From cc57b8984087d1512bb80c10b948681353043c51 Mon Sep 17 00:00:00 2001 From: Reinhard Pointner Date: Sun, 27 Jan 2013 09:56:40 +0000 Subject: [PATCH] * fix TV Movie Lookup --- .../net/sourceforge/filebot/web/IMDbClient.java | 15 ++++++--------- .../sourceforge/filebot/web/IMDbClientTest.java | 10 ++++++++++ 2 files changed, 16 insertions(+), 9 deletions(-) diff --git a/source/net/sourceforge/filebot/web/IMDbClient.java b/source/net/sourceforge/filebot/web/IMDbClient.java index faa21e83..9607991f 100644 --- a/source/net/sourceforge/filebot/web/IMDbClient.java +++ b/source/net/sourceforge/filebot/web/IMDbClient.java @@ -41,6 +41,9 @@ import org.xml.sax.SAXException; public class IMDbClient implements MovieIdentificationService { + private String host = "www.imdb.com"; + + @Override public String getName() { return "IMDb"; @@ -53,12 +56,6 @@ public class IMDbClient implements MovieIdentificationService { } - protected String getHost() { - String host = System.getProperty("imdb.hostname"); // default to akas.imdb.com but allow override via -Dimdb.host - return host == null ? "imdb.com" : host; - } - - protected int getImdbId(String link) { Matcher matcher = Pattern.compile("tt(\\d{7})").matcher(link); @@ -73,7 +70,7 @@ public class IMDbClient implements MovieIdentificationService { @Override public List searchMovie(String query, Locale locale) throws Exception { - Document dom = parsePage(new URL("http", getHost(), "/find?s=tt&q=" + encode(query, false))); + Document dom = parsePage(new URL("http", host, "/find?s=tt&q=" + encode(query, false))); // select movie links followed by year in parenthesis List nodes = selectNodes("//TABLE[@class='findList']//TD/A[substring-after(substring-before(following::text(),')'),'(')]", dom); @@ -118,7 +115,7 @@ public class IMDbClient implements MovieIdentificationService { return null; String name = selectString("//H1/text()", dom).replaceAll("\\s+", " ").trim(); - String year = new Scanner(selectString("//H1//A/text()", dom)).useDelimiter("\\D+").next(); + String year = new Scanner(selectNode("//H1/SPAN", dom).getTextContent()).useDelimiter("\\D+").next(); int imdbid = getImdbId(selectString("//LINK[@rel='canonical']/@href", dom)); return new Movie(name, Pattern.matches("\\d{4}", year) ? Integer.parseInt(year) : -1, imdbid, -1); @@ -132,7 +129,7 @@ public class IMDbClient implements MovieIdentificationService { @Override public Movie getMovieDescriptor(int imdbid, Locale locale) throws Exception { try { - return scrapeMovie(parsePage(new URL("http", getHost(), String.format("/title/tt%07d", imdbid))), locale); + return scrapeMovie(parsePage(new URL("http", host, String.format("/title/tt%07d/", imdbid))), locale); } catch (FileNotFoundException e) { return null; // illegal imdbid } diff --git a/test/net/sourceforge/filebot/web/IMDbClientTest.java b/test/net/sourceforge/filebot/web/IMDbClientTest.java index e0be9188..91845309 100644 --- a/test/net/sourceforge/filebot/web/IMDbClientTest.java +++ b/test/net/sourceforge/filebot/web/IMDbClientTest.java @@ -113,6 +113,16 @@ public class IMDbClientTest { } + @Test + public void getMovieDescriptor3() throws Exception { + Movie movie = imdb.getMovieDescriptor(75610, null); + + assertEquals("21", movie.getName()); + assertEquals(1977, movie.getYear()); + assertEquals(75610, movie.getImdbId(), 0); + } + + @Test public void getAkaMovieDescriptor() throws Exception { Movie movie = imdb.getMovieDescriptor(106559, Locale.ENGLISH);