utils.py

import re

def extract_single_option(video_llm_output) -> Union[str, bool]:
    video_llm_output = video_llm_output.lower()
    pattern = r"(Answer|Assistant)?:?\s*([AB])\b"
    matches = re.findall(pattern, video_llm_output, re.IGNORECASE)
    if matches:
        actual_answer = matches[1] if len(matches) > 1 else matches[0]
        answer = actual_answer[1].lower()
        return answer 
    elif len(video_llm_output) == 1:
        answer = video_llm_output.lower()
        return answer 
    else:
        return False