Rework link handlers to correctly accept external websites

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

@@ -123,6 +123,28 @@ public class BandcampExtractorHelper {
         return "https://f4.bcbits.com/img/" + (album ? 'a' : "") + id + "_10.jpg";
     }
 
+    /**
+     * @return <code>true</code> if the given url looks like it comes from a bandcamp custom domain
+     * or if it comes from bandcamp.com itself
+     */
+    public static boolean isSupportedDomain(final String url) throws ParsingException {
+
+        // Accept all bandcamp.com URLs
+        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
+            return Jsoup.parse(
+                    NewPipe.getDownloader().get(url).responseBody()
+            )
+                    .getElementsByAttributeValue("name", "generator")
+                    .attr("content").equals("Bandcamp");
+        } catch (IOException | ReCaptchaException e) {
+            throw new ParsingException("Could not determine whether URL is custom domain " +
+                    "(not available? network error?)");
+        }
+    }
+
     static DateWrapper parseDate(final String textDate) throws ParsingException {
         try {
             final Date date = new SimpleDateFormat("dd MMM yyyy HH:mm:ss zzz", Locale.ENGLISH).parse(textDate);

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

@@ -24,7 +24,7 @@ public class BandcampChannelLinkHandlerFactory extends ListLinkHandlerFactory {
         try {
             final String response = NewPipe.getDownloader().get(url).responseBody();
 
-            // This variable contains band data!
+            // Use band data embedded in website to extract ID
             final JsonObject bandData = BandcampExtractorHelper.getJsonData(response, "data-band");
 
             return String.valueOf(bandData.getLong("id"));
@@ -51,17 +51,15 @@ public class BandcampChannelLinkHandlerFactory extends ListLinkHandlerFactory {
     }
 
     /**
-     * Matches <code>* .bandcamp.com</code> as well as custom domains
+     * Accepts only pages that do not lead to an album or track. Supports external pages.
-     * where the profile is at <code>* . * /releases</code>
      */
     @Override
-    public boolean onAcceptUrl(final String url) {
+    public boolean onAcceptUrl(final String url) throws ParsingException {
 
-        // Is a subdomain of bandcamp.com?
+        // Exclude URLs that lead to a track or album
-        boolean isBandcampComArtistPage = url.matches("https?://.+\\.bandcamp\\.com/?");
+        if (url.matches(".*/(album|track)/.*")) return false;
 
-        boolean isCustomDomainReleases = url.matches("https?://.+\\..+/releases/?(?!.)");
+        // Test whether domain is supported
-
+        return BandcampExtractorHelper.isSupportedDomain(url);
-        return isBandcampComArtistPage || isCustomDomainReleases;
     }
 }

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

@@ -4,6 +4,7 @@ package org.schabi.newpipe.extractor.services.bandcamp.linkHandler;
 
 import org.schabi.newpipe.extractor.exceptions.ParsingException;
 import org.schabi.newpipe.extractor.linkhandler.ListLinkHandlerFactory;
+import org.schabi.newpipe.extractor.services.bandcamp.extractors.BandcampExtractorHelper;
 
 import java.util.List;
 
@@ -22,8 +23,16 @@ public class BandcampPlaylistLinkHandlerFactory extends ListLinkHandlerFactory {
         return url;
     }
 
+    /**
+     * Accepts all bandcamp URLs that contain /album/ behind their domain name.
+     */
     @Override
-    public boolean onAcceptUrl(final String url) {
+    public boolean onAcceptUrl(final String url) throws ParsingException {
-        return url.toLowerCase().matches("https?://.+\\..+/album/.+");
+
+        // Exclude URLs which do not lead to an album
+        if (!url.toLowerCase().matches("https?://.+\\..+/album/.+")) return false;
+
+        // Test whether domain is supported
+        return BandcampExtractorHelper.isSupportedDomain(url);
     }
 }

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

@@ -4,6 +4,7 @@ package org.schabi.newpipe.extractor.services.bandcamp.linkHandler;
 
 import org.schabi.newpipe.extractor.exceptions.ParsingException;
 import org.schabi.newpipe.extractor.linkhandler.LinkHandlerFactory;
+import org.schabi.newpipe.extractor.services.bandcamp.extractors.BandcampExtractorHelper;
 
