Read vorbis description of local files

8 files changed, 171 insertions, 76 deletions

diff --git a/core/src/main/java/de/danoeh/antennapod/core/feed/LocalFeedUpdater.java b/core/src/main/java/de/danoeh/antennapod/core/feed/LocalFeedUpdater.java
index a63b9d41c..988b7c015 100644
--- a/core/src/main/java/de/danoeh/antennapod/core/feed/LocalFeedUpdater.java
+++ b/core/src/main/java/de/danoeh/antennapod/core/feed/LocalFeedUpdater.java
@@ -38,6 +38,8 @@ import de.danoeh.antennapod.model.playback.MediaType;
 import de.danoeh.antennapod.parser.feed.util.MimeTypeUtils;
 import de.danoeh.antennapod.parser.media.id3.ID3ReaderException;
 import de.danoeh.antennapod.parser.media.id3.Id3MetadataReader;
+import de.danoeh.antennapod.parser.media.vorbis.VorbisCommentMetadataReader;
+import de.danoeh.antennapod.parser.media.vorbis.VorbisCommentReaderException;
 import org.apache.commons.io.input.CountingInputStream;
 
 public class LocalFeedUpdater {
@@ -209,8 +211,15 @@ public class LocalFeedUpdater {
             reader.readInputStream();
             item.setDescriptionIfLonger(reader.getComment());
         } catch (IOException | ID3ReaderException e) {
-            // Do not flood Logcat with full stack traces
             Log.d(TAG, "Unable to parse ID3 of " + file.getUri() + ": " + e.getMessage());
+
+            try (InputStream inputStream = context.getContentResolver().openInputStream(file.getUri())) {
+                VorbisCommentMetadataReader reader = new VorbisCommentMetadataReader(inputStream);
+                reader.readInputStream();
+                item.setDescriptionIfLonger(reader.getDescription());
+            } catch (IOException | VorbisCommentReaderException e2) {
+                Log.d(TAG, "Unable to parse vorbis comments of " + file.getUri() + ": " + e2.getMessage());
+            }
         }
     }
 
diff --git a/core/src/main/java/de/danoeh/antennapod/core/util/ChapterUtils.java b/core/src/main/java/de/danoeh/antennapod/core/util/ChapterUtils.java
index e9f812c7f..723ea1d47 100644
--- a/core/src/main/java/de/danoeh/antennapod/core/util/ChapterUtils.java
+++ b/core/src/main/java/de/danoeh/antennapod/core/util/ChapterUtils.java
@@ -171,8 +171,8 @@ public class ChapterUtils {
 
     @NonNull
     private static List<Chapter> readOggChaptersFromInputStream(InputStream input) throws VorbisCommentReaderException {
-        VorbisCommentChapterReader reader = new VorbisCommentChapterReader();
-        reader.readInputStream(input);
+        VorbisCommentChapterReader reader = new VorbisCommentChapterReader(input);
+        reader.readInputStream();
         List<Chapter> chapters = reader.getChapters();
         if (chapters == null) {
             return Collections.emptyList();
diff --git a/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentChapterReader.java b/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentChapterReader.java
index e3b91a0e7..8290a547a 100644
--- a/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentChapterReader.java
+++ b/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentChapterReader.java
@@ -2,6 +2,7 @@ package de.danoeh.antennapod.parser.media.vorbis;
 
 import android.util.Log;
 
+import java.io.InputStream;
 import java.util.ArrayList;
 import java.util.List;
 import java.util.concurrent.TimeUnit;
@@ -17,25 +18,15 @@ public class VorbisCommentChapterReader extends VorbisCommentReader {
     private static final String CHAPTER_ATTRIBUTE_LINK = "url";
     private static final int CHAPTERXXX_LENGTH = "chapterxxx".length();
 
-    private List<Chapter> chapters;
+    private final List<Chapter> chapters = new ArrayList<>();
 
-    public VorbisCommentChapterReader() {
+    public VorbisCommentChapterReader(InputStream input) {
+        super(input);
     }
 
     @Override
-    public void onVorbisCommentFound() {
-        System.out.println("Vorbis comment found");
-    }
-
-    @Override
-    public void onVorbisCommentHeaderFound(VorbisCommentHeader header) {
-        chapters = new ArrayList<>();
-        System.out.println(header.toString());
-    }
-
-    @Override
-    public boolean onContentVectorKey(String content) {
-        return content.matches(CHAPTER_KEY);
+    public boolean handles(String key) {
+        return key.matches(CHAPTER_KEY);
     }
 
     @Override
@@ -68,19 +59,6 @@ public class VorbisCommentChapterReader extends VorbisCommentReader {
         }
     }
 
