使用Python将SRT (SubRip文本)转换为JSON

import re
import json

def srt_to_json(srt_file):
    with open(srt_file, 'r') as file:
        srt_data = file.read()

    # 使用正则表达式匹配SRT文件中的时间和字幕文本
    pattern = r'(\d+)\n(\d{2}:\d{2}:\d{2},\d{3}) --> (\d{2}:\d{2}:\d{2},\d{3})\n(.+?)\n\n'
    matches = re.findall(pattern, srt_data, re.DOTALL)

    # 将匹配到的数据转换为JSON格式
    subtitles = []
    for match in matches:
        subtitle = {
            'index': int(match[0]),
            'start_time': match[1],
            'end_time': match[2],
            'text': match[3]
        }
        subtitles.append(subtitle)

    json_data = json.dumps(subtitles, ensure_ascii=False)

    return json_data

# 调用函数将SRT转换为JSON
srt_file = 'example.srt'
json_data = srt_to_json(srt_file)
print(json_data)