Add file-like object support to Streaming API (pytorch#2400)

Summary:
This commit adds file-like object support to Streaming API.

## Features
- File-like objects are expected to implement `read(self, n)`.
- Additionally `seek(self, offset, whence)` is used if available.
- Without `seek` method, some formats cannot be decoded properly.
  - To work around this, one can use the existing `decoder` option to tell what decoder it should use.
  - The set of `decoder` and `decoder_option` arguments were added to `add_basic_[audio|video]_stream` method, similar to `add_[audio|video]_stream`.
  - So as to have the arguments common to both audio and video in from of the rest of the arguments, the order of the arguments are changed.
  - Also `dtype` and `format` arguments were changed to make them consistent across audio/video methods.

## Code structure

The approach is very similar to how file-like object is supported in sox-based I/O.
In Streaming API if the input src is string, it is passed to the implementation bound with TorchBind,
if the src has `read` attribute, it is passed to the same implementation bound via PyBind 11.

![Untitled drawing](https://user-images.githubusercontent.com/855818/169098391-6116afee-7b29-460d-b50d-1037bb8a359d.png)

## Refactoring involved
- Extracted to pytorch#2402
  - Some implementation in the original TorchBind surface layer is converted to Wrapper class so that they can be re-used from PyBind11 bindings. The wrapper class serves to simplify the binding.
  - `add_basic_[audio|video]_stream` methods were removed from C++ layer as it was just constructing string and passing it to `add_[audio|video]_stream` method, which is simpler to do in Python.
  - The original core Streamer implementation kept the use of types in `c10` namespace minimum. All the `c10::optional` and `c10::Dict` were converted to the equivalents of `std` at binding layer. But since they work fine with PyBind11, Streamer core methods deal them directly.
- On Python side, the switch of binding happens in the constructor of `StreamReader` class. Since all the methods have to be delegated to the same set of binding, a backend was introduced, which is abstracted away from user code.

## TODO:
- [x] Check if it is possible to stream MP4 (yuv420p) from S3 and directly decode (with/without HW decoding).

Pull Request resolved: pytorch#2400

Differential Revision: D36520073

Pulled By: mthrok

fbshipit-source-id: 9ceb5a2470abf3b764a12f3abe1355311ccc7eb4

examples/tutorials/streaming_api_tutorial.py

@@ -250,21 +250,24 @@
 #   When the StreamReader buffered this number of chunks and is asked to pull
 #   more frames, StreamReader drops the old frames/chunks.
 # - ``stream_index``: The index of the source stream.
+# - ``decoder``: If provided, override the decoder. Useful if it fails to detect
+#   the codec.
+# - ``decoder_option``: The option for the decoder.
 #
 # For audio output stream, you can provide the following additional
 # parameters to change the audio properties.
 #
-# - ``sample_rate``: When provided, StreamReader resamples the audio on-the-fly.
-# - ``dtype``: By default the StreamReader returns tensor of `float32` dtype,
-#   with sample values ranging `[-1, 1]`. By providing ``dtype`` argument
+# - ``format``: By default the StreamReader returns tensor of `float32` dtype,
+#   with sample values ranging `[-1, 1]`. By providing ``format`` argument
 #   the resulting dtype and value range is changed.
+# - ``sample_rate``: When provided, StreamReader resamples the audio on-the-fly.
 #
 # For video output stream, the following parameters are available.
 #
+# - ``format``: Change the image format.
 # - ``frame_rate``: Change the frame rate by dropping or duplicating
 #   frames. No interpolation is performed.
 # - ``width``, ``height``: Change the image size.
-# - ``format``: Change the image format.
 #
 
 ######################################################################
@@ -298,7 +301,7 @@
 #    streamer.add_basic_video_stream(
 #        frames_per_chunk=10,
 #        frame_rate=30,
-#        format="RGB"
+#        format="rgb24"
 #    )
 #
 #    # Stream video from source stream `j`,
@@ -310,7 +313,7 @@
 #        frame_rate=30,
 #        width=128,
 #        height=128,
-#        format="BGR"
+#        format="bgr24"
 #    )
 #
 
@@ -428,7 +431,7 @@
     frame_rate=1,
     width=960,
     height=540,
-    format="RGB",
+    format="rgb24",
 )
 
 # Video stream with 320x320 (stretched) at 3 FPS, grayscale
@@ -437,7 +440,7 @@
     frame_rate=3,
     width=320,
     height=320,
-    format="GRAY",
+    format="gray",
 )
 # fmt: on
 

test/torchaudio_unittest/common_utils/case_utils.py

@@ -36,7 +36,6 @@ def get_base_temp_dir(cls):
 
     @classmethod
     def tearDownClass(cls):
-        super().tearDownClass()
         if cls.temp_dir_ is not None:
             try:
                 cls.temp_dir_.cleanup()
@@ -52,6 +51,7 @@ def tearDownClass(cls):
                 #
                 # Following the above thread, we ignore it.
                 pass
+        super().tearDownClass()
 
     def get_temp_path(self, *paths):
         temp_dir = os.path.join(self.get_base_temp_dir(), self.id())