-    @Override
-    public void onEndOfComment() {
-        System.out.println("End of comment");
-        for (Chapter c : chapters) {
-            System.out.println(c.toString());
-        }
-    }
-
-    @Override
-    public void onError(VorbisCommentReaderException exception) {
-        exception.printStackTrace();
-    }
-
     private Chapter getChapterById(long id) {
         for (Chapter c : chapters) {
             if (("" + id).equals(c.getChapterId())) {
diff --git a/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentMetadataReader.java b/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentMetadataReader.java
new file mode 100644
index 000000000..158a0d8f7
--- /dev/null
+++ b/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentMetadataReader.java
@@ -0,0 +1,32 @@
+package de.danoeh.antennapod.parser.media.vorbis;
+
+import java.io.InputStream;
+
+public class VorbisCommentMetadataReader extends VorbisCommentReader {
+    private static final String KEY_DESCRIPTION = "description";
+    private static final String KEY_COMMENT = "comment";
+
+    private String description = null;
+
+    public VorbisCommentMetadataReader(InputStream input) {
+        super(input);
+    }
+
+    @Override
+    public boolean handles(String key) {
+        return KEY_DESCRIPTION.equals(key) || KEY_COMMENT.equals(key);
+    }
+
+    @Override
+    public void onContentVectorValue(String key, String value) {
+        if (KEY_DESCRIPTION.equals(key) || KEY_COMMENT.equals(key)) {
+            if (description == null || value.length() > description.length()) {
+                description = value;
+            }
+        }
+    }
+
+    public String getDescription() {
+        return description;
+    }
+}
diff --git a/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentReader.java b/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentReader.java
index 3d5f29f17..13126a73d 100644
--- a/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentReader.java
+++ b/parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentReader.java
@@ -3,6 +3,7 @@ package de.danoeh.antennapod.parser.media.vorbis;
 import androidx.annotation.NonNull;
 import org.apache.commons.io.EndianUtils;
 import org.apache.commons.io.IOUtils;
+import android.util.Log;
 
 import java.io.IOException;
 import java.io.InputStream;
@@ -12,51 +13,35 @@ import java.nio.charset.Charset;
 import java.util.Locale;
 
 public abstract class VorbisCommentReader {
-    /** Length of first page in an ogg file in bytes. */
+    private static final String TAG = "VorbisCommentReader";
     private static final int FIRST_OGG_PAGE_LENGTH = 58;
     private static final int FIRST_OPUS_PAGE_LENGTH = 47;
     private static final int SECOND_PAGE_MAX_LENGTH = 64 * 1024 * 1024;
     private static final int PACKET_TYPE_IDENTIFICATION = 1;
     private static final int PACKET_TYPE_COMMENT = 3;
 
-    /** Called when Reader finds identification header. */
-    protected abstract void onVorbisCommentFound();
+    private final InputStream input;
 
-    protected abstract void onVorbisCommentHeaderFound(VorbisCommentHeader header);
-
-    /**
-     * Is called every time the Reader finds a content vector. The handler
-     * should return true if it wants to handle the content vector.
-     */
-    protected abstract boolean onContentVectorKey(String content);
-
-    /**
-     * Is called if onContentVectorKey returned true for the key.
-     */
-    protected abstract void onContentVectorValue(String key, String value) throws VorbisCommentReaderException;
-
-    protected abstract void onEndOfComment();
-
-    protected abstract void onError(VorbisCommentReaderException exception);
+    VorbisCommentReader(InputStream input) {
+        this.input = input;
+    }
 
-    public void readInputStream(InputStream input) throws VorbisCommentReaderException {
+    public void readInputStream() throws VorbisCommentReaderException {
         try {
-            findIdentificationHeader(input);
-            onVorbisCommentFound();
-            findOggPage(input);
-            findCommentHeader(input);
-            VorbisCommentHeader commentHeader = readCommentHeader(input);
-            onVorbisCommentHeaderFound(commentHeader);
+            findIdentificationHeader();
+            findOggPage();
+            findCommentHeader();
+            VorbisCommentHeader commentHeader = readCommentHeader();
+            Log.d(TAG, commentHeader.toString());
             for (int i = 0; i < commentHeader.getUserCommentLength(); i++) {
-                readUserComment(input);
+                readUserComment();
             }
-            onEndOfComment();
         } catch (IOException e) {
-            onError(new VorbisCommentReaderException(e));
+            e.printStackTrace();
         }
     }
 
-    private void findOggPage(InputStream input) throws IOException {
+    private void findOggPage() throws IOException {
         // find OggS
         byte[] buffer = new byte[4];
         final byte[] oggPageHeader = {'O', 'g', 'g', 'S'};
@@ -76,17 +61,19 @@ public abstract class VorbisCommentReader {
         IOUtils.skipFully(input, numSegments);
     }
 
