wrongmove/crawler/4_detect_floorplan.py

import pathlib
from data_access import Listing
from tqdm import tqdm


def detect_floorplan(listing_paths: list[str]):
    listings = Listing.get_all_listings(listing_paths)

    for listing in tqdm(listings):
        tqdm.write(str(listing.identifier))
        # listing.calculate_sqm_model() # using google/deplot model. Too slow, rather use tesseract
        listing.calculate_sqm_ocr(recalculate=False)


def main():
    listing_paths = sorted(list(pathlib.Path("data/rs").glob("*/listing.json")))
    detect_floorplan(listing_paths)


if __name__ == "__main__":
    main()