* added episode list support for IMDb

source/net/sourceforge/filebot/similarity/SeasonEpisodeMatcher.java

@@ -16,10 +16,10 @@ public class SeasonEpisodeMatcher {
 	public SeasonEpisodeMatcher() {
 		patterns = new SeasonEpisodePattern[3];
 		
-		// match patterns like S01E01, s01e02, ... [s01]_[e02], s01.e02, ...
+		// match patterns like S01E01, s01e02, ... [s01]_[e02], s01.e02, s01e02a, ...
 		patterns[0] = new SeasonEpisodePattern("(?<!\\p{Alnum})[Ss](\\d{1,2})[^\\p{Alnum}]{0,3}[Ee](\\d{1,3})(?!\\p{Digit})");
 		
-		// match patterns like 1x01, 1.02, ... 10x01, 10.02, ...
+		// match patterns like 1x01, 1.02, ..., 1x01a, 10x01, 10.02, ...
 		patterns[1] = new SeasonEpisodePattern("(?<!\\p{Alnum})(\\d{1,2})[x\\.](\\d{1,3})(?!\\p{Digit})");
 		
 		// match patterns like 01, 102, 1003 (enclosed in separators)

source/net/sourceforge/filebot/ui/panel/episodelist/EpisodeListPanel.java

@@ -30,6 +30,7 @@ import net.sourceforge.filebot.ui.transfer.SaveAction;
 import net.sourceforge.filebot.web.AnidbClient;
 import net.sourceforge.filebot.web.Episode;
 import net.sourceforge.filebot.web.EpisodeListClient;
+import net.sourceforge.filebot.web.IMDbClient;
 import net.sourceforge.filebot.web.SearchResult;
 import net.sourceforge.filebot.web.TVDotComClient;
 import net.sourceforge.filebot.web.TVRageClient;
@@ -74,6 +75,7 @@ public class EpisodeListPanel extends AbstractSearchPanel<EpisodeListClient, Epi
 		engines.add(new TVRageClient());
 		engines.add(new AnidbClient());
 		engines.add(new TVDotComClient());
+		engines.add(new IMDbClient());
 		engines.add(new TheTVDBClient(Settings.userRoot().get("thetvdb.apikey")));
 		
 		return engines;
@@ -220,10 +222,11 @@ public class EpisodeListPanel extends AbstractSearchPanel<EpisodeListClient, Epi
 
 		@Override
 		public URI getLink() {
-			if (request.getSeason() != ALL_SEASONS)
+			if (request.getSeason() != ALL_SEASONS) {
 				return request.getClient().getEpisodeListLink(getSearchResult(), request.getSeason());
-			else
-				return request.getClient().getEpisodeListLink(getSearchResult());
+			}
+			
+			return request.getClient().getEpisodeListLink(getSearchResult());
 		}
 		
 

source/net/sourceforge/filebot/ui/panel/rename/RenameListCellRenderer.java

@@ -79,7 +79,7 @@ class RenameListCellRenderer extends DefaultFancyListCellRenderer {
 		
 		String extension = FileUtilities.getExtension(file);
 		
-		if (!extension.isEmpty())
+		if (extension != null)
 			return extension;
 		
 		// some file with no extension

source/net/sourceforge/filebot/ui/panel/rename/RenamePanel.java

@@ -42,6 +42,7 @@ import net.sourceforge.filebot.ui.SelectDialog;
 import net.sourceforge.filebot.web.AnidbClient;
 import net.sourceforge.filebot.web.Episode;
 import net.sourceforge.filebot.web.EpisodeListClient;
+import net.sourceforge.filebot.web.IMDbClient;
 import net.sourceforge.filebot.web.SearchResult;
 import net.sourceforge.filebot.web.TVDotComClient;
 import net.sourceforge.filebot.web.TVRageClient;
@@ -143,6 +144,7 @@ public class RenamePanel extends JComponent {
 		actionPopup.add(new AutoFetchEpisodeListAction(new TVRageClient()));
 		actionPopup.add(new AutoFetchEpisodeListAction(new AnidbClient()));
 		actionPopup.add(new AutoFetchEpisodeListAction(new TVDotComClient()));
+		actionPopup.add(new AutoFetchEpisodeListAction(new IMDbClient()));
 		actionPopup.add(new AutoFetchEpisodeListAction(new TheTVDBClient(Settings.userRoot().get("thetvdb.apikey"))));
 		
