(?<!\\p{Alnum})(\\d{1,2})[^\\p{Alnum}](\\d{1,2})[^\\p{Alnum}](\\d{4})(?!\\p{Alnum})", new int[] { 3, 2, 1 });
+	}
+	
+	
+	public DateMatcher(DatePattern... patterns) {
+		this.patterns = patterns;
+	}
+	
+	
+	public Date match(CharSequence seq) {
+		for (DatePattern pattern : patterns) {
+			Date match = pattern.match(seq);
+			
+			if (match != null) {
+				return match;
+			}
+		}
+		
+		return null;
+	}
+	
+	
+	public int find(CharSequence seq, int fromIndex) {
+		for (DatePattern pattern : patterns) {
+			int pos = pattern.find(seq, fromIndex);
+			
+			if (pos >= 0) {
+				return pos;
+			}
+		}
+		
+		return -1;
+	}
+	
+	
+	private static class DatePattern {
+		
+		protected final Pattern pattern;
+		protected final int[] order;
+		
+		
+		public DatePattern(String pattern, int[] order) {
+			this.pattern = Pattern.compile(pattern);
+			this.order = order;
+		}
+		
+		
+		protected Date process(MatchResult match) {
+			return new Date(Integer.parseInt(match.group(order[0])), Integer.parseInt(match.group(order[1])), Integer.parseInt(match.group(order[2])));
+		}
+		
+		
+		public Date match(CharSequence seq) {
+			Matcher matcher = pattern.matcher(seq);
+			
+			if (matcher.find()) {
+				return process(matcher);
+			}
+			
+			return null;
+		}
+		
+		
+		public int find(CharSequence seq, int fromIndex) {
+			Matcher matcher = pattern.matcher(seq).region(fromIndex, seq.length());
+			
+			if (matcher.find()) {
+				return matcher.start();
+			}
+			
+			return -1;
+		}
+	}
+	
+}
diff --git a/source/net/sourceforge/filebot/similarity/DateMetric.java b/source/net/sourceforge/filebot/similarity/DateMetric.java
index f58ac212..68f8e6ed 100644
--- a/source/net/sourceforge/filebot/similarity/DateMetric.java
+++ b/source/net/sourceforge/filebot/similarity/DateMetric.java
@@ -3,26 +3,22 @@ package net.sourceforge.filebot.similarity;
 
 
 import java.io.File;
-import java.util.regex.MatchResult;
-import java.util.regex.Matcher;
-import java.util.regex.Pattern;
 
 import net.sourceforge.filebot.web.Date;
 
 
 public class DateMetric implements SimilarityMetric {
 	
-	private final DatePattern[] patterns;
+	private final DateMatcher matcher;
 	
 	
 	public DateMetric() {
-		patterns = new DatePattern[2];
-		
-		// match yyyy-mm-dd patterns like 2010-10-24, 2009/6/1, etc.
-		patterns[0] = new DatePattern("(?<!\\p{Alnum})(\\d{4})[^\\p{Alnum}](\\d{1,2})[^\\p{Alnum}](\\d{1,2})(?!\\p{Alnum})", new int[] { 1, 2, 3 });
-		
-		// match dd-mm-yyyy patterns like 1.1.2010, 01/06/2010, etc.
-		patterns[1] = new DatePattern("(?<!\\p{Alnum})(\\d{1,2})[^\\p{Alnum}](\\d{1,2})[^\\p{Alnum}](\\d{4})(?!\\p{Alnum})", new int[] { 3, 2, 1 });
+		this.matcher = new DateMatcher();
+	}
+	
+	
+	public DateMetric(DateMatcher matcher) {
+		this.matcher = matcher;
 	}
 	
 	
@@ -46,49 +42,7 @@ public class DateMetric implements SimilarityMetric {
 			object = ((File) object).getName();
 		}
 		