-    private void readUserComment(InputStream input) throws VorbisCommentReaderException {
+    private void readUserComment() throws VorbisCommentReaderException {
         try {
             long vectorLength = EndianUtils.readSwappedUnsignedInteger(input);
             if (vectorLength > 20 * 1024 * 1024) {
-                // Avoid reading entire file if it is encoded incorrectly
-                throw new VorbisCommentReaderException("User comment unrealistically long: " + vectorLength);
+                String keyPart = readUtf8String(10);
+                throw new VorbisCommentReaderException("User comment unrealistically long. "
+                        + "key=" + keyPart + ", length=" + vectorLength);
             }
-            String key = readContentVectorKey(input, vectorLength).toLowerCase(Locale.US);
-            boolean readValue = onContentVectorKey(key);
-            if (readValue) {
-                String value = readUtf8String(input, vectorLength - key.length() - 1);
+            String key = readContentVectorKey(vectorLength).toLowerCase(Locale.US);
+            boolean shouldReadValue = handles(key);
+            Log.d(TAG, "key=" + key + ", length=" + vectorLength + ", handles=" + shouldReadValue);
+            if (shouldReadValue) {
+                String value = readUtf8String(vectorLength - key.length() - 1);
                 onContentVectorValue(key, value);
             } else {
                 IOUtils.skipFully(input, vectorLength - key.length() - 1);
@@ -96,7 +83,7 @@ public abstract class VorbisCommentReader {
         }
     }
 
-    private String readUtf8String(InputStream input, long length) throws IOException {
+    private String readUtf8String(long length) throws IOException {
         byte[] buffer = new byte[(int) length];
         IOUtils.readFully(input, buffer);
         Charset charset = Charset.forName("UTF-8");
@@ -107,7 +94,7 @@ public abstract class VorbisCommentReader {
      * Looks for an identification header in the first page of the file. If an
      * identification header is found, it will be skipped completely
      */
-    private void findIdentificationHeader(InputStream input) throws IOException {
+    private void findIdentificationHeader() throws IOException {
         byte[] buffer = new byte[FIRST_OPUS_PAGE_LENGTH];
         IOUtils.readFully(input, buffer);
         final byte[] oggIdentificationHeader = new byte[]{ PACKET_TYPE_IDENTIFICATION, 'v', 'o', 'r', 'b', 'i', 's' };
@@ -122,7 +109,7 @@ public abstract class VorbisCommentReader {
         throw new IOException("No vorbis identification header found");
     }
 
-    private void findCommentHeader(InputStream input) throws IOException {
+    private void findCommentHeader() throws IOException {
         byte[] buffer = new byte[64]; // Enough space for some bytes. Used circularly.
         final byte[] oggCommentHeader = new byte[]{ PACKET_TYPE_COMMENT, 'v', 'o', 'r', 'b', 'i', 's' };
         for (int bytesRead = 0; bytesRead < SECOND_PAGE_MAX_LENGTH; bytesRead++) {
@@ -155,10 +142,10 @@ public abstract class VorbisCommentReader {
     }
 
     @NonNull
-    private VorbisCommentHeader readCommentHeader(InputStream input) throws IOException, VorbisCommentReaderException {
+    private VorbisCommentHeader readCommentHeader() throws IOException, VorbisCommentReaderException {
         try {
             long vendorLength = EndianUtils.readSwappedUnsignedInteger(input);
-            String vendorName = readUtf8String(input, vendorLength);
+            String vendorName = readUtf8String(vendorLength);
             long userCommentLength = EndianUtils.readSwappedUnsignedInteger(input);
             return new VorbisCommentHeader(vendorName, userCommentLength);
         } catch (UnsupportedEncodingException e) {
@@ -166,7 +153,7 @@ public abstract class VorbisCommentReader {
         }
     }
 
