webrtc/modules/video_coding/codecs/h264/h264_video_toolbox_decoder.cc

/*
 *  Copyright (c) 2015 The WebRTC project authors. All Rights Reserved.
 *
 *  Use of this source code is governed by a BSD-style license
 *  that can be found in the LICENSE file in the root of the source
 *  tree. An additional intellectual property rights grant can be found
 *  in the file PATENTS.  All contributing project authors may
 *  be found in the AUTHORS file in the root of the source tree.
 *
 */

#include "webrtc/modules/video_coding/codecs/h264/h264_video_toolbox_decoder.h"

#if defined(WEBRTC_VIDEO_TOOLBOX_SUPPORTED)

#include "libyuv/convert.h"
#include "webrtc/base/checks.h"
#include "webrtc/base/logging.h"
#include "webrtc/common_video/interface/video_frame_buffer.h"
#include "webrtc/modules/video_coding/codecs/h264/h264_video_toolbox_nalu.h"
#include "webrtc/video_frame.h"

namespace internal {

// Convenience function for creating a dictionary.
inline CFDictionaryRef CreateCFDictionary(CFTypeRef* keys,
                                          CFTypeRef* values,
                                          size_t size) {
  return CFDictionaryCreate(nullptr, keys, values, size,
                            &kCFTypeDictionaryKeyCallBacks,
                            &kCFTypeDictionaryValueCallBacks);
}

// Struct that we pass to the decoder per frame to decode. We receive it again
// in the decoder callback.
struct FrameDecodeParams {
  FrameDecodeParams(webrtc::DecodedImageCallback* cb, int64_t ts)
      : callback(cb), timestamp(ts) {}
  webrtc::DecodedImageCallback* callback;
  int64_t timestamp;
};

// On decode we receive a CVPixelBuffer, which we need to convert to a frame
// buffer for use in the rest of WebRTC. Unfortunately this involves a frame
// copy.
// TODO(tkchin): Stuff CVPixelBuffer into a TextureBuffer and pass that along
// instead once the pipeline supports it.
rtc::scoped_refptr<webrtc::VideoFrameBuffer> VideoFrameBufferForPixelBuffer(
    CVPixelBufferRef pixel_buffer) {
  DCHECK(pixel_buffer);
  DCHECK(CVPixelBufferGetPixelFormatType(pixel_buffer) ==
         kCVPixelFormatType_420YpCbCr8BiPlanarFullRange);
  size_t width = CVPixelBufferGetWidthOfPlane(pixel_buffer, 0);
  size_t height = CVPixelBufferGetHeightOfPlane(pixel_buffer, 0);
  // TODO(tkchin): Use a frame buffer pool.
  rtc::scoped_refptr<webrtc::VideoFrameBuffer> buffer =
      new rtc::RefCountedObject<webrtc::I420Buffer>(width, height);
  CVPixelBufferLockBaseAddress(pixel_buffer, kCVPixelBufferLock_ReadOnly);
  const uint8* src_y = reinterpret_cast<const uint8*>(
      CVPixelBufferGetBaseAddressOfPlane(pixel_buffer, 0));
  int src_y_stride = CVPixelBufferGetBytesPerRowOfPlane(pixel_buffer, 0);
  const uint8* src_uv = reinterpret_cast<const uint8*>(
      CVPixelBufferGetBaseAddressOfPlane(pixel_buffer, 1));
  int src_uv_stride = CVPixelBufferGetBytesPerRowOfPlane(pixel_buffer, 1);
  int ret = libyuv::NV12ToI420(
      src_y, src_y_stride, src_uv, src_uv_stride,
      buffer->data(webrtc::kYPlane), buffer->stride(webrtc::kYPlane),
      buffer->data(webrtc::kUPlane), buffer->stride(webrtc::kUPlane),
      buffer->data(webrtc::kVPlane), buffer->stride(webrtc::kVPlane),
      width, height);
  CVPixelBufferUnlockBaseAddress(pixel_buffer, kCVPixelBufferLock_ReadOnly);
  if (ret) {
    LOG(LS_ERROR) << "Error converting NV12 to I420: " << ret;
    return nullptr;
  }
  return buffer;
}

// This is the callback function that VideoToolbox calls when decode is
// complete.
void VTDecompressionOutputCallback(void* decoder,
                                   void* params,
                                   OSStatus status,
                                   VTDecodeInfoFlags info_flags,
                                   CVImageBufferRef image_buffer,
                                   CMTime timestamp,
                                   CMTime duration) {
  rtc::scoped_ptr<FrameDecodeParams> decode_params(
      reinterpret_cast<FrameDecodeParams*>(params));
  if (status != noErr) {
    LOG(LS_ERROR) << "Failed to decode frame. Status: " << status;
    return;
  }
  // TODO(tkchin): Handle CVO properly.
  rtc::scoped_refptr<webrtc::VideoFrameBuffer> buffer =
      VideoFrameBufferForPixelBuffer(image_buffer);
  webrtc::VideoFrame decoded_frame(buffer, decode_params->timestamp, 0,
                                   webrtc::kVideoRotation_0);
  decode_params->callback->Decoded(decoded_frame);
}

}  // namespace internal

namespace webrtc {

H264VideoToolboxDecoder::H264VideoToolboxDecoder()
    : callback_(nullptr),
      video_format_(nullptr),
      decompression_session_(nullptr) {
}

H264VideoToolboxDecoder::~H264VideoToolboxDecoder() {
  DestroyDecompressionSession();
  SetVideoFormat(nullptr);
}

int H264VideoToolboxDecoder::InitDecode(const VideoCodec* video_codec,
                                        int number_of_cores) {
  return WEBRTC_VIDEO_CODEC_OK;
}

int H264VideoToolboxDecoder::Decode(
    const EncodedImage& input_image,
    bool missing_frames,
    const RTPFragmentationHeader* fragmentation,
    const CodecSpecificInfo* codec_specific_info,
    int64_t render_time_ms) {
  DCHECK(input_image._buffer);

