test

.github/workflows/check-code.yml

@@ -50,4 +50,6 @@ jobs:
       - name: Install test dependencies
         run: python -m pip install -r dev-requirements.txt
       - name: Test
-        run: pytest tests --full-trace
+        run: |
+          $Env:PYTHONASYNCIODEBUG = 1
+          pytest tests --full-trace --log_cli_level=DEBUG

buganime/transcode.py

@@ -67,19 +67,25 @@ async def __read_input_frames(self) -> AsyncIterator[bytes]:
         args = ('-i', self.__input_path,
                 '-f', 'rawvideo', '-pix_fmt', 'rgb24', 'pipe:',
                 '-loglevel', 'warning')
-        proc = await asyncio.subprocess.create_subprocess_exec('ffmpeg', *args, stdout=asyncio.subprocess.PIPE, stderr=asyncio.subprocess.PIPE)
+        logging.warning('c0')
+        proc = await asyncio.subprocess.create_subprocess_exec('ffmpeg', *args)
+        logging.warning('c-1')
         assert proc.stdout
         assert proc.stderr
         try:
             frame_length = self.__video_info.width * self.__video_info.height * 3
             with contextlib.suppress(asyncio.IncompleteReadError):
                 while True:
+                    logging.warning('c1')
                     yield await proc.stdout.readexactly(frame_length)
+                    logging.warning('c2')
         finally:
+            logging.warning('c3')
             with contextlib.suppress(ProcessLookupError):
                 proc.terminate()
             logging.info('ffmpeg input: %s', str(await proc.stderr.read()))
             await proc.wait()
+            logging.warning('c4')
 
     async def __write_output_frames(self, frames: AsyncIterator[bytes]) -> None:
         with tempfile.TemporaryDirectory() as temp_dir:
@@ -115,29 +121,43 @@ async def __write_output_frames(self, frames: AsyncIterator[bytes]) -> None:
     @retry.retry(RuntimeError, tries=10, delay=1)
     def __gpu_upscale(self, frame: torch.Tensor) -> torch.Tensor:
         with torch.no_grad():
+            logging.warning('x')
             if torch.cuda.is_available():
                 frame = frame.cuda()
+            logging.warning('a')
             frame_float = frame.permute(2, 0, 1).half() / 255
+            logging.warning('b')
             frame_upscaled_float = self.__model(frame_float.unsqueeze(0)).data.squeeze().clamp_(0, 1)
+            logging.warning('c')
             return cast(torch.Tensor, (frame_upscaled_float * 255.0).round().byte().permute(1, 2, 0).cpu())
 
     async def __upscale_frame(self, frame: bytes) -> bytes:
+        logging.warning('a1')
         if self.__video_info.height == self.__height_out:
             return frame
+        logging.warning('a2')
         with torch.no_grad():
             with warnings.catch_warnings(action='ignore'):
                 frame_arr = torch.frombuffer(frame, dtype=torch.uint8).reshape([self.__video_info.height, self.__video_info.width, 3])
+        logging.warning('a3')
         assert self.__gpu_lock
         async with self.__gpu_lock:
+            logging.warning('a4')
             frame_cpu = await asyncio.to_thread(self.__gpu_upscale, frame_arr)
+            logging.warning('a5')
+        logging.warning('a6')
         return cast(bytes, await asyncio.to_thread(
             lambda: cv2.resize(frame_cpu.numpy(), (self.__width_out, self.__height_out), interpolation=cv2.INTER_LANCZOS4).tobytes()))
 
     async def __generate_upscaling_tasks(self) -> None:
         assert self.__frame_tasks_queue
+        logging.warning('b2')
         async for frame in self.__read_input_frames():
+            logging.warning('b3')
             await self.__frame_tasks_queue.put(asyncio.create_task(self.__upscale_frame(frame)))
+        logging.warning('b4')
         await self.__frame_tasks_queue.put(None)
+        logging.warning('b5')
 
     async def __get_output_frames(self) -> AsyncIterator[bytes]:
         assert self.__frame_tasks_queue