 		actionPopup.addSeparator();

source/net/sourceforge/filebot/web/AnidbClient.java

@@ -141,7 +141,7 @@ public class AnidbClient implements EpisodeListClient {
 
 	@Override
 	public URI getEpisodeListLink(SearchResult searchResult, int season) {
-		throw new UnsupportedOperationException();
+		return null;
 	}
 	
 }

source/net/sourceforge/filebot/web/IMDbClient.java

@@ -0,0 +1,162 @@
+
+package net.sourceforge.filebot.web;
+
+
+import static net.sourceforge.filebot.web.WebRequest.getHtmlDocument;
+import static net.sourceforge.tuned.XPathUtilities.getAttribute;
+import static net.sourceforge.tuned.XPathUtilities.selectNodes;
+import static net.sourceforge.tuned.XPathUtilities.selectString;
+
+import java.io.IOException;
+import java.net.URI;
+import java.net.URL;
+import java.net.URLConnection;
+import java.net.URLEncoder;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Scanner;
+import java.util.logging.Level;
+import java.util.logging.Logger;
+
+import javax.swing.Icon;
+
+import net.sourceforge.filebot.ResourceManager;
+
+import org.w3c.dom.Document;
+import org.w3c.dom.Node;
+import org.xml.sax.SAXException;
+
+
+public class IMDbClient implements EpisodeListClient {
+	
+	private static final String host = "www.imdb.com";
+	
+	
+	@Override
+	public String getName() {
+		return "IMDb";
+	}
+	
+
+	@Override
+	public Icon getIcon() {
+		return ResourceManager.getIcon("search.imdb");
+	}
+	
+
+	@Override
+	public boolean hasSingleSeasonSupport() {
+		return true;
+	}
+	
+
+	@Override
+	public List<SearchResult> search(String query) throws IOException, SAXException {
+		
+		URL searchUrl = new URL("http", host, "/find?s=tt&q=" + URLEncoder.encode(query, "UTF-8"));
+		
+		Document dom = getHtmlDocument(openConnection(searchUrl));
+		
+		List<Node> nodes = selectNodes("//TABLE//A[following-sibling::SMALL[contains(.,'TV series')]]", dom);
+		
+		List<SearchResult> results = new ArrayList<SearchResult>(nodes.size());
+		
+		for (Node node : nodes) {
+			String name = removeQuotationMarks(node.getTextContent().trim());
+			String year = node.getNextSibling().getTextContent().trim();
+			String href = getAttribute("href", node);
+			
+			String nameAndYear = String.format("%s %s", name, year).trim();
+			int imdbId = new Scanner(href).useDelimiter("\\D+").nextInt();
+			
+			results.add(new MovieDescriptor(nameAndYear, imdbId));
+		}
+		
+		return results;
+	}
+	
+
+	@Override
+	public List<Episode> getEpisodeList(SearchResult searchResult) throws IOException, SAXException {
+		Document dom = getHtmlDocument(openConnection(getEpisodeListLink(searchResult).toURL()));
+		
+		String seriesName = removeQuotationMarks(selectString("//H1/A", dom));
+		
+		List<Node> nodes = selectNodes("//TABLE//H3/A[preceding-sibling::text()]", dom);
+		
+		List<Episode> episodes = new ArrayList<Episode>(nodes.size());
+		
+		for (Node node : nodes) {
+			String title = node.getTextContent().trim();
+			
+			Scanner numberScanner = new Scanner(node.getPreviousSibling().getTextContent()).useDelimiter("\\D+");
+			String season = numberScanner.next();
+			String episode = numberScanner.next();
+			
+			episodes.add(new Episode(seriesName, season, episode, title));
+		}
+		
+		return episodes;
+	}
+	
+
+	@Override
+	public List<Episode> getEpisodeList(SearchResult searchResult, int season) throws Exception {
+		
+		List<Episode> episodes = new ArrayList<Episode>(25);
+		
+		// remember max. season, so we can throw a proper exception, in case an illegal season number was requested
+		int maxSeason = 0;
+		
+		// filter given season from all seasons
+		for (Episode episode : getEpisodeList(searchResult)) {
+			try {
+				int seasonNumber = Integer.parseInt(episode.getSeasonNumber());
+				
+				if (season == seasonNumber) {
+					episodes.add(episode);
+				}
+				
+				if (seasonNumber > maxSeason) {
+					maxSeason = seasonNumber;
+				}
+			} catch (NumberFormatException e) {
+				Logger.getLogger(getClass().getName()).log(Level.WARNING, "Illegal season number", e);
+			}
+		}
+		
+		if (episodes.isEmpty()) {
+			throw new SeasonOutOfBoundsException(searchResult.getName(), season, maxSeason);
+		}
+		
+		return episodes;
+	}
+	
+
+	protected URLConnection openConnection(URL url) throws IOException {
+		URLConnection connection = url.openConnection();
+		
+		// IMDb refuses default user agent (Java/1.6.0_12)
+		connection.addRequestProperty("User-Agent", "Scraper");
+		
+		return connection;
+	}
+	
+
+	protected String removeQuotationMarks(String name) {
+		return name.replaceAll("^\"|\"$", "");
+	}
+	
+
+	@Override
+	public URI getEpisodeListLink(SearchResult searchResult) {
+		return URI.create("http://" + host + String.format("/title/tt%07d/episodes", ((MovieDescriptor) searchResult).getImdbId()));
+	}
+	
+
+	@Override
+	public URI getEpisodeListLink(SearchResult searchResult, int season) {
+		return null;
+	}
+	
+}

