more common sense to that

preprocessor/__main__.py

@@ -1,20 +1,28 @@
 import argparse
+import logging
+from pathlib import Path
 
 from transciption_generator import TranscriptionGenerator
 from video_transcoder import VideoTranscoder
 
 
-def main() -> None:
+if __name__ == "__main__":
+    logging.basicConfig(format="%(asctime)s | %(levelname)s | %(message)s", level=logging.DEBUG)
+
     parser = argparse.ArgumentParser()
-    parser.add_argument("videos", type=argparse.FileType("r"), help="Path to input videos for preprocessing")
-    parser.add_argument("--transcoded-videos-dir", "-v", type=argparse.FileType("w"), default="transcoded_videos", help="Path for output videos after transcoding")
-    parser.add_argument("--transcription-jsons-dir", "-j", type=argparse.FileType("w"), default="transcriptions", help="Path for output transcriptions JSONs")
+    parser.add_argument("videos", type=Path, help="Path to input videos for preprocessing")
 
+    # 2 subparsers to split stuff
+    # add defaults from classes here
+    parser.add_argument("--transcoded-videos-dir", "-v", type=Path, default="transcoded_videos", help="Path for output videos after transcoding")
+    parser.add_argument("--transcription-jsons-dir", "-j", type=Path, default="transcriptions", help="Path for output transcriptions JSONs")
 
     args = parser.parse_args()
 
+
     TranscriptionGenerator(args.videos, args.transcription_jsons_dir).transcribe()
     VideoTranscoder(args.videos, args.transcoded_videos_dir).transcode()
 
     # pass transcriptions to elastic
     # split two paths to be async
+

preprocessor/transciption_generator.py

@@ -5,6 +5,8 @@ def __init__(self, input_videos: Path, output_jsons: Path):
         self.__input_videos = input_videos
         self.__output_jsons = output_jsons
 
+        # normalizer -> audio processor -> json processor
+
     def transcribe(self) -> None:
         self.__get_best_audio_path()
         self.__normalize()

preprocessor/video_transcoder.py

@@ -1,10 +1,31 @@
 from pathlib import Path
 
 class VideoTranscoder:
-    def __init__(self, input_videos: Path, output_videos: Path):
+    DEFAULT_CODEC: str = "h264_nvenc"
+    DEFAULT_PRESET: str = "slow"
+    DEFAULT_CRF: int = 31
+    DEFAULT_GOP_SIZE: float = 0.5
+
+    def __init__(
+            self,
+            input_videos: Path,
+            output_videos: Path,
+            codec: str = DEFAULT_CODEC,
+            preset: str = DEFAULT_PRESET,
+            crf: int = DEFAULT_CRF,
+            gop_size: float = DEFAULT_GOP_SIZE,
+    ):
         self.__input_videos = input_videos
         self.__output_videos = output_videos
 
+        self.__codec = codec
+        self.__preset = preset
+        self.__crf = crf
+        self.__gop_size = gop_size
+
+
+        # video converter
+
     def transcode(self):
         self.__prepare_videos()
         self.__do_transcoding()