Parse podcast:transcript url and store in SQLite (#6739)

7 files changed, 106 insertions, 4 deletions

diff --git a/model/src/main/java/de/danoeh/antennapod/model/feed/FeedItem.java b/model/src/main/java/de/danoeh/antennapod/model/feed/FeedItem.java
index df4cc8f9c..bc95dc85f 100644
--- a/model/src/main/java/de/danoeh/antennapod/model/feed/FeedItem.java
+++ b/model/src/main/java/de/danoeh/antennapod/model/feed/FeedItem.java
@@ -43,6 +43,8 @@ public class FeedItem implements Serializable {
     private transient Feed feed;
     private long feedId;
     private String podcastIndexChapterUrl;
+    private String podcastIndexTranscriptUrl;
+    private String podcastIndexTranscriptType;
 
     private int state;
     public static final int NEW = -1;
@@ -83,7 +85,8 @@ public class FeedItem implements Serializable {
      * */
     public FeedItem(long id, String title, String link, Date pubDate, String paymentLink, long feedId,
                     boolean hasChapters, String imageUrl, int state,
-                    String itemIdentifier, boolean autoDownloadEnabled, String podcastIndexChapterUrl) {
+                    String itemIdentifier, boolean autoDownloadEnabled, String podcastIndexChapterUrl,
+                    String transcriptType, String transcriptUrl) {
         this.id = id;
         this.title = title;
         this.link = link;
@@ -96,6 +99,10 @@ public class FeedItem implements Serializable {
         this.itemIdentifier = itemIdentifier;
         this.autoDownloadEnabled = autoDownloadEnabled;
         this.podcastIndexChapterUrl = podcastIndexChapterUrl;
+        if (transcriptUrl != null) {
+            this.podcastIndexTranscriptUrl = transcriptUrl;
+            this.podcastIndexTranscriptType = transcriptType;
+        }
     }
 
     /**
@@ -162,6 +169,9 @@ public class FeedItem implements Serializable {
         if (other.podcastIndexChapterUrl != null) {
             podcastIndexChapterUrl = other.podcastIndexChapterUrl;
         }
+        if (other.getPodcastIndexTranscriptUrl() != null) {
+            podcastIndexTranscriptUrl = other.podcastIndexTranscriptUrl;
+        }
     }
 
     public long getId() {
@@ -413,6 +423,45 @@ public class FeedItem implements Serializable {
         podcastIndexChapterUrl = url;
     }
 
+    public void setPodcastIndexTranscriptUrl(String type, String url) {
+        updateTranscriptPreferredFormat(type, url);
+    }
+
+    public String getPodcastIndexTranscriptUrl() {
+        return podcastIndexTranscriptUrl;
+    }
+
+    public String getPodcastIndexTranscriptType() {
+        return podcastIndexTranscriptType;
+    }
+
+    public void updateTranscriptPreferredFormat(String type, String url) {
+        if (StringUtils.isEmpty(type) || StringUtils.isEmpty(url)) {
+            return;
+        }
+
+        String canonicalSrr = "application/srr";
+        String jsonType = "application/json";
+
+        switch (type) {
+            case "application/json":
+                podcastIndexTranscriptUrl = url;
+                podcastIndexTranscriptType = type;
+                break;
+            case "application/srr":
+            case "application/srt":
+            case "application/x-subrip":
+                if (podcastIndexTranscriptUrl == null || !podcastIndexTranscriptType.equals(jsonType)) {
+                    podcastIndexTranscriptUrl = url;
+                    podcastIndexTranscriptType = canonicalSrr;
+                }
+                break;
+            default:
+                System.out.println("Invalid format for transcript " + type);
+                break;
+        }
+    }
+
     @NonNull
     @Override
     public String toString() {
diff --git a/parser/feed/src/main/java/de/danoeh/antennapod/parser/feed/namespace/PodcastIndex.java b/parser/feed/src/main/java/de/danoeh/antennapod/parser/feed/namespace/PodcastIndex.java
index 1f543a5ae..dd11fb4e7 100644
--- a/parser/feed/src/main/java/de/danoeh/antennapod/parser/feed/namespace/PodcastIndex.java
+++ b/parser/feed/src/main/java/de/danoeh/antennapod/parser/feed/namespace/PodcastIndex.java
@@ -14,6 +14,8 @@ public class PodcastIndex extends Namespace {
     private static final String URL = "url";
     private static final String FUNDING = "funding";
     private static final String CHAPTERS = "chapters";
+    private static final String TRANSCRIPT = "transcript";
+    private static final String TYPE = "type";
 
