Rohit-Katkar2003's picture
Upload 7 files
89643c6 verified
Raw
History Blame Contribute Delete
400 Bytes
import re
def parse_script(script):
pattern = r"\[(.*?)\]:\s*(.*?)(?=\n\[|$)"
matches = re.findall(pattern, script, re.DOTALL)
parsed = []
for speaker_info, text in matches:
name = speaker_info.split("(")[0].strip()
gender = "female" if "WOMAN" in speaker_info else "male"
parsed.append((name, gender, text.strip()))
return parsed