wrongmove/crawler/ui_exporter.py

53 lines
1.8 KiB
Python

import json
import pathlib
from data_access import Listing
from rec.query import QueryParameters, filter_listings
async def export_immoweb(
ctx,
output_file: str,
query_parameters: QueryParameters | None = None,
):
data_dir = ctx.obj["data_dir"]
output_file_path = pathlib.Path(output_file)
listing_paths = sorted(list(pathlib.Path(data_dir).glob("*/listing.json")))
# listing_paths = listing_paths[:10]
listings = Listing.get_all_listings([str(path) for path in listing_paths])
if query_parameters is not None:
listings = await filter_listings(listings, query_parameters)
# Convert listings to immoweb format
immoweb_listings = []
for listing in listings:
immoweb_listing = {
"type": "Feature",
"properties": {
"city": "London", # change me
"country": "United Kingdom",
"qm": await listing.sqm_ocr(),
"qmprice": round(await listing.price_per_sqm(), 2),
"rooms": listing.bedrooms,
"total_price": listing.price,
"url": listing.url,
# Additional info; the above is GeoJSON format
# Below is all other crap we want in the UI
"info": await listing.dict_nicely(),
},
"geometry": {
"coordinates": [
listing.longitude,
listing.latitude,
],
"type": "Point",
},
}
immoweb_listings.append(immoweb_listing)
prefix = "var data = "
serialized_data = {"type": "FeatureCollection", "features": immoweb_listings}
result = prefix + json.dumps(serialized_data, indent=4)
with open(output_file_path, "w") as f:
f.write(result)
# json.dump(serialized_data, f, indent=4)