     @Override
     public SyndElement handleElementStart(String localName, HandlerState state,
@@ -28,6 +30,12 @@ public class PodcastIndex extends Namespace {
             if (!TextUtils.isEmpty(href)) {
                 state.getCurrentItem().setPodcastIndexChapterUrl(href);
             }
+        } else if (TRANSCRIPT.equals(localName)) {
+            String href = attributes.getValue(URL);
+            String type = attributes.getValue(TYPE);
+            if (!TextUtils.isEmpty(href) && !TextUtils.isEmpty(type)) {
+                state.getCurrentItem().setPodcastIndexTranscriptUrl(type, href);
+            }
         }
         return new SyndElement(localName, this);
     }
diff --git a/parser/feed/src/test/java/de/danoeh/antennapod/parser/feed/element/namespace/RssParserTest.java b/parser/feed/src/test/java/de/danoeh/antennapod/parser/feed/element/namespace/RssParserTest.java
index bc30f2d7c..983766f12 100644
--- a/parser/feed/src/test/java/de/danoeh/antennapod/parser/feed/element/namespace/RssParserTest.java
+++ b/parser/feed/src/test/java/de/danoeh/antennapod/parser/feed/element/namespace/RssParserTest.java
@@ -98,6 +98,14 @@ public class RssParserTest {
     }
 
     @Test
+    public void testPodcastIndexTranscript() throws Exception {
+        File feedFile = FeedParserTestHelper.getFeedFile("feed-rss-testPodcastIndexTranscript.xml");
+        Feed feed = FeedParserTestHelper.runFeedParser(feedFile);
+        assertEquals("https://podnews.net/audio/podnews231011.mp3.json", feed.getItems().get(0).getPodcastIndexTranscriptUrl());
+        assertEquals("application/json", feed.getItems().get(0).getPodcastIndexTranscriptType());
+    }
+
+    @Test
     public void testUnsupportedElements() throws Exception {
         File feedFile = FeedParserTestHelper.getFeedFile("feed-rss-testUnsupportedElements.xml");
         Feed feed = FeedParserTestHelper.runFeedParser(feedFile);
diff --git a/parser/feed/src/test/resources/feed-rss-testPodcastIndexTranscript.xml b/parser/feed/src/test/resources/feed-rss-testPodcastIndexTranscript.xml
new file mode 100644
index 000000000..f972ee207
--- /dev/null
+++ b/parser/feed/src/test/resources/feed-rss-testPodcastIndexTranscript.xml
@@ -0,0 +1,13 @@
+<?xml version='1.0' encoding='UTF-8' ?>
+<rss version="2.0" xmlns:podcast="https://podcastindex.org/namespace/1.0">
+    <channel>
+        <title>title</title>
+        <item>
+            <title>Podcasts in YouTube make it to the UK</title>
+            <link>https://podnews.net/update/youtube-podcasts-uk</link>
+            <pubDate>Tue, 10 Oct 2023 08:46:31 +0000</pubDate>
+            <podcast:transcript url="https://podnews.net/audio/podnews231011.mp3.json" type="application/json" />
+            <podcast:transcript url="https://podnews.net/audio/podnews231011.mp3.srt" type="application/srt" />
+        </item>
+    </channel>
+</rss>
diff --git a/storage/database/src/main/java/de/danoeh/antennapod/storage/database/DBUpgrader.java b/storage/database/src/main/java/de/danoeh/antennapod/storage/database/DBUpgrader.java
index be09a5132..a3b86be74 100644
--- a/storage/database/src/main/java/de/danoeh/antennapod/storage/database/DBUpgrader.java
+++ b/storage/database/src/main/java/de/danoeh/antennapod/storage/database/DBUpgrader.java
@@ -342,6 +342,10 @@ class DBUpgrader {
         if (oldVersion < 3050000) {
             db.execSQL("ALTER TABLE " + PodDBAdapter.TABLE_NAME_FEEDS
                     + " ADD COLUMN " + PodDBAdapter.KEY_STATE + " INTEGER DEFAULT " + Feed.STATE_SUBSCRIBED);
+            db.execSQL("ALTER TABLE " + PodDBAdapter.TABLE_NAME_FEED_ITEMS
+                    + " ADD COLUMN " + PodDBAdapter.KEY_PODCASTINDEX_TRANSCRIPT_URL + " TEXT");
+            db.execSQL("ALTER TABLE " + PodDBAdapter.TABLE_NAME_FEED_ITEMS
+                    + " ADD COLUMN " + PodDBAdapter.KEY_PODCASTINDEX_TRANSCRIPT_TYPE + " TEXT");
         }
     }
 
