Merge pull request #579 from fynngodau/bandcamp-fix-linkhandler

[Bandcamp] Fix link handler acceptance behaviour

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

@@ -7,20 +7,15 @@ import com.grack.nanojson.JsonParser;
 import com.grack.nanojson.JsonParserException;
 import com.grack.nanojson.JsonWriter;
 import org.jsoup.Jsoup;
-import org.jsoup.nodes.Document;
 import org.schabi.newpipe.extractor.NewPipe;
 import org.schabi.newpipe.extractor.exceptions.ParsingException;
 import org.schabi.newpipe.extractor.exceptions.ReCaptchaException;
 import org.schabi.newpipe.extractor.localization.DateWrapper;
-import org.schabi.newpipe.extractor.utils.Utils;
 
 import java.io.IOException;
 import java.time.DateTimeException;
 import java.time.ZonedDateTime;
 import java.time.format.DateTimeFormatter;
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.List;
 import java.util.Locale;
 
 public class BandcampExtractorHelper {
@@ -95,12 +90,18 @@ public class BandcampExtractorHelper {
         if (url.toLowerCase().matches("https?://.+\\.bandcamp\\.com(/.*)?")) return true;
 
         try {
-            // Accept all other URLs if they contain a <meta> tag that says they are generated by bandcamp
+            // Test other URLs for whether they contain a footer that links to bandcamp
             return Jsoup.parse(
                     NewPipe.getDownloader().get(url).responseBody()
             )
-                    .getElementsByAttributeValue("name", "generator")
+                    .getElementById("pgFt")
-                    .attr("content").equals("Bandcamp");
+                    .getElementById("pgFt-inner")
+                    .getElementById("footer-logo-wrapper")
+                    .getElementById("footer-logo")
+                    .getElementsByClass("hiddenAccess")
+                    .text().equals("Bandcamp");
+        } catch (NullPointerException e) {
+            return false;
         } catch (IOException | ReCaptchaException e) {
             throw new ParsingException("Could not determine whether URL is custom domain " +
                     "(not available? network error?)");

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

@@ -55,7 +55,9 @@ public class BandcampChannelLinkHandlerFactory extends ListLinkHandlerFactory {
      * Accepts only pages that lead to the root of an artist profile. Supports external pages.
      */
     @Override
-    public boolean onAcceptUrl(final String url) throws ParsingException {
+    public boolean onAcceptUrl(String url) throws ParsingException {
+
+        url = url.toLowerCase();
 
         // https: | | artist.bandcamp.com | releases
         //  0      1           2               3
@@ -64,8 +66,10 @@ public class BandcampChannelLinkHandlerFactory extends ListLinkHandlerFactory {
         // URL is too short
         if (splitUrl.length < 3) return false;
 
-        // Must have "releases" as segment after url or none at all
+        // Must have "releases" or "music" as segment after url or none at all
-        if (splitUrl.length > 3 && !splitUrl[3].equals("releases")) {
+        if (splitUrl.length > 3 && !(
+                splitUrl[3].equals("releases") || splitUrl[3].equals("music")
+        )) {
 
             return false;
 

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

@@ -29,17 +29,24 @@ public class BandcampChannelLinkHandlerFactoryTest {
         assertTrue(linkHandler.acceptUrl("http://zachbenson.bandcamp.com"));
         assertTrue(linkHandler.acceptUrl("https://zachbenson.bandcamp.com/"));
         assertTrue(linkHandler.acceptUrl("https://billwurtz.bandcamp.com/releases"));
+        assertTrue(linkHandler.acceptUrl("https://interovgm.bandcamp.com/releases"));
+        assertTrue(linkHandler.acceptUrl("https://interovgm.bandcamp.com/releases/"));
 
         assertTrue(linkHandler.acceptUrl("http://zachbenson.bandcamp.com/"));
 
         assertFalse(linkHandler.acceptUrl("https://bandcamp.com"));
         assertFalse(linkHandler.acceptUrl("https://zachbenson.bandcamp.com/track/kitchen"));
         assertFalse(linkHandler.acceptUrl("https://daily.bandcamp.com/"));
+        assertFalse(linkHandler.acceptUrl("https://DAILY.BANDCAMP.COM"));
         assertFalse(linkHandler.acceptUrl("https://daily.bandcamp.com/best-of-2020/bandcamp-daily-staffers-on-their-favorite-albums-of-2020"));
 
         // External URLs
-        assertTrue(linkHandler.acceptUrl("https://interovgm.bandcamp.com/releases"));
+        assertTrue(linkHandler.acceptUrl("https://lobstertheremin.com"));
-        assertTrue(linkHandler.acceptUrl("https://interovgm.bandcamp.com/releases/"));
+        assertTrue(linkHandler.acceptUrl("https://lobstertheremin.com/music"));
+        assertTrue(linkHandler.acceptUrl("https://lobstertheremin.com/music/"));
+        assertTrue(linkHandler.acceptUrl("https://diskak.usopop.com/"));
+        assertTrue(linkHandler.acceptUrl("https://diskak.usopop.com/releases"));
+        assertTrue(linkHandler.acceptUrl("https://diskak.usopop.com/RELEASES"));
 
         assertFalse(linkHandler.acceptUrl("https://example.com/releases"));
     }
@@ -51,6 +58,10 @@ public class BandcampChannelLinkHandlerFactoryTest {
         assertEquals("1581461772", linkHandler.getId("https://interovgm.bandcamp.com/releases"));
         assertEquals("3321800855", linkHandler.getId("https://infiniteammo.bandcamp.com/"));
         assertEquals("3775652329", linkHandler.getId("https://npet.bandcamp.com/"));
+
+        assertEquals("2735462545", linkHandler.getId("http://lobstertheremin.com/"));
+        assertEquals("2735462545", linkHandler.getId("https://lobstertheremin.com/music/"));
+        assertEquals("3826445168", linkHandler.getId("https://diskak.usopop.com/releases"));
     }
 
     @Test
@@ -58,6 +69,8 @@ public class BandcampChannelLinkHandlerFactoryTest {
         assertEquals("https://macbenson.bandcamp.com", linkHandler.getUrl("1196681540"));
         assertEquals("https://interovgm.bandcamp.com", linkHandler.getUrl("1581461772"));
         assertEquals("https://infiniteammo.bandcamp.com", linkHandler.getUrl("3321800855"));
+
+        assertEquals("https://lobstertheremin.com", linkHandler.getUrl("2735462545"));
     }
 
     @Test(expected = ParsingException.class)

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

@@ -49,5 +49,6 @@ public class BandcampStreamLinkHandlerFactoryTest {
         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"));
     }
 }