-		return match(object.toString());
-	}
-	
-	
-	public Date match(CharSequence name) {
-		for (DatePattern pattern : patterns) {
-			Date match = pattern.match(name);
-			
-			if (match != null) {
-				return match;
-			}
-		}
-		
-		return null;
-	}
-	
-	
-	protected static class DatePattern {
-		
-		protected final Pattern pattern;
-		protected final int[] order;
-		
-		
-		public DatePattern(String pattern, int[] order) {
-			this.pattern = Pattern.compile(pattern);
-			this.order = order;
-		}
-		
-		
-		protected Date process(MatchResult match) {
-			return new Date(Integer.parseInt(match.group(order[0])), Integer.parseInt(match.group(order[1])), Integer.parseInt(match.group(order[2])));
-		}
-		
-		
-		public Date match(CharSequence name) {
-			Matcher matcher = pattern.matcher(name);
-			
-			if (matcher.find()) {
-				return process(matcher);
-			}
-			
-			return null;
-		}
+		return matcher.match(object.toString());
 	}
 	
 }
diff --git a/source/net/sourceforge/filebot/similarity/SeasonEpisodeMatcher.java b/source/net/sourceforge/filebot/similarity/SeasonEpisodeMatcher.java
index a443a2ab..c03006eb 100644
--- a/source/net/sourceforge/filebot/similarity/SeasonEpisodeMatcher.java
+++ b/source/net/sourceforge/filebot/similarity/SeasonEpisodeMatcher.java
@@ -24,7 +24,7 @@ public class SeasonEpisodeMatcher {
 		patterns[0] = new SeasonEpisodePattern(sanity, "(?<!\\p{Alnum})[Ss](\\d{1,2}|\\d{4})[^\\p{Alnum}]{0,3}[Ee](\\d{1,3})(?!\\p{Digit})");
 		
 		// match patterns like 1x01, 1.02, ..., 1x01a, 10x01, 10.02, ...
-		patterns[1] = new SeasonEpisodePattern(sanity, "(?<!\\p{Alnum})(\\d{1,2})[x.](\\d{2,3})(?!\\p{Digit})");
+		patterns[1] = new SeasonEpisodePattern(sanity, "(?<!\\p{Alnum}|\\d{4}[.])(\\d{1,2})[x.](\\d{2,3})(?!\\p{Digit})");
 		
 		// match patterns like 01, 102, 1003 (enclosed in separators)
 		patterns[2] = new SeasonEpisodePattern(sanity, "(?<!\\p{Alnum})([0-1]?\\d?)(\\d{2})(?!\\p{Alnum})") {
diff --git a/source/net/sourceforge/filebot/similarity/SeriesNameMatcher.java b/source/net/sourceforge/filebot/similarity/SeriesNameMatcher.java
index e5cf2f21..07ddd4eb 100644
--- a/source/net/sourceforge/filebot/similarity/SeriesNameMatcher.java
+++ b/source/net/sourceforge/filebot/similarity/SeriesNameMatcher.java
@@ -31,6 +31,7 @@ import net.sourceforge.tuned.FileUtilities;
 public class SeriesNameMatcher {
 	
 	protected SeasonEpisodeMatcher seasonEpisodeMatcher = new SeasonEpisodeMatcher(new SeasonEpisodeFilter(30, 50, -1), true);
+	protected DateMatcher dateMatcher = new DateMatcher();
 	protected NameSimilarityMetric nameSimilarityMetric = new NameSimilarityMetric();
 	
 	protected int commonWordSequenceMaxStartIndex;
@@ -83,12 +84,17 @@ public class SeriesNameMatcher {
 		// match common word sequences (likely series names)
 		SeriesNameCollection whitelist = new SeriesNameCollection();
 		
-		// focus chars before the SxE pattern when matching by common word sequence
+		// focus chars before the SxE / Date pattern when matching by common word sequence
 		String[] focus = Arrays.copyOf(names, names.length);
 		for (int i = 0; i < focus.length; i++) {
-			int pos = seasonEpisodeMatcher.find(focus[i], 0);
-			if (pos >= 0) {
-				focus[i] = focus[i].substring(0, pos);
+			int sxePos = seasonEpisodeMatcher.find(focus[i], 0);
+			if (sxePos >= 0) {
+				focus[i] = focus[i].substring(0, sxePos);
+			} else {
+				int datePos = dateMatcher.find(focus[i], 0);
+				if (datePos >= 0) {
+					focus[i] = focus[i].substring(0, datePos);
+				}
 			}
 		}
 		whitelist.addAll(deepMatchAll(focus, threshold));
@@ -118,8 +124,9 @@ public class SeriesNameMatcher {
 			name = normalize(name);
 			