diff --git a/storage/database/src/main/java/de/danoeh/antennapod/storage/database/PodDBAdapter.java b/storage/database/src/main/java/de/danoeh/antennapod/storage/database/PodDBAdapter.java
index aec136457..c7204b567 100644
--- a/storage/database/src/main/java/de/danoeh/antennapod/storage/database/PodDBAdapter.java
+++ b/storage/database/src/main/java/de/danoeh/antennapod/storage/database/PodDBAdapter.java
@@ -122,6 +122,8 @@ public class PodDBAdapter {
     public static final String KEY_NEW_EPISODES_ACTION = "new_episodes_action";
     public static final String KEY_PODCASTINDEX_CHAPTER_URL = "podcastindex_chapter_url";
     public static final String KEY_STATE = "state";
+    public static final String KEY_PODCASTINDEX_TRANSCRIPT_URL = "podcastindex_transcript_url";
+    public static final String KEY_PODCASTINDEX_TRANSCRIPT_TYPE = "podcastindex_transcript_type";
 
     // Table names
     public static final String TABLE_NAME_FEEDS = "Feeds";
@@ -184,7 +186,9 @@ public class PodDBAdapter {
             + KEY_HAS_CHAPTERS + " INTEGER," + KEY_ITEM_IDENTIFIER + " TEXT,"
             + KEY_IMAGE_URL + " TEXT,"
             + KEY_AUTO_DOWNLOAD_ENABLED + " INTEGER,"
-            + KEY_PODCASTINDEX_CHAPTER_URL + " TEXT)";
+            + KEY_PODCASTINDEX_CHAPTER_URL + " TEXT,"
+            + KEY_PODCASTINDEX_TRANSCRIPT_TYPE + " TEXT,"
+            + KEY_PODCASTINDEX_TRANSCRIPT_URL + " TEXT" + ")";
 
     private static final String CREATE_TABLE_FEED_MEDIA = "CREATE TABLE "
             + TABLE_NAME_FEED_MEDIA + " (" + TABLE_PRIMARY_KEY + KEY_DURATION
@@ -272,7 +276,9 @@ public class PodDBAdapter {
             + TABLE_NAME_FEED_ITEMS + "." + KEY_ITEM_IDENTIFIER + ", "
             + TABLE_NAME_FEED_ITEMS + "." + KEY_IMAGE_URL + ", "
             + TABLE_NAME_FEED_ITEMS + "." + KEY_AUTO_DOWNLOAD_ENABLED + ", "
-            + TABLE_NAME_FEED_ITEMS + "." + KEY_PODCASTINDEX_CHAPTER_URL;
+            + TABLE_NAME_FEED_ITEMS + "." + KEY_PODCASTINDEX_CHAPTER_URL + ", "
+            + TABLE_NAME_FEED_ITEMS + "." + KEY_PODCASTINDEX_TRANSCRIPT_TYPE + ", "
+            + TABLE_NAME_FEED_ITEMS + "." + KEY_PODCASTINDEX_TRANSCRIPT_URL;
 