-    private String readContentVectorKey(InputStream input, long vectorLength) throws IOException {
+    private String readContentVectorKey(long vectorLength) throws IOException {
         StringBuilder builder = new StringBuilder();
         for (int i = 0; i < vectorLength; i++) {
             char c = (char) input.read();
@@ -178,4 +165,15 @@ public abstract class VorbisCommentReader {
         }
         return null; // no key found
     }
+
+    /**
+     * Is called every time the Reader finds a content vector. The handler
+     * should return true if it wants to handle the content vector.
+     */
+    protected abstract boolean handles(String key);
+
+    /**
+     * Is called if onContentVectorKey returned true for the key.
+     */
+    protected abstract void onContentVectorValue(String key, String value) throws VorbisCommentReaderException;
 }
diff --git a/parser/media/src/test/java/de/danoeh/antennapod/parser/media/id3/MetadataReaderTest.java b/parser/media/src/test/java/de/danoeh/antennapod/parser/media/id3/MetadataReaderTest.java
new file mode 100644
index 000000000..5b6b06ea7
--- /dev/null
+++ b/parser/media/src/test/java/de/danoeh/antennapod/parser/media/id3/MetadataReaderTest.java
@@ -0,0 +1,50 @@
+package de.danoeh.antennapod.parser.media.id3;
+
+import org.apache.commons.io.input.CountingInputStream;
+import org.junit.Test;
+import org.junit.runner.RunWith;
+import org.robolectric.RobolectricTestRunner;
+
+import java.io.IOException;
+
+import static org.junit.Assert.assertEquals;
+
+@RunWith(RobolectricTestRunner.class)
+public class MetadataReaderTest {
+    @Test
+    public void testRealFileUltraschall() throws IOException, ID3ReaderException {
+        CountingInputStream inputStream = new CountingInputStream(getClass().getClassLoader()
+                .getResource("ultraschall5.mp3").openStream());
+        Id3MetadataReader reader = new Id3MetadataReader(inputStream);
+        reader.readInputStream();
+        assertEquals("Description", reader.getComment());
+    }
+
+    @Test
+    public void testRealFileAuphonic() throws IOException, ID3ReaderException {
+        CountingInputStream inputStream = new CountingInputStream(getClass().getClassLoader()
+                .getResource("auphonic.mp3").openStream());
+        Id3MetadataReader reader = new Id3MetadataReader(inputStream);
+        reader.readInputStream();
+        assertEquals("Summary", reader.getComment());
+    }
+
+    @Test
+    public void testRealFileHindenburgJournalistPro() throws IOException, ID3ReaderException {
+        CountingInputStream inputStream = new CountingInputStream(getClass().getClassLoader()
+                .getResource("hindenburg-journalist-pro.mp3").openStream());
+        Id3MetadataReader reader = new Id3MetadataReader(inputStream);
+        reader.readInputStream();
+        assertEquals("This is the summary of this podcast episode. This file was made with"
+                + " Hindenburg Journalist Pro version 1.85, build number 2360.", reader.getComment());
+    }
+
+    @Test
+    public void testRealFileMp3chapsPy() throws IOException, ID3ReaderException {
+        CountingInputStream inputStream = new CountingInputStream(getClass().getClassLoader()
+                .getResource("mp3chaps-py.mp3").openStream());
+        Id3MetadataReader reader = new Id3MetadataReader(inputStream);
+        reader.readInputStream();
+        assertEquals("2021.08.13", reader.getComment());
+    }
+}
diff --git a/parser/media/src/test/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentChapterReaderTest.java b/parser/media/src/test/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentChapterReaderTest.java
index 6ebe875d9..b19be324d 100644
--- a/parser/media/src/test/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentChapterReaderTest.java
+++ b/parser/media/src/test/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentChapterReaderTest.java
@@ -23,8 +23,8 @@ public class VorbisCommentChapterReaderTest {
     public void testRealFileAuphonic(String filename) throws IOException, VorbisCommentReaderException {
         InputStream inputStream = getClass().getClassLoader()
                 .getResource(filename).openStream();
-        VorbisCommentChapterReader reader = new VorbisCommentChapterReader();
-        reader.readInputStream(inputStream);
+        VorbisCommentChapterReader reader = new VorbisCommentChapterReader(inputStream);
+        reader.readInputStream();
         List<Chapter> chapters = reader.getChapters();
 
         assertEquals(4, chapters.size());
diff --git a/parser/media/src/test/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentMetadataReaderTest.java b/parser/media/src/test/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentMetadataReaderTest.java
new file mode 100644
index 000000000..7f71c8fa9
--- /dev/null
+++ b/parser/media/src/test/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentMetadataReaderTest.java
@@ -0,0 +1,28 @@
+package de.danoeh.antennapod.parser.media.vorbis;
+
+import org.junit.Test;
+import org.junit.runner.RunWith;
+import org.robolectric.RobolectricTestRunner;
+
+import java.io.IOException;
+import java.io.InputStream;
+
+import static org.junit.Assert.assertEquals;
+
+@RunWith(RobolectricTestRunner.class)
+public class VorbisCommentMetadataReaderTest {
+
+    @Test
+    public void testRealFilesAuphonic() throws IOException, VorbisCommentReaderException {
+        testRealFileAuphonic("auphonic.ogg");
+        testRealFileAuphonic("auphonic.opus");
+    }
+
+    public void testRealFileAuphonic(String filename) throws IOException, VorbisCommentReaderException {
+        InputStream inputStream = getClass().getClassLoader()
+                .getResource(filename).openStream();
+        VorbisCommentMetadataReader reader = new VorbisCommentMetadataReader(inputStream);
+        reader.readInputStream();
+        assertEquals("Summary", reader.getDescription());
+    }
+}