VTT to TXT parser (serverless)

handler.py

import re
import json

def convert_vtt_to_txt(raw_text):
    # Improved regex to handle closing tags properly
    pattern = re.compile(r"<v ([^>]+)>(.+?)</v>", re.DOTALL)
    matches = pattern.findall(raw_text)

    # Clean up whitespaces in dialogues
    structured_lines = [
        (match[0].strip(), re.sub(r'\s+', ' ', match[1].strip()))
        for match in matches
    ]

    combined_lines = []
    previous_speaker = None
    current_dialogue = []

    for speaker, dialogue in structured_lines:
        if speaker == previous_speaker:
            current_dialogue.append(dialogue)
        else:
            if previous_speaker is not None:
                combined_lines.append(f"{previous_speaker}: {' '.join(current_dialogue)}")
            previous_speaker = speaker
            current_dialogue = [dialogue]

    if previous_speaker is not None:
        combined_lines.append(f"{previous_speaker}: {' '.join(current_dialogue)}")

    return "\n\n".join(combined_lines)

def handle(event, context):
    try:
        body = json.loads(event["body"])
        text = body.get("text", "")

        if not text:
            raise ValueError("No text provided.")

        cleaned_text = convert_vtt_to_txt(text)

        return {
            "statusCode": 200,
            "headers": {"Content-Type": "application/json"},
            "body": json.dumps({"result": cleaned_text})
        }
    except Exception as e:
        return {
            "statusCode": 400,
            "headers": {"Content-Type": "application/json"},
            "body": json.dumps({"error": str(e)})
        }