     private static final String KEYS_FEED_MEDIA =
             TABLE_NAME_FEED_MEDIA + "." + KEY_ID + " AS " + SELECT_KEY_MEDIA_ID + ", "
@@ -675,6 +681,14 @@ public class PodDBAdapter {
         values.put(KEY_IMAGE_URL, item.getImageUrl());
         values.put(KEY_PODCASTINDEX_CHAPTER_URL, item.getPodcastIndexChapterUrl());
 
+        // We only store one transcript url, we prefer JSON if it exists
+        String type = item.getPodcastIndexTranscriptType();
+        String url = item.getPodcastIndexTranscriptUrl();
+        if (url != null) {
+            values.put(KEY_PODCASTINDEX_TRANSCRIPT_TYPE, type);
+            values.put(KEY_PODCASTINDEX_TRANSCRIPT_URL, url);
+        }
+
         if (item.getId() == 0) {
             item.setId(db.insert(TABLE_NAME_FEED_ITEMS, null, values));
         } else {
diff --git a/storage/database/src/main/java/de/danoeh/antennapod/storage/database/mapper/FeedItemCursor.java b/storage/database/src/main/java/de/danoeh/antennapod/storage/database/mapper/FeedItemCursor.java
index d526299e4..41e4d474a 100644
--- a/storage/database/src/main/java/de/danoeh/antennapod/storage/database/mapper/FeedItemCursor.java
+++ b/storage/database/src/main/java/de/danoeh/antennapod/storage/database/mapper/FeedItemCursor.java
@@ -26,6 +26,8 @@ public class FeedItemCursor extends CursorWrapper {
     private final int indexImageUrl;
     private final int indexPodcastIndexChapterUrl;
     private final int indexMediaId;
+    private final int indexPodcastIndexTranscriptUrl;
+    private final int indexPodcastIndexTranscriptType;
 
     public FeedItemCursor(Cursor cursor) {
         super(new FeedMediaCursor(cursor));
@@ -43,6 +45,8 @@ public class FeedItemCursor extends CursorWrapper {
         indexImageUrl = cursor.getColumnIndexOrThrow(PodDBAdapter.KEY_IMAGE_URL);
         indexPodcastIndexChapterUrl = cursor.getColumnIndexOrThrow(PodDBAdapter.KEY_PODCASTINDEX_CHAPTER_URL);
         indexMediaId = cursor.getColumnIndexOrThrow(PodDBAdapter.SELECT_KEY_MEDIA_ID);
+        indexPodcastIndexTranscriptUrl = cursor.getColumnIndexOrThrow(PodDBAdapter.KEY_PODCASTINDEX_TRANSCRIPT_URL);
+        indexPodcastIndexTranscriptType = cursor.getColumnIndexOrThrow(PodDBAdapter.KEY_PODCASTINDEX_TRANSCRIPT_TYPE);
     }
 
     /**
@@ -62,7 +66,9 @@ public class FeedItemCursor extends CursorWrapper {
                 getInt(indexRead),
                 getString(indexItemIdentifier),
                 getLong(indexAutoDownload) > 0,
-                getString(indexPodcastIndexChapterUrl));
+                getString(indexPodcastIndexChapterUrl),
+                getString(indexPodcastIndexTranscriptUrl),
+                getString(indexPodcastIndexTranscriptType));
         if (!isNull(indexMediaId)) {
             item.setMedia(feedMediaCursor.getFeedMedia());
         }