source/net/sourceforge/filebot/web/SubsceneSubtitleClient.java

@@ -117,7 +117,7 @@ public class SubsceneSubtitleClient implements SubtitleClient {
 		Document subtitleListDocument = getSubtitleListDocument(subtitleListUrl, languageFilter);
 		
 		// let's update language filters if they are not known yet
-		if (languageFilterMap.isEmpty()) {
+		if (languageName != null && languageFilter == null) {
 			synchronized (languageFilterMap) {
 				languageFilterMap.putAll(getLanguageFilterMap(subtitleListDocument));
 			}

source/net/sourceforge/filebot/web/TheTVDBClient.java

@@ -10,7 +10,6 @@ import static net.sourceforge.tuned.XPathUtilities.selectString;
 import java.io.FileNotFoundException;
 import java.io.IOException;
 import java.net.URI;
-import java.net.URISyntaxException;
 import java.net.URL;
 import java.net.URLEncoder;
 import java.util.ArrayList;
@@ -205,11 +204,7 @@ public class TheTVDBClient implements EpisodeListClient {
 	public URI getEpisodeListLink(SearchResult searchResult) {
 		int seriesId = ((TheTVDBSearchResult) searchResult).getSeriesId();
 		
-		try {
-			return new URI("http://" + host + "/?tab=seasonall&id=" + seriesId);
-		} catch (URISyntaxException e) {
-			throw new RuntimeException(e);
-		}
+		return URI.create("http://" + host + "/?tab=seasonall&id=" + seriesId);
 	}
 	
