Merge pull request #1199 from TeamNewPipe/bandcamp-update-artist-detection

[Bandcamp] Update artist page detection

Author: TobiGr

extractor/src/main/java/org/schabi/newpipe/extractor/services/bandcamp/extractors/BandcampExtractorHelper.java

@@ -155,25 +155,30 @@ public static String getImageUrl(final long id, final boolean isAlbum) {
 
     /**
      * @return <code>true</code> if the given URL looks like it comes from a bandcamp custom domain
-     * or if it comes from <code>bandcamp.com</code> itself
+     * or a <code>*.bandcamp.com</code> subdomain
      */
-    public static boolean isSupportedDomain(final String url) throws ParsingException {
+    public static boolean isArtistDomain(final String url) throws ParsingException {
 
         // Accept all bandcamp.com URLs
         if (url.toLowerCase().matches("https?://.+\\.bandcamp\\.com(/.*)?")) {
             return true;
         }
 
+        // Reject non-artist bandcamp.com URLs
+        if (url.toLowerCase().matches("https?://bandcamp\\.com(/.*)?")) {
+            return false;
+        }
+
         try {
             // Test other URLs for whether they contain a footer that links to bandcamp
             return Jsoup.parse(NewPipe.getDownloader().get(url).responseBody())
-                    .getElementById("pgFt")
-                    .getElementById("pgFt-inner")
-                    .getElementById("footer-logo-wrapper")
-                    .getElementById("footer-logo")
-                    .getElementsByClass("hiddenAccess")
-                    .text().equals("Bandcamp");
-        } catch (final NullPointerException e) {
+                    .getElementsByClass("cart-wrapper")
+                    .get(0)
+                    .getElementsByTag("a")
+                    .get(0)
+                    .attr("href")
+                    .equals("https://bandcamp.com/cart");
+        } catch (final NullPointerException | IndexOutOfBoundsException e) {
             return false;
         } catch (final IOException | ReCaptchaException e) {
             throw new ParsingException("Could not determine whether URL is custom domain "

extractor/src/main/java/org/schabi/newpipe/extractor/services/bandcamp/linkHandler/BandcampChannelLinkHandlerFactory.java

@@ -90,7 +90,7 @@ public boolean onAcceptUrl(final String url) throws ParsingException {
             }
 
             // Test whether domain is supported
-            return BandcampExtractorHelper.isSupportedDomain(lowercaseUrl);
+            return BandcampExtractorHelper.isArtistDomain(lowercaseUrl);
         }
     }
 }

extractor/src/main/java/org/schabi/newpipe/extractor/services/bandcamp/linkHandler/BandcampCommentsLinkHandlerFactory.java

@@ -39,7 +39,7 @@ public boolean onAcceptUrl(final String url) throws ParsingException {
         }
 
         // Test whether domain is supported
-        return BandcampExtractorHelper.isSupportedDomain(url);
+        return BandcampExtractorHelper.isArtistDomain(url);
     }
 
     @Override

extractor/src/main/java/org/schabi/newpipe/extractor/services/bandcamp/linkHandler/BandcampPlaylistLinkHandlerFactory.java

@@ -48,6 +48,6 @@ public boolean onAcceptUrl(final String url) throws ParsingException {
         }
 
         // Test whether domain is supported
-        return BandcampExtractorHelper.isSupportedDomain(url);
+        return BandcampExtractorHelper.isArtistDomain(url);
     }
 }

extractor/src/main/java/org/schabi/newpipe/extractor/services/bandcamp/linkHandler/BandcampStreamLinkHandlerFactory.java

@@ -71,6 +71,6 @@ public boolean onAcceptUrl(final String url) throws ParsingException {
         }
 
         // Test whether domain is supported
-        return BandcampExtractorHelper.isSupportedDomain(url);
+        return BandcampExtractorHelper.isArtistDomain(url);
     }
 }

extractor/src/test/java/org/schabi/newpipe/extractor/services/bandcamp/BandcampCommentsLinkHandlerFactoryTest.java

@@ -39,6 +39,5 @@ public void testAcceptUrl() throws ParsingException {
         assertTrue(linkHandler.acceptUrl("http://ZachBenson.Bandcamp.COM/Track/U-I-Tonite/"));
         assertTrue(linkHandler.acceptUrl("https://interovgm.bandcamp.com/track/title"));
         assertTrue(linkHandler.acceptUrl("https://goodgoodblood-tl.bandcamp.com/track/when-it-all-wakes-up"));
-        assertTrue(linkHandler.acceptUrl("https://lobstertheremin.com/track/unfinished"));
     }
 }

extractor/src/test/java/org/schabi/newpipe/extractor/services/bandcamp/BandcampStreamLinkHandlerFactoryTest.java

@@ -49,6 +49,5 @@ public void testAcceptUrl() throws ParsingException {
         assertTrue(linkHandler.acceptUrl("https://interovgm.bandcamp.com/track/title"));
         assertTrue(linkHandler.acceptUrl("http://bandcamP.com/?show=38"));
         assertTrue(linkHandler.acceptUrl("https://goodgoodblood-tl.bandcamp.com/track/when-it-all-wakes-up"));
-        assertTrue(linkHandler.acceptUrl("https://lobstertheremin.com/track/unfinished"));
     }
 }