wrongmove/crawler/4_detect_floorplan.py

import asyncio
from data_access import Listing
from tqdm.asyncio import tqdm
import multiprocessing


async def detect_floorplan(listing_paths: list[str]):
    listings = Listing.get_all_listings(listing_paths)
    cpu_count = multiprocessing.cpu_count() // 4
    semaphore = asyncio.Semaphore(cpu_count)

    await tqdm.gather(
        *[_detect_floorplan_with_semaphore(listing, semaphore) for listing in listings]
    )


async def _detect_floorplan_with_semaphore(
    listing: Listing, semaphore: asyncio.Semaphore
):
    async with semaphore:
        return await listing.calculate_sqm_ocr(recalculate=False)