From 284d53bee5f23bbd5b12530f8230b7f556f400b7 Mon Sep 17 00:00:00 2001
From: Reinhard Pointner <rednoah@users.noreply.github.com>
Date: Mon, 6 Jan 2014 23:22:31 +0000
Subject: [PATCH] + make movie matching properly aware of alias names

---
 BuildData.groovy                              | 32 +++++++++++++------
 .../filebot/cli/ScriptShell.lib.groovy        |  2 +-
 .../filebot/media/MediaDetection.java         |  4 ++-
 website/data/query-blacklist.txt              |  3 ++
 4 files changed, 30 insertions(+), 11 deletions(-)

diff --git a/BuildData.groovy b/BuildData.groovy
index 2d56a637..e70e68f2 100644
--- a/BuildData.groovy
+++ b/BuildData.groovy
@@ -62,6 +62,20 @@ def isValidMovieName(s) {
 	return s=~ /^[A-Z0-9]/ && s =~ /[\p{Alpha}]{3}/
 }
 
+def getNamePermutations(names) {
+	def fn1 = { s -> s.replaceAll(/^(?i)(The|A)\s/, '') }
+	def fn2 = { s -> s.replaceAll(/\s&\s/, ' and ') }
+	def fn3 = { s -> s.replaceAll(/\([^\)]*\)$/, '') }
+
+	def out = new LinkedHashSet()
+	out += names*.trim()
+	[fn1, fn2, fn3].each{ fn ->
+		def results = out.findResults{ fn(it) }
+		out += results*.trim();
+	}
+	return out.unique{ it.toLowerCase().normalizePunctuation() }.findAll{ it.length() > 0 }
+}
+
 def treeSort(list, keyFunction) {
 	def sorter = new TreeMap(String.CASE_INSENSITIVE_ORDER)
 	list.each{
@@ -115,7 +129,7 @@ tmdb*.join('\t').join('\n').saveAs(tmdb_txt)
 
 movies = tmdb.findResults{
 	def ity = it[1..3] // imdb id, tmdb id, year
-	def names = it[4..-2].findAll{ isValidMovieName(it) }.unique{ it.toLowerCase().normalizePunctuation() }
+	def names = getNamePermutations(it[4..-1]).findAll{ isValidMovieName(it) }
 	if (ity[0].toInteger() > 0 && ity[1].toInteger() > 0 && names.size() > 0)
 		return ity + names
 	else
@@ -181,24 +195,24 @@ tvdb.values().findResults{ it.join('\t') }.join('\n').saveAs(tvdb_txt)
 
 
 def thetvdb_index = []
-tvdb.values().each{
-	def n1 = it[2].trim()
-	def n2 = it[3].replaceAll(/^(?i)(The|A)\s/, '').replaceAll(/\s&\s/, ' and ').replaceAll(/\([^\)]*\)$/, '').trim()
+tvdb.values().each{ r ->
+	def tvdb_name = r[2]
+	def imdb_name = r[3].replaceAll(/\([^\)]*\)$/, '').trim()
 	
-	thetvdb_index << [it[0], n1]
-	if (similarity(n1,n2) < 1) {
-		thetvdb_index << [it[0], n2]
+	getNamePermutations([tvdb_name, imdb_name]).each{ n ->
+		thetvdb_index << [r[0], n]
 	}
 }
 
 def addSeriesAlias = { from, to ->
 	def se = thetvdb_index.find{ from == it[1] && !it.contains(to) }
 	if (se == null) throw new Exception("Unabled to find series '${from}'")
-	thetvdb_index << [se[0], to]
+	getNamePermutations([to]).each{ n ->
+		thetvdb_index << [se[0], n]
+	}
 }
 
 // additional custom mappings
-addSeriesAlias('Law & Order: Special Victims Unit', 'Law and Order SVU')
 addSeriesAlias('Law & Order: Special Victims Unit', 'Law & Order SVU')
 addSeriesAlias('CSI: Crime Scene Investigation', 'CSI')
 addSeriesAlias('M*A*S*H', 'MASH')
diff --git a/source/net/sourceforge/filebot/cli/ScriptShell.lib.groovy b/source/net/sourceforge/filebot/cli/ScriptShell.lib.groovy
index 4c5985ea..b04d8a93 100644
--- a/source/net/sourceforge/filebot/cli/ScriptShell.lib.groovy
+++ b/source/net/sourceforge/filebot/cli/ScriptShell.lib.groovy
@@ -235,7 +235,7 @@ def detectSeriesName(files, locale = Locale.ENGLISH) {
 }
 
 def detectMovie(File file, strict = true, queryLookupService = TheMovieDB, hashLookupService = OpenSubtitles, locale = Locale.ENGLISH) {
-	def movies = MediaDetection.matchMovieName([file.name, file.parentFile.name], true, 0) ?: MediaDetection.detectMovie(file, hashLookupService, queryLookupService, locale, strict)
+	def movies = MediaDetection.matchMovieName(file.listPath(3, true)*.name, true, 0) ?: MediaDetection.detectMovie(file, hashLookupService, queryLookupService, locale, strict)
 	return movies == null || movies.isEmpty() ? null : movies.toList()[0]
 }
 
diff --git a/source/net/sourceforge/filebot/media/MediaDetection.java b/source/net/sourceforge/filebot/media/MediaDetection.java
index 1de00b8f..74f0a576 100644
--- a/source/net/sourceforge/filebot/media/MediaDetection.java
+++ b/source/net/sourceforge/filebot/media/MediaDetection.java
@@ -752,7 +752,9 @@ public class MediaDetection {
 		if (movieIndex.isEmpty()) {
 			try {
 				for (Movie movie : releaseInfo.getMovieList()) {
-					movieIndex.add(new SimpleEntry<String, Movie>(normalizePunctuation(movie.getName()).toLowerCase(), movie));
+					for (String name : movie.getEffectiveNames()) {
+						movieIndex.add(new SimpleEntry<String, Movie>(normalizePunctuation(name).toLowerCase(), movie));
+					}
 				}
 			} catch (Exception e) {
 				// can't load movie index, just try again next time
diff --git a/website/data/query-blacklist.txt b/website/data/query-blacklist.txt
index 56f43619..bb716fa5 100644
--- a/website/data/query-blacklist.txt
+++ b/website/data/query-blacklist.txt
@@ -137,6 +137,7 @@ AMC-BL
 AMC-TEST
 Anaglyph
 Arte
+Best.of.\d{4}
 Best.Quality
 bt.downloads
 btarena.org
@@ -209,6 +210,7 @@ info
 iNT
 iNTERNAL
 iPod
+IPT
 ISO
 iTA
 iTALIA
@@ -225,6 +227,7 @@ lol
 Los.Sustitutos
 M.HD
 mkvonly
+Movie.Pack
 mpg$
 MultiSub
 MVGroup.org