change: refactor video transforms

Author: boomb0om

DPF/processors/processor_mixins.py

@@ -25,6 +25,8 @@ def apply_transform(self, transforms: BaseFilesTransforms) -> None:
 
  metadata_lists = None
  if len(transforms.required_metadata) > 0:
+ assert all(col in self._df.columns for col in transforms.required_metadata), \
+ f"Columns {[col not in self._df.columns for col in transforms.required_metadata]} are not presented"
  metadata_lists = {
  col: self._df[col].tolist()
  for col in transforms.required_metadata

DPF/transforms/__init__.py

@@ -1,5 +1,4 @@
 from .base_file_transforms import BaseFilesTransforms, TransformsFileData
 from .image_resize_transforms import ImageResizeTransforms
 from .resizer import Resizer, ResizerModes
-from .video_fps_transforms import VideoFPSTransforms
-from .video_resize_transforms import VideoResizeTransforms
+from .video_ffmpeg_transforms import VideoFFMPEGTransforms

DPF/transforms/video_ffmpeg_transforms.py

@@ -0,0 +1,120 @@
+import shutil
+import subprocess
+import uuid
+from typing import Any, Optional
+
+from DPF.transforms.base_file_transforms import (
+ BaseFilesTransforms,
+ PoolOptions,
+ TransformsFileData,
+)
+from DPF.transforms.resizer import Resizer
+
+
+def is_ffmpeg_installed() -> bool:
+ try:
+ subprocess.run('ffmpeg -version', shell=True, capture_output=True, check=True)
+ return True
+ except subprocess.CalledProcessError:
+ return False
+
+
+def convert_ffmpeg_args_to_str(args: dict[str, list[str]]) -> str:
+ arg_strings = []
+ for k, v in args.items():
+ s = ','.join(v)
+ arg_strings.append(f"{k} {s}")
+ return ' '.join(arg_strings)
+
+
+class VideoFFMPEGTransforms(BaseFilesTransforms):
+
+ def __init__(
+ self,
+ resizer: Optional[Resizer] = None,
+ fps: Optional[int] = None,
+ fps_eps: float = 0.1,
+ pool_type: PoolOptions = 'threads',
+ workers: int = 16,
+ pbar: bool = True,
+ preset: Optional[str] = None,
+ crf: Optional[int] = None,
+ copy_audio_stream: bool = True
+ ):
+ super().__init__(pool_type, workers, pbar)
+ self.resizer = resizer
+ self.fps = fps
+ self.fps_eps = fps_eps
+
+ self.preset = preset
+ self.crf = crf
+ self.copy_audio_stream = copy_audio_stream
+
+ self.default_args = ' '.join(self.get_default_ffmpeg_args())
+
+ assert is_ffmpeg_installed(), "Install ffmpeg first"
+ assert self.resizer or self.fps, "At least one transform should be specified"
+
+ def get_default_ffmpeg_args(self) -> list[str]:
+ args = []
+ if self.preset:
+ args.append(f"-preset {self.preset}")
+ if self.crf:
+ args.append(f'-crf {self.crf}')
+ if self.copy_audio_stream:
+ args.append('-c:a copy')
+ return args
+
+ @property
+ def required_metadata(self) -> list[str]:
+ meta = []
+ if self.resizer:
+ meta += ['width', 'height']
+ if self.fps:
+ meta += ['fps']
+ return meta
+
+ @property
+ def metadata_to_change(self) -> list[str]:
+ meta = []
+ if self.resizer:
+ meta += ['width', 'height']
+ if self.fps:
+ meta += ['fps']
+ return meta
+
+ @property
+ def modality(self) -> str:
+ return 'video'
+
+ def _process_filepath(self, data: TransformsFileData) -> TransformsFileData:
+ filepath = data.filepath
+ ext = filepath.split('.')[-1]
+ ffmpeg_args_map: dict[str, list[str]] = {}
+ result_metadata: dict[str, Any] = {}
+
+ if self.resizer:
+ width, height = data.metadata['width'], data.metadata['height']
+ new_width, new_height = self.resizer.get_new_size(width, height)
+ if (new_width, new_height) != (width, height):
+ new_width += new_width % 2
+ new_height += new_height % 2
+ ffmpeg_args_map['-vf'] = ffmpeg_args_map.get('-vf', []) + [f"scale={new_width}:{new_height}"]
+ result_metadata['width'] = new_width
+ result_metadata['height'] = new_height
+
+ if self.fps:
+ video_fps = data.metadata['fps']
+ if (video_fps < (self.fps - self.fps_eps)) or (video_fps > (self.fps + self.fps_eps)):
+ ffmpeg_args_map['-vf'] = ffmpeg_args_map.get('-vf', []) + [f"fps={self.fps}"]
+ video_fps = float(self.fps)
+ result_metadata['fps'] = video_fps
+
+ if len(ffmpeg_args_map) > 0:
+ args_str = convert_ffmpeg_args_to_str(ffmpeg_args_map)
+ temp_filename = str(uuid.uuid4()) + '.' + ext
+ ffmpeg_command = f'ffmpeg -hide_banner -i {filepath} {args_str} {self.default_args} {temp_filename} -y'
+ subprocess.run(ffmpeg_command, shell=True, capture_output=True, check=True)
+ shutil.move(temp_filename, filepath)
+
+ return TransformsFileData(filepath, result_metadata)

DPF/transforms/video_fps_transforms.py

@@ -8,7 +8,7 @@ You can use `DPF.transforms` for these tasks.
 
 List of implemented transforms:
 - [ImageResizeTransforms](../DPF/transforms/image_resize_transforms.py) - transforms that resizes images
-- [VideoResizeTransforms](../DPF/transforms/video_resize_transforms.py) - transforms that resizes videos
+- [VideoFFMPEGTransforms](../DPF/transforms/video_ffmpeg_transforms.py) - transforms that resizes and changing fps of videos using ffmpeg
 
 ### Examples
 
@@ -26,4 +26,16 @@ from DPF.transforms import VideoResizeTransforms, Resizer, ResizerModes
 
 transforms = VideoResizeTransforms(Resizer(ResizerModes.MAX_SIZE, size=768))
 processor.apply_transform(transforms)
+```
+
+Change fps to 24 and resize all videos to 768 pixels on the minimum side while maintaining the aspect ratio:
+```python
+from DPF.transforms import VideoFFMPEGTransforms, Resizer, ResizerModes
+
+transforms = VideoFFMPEGTransforms(
+ resizer=Resizer(ResizerModes.MIN_SIZE, size=768),
+ fps=24,
+ workers=8
+)
+processor.apply_transform(transforms)
 ```