This is a script that converts all ASS subtitles in MKV files to SRT.
Usage: python main.py -p PATH_TO_VIDEO_FOLDER
Python 3.5 or above is required, and the script also requires mkvtoolnix binaries (mkvmerge and mkvextract) to be in PATH (a folder contaning them can also be specified with the -mkv or --mkvtoolnix-path flag).
Note: Install requirements.txt before running.
License: MIT
| import sys | |
| import argparse | |
| import os | |
| import subprocess | |
| import json | |
| import asstosrt | |
| import shutil | |
| def main(): | |
| parser = argparse.ArgumentParser(description="Python Script to convert ASS subtitles to SRT in MKV files.") | |
| parser.add_argument('-p', '--path', help="Path to the folder where the MKV files are located (or MKV file).", required=True) | |
| parser.add_argument('-o', '--output', help="Path to the output folder (or output MKV file). (Default: [current directory]/output)", default="output") | |
| parser.add_argument('-s', '--subs-path', help="Path to the temporary folder where the subtitles are going to be stored. (Default: [current directory]/subs)", default="subs") | |
| parser.add_argument('--no-delete', help="Keep the temporary subtitle folder after the script is finished.", action='store_true') | |
| parser.add_argument('--keep-ass', help="Keep only ASS files in the temporary subtitle folder and delete converted ones.", action='store_true') | |
| parser.add_argument('--keep-fonts', help="The resulting MKV files will contain fonts.", action='store_true') | |
| parser.add_argument('-mkv', '--mkvtoolnix-path', help="Path to mkvtoolnix binary folder.", default="") | |
| args = parser.parse_args() | |
| files = [] | |
| if args.path.endswith('mkv'): | |
| if not os.path.isfile(args.path): | |
| print("Error: path is not a valid file") | |
| quit() | |
| else: | |
| files.append(os.path.basename(args.path)) | |
| args.path = os.path.dirname(args.path) | |
| else: | |
| if not os.path.isdir(args.path): | |
| print("Error: path is not a valid directory") | |
| quit() | |
| else: | |
| files += os.listdir(args.path) | |
| if args.mkvtoolnix_path != "" and not os.path.isdir(args.mkvtoolnix_path): | |
| print("Error: mkvtoolnix path is not a valid directory") | |
| quit() | |
| subtitles = {} | |
| exe_str = "" | |
| if os.name == 'nt': # Check for Windows to add ".exe" at the end of the path of executables | |
| exe_str += ".exe" | |
| if not os.path.exists(args.subs_path): | |
| os.makedirs(args.subs_path) | |
| if (not args.output.endswith(".mkv") or len(files) > 1) and not os.path.exists(args.output): | |
| os.makedirs(args.output) | |
| for file in files: | |
| if file.endswith(".mkv"): | |
| print(f"Processing {file}...") | |
| subtitles[file] = {"ass": [], "others": []} | |
| result = subprocess.run( | |
| [os.path.join(args.mkvtoolnix_path, f"mkvmerge{exe_str}"), "-J", os.path.join(args.path, file)], | |
| shell=False, stdout=subprocess.PIPE) | |
| raw_info = result.stdout.decode('utf-8') | |
| json_info = json.loads(raw_info) | |
| for track in json_info["tracks"]: | |
| append_to = "" | |
| if track["codec"] == "SubStationAlpha": | |
| append_to = "ass" | |
| elif track["type"] == "subtitles": | |
| append_to = "others" | |
| else: | |
| continue | |
| track_name = "" | |
| if "track_name" in track["properties"]: | |
| track_name = track["properties"]["track_name"] | |
| subtitles[file][append_to].append((track["id"], track["properties"]["language"], track["properties"]["default_track"], | |