 /**
  * <p>Tracks don't have standalone ids, they are always in combination with the band id.
@@ -40,16 +41,19 @@ public class BandcampStreamLinkHandlerFactory extends LinkHandlerFactory {
     }
 
     /**
-     * Sometimes, the root page of an artist is also an album or track
+     * Accepts URLs that point to a bandcamp radio show or that are a bandcamp
-     * page. In that case, it is assumed that one actually wants to open
+     * domain and point to a track.
-     * the profile and not the track it has set as the default one.
-     * <p>Urls are expected to be in this format to account for
-     * custom domains:</p>
-     * <code>https:// * . * /track/ *</code>
      */
     @Override
-    public boolean onAcceptUrl(final String url) {
+    public boolean onAcceptUrl(final String url) throws ParsingException {
-        return url.toLowerCase().matches("https?://.+\\..+/track/.+")
+
-                || url.toLowerCase().matches("https?://bandcamp\\.com/\\?show=\\d+");
+        // Accept Bandcamp radio
+        if (url.toLowerCase().matches("https?://bandcamp\\.com/\\?show=\\d+")) return true;
+
+        // Don't accept URLs that don't point to a track
+        if (!url.toLowerCase().matches("https?://.+\\..+/track/.+")) return false;
+
+        // Test whether domain is supported
+        return BandcampExtractorHelper.isSupportedDomain(url);
     }
 }

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

@@ -26,13 +26,19 @@ public class BandcampChannelLinkHandlerFactoryTest {
 
     @Test
     public void testAcceptUrl() throws ParsingException {
-        assertTrue(linkHandler.acceptUrl("http://interovgm.com/releases/"));
+        // Bandcamp URLs
-        assertTrue(linkHandler.acceptUrl("https://interovgm.com/releases"));
         assertTrue(linkHandler.acceptUrl("http://zachbenson.bandcamp.com"));
         assertTrue(linkHandler.acceptUrl("https://zachbenson.bandcamp.com/"));
+        assertTrue(linkHandler.acceptUrl("https://billwurtz.bandcamp.com/releases"));
 
         assertFalse(linkHandler.acceptUrl("https://bandcamp.com"));
         assertFalse(linkHandler.acceptUrl("https://zachbenson.bandcamp.com/track/kitchen"));
+
+        // External URLs
+        assertTrue(linkHandler.acceptUrl("http://interovgm.com/releases/"));
+        assertTrue(linkHandler.acceptUrl("https://interovgm.com/releases"));
+
+        assertFalse(linkHandler.acceptUrl("https://example.com/releases"));
     }
 
     @Test

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

@@ -35,6 +35,7 @@ public class BandcampPlaylistLinkHandlerFactoryTest {
         assertFalse(linkHandler.acceptUrl("https://zachbenson.bandcamp.com/"));
         assertFalse(linkHandler.acceptUrl("https://zachbenson.bandcamp.com/track/kitchen"));
         assertFalse(linkHandler.acceptUrl("https://interovgm.com/track/title"));
+        assertFalse(linkHandler.acceptUrl("https://example.com/album/samplealbum"));
 
         assertTrue(linkHandler.acceptUrl("https://powertothequeerkids.bandcamp.com/album/power-to-the-queer-kids"));
         assertTrue(linkHandler.acceptUrl("https://zachbenson.bandcamp.com/album/prom"));

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

@@ -43,6 +43,7 @@ public class BandcampStreamLinkHandlerFactoryTest {
         assertFalse(linkHandler.acceptUrl("https://bandcamp.com"));
         assertFalse(linkHandler.acceptUrl("https://zachbenson.bandcamp.com/"));
         assertFalse(linkHandler.acceptUrl("https://powertothequeerkids.bandcamp.com/album/power-to-the-queer-kids"));
+        assertFalse(linkHandler.acceptUrl("https://example.com/track/sampletrack"));
 
         assertTrue(linkHandler.acceptUrl("https://zachbenson.bandcamp.com/track/kitchen"));
         assertTrue(linkHandler.acceptUrl("http://ZachBenson.Bandcamp.COM/Track/U-I-Tonite/"));