Examples with AvcConfig androidx.media3.extractor.AvcConfig used on opensource projects

Example 1 with AvcConfig

use of androidx.media3.extractor.AvcConfig in project media by androidx.

the class AtomParsers method parseVideoSampleEntry.

// hdrStaticInfo is allocated using allocate() in allocateHdrStaticInfo().
@SuppressWarnings("ByteBufferBackingArray")
private static void parseVideoSampleEntry(ParsableByteArray parent, int atomType, int position, int size, int trackId, int rotationDegrees, @Nullable DrmInitData drmInitData, StsdData out, int entryIndex) throws ParserException {
    parent.setPosition(position + Atom.HEADER_SIZE + StsdData.STSD_HEADER_SIZE);
    parent.skipBytes(16);
    int width = parent.readUnsignedShort();
    int height = parent.readUnsignedShort();
    boolean pixelWidthHeightRatioFromPasp = false;
    float pixelWidthHeightRatio = 1;
    parent.skipBytes(50);
    int childPosition = parent.getPosition();
    if (atomType == Atom.TYPE_encv) {
        @Nullable Pair<Integer, TrackEncryptionBox> sampleEntryEncryptionData = parseSampleEntryEncryptionData(parent, position, size);
        if (sampleEntryEncryptionData != null) {
            atomType = sampleEntryEncryptionData.first;
            drmInitData = drmInitData == null ? null : drmInitData.copyWithSchemeType(sampleEntryEncryptionData.second.schemeType);
            out.trackEncryptionBoxes[entryIndex] = sampleEntryEncryptionData.second;
        }
        parent.setPosition(childPosition);
    }
    // TODO: Uncomment when [Internal: b/63092960] is fixed.
    // else {
    // drmInitData = null;
    // }
    @Nullable String mimeType = null;
    if (atomType == Atom.TYPE_m1v_) {
        mimeType = MimeTypes.VIDEO_MPEG;
    } else if (atomType == Atom.TYPE_H263) {
        mimeType = MimeTypes.VIDEO_H263;
    }
    @Nullable List<byte[]> initializationData = null;
    @Nullable String codecs = null;
    @Nullable byte[] projectionData = null;
    @C.StereoMode int stereoMode = Format.NO_VALUE;
    // HDR related metadata.
    @C.ColorSpace int colorSpace = Format.NO_VALUE;
    @C.ColorRange int colorRange = Format.NO_VALUE;
    @C.ColorTransfer int colorTransfer = Format.NO_VALUE;
    // The format of HDR static info is defined in CTA-861-G:2017, Table 45.
    @Nullable ByteBuffer hdrStaticInfo = null;
    while (childPosition - position < size) {
        parent.setPosition(childPosition);
        int childStartPosition = parent.getPosition();
        int childAtomSize = parent.readInt();
        if (childAtomSize == 0 && parent.getPosition() - position == size) {
            // Handle optional terminating four zero bytes in MOV files.
            break;
        }
        ExtractorUtil.checkContainerInput(childAtomSize > 0, "childAtomSize must be positive");
        int childAtomType = parent.readInt();
        if (childAtomType == Atom.TYPE_avcC) {
            ExtractorUtil.checkContainerInput(mimeType == null, /* message= */
            null);
            mimeType = MimeTypes.VIDEO_H264;
            parent.setPosition(childStartPosition + Atom.HEADER_SIZE);
            AvcConfig avcConfig = AvcConfig.parse(parent);
            initializationData = avcConfig.initializationData;
            out.nalUnitLengthFieldLength = avcConfig.nalUnitLengthFieldLength;
            if (!pixelWidthHeightRatioFromPasp) {
                pixelWidthHeightRatio = avcConfig.pixelWidthHeightRatio;
            }
            codecs = avcConfig.codecs;
        } else if (childAtomType == Atom.TYPE_hvcC) {
            ExtractorUtil.checkContainerInput(mimeType == null, /* message= */
            null);
            mimeType = MimeTypes.VIDEO_H265;
            parent.setPosition(childStartPosition + Atom.HEADER_SIZE);
            HevcConfig hevcConfig = HevcConfig.parse(parent);
            initializationData = hevcConfig.initializationData;
