#!/usr/bin/python3
import os

subtitle_dir = r"."

"""
- Projde rekurzivně daný adresář a hledá soubory s koncovkou *.srt a *.sub.
- Ty se snaží otevřít postupně v kódování windows-1250, utf-8-sig a utf-8.
- Obsah načte (převede do utf-8) a zapíše zpět do původního souboru s
  kódováním utf-8-sig, což je BOM verze.
"""

def readContent(fn, encodings=['utf-8-sig', 'windows-1250', 'utf-8']):
  for enc in encodings:
    try:
      with open(fn, "rt", encoding=enc) as f:
        return enc, f.readlines()
    except UnicodeDecodeError as e:
      continue
  raise UnicodeDecodeError('Unknown encoding: ' + fn)

def writeContent(fn, lines, encoding='utf-8-sig'):
  with open(fn, "wt", encoding=encoding) as f:
    for line in lines:
      f.write(line)

# ----------------------------------------------------------------------
# MAIN
# ----------------------------------------------------------------------
if __name__ == '__main__':
  for root, dirs, files in os.walk(subtitle_dir):
    print(f"[{root}]")
    for file in files:
      src = os.path.join(root, file)
      fn, ext = os.path.splitext(src)
      if ext.lower() in (".srt", ".sub"):
        print(f"\t{file}")
        try:
          enc, lines = readContent(src)
          if enc != 'utf-8-sig':
            print(f"\t\tencoding: {enc}")
            writeContent(src, lines, 'utf-8-sig')
        except UnicodeDecodeError as e:
          print("\t\tunknown encoding - unchanged.")
    print()
  input("Press Enter to continue...")