Add Sink class (#2111)

Summary:
Add Sink class that bundles FilterGraph and Buffer. Part of https://github.com/pytorch/audio/issues/1986. Splitting the PR for easier review.

For the overall architecture, see https://github.com/mthrok/audio/blob/ffmpeg/torchaudio/csrc/ffmpeg/README.md.

Note: Without a change to build process, the code added here won't be compiled. The build process will be updated later.

Pull Request resolved: https://github.com/pytorch/audio/pull/2111

Reviewed By: carolineechen

Differential Revision: D33350388

Pulled By: mthrok

fbshipit-source-id: 8f42c5fe4be39ef2432c51fc0d0ac72ba3f06a26

torchaudio/csrc/ffmpeg/sink.cpp

+#include <torchaudio/csrc/ffmpeg/sink.h>
+
+namespace torchaudio {
+namespace ffmpeg {
+
+namespace {
+std::unique_ptr<Buffer> get_buffer(
+    AVMediaType type,
+    int frames_per_chunk,
+    int num_chunks) {
+  switch (type) {
+    case AVMEDIA_TYPE_AUDIO:
+      return std::unique_ptr<Buffer>(
+          new AudioBuffer(frames_per_chunk, num_chunks));
+    case AVMEDIA_TYPE_VIDEO:
+      return std::unique_ptr<Buffer>(
+          new VideoBuffer(frames_per_chunk, num_chunks));
+    default:
+      throw std::runtime_error(
+          std::string("Unsupported media type: ") +
+          av_get_media_type_string(type));
+  }
+}
+} // namespace
+
+Sink::Sink(
+    AVRational input_time_base,
+    AVCodecParameters* codecpar,
+    int frames_per_chunk,
+    int num_chunks,
+    double output_time_base,
+    std::string filter_description)
+    : filter(input_time_base, codecpar, filter_description),
+      buffer(get_buffer(codecpar->codec_type, frames_per_chunk, num_chunks)),
+      time_base(output_time_base) {}
+
+// 0: some kind of success
+// <0: Some error happened
+int Sink::process_frame(AVFrame* pFrame) {
+  int ret = filter.add_frame(pFrame);
+  while (ret >= 0) {
+    ret = filter.get_frame(frame);
+    //  AVERROR(EAGAIN) means that new input data is required to return new
+    //  output.
+    if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF)
+      return 0;
+    if (ret >= 0)
+      buffer->push_frame(frame);
+    av_frame_unref(frame);
+  }
+  return ret;
+}
+
+bool Sink::is_buffer_ready() const {
+  return buffer->is_ready();
+}
+} // namespace ffmpeg
+} // namespace torchaudio

torchaudio/csrc/ffmpeg/sink.h

+#pragma once
+
+#include <torchaudio/csrc/ffmpeg/buffer.h>
+#include <torchaudio/csrc/ffmpeg/ffmpeg.h>
+#include <torchaudio/csrc/ffmpeg/filter_graph.h>
+
+namespace torchaudio {
+namespace ffmpeg {
+
+class Sink {
+  AVFramePtr frame;
+
+ public:
+  FilterGraph filter;
+  std::unique_ptr<Buffer> buffer;
+  double time_base;
+  Sink(
+      AVRational input_time_base,
+      AVCodecParameters* codecpar,
+      int frames_per_chunk,
+      int num_chunks,
+      double output_time_base,
+      std::string filter_description);
+
+  int process_frame(AVFrame* frame);
+  bool is_buffer_ready() const;
+};
+
+} // namespace ffmpeg
+} // namespace torchaudio