From c738e8783c2157972a86095ae2ff55addd7de858 Mon Sep 17 00:00:00 2001
From: Reinhard Pointner <reinhard.pointner@gmail.com>
Date: Thu, 30 Jun 2016 17:51:07 +0800
Subject: [PATCH] Simplify NFO parser

---
 source/net/filebot/media/MediaDetection.java  | 44 +++++--------------
 source/net/filebot/util/StringUtilities.java  |  7 ++-
 .../net/filebot/media/MediaDetectionTest.java | 10 +++++
 3 files changed, 28 insertions(+), 33 deletions(-)

diff --git a/source/net/filebot/media/MediaDetection.java b/source/net/filebot/media/MediaDetection.java
index 1f197f60..769a95ef 100644
--- a/source/net/filebot/media/MediaDetection.java
+++ b/source/net/filebot/media/MediaDetection.java
@@ -17,7 +17,6 @@ import java.io.File;
 import java.io.FileFilter;
 import java.io.IOException;
 import java.io.Serializable;
-import java.net.URL;
 import java.text.CollationKey;
 import java.text.Collator;
 import java.util.ArrayList;
@@ -1207,47 +1206,28 @@ public class MediaDetection {
 		return names;
 	}
 
-	public static Set<Integer> grepImdbId(CharSequence text) {
+	public static List<Integer> grepImdbId(CharSequence text) {
 		// scan for imdb id patterns like tt1234567
-		Matcher imdbMatch = Pattern.compile("(?<!\\p{Alnum})tt(\\d{7})(?!\\p{Alnum})", Pattern.CASE_INSENSITIVE).matcher(text);
-		Set<Integer> collection = new LinkedHashSet<Integer>();
-
-		while (imdbMatch.find()) {
-			collection.add(Integer.parseInt(imdbMatch.group(1)));
-		}
-
-		return collection;
+		Pattern imdbId = Pattern.compile("(?<!\\p{Alnum})tt(\\d{7})(?!\\p{Alnum})", Pattern.CASE_INSENSITIVE);
+		return streamMatches(text, imdbId, m -> m.group(1)).map(Integer::new).collect(toList());
 	}
 
-	public static Set<Integer> grepTheTvdbId(CharSequence text) {
+	public static List<Integer> grepTheTvdbId(CharSequence text) {
 		// scan for thetvdb id patterns like http://www.thetvdb.com/?tab=series&id=78874&lid=14
-		Set<Integer> collection = new LinkedHashSet<Integer>();
-		for (String token : Pattern.compile("[\\s\"<>|]+").split(text)) {
-			try {
-				URL url = new URL(token);
-				if (url.getHost().contains("thetvdb") && url.getQuery() != null && url.getQuery().contains("tab=series")) {
-					Matcher m = Pattern.compile("\\Wid=(\\d+)").matcher(url.getQuery());
-					while (m.find()) {
-						collection.add(Integer.parseInt(m.group(1)));
-					}
-				}
-			} catch (Exception e) {
-				debug.finest(e::toString);
-			}
-		}
-		return collection;
+		Pattern tvdbUrl = Pattern.compile("http[s]?://www.thetvdb.com/[?]tab=series&id=(\\d+)", Pattern.CASE_INSENSITIVE);
+		return streamMatches(text, tvdbUrl, m -> m.group(1)).map(Integer::new).collect(toList());
 	}
 
 	public static Movie grepMovie(File nfo, MovieIdentificationService resolver, Locale locale) throws Exception {
-		String contents = new String(readFile(nfo), "UTF-8");
-		int imdbid = grepImdbId(contents).iterator().next();
-		return resolver.getMovieDescriptor(new Movie(imdbid), locale);
+		String text = readTextFile(nfo);
+		int imdbId = grepImdbId(text).get(0);
+		return resolver.getMovieDescriptor(new Movie(imdbId), locale);
 	}
 
 	public static SeriesInfo grepSeries(File nfo, Locale locale) throws Exception {
-		String contents = new String(readFile(nfo), "UTF-8");
-		int thetvdbid = grepTheTvdbId(contents).iterator().next();
-		return WebServices.TheTVDB.getSeriesInfo(thetvdbid, locale);
+		String text = readTextFile(nfo);
+		int tvdbId = grepTheTvdbId(text).get(0);
+		return WebServices.TheTVDB.getSeriesInfo(tvdbId, locale);
 	}
 
 	public static List<SearchResult> getProbableMatches(String query, Collection<? extends SearchResult> options, boolean alias, boolean strict) {
diff --git a/source/net/filebot/util/StringUtilities.java b/source/net/filebot/util/StringUtilities.java
index afed651f..58733a0d 100644
--- a/source/net/filebot/util/StringUtilities.java
+++ b/source/net/filebot/util/StringUtilities.java
@@ -13,6 +13,7 @@ import java.util.Objects;
 import java.util.Optional;
 import java.util.Spliterators.AbstractSpliterator;
 import java.util.function.Consumer;
+import java.util.function.Function;
 import java.util.regex.MatchResult;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
@@ -54,7 +55,11 @@ public final class StringUtilities {
 	}
 
 	public static Stream<String> streamMatches(CharSequence s, Pattern pattern) {
-		return stream(new MatcherSpliterator(pattern.matcher(s)), false).map(MatchResult::group);
+		return streamMatches(s, pattern, MatchResult::group);
+	}
+
+	public static <T> Stream<T> streamMatches(CharSequence s, Pattern pattern, Function<MatchResult, T> mapper) {
+		return stream(new MatcherSpliterator(pattern.matcher(s)), false).map(mapper);
 	}
 
 	public static boolean find(String s, Pattern pattern) {
diff --git a/test/net/filebot/media/MediaDetectionTest.java b/test/net/filebot/media/MediaDetectionTest.java
index 5b6ec36f..a97a4edb 100644
--- a/test/net/filebot/media/MediaDetectionTest.java
+++ b/test/net/filebot/media/MediaDetectionTest.java
@@ -26,4 +26,14 @@ public class MediaDetectionTest {
 		assertEquals("[]", MediaDetection.detectSeriesNames(singleton(new File("Movie/LOTR.2001.AVC-1080")), false, Locale.ENGLISH).toString());
 	}
 
+	@Test
+	public void grepImdbId() throws Exception {
+		assertEquals("[499549]", MediaDetection.grepImdbId("@see http://www.imdb.com/title/tt0499549/").toString());
+	}
+
+	@Test
+	public void grepTheTvdbId() throws Exception {
+		assertEquals("[78874]", MediaDetection.grepTheTvdbId("@see http://www.thetvdb.com/?tab=series&id=78874&lid=14").toString());
+	}
+
 }