 			Matcher prefix = prefixPattern.matcher(name);
-			int sxePosition = seasonEpisodeMatcher.find(name, prefix.find() ? prefix.end() : 0);
+			int prefixEnd = prefix.find() ? prefix.end() : 0;
 			
+			int sxePosition = seasonEpisodeMatcher.find(name, prefixEnd);
 			if (sxePosition > 0) {
 				String hit = name.substring(0, sxePosition).trim();
 				List<SxE> sxe = seasonEpisodeMatcher.match(name.substring(sxePosition));
@@ -131,7 +138,14 @@ public class SeriesNameMatcher {
 					// require multiple matches, if hit might be a false match
 					thresholdCollection.add(hit);
 				}
+			} else {
+				// try date pattern as fallback
+				int datePosition = dateMatcher.find(name, prefixEnd);
+				if (datePosition > 0) {
+					thresholdCollection.addDirect(name.substring(0, datePosition).trim());
+				}
 			}
+			
 		}
 		
 		return thresholdCollection;
@@ -176,14 +190,19 @@ public class SeriesNameMatcher {
 	 * @return a substring of the given name that ends before the first occurrence of a season
 	 *         episode pattern, or null if there is no such pattern
 	 */
-	public String matchBySeasonEpisodePattern(String name) {
+	public String matchByEpisodeIdentifier(String name) {
 		int seasonEpisodePosition = seasonEpisodeMatcher.find(name, 0);
-		
 		if (seasonEpisodePosition > 0) {
 			// series name ends at the first season episode pattern
 			return normalize(name.substring(0, seasonEpisodePosition));
 		}
 		
+		int datePosition = dateMatcher.find(name, 0);
+		if (datePosition > 0) {
+			// series name ends at the first season episode pattern
+			return normalize(name.substring(0, datePosition));
+		}
+		
 		return null;
 	}
 	
diff --git a/source/net/sourceforge/filebot/ui/rename/EpisodeListMatcher.java b/source/net/sourceforge/filebot/ui/rename/EpisodeListMatcher.java
index aa722e15..f1965f1a 100644
--- a/source/net/sourceforge/filebot/ui/rename/EpisodeListMatcher.java
+++ b/source/net/sourceforge/filebot/ui/rename/EpisodeListMatcher.java
@@ -233,7 +233,7 @@ class EpisodeListMatcher implements AutoCompleteMatcher {
 		
 		// require user input if auto-detection has failed or has been disabled 
 		if (episodes.isEmpty()) {
-			String suggestion = new SeriesNameMatcher().matchBySeasonEpisodePattern(getName(files.get(0)));
+			String suggestion = new SeriesNameMatcher().matchByEpisodeIdentifier(getName(files.get(0)));
 			if (suggestion != null) {
 				// clean media info / release group info / etc 
 				suggestion = stripReleaseInfo(suggestion);
diff --git a/test/net/sourceforge/filebot/similarity/SeriesNameMatcherTest.java b/test/net/sourceforge/filebot/similarity/SeriesNameMatcherTest.java
index b44eb46b..6e791de5 100644
--- a/test/net/sourceforge/filebot/similarity/SeriesNameMatcherTest.java
+++ b/test/net/sourceforge/filebot/similarity/SeriesNameMatcherTest.java
@@ -34,10 +34,10 @@ public class SeriesNameMatcherTest {
 
 	@Test
 	public void matchBeforeSeasonEpisodePattern() {
-		assertEquals("The Test", matcher.matchBySeasonEpisodePattern("The Test - 1x01"));
+		assertEquals("The Test", matcher.matchByEpisodeIdentifier("The Test - 1x01"));
 		
 		// real world test
-		assertEquals("Mushishi", matcher.matchBySeasonEpisodePattern("[niizk]_Mushishi_-_01_-_The_Green_Gathering"));
+		assertEquals("Mushishi", matcher.matchByEpisodeIdentifier("[niizk]_Mushishi_-_01_-_The_Green_Gathering"));
 	}