 
@@ -230,11 +225,9 @@ public class TheTVDBClient implements EpisodeListClient {
 			}
 			
 			return new URI("http://" + host + "/?tab=season&seriesid=" + seriesId + "&seasonid=" + seasonId);
-		} catch (IOException e) {
+		} catch (Exception e) {
 			// log and ignore any IOException
 			Logger.getLogger(getClass().getName()).log(Level.WARNING, "Failed to retrieve season id", e);
-		} catch (Exception e) {
-			throw new RuntimeException(e);
 		}
 		
 		return null;

test/net/sourceforge/filebot/web/IMDbClientTest.java

@@ -0,0 +1,79 @@
+
+package net.sourceforge.filebot.web;
+
+
+import static org.junit.Assert.assertEquals;
+
+import java.util.List;
+
+import org.junit.Test;
+
+
+public class IMDbClientTest {
+	
+	private final IMDbClient imdb = new IMDbClient();
+	
+	
+	@Test
+	public void search() throws Exception {
+		List<SearchResult> results = imdb.search("battlestar");
+		
+		MovieDescriptor movie = (MovieDescriptor) results.get(0);
+		
+		assertEquals("Battlestar Galactica (2004)", movie.getName());
+		assertEquals(407362, movie.getImdbId(), 0);
+		
+		assertEquals(6, results.size(), 0);
+	}
+	
+
+	@Test
+	public void getEpisodeList() throws Exception {
+		List<Episode> list = imdb.getEpisodeList(new MovieDescriptor("Buffy", 118276));
+		
+		assertEquals(145, list.size());
+		
+		Episode first = list.get(0);
+		
+		assertEquals("Buffy the Vampire Slayer", first.getSeriesName());
+		assertEquals("Unaired Pilot", first.getTitle());
+		assertEquals("0", first.getEpisodeNumber());
+		assertEquals("1", first.getSeasonNumber());
+		
+		Episode last = list.get(144);
+		
+		assertEquals("Buffy the Vampire Slayer", last.getSeriesName());
+		assertEquals("Chosen", last.getTitle());
+		assertEquals("22", last.getEpisodeNumber());
+		assertEquals("7", last.getSeasonNumber());
+	}
+	
+
+	@Test
+	public void getEpisodeListWithUnknownSeason() throws Exception {
+		List<Episode> list = imdb.getEpisodeList(new MovieDescriptor("Mushishi", 807832));
+		
+		assertEquals(26, list.size());
+		
+		Episode first = list.get(0);
+		
+		assertEquals("Mushishi", first.getSeriesName());
+		assertEquals("Midori no za", first.getTitle());
+		assertEquals("1", first.getEpisodeNumber());
+		assertEquals("1", first.getSeasonNumber());
+	}
+	
+
+	@Test
+	public void getEpisodeListLink() throws Exception {
+		assertEquals("http://www.imdb.com/title/tt0407362/episodes", imdb.getEpisodeListLink(new MovieDescriptor("Battlestar Galactica", 407362)).toString());
+	}
+	
+
+	@Test
+	public void removeQuotationMarks() throws Exception {
+		assertEquals("test", imdb.removeQuotationMarks("\"test\""));
+		
+		assertEquals("inner \"quotation marks\"", imdb.removeQuotationMarks("\"inner \"quotation marks\"\""));
+	}
+}

test/net/sourceforge/filebot/web/WebTestSuite.java

@@ -8,7 +8,7 @@ import org.junit.runners.Suite.SuiteClasses;
 
 
 @RunWith(Suite.class)
-@SuiteClasses( { TVDotComClientTest.class, AnidbClientTest.class, TVRageClientTest.class, TheTVDBClientTest.class, SubsceneSubtitleClientTest.class, SubtitleSourceClientTest.class, OpenSubtitlesHasherTest.class })
+@SuiteClasses( { TVDotComClientTest.class, AnidbClientTest.class, TVRageClientTest.class, TheTVDBClientTest.class, IMDbClientTest.class, SubsceneSubtitleClientTest.class, SubtitleSourceClientTest.class, OpenSubtitlesHasherTest.class })
 public class WebTestSuite {
 	
 }