Support parsing vorbis comments spanning across pages (#7215)

parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentReader.java

@@ -20,16 +20,14 @@ public abstract class VorbisCommentReader {
     private static final int PACKET_TYPE_IDENTIFICATION = 1;
     private static final int PACKET_TYPE_COMMENT = 3;
 
-    private final InputStream input;
+    private final VorbisInputStream input;
 
     VorbisCommentReader(InputStream input) {
-        this.input = input;
+        this.input = new VorbisInputStream(input);
     }
 
     public void readInputStream() throws VorbisCommentReaderException {
         try {
-            findIdentificationHeader();
-            findOggPage();
             findCommentHeader();
             VorbisCommentHeader commentHeader = readCommentHeader();
             Log.d(TAG, commentHeader.toString());
@@ -41,26 +39,6 @@ public abstract class VorbisCommentReader {
         }
     }
 
-    private void findOggPage() throws IOException {
-        // find OggS
-        byte[] buffer = new byte[4];
-        final byte[] oggPageHeader = {'O', 'g', 'g', 'S'};
-        for (int bytesRead = 0; bytesRead < SECOND_PAGE_MAX_LENGTH; bytesRead++) {
-            int data = input.read();
-            if (data == -1) {
-                throw new IOException("EOF while trying to find vorbis page");
-            }
-            buffer[bytesRead % buffer.length] = (byte) data;
-            if (bufferMatches(buffer, oggPageHeader, bytesRead)) {
-                break;
-            }
-        }
-        // read segments
-        IOUtils.skipFully(input, 22);
-        int numSegments = input.read();
-        IOUtils.skipFully(input, numSegments);
-    }
-
     private void readUserComment() throws VorbisCommentReaderException {
         try {
             long vectorLength = EndianUtils.readSwappedUnsignedInteger(input);
@@ -90,25 +68,6 @@ public abstract class VorbisCommentReader {
         return charset.newDecoder().decode(ByteBuffer.wrap(buffer)).toString();
     }
 
-    /**
-     * 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() 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' };
-        for (int i = 6; i < buffer.length; i++) {
-            if (bufferMatches(buffer, oggIdentificationHeader, i)) {
-                IOUtils.skip(input, FIRST_OGG_PAGE_LENGTH - FIRST_OPUS_PAGE_LENGTH);
-                return;
-            } else if (bufferMatches(buffer, "OpusHead".getBytes(), i)) {
-                return;
-            }
-        }
-        throw new IOException("No vorbis identification header found");
-    }
-
     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' };

parser/media/src/main/java/de/danoeh/antennapod/parser/media/vorbis/VorbisInputStream.java

@@ -0,0 +1,69 @@
+package de.danoeh.antennapod.parser.media.vorbis;
+
+import org.apache.commons.io.IOUtils;
+
+import java.io.BufferedInputStream;
+import java.io.FilterInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.Arrays;
+
+
+public class VorbisInputStream extends FilterInputStream {
+    private static final byte[] CAPTURE_PATTERN = {'O', 'g', 'g', 'S'};
+    private static final int HEADER_SKIP_LENGTH = 1 + 1 + 8 + 4 + 4 + 4;
+
+    private final BufferedInputStream inputStream;
+    private int pageRemainBytes = 0;
+
+    protected VorbisInputStream(InputStream in) {
+        super(in);
+        inputStream = new BufferedInputStream(in);
+    }
+
+    private int parsePageHeader(InputStream in) throws IOException {
+        byte[] capturePattern = new byte[4];
+
+        IOUtils.readFully(in, capturePattern, 0, 4);
+        if (!Arrays.equals(CAPTURE_PATTERN, capturePattern)) {
+            throw new IOException("Invalid page header");
+        }
+
+        IOUtils.skipFully(in, HEADER_SKIP_LENGTH);
+
+        int pageSegments = in.read();
+        byte[] segmentTable = new byte[pageSegments];
+        int pageLength = 0;
+        IOUtils.readFully(in, segmentTable);
+        for (byte segment:segmentTable) {
+            pageLength += (segment & 0xff);
+        }
+
+        return pageLength;
+    }
+
+    /** check and update remaining bytes **/
+    private void updateRemainBytes() throws IOException {
+        if (pageRemainBytes == 0) {
+            pageRemainBytes = parsePageHeader(inputStream);
+        } else if (pageRemainBytes < 0) {
+            throw new IOException("Page remain bytes less than 0");
+        }
+    }
+
+    @Override
+    public int read() throws IOException {
+        updateRemainBytes();
+        pageRemainBytes--;
+        return inputStream.read();
+    }
+
+    @Override
+    public int read(byte[] b, int off, int len) throws IOException {
+        updateRemainBytes();
+        int bytesToRead = Math.min(len, pageRemainBytes);
+        IOUtils.readFully(inputStream, b, off, bytesToRead);
+        this.pageRemainBytes -= bytesToRead;
+        return bytesToRead;
+    }
+}

parser/media/src/test/java/de/danoeh/antennapod/parser/media/vorbis/VorbisCommentMetadataReaderTest.java

@@ -16,6 +16,7 @@ public class VorbisCommentMetadataReaderTest {
     public void testRealFilesAuphonic() throws IOException, VorbisCommentReaderException {
         testRealFileAuphonic("auphonic.ogg");
         testRealFileAuphonic("auphonic.opus");
+        testRealFileAuphonic("opus-comment.opus");
     }
 
     public void testRealFileAuphonic(String filename) throws IOException, VorbisCommentReaderException {