Spaces:

AdarshDS
/

ThesisBackend

Build error

ThesisBackend / src /ingestion /caption_extractor.py

AdarshRajDS

stable multimodal supabase ingestion milestone

5484978 7 months ago

267 Bytes

import re

CAPTION_PATTERN = r"(Figure\s+\d+(?:\.\d+)?[:.]?\s.*)"

def extract_caption(text: str):

    if not text:
        return None

    match = re.search(CAPTION_PATTERN, text, re.IGNORECASE)

    if match:
        return match.group(1).strip()

    return None

	import re

	CAPTION_PATTERN = r"(Figure\s+\d+(?:\.\d+)?[:.]?\s.*)"

	def extract_caption(text: str):

	if not text:
	return None

	match = re.search(CAPTION_PATTERN, text, re.IGNORECASE)

	if match:
	return match.group(1).strip()

	return None