  CMSampleBufferRef sample_buffer = nullptr;
  if (!H264AnnexBBufferToCMSampleBuffer(input_image._buffer,
                                        input_image._length,
                                        video_format_,
                                        &sample_buffer)) {
    return WEBRTC_VIDEO_CODEC_ERROR;
  }
  DCHECK(sample_buffer);
  // Check if the video format has changed, and reinitialize decoder if needed.
  CMVideoFormatDescriptionRef description =
      CMSampleBufferGetFormatDescription(sample_buffer);
  if (!CMFormatDescriptionEqual(description, video_format_)) {
    SetVideoFormat(description);
    ResetDecompressionSession();
  }
  VTDecodeFrameFlags decode_flags =
      kVTDecodeFrame_EnableAsynchronousDecompression;
  rtc::scoped_ptr<internal::FrameDecodeParams> frame_decode_params;
  frame_decode_params.reset(
      new internal::FrameDecodeParams(callback_, input_image._timeStamp));
  OSStatus status = VTDecompressionSessionDecodeFrame(
      decompression_session_, sample_buffer, decode_flags,
      frame_decode_params.release(), nullptr);
  CFRelease(sample_buffer);
  if (status != noErr) {
    LOG(LS_ERROR) << "Failed to decode frame with code: " << status;
    return WEBRTC_VIDEO_CODEC_ERROR;
  }
  return WEBRTC_VIDEO_CODEC_OK;
}

int H264VideoToolboxDecoder::RegisterDecodeCompleteCallback(
    DecodedImageCallback* callback) {
  DCHECK(!callback_);
  callback_ = callback;
  return WEBRTC_VIDEO_CODEC_OK;
}

int H264VideoToolboxDecoder::Release() {
  callback_ = nullptr;
  return WEBRTC_VIDEO_CODEC_OK;
}

int H264VideoToolboxDecoder::Reset() {
  ResetDecompressionSession();
  return WEBRTC_VIDEO_CODEC_OK;
}

int H264VideoToolboxDecoder::ResetDecompressionSession() {
  DestroyDecompressionSession();

  // Need to wait for the first SPS to initialize decoder.
  if (!video_format_) {
    return WEBRTC_VIDEO_CODEC_OK;
  }

  // Set keys for OpenGL and IOSurface compatibilty, which makes the encoder
  // create pixel buffers with GPU backed memory. The intent here is to pass
  // the pixel buffers directly so we avoid a texture upload later during
  // rendering. This currently is moot because we are converting back to an
  // I420 frame after decode, but eventually we will be able to plumb
  // CVPixelBuffers directly to the renderer.
  // TODO(tkchin): Maybe only set OpenGL/IOSurface keys if we know that that
  // we can pass CVPixelBuffers as native handles in decoder output.
  static size_t const attributes_size = 3;
  CFTypeRef keys[attributes_size] = {
#if defined(WEBRTC_IOS)
    kCVPixelBufferOpenGLESCompatibilityKey,
#elif defined(WEBRTC_MAC)
    kCVPixelBufferOpenGLCompatibilityKey,
#endif
    kCVPixelBufferIOSurfacePropertiesKey,
    kCVPixelBufferPixelFormatTypeKey
  };
  CFDictionaryRef io_surface_value =
      internal::CreateCFDictionary(nullptr, nullptr, 0);
  int64_t nv12type = kCVPixelFormatType_420YpCbCr8BiPlanarFullRange;
  CFNumberRef pixel_format =
      CFNumberCreate(nullptr, kCFNumberLongType, &nv12type);
  CFTypeRef values[attributes_size] = {
    kCFBooleanTrue,
    io_surface_value,
    pixel_format
  };
  CFDictionaryRef attributes =
      internal::CreateCFDictionary(keys, values, attributes_size);
  if (io_surface_value) {
    CFRelease(io_surface_value);
    io_surface_value = nullptr;
  }
  if (pixel_format) {
    CFRelease(pixel_format);
    pixel_format = nullptr;
  }
  VTDecompressionOutputCallbackRecord record = {
      internal::VTDecompressionOutputCallback, this,
  };
  OSStatus status =
      VTDecompressionSessionCreate(nullptr, video_format_, nullptr, attributes,
                                   &record, &decompression_session_);
  CFRelease(attributes);
  if (status != noErr) {
    DestroyDecompressionSession();
    return WEBRTC_VIDEO_CODEC_ERROR;
  }
  ConfigureDecompressionSession();

  return WEBRTC_VIDEO_CODEC_OK;
}

void H264VideoToolboxDecoder::ConfigureDecompressionSession() {
  DCHECK(decompression_session_);
#if defined(WEBRTC_IOS)
  VTSessionSetProperty(decompression_session_,
                       kVTDecompressionPropertyKey_RealTime, kCFBooleanTrue);
#endif
}

void H264VideoToolboxDecoder::DestroyDecompressionSession() {
  if (decompression_session_) {
    VTDecompressionSessionInvalidate(decompression_session_);
    decompression_session_ = nullptr;
  }
}

void H264VideoToolboxDecoder::SetVideoFormat(
    CMVideoFormatDescriptionRef video_format) {
  if (video_format_ == video_format) {
    return;
  }
  if (video_format_) {
    CFRelease(video_format_);
  }
  video_format_ = video_format;
  if (video_format_) {
    CFRetain(video_format_);
  }
}

}  // namespace webrtc

#endif  // defined(WEBRTC_VIDEO_TOOLBOX_SUPPORTED)