| track["properties"]["forced_track"], track_name)) | |
| if append_to == "ass": | |
| subtitle_filename = os.path.join(args.subs_path, os.path.splitext(file)[0] + "_" + str(track["id"])) | |
| subprocess.run( | |
| [os.path.join(args.mkvtoolnix_path, f"mkvextract{exe_str}"), os.path.join(args.path, file), | |
| "tracks", str(track["id"]) + ":" + subtitle_filename + ".ass"], | |
| shell=False, stdout=subprocess.DEVNULL) | |
| ass_file = open(subtitle_filename + ".ass", "r", encoding="utf-8", errors="ignore") | |
| srt_data = asstosrt.convert(ass_file) | |
| ass_file.close() | |
| srt_data = patch_duplicated_dialogues(srt_data) | |
| srt_file = open(subtitle_filename + ".srt", "w+", encoding="utf-8") | |
| srt_file.write(srt_data) | |
| srt_file.close() | |
| output_file = "" | |
| if args.output.endswith(".mkv") and not len(files) > 1: | |
| output_file = args.output | |
| else: | |
| output_file = os.path.join(args.output, file) | |
| if os.path.exists(output_file): | |
| output_file = os.path.join(os.path.dirname(output_file), os.path.splitext(os.path.basename(output_file))[0] + "_p.mkv") | |
| command = [os.path.join(args.mkvtoolnix_path, f"mkvmerge{exe_str}"), | |
| "-o", output_file, "-S", "-M", os.path.join(args.path, file)] | |
| if args.keep_fonts: | |
| command = [os.path.join(args.mkvtoolnix_path, f"mkvmerge{exe_str}"), | |
| "-o", output_file, "-S", os.path.join(args.path, file)] | |
| for track in subtitles[file]["ass"]: | |
| subtitle_filename = os.path.join(args.subs_path, os.path.splitext(file)[0] + "_" + str(track[0]) + ".srt") | |
| command += ["--language", "0:" + track[1]] | |
| command += ["--default-track", "0:" + str(int(track[2]))] | |
| command += ["--forced-track", "0:" + str(int(track[3]))] | |
| if track[4] != "": | |
| command += ["--track-name", "0:" + track[4]] | |
| command += [subtitle_filename] | |
| for track in subtitles[file]["others"]: | |
| command += ["-A", "-D", "-B", "-M", "--no-chapters", "--no-global-tags"] # Remove everything that is not a subtitle track | |
| command += ["-s", str(track[0])] | |
| command += [os.path.join(args.path, file)] | |
| subprocess.run(command, shell=False, stdout=subprocess.DEVNULL) | |
| if not args.no_delete: | |
| if not args.keep_ass: | |
| shutil.rmtree(args.subs_path) | |
| else: | |
| for file in os.listdir(args.subs_path): | |
| if file.endswith(".srt"): | |
| os.remove(os.path.join(args.subs_path, file)) | |
| print("Done!") | |
| def patch_duplicated_dialogues(srt_data): | |
| data_dialogues = srt_data.split("\r\n\r\n") | |
| data_dialogues = list(filter(None, data_dialogues)) # Remove empty strings from list | |
| checked_parts = [] | |
| indexes = [] | |
| i = 0 | |
| for dialogue in data_dialogues: | |
| dialogue_parts = dialogue.split("\r\n") | |
| if len(dialogue_parts) < 3: | |
| indexes.append(i) | |
| i += 1 | |
| break | |
| if (dialogue_parts[1], dialogue_parts[2]) in checked_parts: | |
| indexes.append(i) | |
| else: | |
| checked_parts.append((dialogue_parts[1], dialogue_parts[2])) | |
| i += 1 | |
| for index in sorted(indexes, reverse=True): | |
| del data_dialogues[index] | |
| str_patched = "" | |
| i = 0 | |
| for dialogue in data_dialogues: | |
| str_patched += dialogue | |
| if i != len(data_dialogues) - 1: | |
| str_patched += "\r\n\r\n" | |
| i += 1 | |
| return str_patched | |
| if __name__ == "__main__": | |
| if not sys.version_info >= (3, 5): | |
| print("Python 3.5 or above is required.") | |
| quit() | |
| main() |
| asstosrt==0.1.6 |