Match filename of videos and subtitles.

sub_match.py

import sys
import os
import re


"""
USAGE: python sub_match.py DIR
"""


# Use season and episode as identifier.
KEYWORD_PATTERN = r'[Ss]\d+[Ee]\d+'

# Video and subtitle extensions
VID_EXT = ['mkv', 'mp4']
SUB_EXT = ['srt', 'ass']


def match_subs(dir_path: str) -> None:
    # {keyword: filename, ...}
    videos: dict[str, str] = {}
    subtitles: dict[str, str] = {}

    # Enter dir
    os.chdir(dir_path)

    # Find videos and subtitles
    for f in filter(os.path.isfile, os.listdir()):
        ext = f.rsplit('.', maxsplit=1)[-1].lower()
        if not (ext in SUB_EXT or ext in VID_EXT):
            continue

        keyword = re.findall(KEYWORD_PATTERN, f)[0]

        if ext in VID_EXT:
            videos[keyword] = f
        elif ext in SUB_EXT:
            subtitles[keyword] = f

    print('# Generated shell script.')

    # Generate script
    for kw, video_full_name in videos.items():
        if kw in subtitles:
            # `mv` if video-sub pair found
            video_name = video_full_name.rsplit('.', maxsplit=1)[0]
            sub_extension = subtitles[kw].rsplit('.', maxsplit=1)[-1]
            sub_new_path = f'{video_name}.{sub_extension}'
            # print(kw, sub_new_path)
            print(f'mv '
                  f'"{os.path.abspath(subtitles[kw])}" '
                  f'"{os.path.abspath(sub_new_path)}"'
                  )
        else:
            # warn if no sub matched
            print(f'echo "!!! Subtitle for {video_full_name} not found !!!"')


if __name__ == '__main__':
    match_subs(sys.argv[1])