batch_sub_converter.py

import os, sys, re
from glob import glob
import pathlib

class substation:

    def __init__(self, filepath):
        self.filepath = filepath
        self.raw_text = self.get_text()
        self.full_info = {}

    def get_text(self):
        return pathlib.Path(self.filepath).read_text(encoding="utf8")

    def analyze(self):
        toremove = re.compile(r"\{\\fad\(\d+,\d+\)\\.*\}")
        dialogRe = re.compile("Dialogue: \d,(\d:\d{2}:\d{2}.\d{2}),(\d:\d{2}:\d{2}.\d{2}),.*,.*.*,.*,.*,.*,(.*)")
        zerotime = re.compile("0:00:00\.00")
        dialogs = re.findall(dialogRe, re.sub(toremove,"",self.raw_text))
        dialogs = sorted(list(filter(lambda x: x[2],
                              dialogs)))
        self.format_subs(dialogs)

    def format_subs(self,dialogs):
        self.full_info = {start.replace(".", ",") + " --> " +
                          end.replace(".", ","): text for start, end, text in dialogs}

    def translate(self, outputDir):
        self.analyze()
        with open(os.path.join(outputDir,self.filepath[:-4] + ".srt"),encoding="utf8",mode="w") as writer:
            entry = 1
            for timestamp, text in self.full_info.items():
                sub = f"{entry}\n{timestamp}\n{text}\n\n"
                writer.write(sub)
                entry += 1

if __name__ == "__main__":
    for subsalpha in glob(sys.argv[1] + "/*.ass"):
        subs = substation(subsalpha)
        subs.translate(sys.argv[2])