From 1dd6f6e3dd804905226e8eecbcf086ce36009941 Mon Sep 17 00:00:00 2001
From: Reinhard Pointner <rednoah@users.noreply.github.com>
Date: Mon, 2 Dec 2013 18:25:06 +0000
Subject: [PATCH] * account for TVDB entries being removed sometimes because of
 duplicates

---
 BuildData.groovy                                   | 12 +++++++++++-
 source/net/sourceforge/filebot/web/WebRequest.java |  6 ++++++
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/BuildData.groovy b/BuildData.groovy
index a1d0a84a..077e6aac 100644
--- a/BuildData.groovy
+++ b/BuildData.groovy
@@ -134,8 +134,8 @@ if (movies.size() < 50000) { throw new Exception('Movie index sanity failed') }
 
 
 // BUILD tvdb index
-def tvdb = new HashMap()
 def tvdb_txt = new File('tvdb.txt')
+def tvdb = [:]
 new File('tvdb.txt').eachLine{
 	def line = it.split('\t', 5).toList()
 	tvdb.put(line[0] as Integer, [line[0] as Integer, line[1], line[2], line[3], line[4] as Integer])
@@ -167,6 +167,16 @@ tvdb_updates.each{ update ->
 		}
 	}
 }
+
+// remove entries that have become invalid
+def tvdb_ids = tvdb_updates.findResults{ it.id } as HashSet
+tvdb.keySet().toList().each{ id ->
+	if (!tvdb_ids.contains(id)) {
+		println "Invalid ID found: ${tvdb[id]}"
+		tvdb.remove(id)
+	}
+}
+
 tvdb.values().findResults{ it.join('\t') }.join('\n').saveAs(tvdb_txt)
 
 
diff --git a/source/net/sourceforge/filebot/web/WebRequest.java b/source/net/sourceforge/filebot/web/WebRequest.java
index a1dc934a..f7d65d9b 100644
--- a/source/net/sourceforge/filebot/web/WebRequest.java
+++ b/source/net/sourceforge/filebot/web/WebRequest.java
@@ -216,6 +216,12 @@ public final class WebRequest {
 		return buffer.getByteBuffer();
 	}
 
+	public static int head(URL url) throws IOException {
+		HttpURLConnection c = (HttpURLConnection) url.openConnection();
+		c.setRequestMethod("HEAD");
+		return c.getResponseCode();
+	}
+
 	private static Charset getCharset(String contentType) {
 		if (contentType != null) {
 			// e.g. Content-Type: text/html; charset=iso-8859-1