            out.nalUnitLengthFieldLength = hevcConfig.nalUnitLengthFieldLength;
            if (!pixelWidthHeightRatioFromPasp) {
                pixelWidthHeightRatio = hevcConfig.pixelWidthHeightRatio;
            }
            codecs = hevcConfig.codecs;
        } else if (childAtomType == Atom.TYPE_dvcC || childAtomType == Atom.TYPE_dvvC) {
            @Nullable DolbyVisionConfig dolbyVisionConfig = DolbyVisionConfig.parse(parent);
            if (dolbyVisionConfig != null) {
                codecs = dolbyVisionConfig.codecs;
                mimeType = MimeTypes.VIDEO_DOLBY_VISION;
            }
        } else if (childAtomType == Atom.TYPE_vpcC) {
            ExtractorUtil.checkContainerInput(mimeType == null, /* message= */
            null);
            mimeType = (atomType == Atom.TYPE_vp08) ? MimeTypes.VIDEO_VP8 : MimeTypes.VIDEO_VP9;
        } else if (childAtomType == Atom.TYPE_av1C) {
            ExtractorUtil.checkContainerInput(mimeType == null, /* message= */
            null);
            mimeType = MimeTypes.VIDEO_AV1;
        } else if (childAtomType == Atom.TYPE_clli) {
            if (hdrStaticInfo == null) {
                hdrStaticInfo = allocateHdrStaticInfo();
            }
            // The contents of the clli box occupy the last 4 bytes of the HDR static info array. Note
            // that each field is read in big endian and written in little endian.
            hdrStaticInfo.position(21);
            // max_content_light_level.
            hdrStaticInfo.putShort(parent.readShort());
            // max_pic_average_light_level.
            hdrStaticInfo.putShort(parent.readShort());
        } else if (childAtomType == Atom.TYPE_mdcv) {
            if (hdrStaticInfo == null) {
                hdrStaticInfo = allocateHdrStaticInfo();
            }
            // The contents of the mdcv box occupy 20 bytes after the first byte of the HDR static info
            // array. Note that each field is read in big endian and written in little endian.
            short displayPrimariesGX = parent.readShort();
            short displayPrimariesGY = parent.readShort();
            short displayPrimariesBX = parent.readShort();
            short displayPrimariesBY = parent.readShort();
            short displayPrimariesRX = parent.readShort();
            short displayPrimariesRY = parent.readShort();
            short whitePointX = parent.readShort();
            short whitePointY = parent.readShort();
            long maxDisplayMasteringLuminance = parent.readUnsignedInt();
            long minDisplayMasteringLuminance = parent.readUnsignedInt();
            hdrStaticInfo.position(1);
            hdrStaticInfo.putShort(displayPrimariesRX);
            hdrStaticInfo.putShort(displayPrimariesRY);
            hdrStaticInfo.putShort(displayPrimariesGX);
            hdrStaticInfo.putShort(displayPrimariesGY);
            hdrStaticInfo.putShort(displayPrimariesBX);
            hdrStaticInfo.putShort(displayPrimariesBY);
            hdrStaticInfo.putShort(whitePointX);
            hdrStaticInfo.putShort(whitePointY);
            hdrStaticInfo.putShort((short) (maxDisplayMasteringLuminance / 10000));
            hdrStaticInfo.putShort((short) (minDisplayMasteringLuminance / 10000));
        } else if (childAtomType == Atom.TYPE_d263) {
            ExtractorUtil.checkContainerInput(mimeType == null, /* message= */
            null);
            mimeType = MimeTypes.VIDEO_H263;
        } else if (childAtomType == Atom.TYPE_esds) {
            ExtractorUtil.checkContainerInput(mimeType == null, /* message= */
            null);
            Pair<@NullableType String, byte @NullableType []> mimeTypeAndInitializationDataBytes = parseEsdsFromParent(parent, childStartPosition);
            mimeType = mimeTypeAndInitializationDataBytes.first;
            @Nullable byte[] initializationDataBytes = mimeTypeAndInitializationDataBytes.second;
