Merge pull request #160 from Stypox/invalid-youtube-subscription-fix

TobiGr · web-flow · commit 2ac713e70f26 · 2019-05-14T22:26:25.000+02:00
Fixed youtube subscription import: ignore ones with invalid url and keep ones with empty title.
diff --git a/extractor/src/main/java/org/schabi/newpipe/extractor/services/youtube/extractors/YoutubeSubscriptionExtractor.java b/extractor/src/main/java/org/schabi/newpipe/extractor/services/youtube/extractors/YoutubeSubscriptionExtractor.java
@@ -63,16 +63,10 @@ private List<SubscriptionItem> getItemsFromOPML(InputStream contentInputStream)
             String title = outline.attr("title");
             String xmlUrl = outline.attr("abs:xmlUrl");
 
-            if (title.isEmpty() || xmlUrl.isEmpty()) {
-                throw new InvalidSourceException("document has invalid entries");
-            }
-
             try {
                 String id = Parser.matchGroup1(ID_PATTERN, xmlUrl);
                 result.add(new SubscriptionItem(service.getServiceId(), BASE_CHANNEL_URL + id, title));
-            } catch (Parser.RegexException e) {
-                throw new InvalidSourceException("document has invalid entries", e);
-            }
+            } catch (Parser.RegexException ignored) { /* ignore invalid subscriptions */ }
         }
 
         return result;
diff --git a/extractor/src/test/java/org/schabi/newpipe/extractor/services/youtube/YoutubeSubscriptionExtractorTest.java b/extractor/src/test/java/org/schabi/newpipe/extractor/services/youtube/YoutubeSubscriptionExtractorTest.java
@@ -59,15 +59,38 @@ public void testEmptySourceException() throws Exception {
         assertTrue(items.isEmpty());
     }
 
+    @Test
+    public void testSubscriptionWithEmptyTitleInSource() throws Exception {
+        String channelId = "AA0AaAa0AaaaAAAAAA0aa0AA";
+        String source = "<opml version=\"1.1\"><body><outline text=\"YouTube Subscriptions\" title=\"YouTube Subscriptions\">" +
+                "<outline text=\"\" title=\"\" type=\"rss\" xmlUrl=\"https://www.youtube.com/feeds/videos.xml?channel_id=" + channelId + "\" />" +
+                "</outline></body></opml>";
+
+        List<SubscriptionItem> items = subscriptionExtractor.fromInputStream(new ByteArrayInputStream(source.getBytes("UTF-8")));
+        assertTrue("List doesn't have exactly 1 item (had " + items.size() + ")", items.size() == 1);
+        assertTrue("Item does not have an empty title (had \"" + items.get(0).getName() + "\")", items.get(0).getName().isEmpty());
+        assertTrue("Item does not have the right channel id \"" + channelId + "\" (the whole url is \"" + items.get(0).getUrl() + "\")", items.get(0).getUrl().endsWith(channelId));
+    }
+
+    @Test
+    public void testSubscriptionWithInvalidUrlInSource() throws Exception {
+        String source = "<opml version=\"1.1\"><body><outline text=\"YouTube Subscriptions\" title=\"YouTube Subscriptions\">" +
+                "<outline text=\"invalid\" title=\"url\" type=\"rss\" xmlUrl=\"https://www.youtube.com/feeds/videos.xml?channel_not_id=|||||||\"/>" +
+                "<outline text=\"fail\" title=\"fail\" type=\"rss\" xmlUgrl=\"invalidTag\"/>" +
+                "<outline text=\"invalid\" title=\"url\" type=\"rss\" xmlUrl=\"\"/>" +
+                "<outline text=\"\" title=\"\" type=\"rss\" xmlUrl=\"\"/>" +
+                "</outline></body></opml>";
+
+        List<SubscriptionItem> items = subscriptionExtractor.fromInputStream(new ByteArrayInputStream(source.getBytes("UTF-8")));
+        assertTrue(items.isEmpty());
+    }
+
     @Test
     public void testInvalidSourceException() {
         List<String> invalidList = Arrays.asList(
                 "<xml><notvalid></notvalid></xml>",
                 "<opml><notvalid></notvalid></opml>",
                 "<opml><body></body></opml>",
-                "<opml><body><outline text=\"fail\" title=\"fail\" type=\"rss\" xmlUgrl=\"invalidTag\"/></outline></body></opml>",
-                "<opml><body><outline><outline text=\"invalid\" title=\"url\" type=\"rss\"" +
-                        " xmlUrl=\"https://www.youtube.com/feeds/videos.xml?channel_not_id=|||||||\"/></outline></body></opml>",
                 "",
                 null,
                 "\uD83D\uDC28\uD83D\uDC28\uD83D\uDC28",
@@ -78,11 +101,11 @@ public void testInvalidSourceException() {
                 if (invalidContent != null) {
                     byte[] bytes = invalidContent.getBytes("UTF-8");
                     subscriptionExtractor.fromInputStream(new ByteArrayInputStream(bytes));
+                    fail("Extracting from \"" + invalidContent + "\" didn't throw an exception");
                 } else {
                     subscriptionExtractor.fromInputStream(null);
+                    fail("Extracting from null String didn't throw an exception");
                 }
-
-                fail("didn't throw exception");
             } catch (Exception e) {
                 // System.out.println(" -> " + e);
                 boolean isExpectedException = e instanceof SubscriptionExtractor.InvalidSourceException;