            if (initializationDataBytes != null) {
                initializationData = ImmutableList.of(initializationDataBytes);
            }
        } else if (childAtomType == Atom.TYPE_pasp) {
            pixelWidthHeightRatio = parsePaspFromParent(parent, childStartPosition);
            pixelWidthHeightRatioFromPasp = true;
        } else if (childAtomType == Atom.TYPE_sv3d) {
            projectionData = parseProjFromParent(parent, childStartPosition, childAtomSize);
        } else if (childAtomType == Atom.TYPE_st3d) {
            int version = parent.readUnsignedByte();
            // Flags.
            parent.skipBytes(3);
            if (version == 0) {
                int layout = parent.readUnsignedByte();
                switch(layout) {
                    case 0:
                        stereoMode = C.STEREO_MODE_MONO;
                        break;
                    case 1:
                        stereoMode = C.STEREO_MODE_TOP_BOTTOM;
                        break;
                    case 2:
                        stereoMode = C.STEREO_MODE_LEFT_RIGHT;
                        break;
                    case 3:
                        stereoMode = C.STEREO_MODE_STEREO_MESH;
                        break;
                    default:
                        break;
                }
            }
        } else if (childAtomType == Atom.TYPE_colr) {
            int colorType = parent.readInt();
            if (colorType == TYPE_nclx || colorType == TYPE_nclc) {
                // For more info on syntax, see Section 8.5.2.2 in ISO/IEC 14496-12:2012(E) and
                // https://developer.apple.com/library/archive/documentation/QuickTime/QTFF/QTFFChap3/qtff3.html.
                int colorPrimaries = parent.readUnsignedShort();
                int transferCharacteristics = parent.readUnsignedShort();
                // matrix_coefficients.
                parent.skipBytes(2);
                // Only try and read full_range_flag if the box is long enough. It should be present in
                // all colr boxes with type=nclx (Section 8.5.2.2 in ISO/IEC 14496-12:2012(E)) but some
                // device cameras record videos with type=nclx without this final flag (and therefore
                // size=18): https://github.com/google/ExoPlayer/issues/9332
                boolean fullRangeFlag = childAtomSize == 19 && (parent.readUnsignedByte() & 0b10000000) != 0;
                colorSpace = ColorInfo.isoColorPrimariesToColorSpace(colorPrimaries);
                colorRange = fullRangeFlag ? C.COLOR_RANGE_FULL : C.COLOR_RANGE_LIMITED;
                colorTransfer = ColorInfo.isoTransferCharacteristicsToColorTransfer(transferCharacteristics);
            } else {
                Log.w(TAG, "Unsupported color type: " + Atom.getAtomTypeString(colorType));
            }
        }
        childPosition += childAtomSize;
    }
    // If the media type was not recognized, ignore the track.
    if (mimeType == null) {
        return;
    }
    Format.Builder formatBuilder = new Format.Builder().setId(trackId).setSampleMimeType(mimeType).setCodecs(codecs).setWidth(width).setHeight(height).setPixelWidthHeightRatio(pixelWidthHeightRatio).setRotationDegrees(rotationDegrees).setProjectionData(projectionData).setStereoMode(stereoMode).setInitializationData(initializationData).setDrmInitData(drmInitData);
    if (colorSpace != Format.NO_VALUE || colorRange != Format.NO_VALUE || colorTransfer != Format.NO_VALUE || hdrStaticInfo != null) {
        // Note that if either mdcv or clli are missing, we leave the corresponding HDR static
        // metadata bytes with value zero. See [Internal ref: b/194535665].
        formatBuilder.setColorInfo(new ColorInfo(colorSpace, colorRange, colorTransfer, hdrStaticInfo != null ? hdrStaticInfo.array() : null));
    }
    out.format = formatBuilder.build();
}

Example 2 with AvcConfig

use of androidx.media3.extractor.AvcConfig in project media by androidx.

the class VideoTagPayloadReader method parsePayload.

@Override
protected boolean parsePayload(ParsableByteArray data, long timeUs) throws ParserException {
    int packetType = data.readUnsignedByte();
    int compositionTimeMs = data.readInt24();
    timeUs += compositionTimeMs * 1000L;
    // Parse avc sequence header in case this was not done before.
    if (packetType == AVC_PACKET_TYPE_SEQUENCE_HEADER && !hasOutputFormat) {
        ParsableByteArray videoSequence = new ParsableByteArray(new byte[data.bytesLeft()]);
        data.readBytes(videoSequence.getData(), 0, data.bytesLeft());
        AvcConfig avcConfig = AvcConfig.parse(videoSequence);
        nalUnitLengthFieldLength = avcConfig.nalUnitLengthFieldLength;
        // Construct and output the format.
        Format format = new Format.Builder().setSampleMimeType(MimeTypes.VIDEO_H264).setCodecs(avcConfig.codecs).setWidth(avcConfig.width).setHeight(avcConfig.height).setPixelWidthHeightRatio(avcConfig.pixelWidthHeightRatio).setInitializationData(avcConfig.initializationData).build();
        output.format(format);
        hasOutputFormat = true;
        return false;
    } else if (packetType == AVC_PACKET_TYPE_AVC_NALU && hasOutputFormat) {
        boolean isKeyframe = frameType == VIDEO_FRAME_KEYFRAME;
        if (!hasOutputKeyframe && !isKeyframe) {
            return false;
        }
        // TODO: Deduplicate with Mp4Extractor.
        // Zero the top three bytes of the array that we'll use to decode nal unit lengths, in case
        // they're only 1 or 2 bytes long.
        byte[] nalLengthData = nalLength.getData();
        nalLengthData[0] = 0;
        nalLengthData[1] = 0;
        nalLengthData[2] = 0;
        int nalUnitLengthFieldLengthDiff = 4 - nalUnitLengthFieldLength;
        // NAL units are length delimited, but the decoder requires start code delimited units.
        // Loop until we've written the sample to the track output, replacing length delimiters with
        // start codes as we encounter them.
        int bytesWritten = 0;
        int bytesToWrite;
        while (data.bytesLeft() > 0) {
            // Read the NAL length so that we know where we find the next one.
            data.readBytes(nalLength.getData(), nalUnitLengthFieldLengthDiff, nalUnitLengthFieldLength);
            nalLength.setPosition(0);
            bytesToWrite = nalLength.readUnsignedIntToInt();
            // Write a start code for the current NAL unit.
            nalStartCode.setPosition(0);
            output.sampleData(nalStartCode, 4);
            bytesWritten += 4;
            // Write the payload of the NAL unit.
            output.sampleData(data, bytesToWrite);
            bytesWritten += bytesToWrite;
        }
        output.sampleMetadata(timeUs, isKeyframe ? C.BUFFER_FLAG_KEY_FRAME : 0, bytesWritten, 0, null);
        hasOutputKeyframe = true;
        return true;
    } else {
        return false;
    }
}

Example 3 with AvcConfig

use of androidx.media3.extractor.AvcConfig in project media by androidx.

the class AvcConfig method parse.

/**
 * Parses AVC configuration data.
 *
 * @param data A {@link ParsableByteArray}, whose position is set to the start of the AVC
 *     configuration data to parse.
 * @return A parsed representation of the AVC configuration data.
 * @throws ParserException If an error occurred parsing the data.
 */
public static AvcConfig parse(ParsableByteArray data) throws ParserException {
    try {
        // Skip to the AVCDecoderConfigurationRecord (defined in 14496-15)
        data.skipBytes(4);
        int nalUnitLengthFieldLength = (data.readUnsignedByte() & 0x3) + 1;
        if (nalUnitLengthFieldLength == 3) {
            throw new IllegalStateException();
        }
        List<byte[]> initializationData = new ArrayList<>();
        int numSequenceParameterSets = data.readUnsignedByte() & 0x1F;
        for (int j = 0; j < numSequenceParameterSets; j++) {
            initializationData.add(buildNalUnitForChild(data));
        }
        int numPictureParameterSets = data.readUnsignedByte();
        for (int j = 0; j < numPictureParameterSets; j++) {
            initializationData.add(buildNalUnitForChild(data));
        }
        int width = Format.NO_VALUE;
        int height = Format.NO_VALUE;
        float pixelWidthHeightRatio = 1;
        @Nullable String codecs = null;
        if (numSequenceParameterSets > 0) {
            byte[] sps = initializationData.get(0);
            SpsData spsData = NalUnitUtil.parseSpsNalUnit(initializationData.get(0), nalUnitLengthFieldLength, sps.length);
            width = spsData.width;
            height = spsData.height;
            pixelWidthHeightRatio = spsData.pixelWidthHeightRatio;
            codecs = CodecSpecificDataUtil.buildAvcCodecString(spsData.profileIdc, spsData.constraintsFlagsAndReservedZero2Bits, spsData.levelIdc);
        }
        return new AvcConfig(initializationData, nalUnitLengthFieldLength, width, height, pixelWidthHeightRatio, codecs);
    } catch (ArrayIndexOutOfBoundsException e) {
        throw ParserException.createForMalformedContainer("Error parsing AVC config", e);
    }
}