83 lines
2.9 KiB
Python
Executable file
83 lines
2.9 KiB
Python
Executable file
#!/usr/bin/python
|
|
|
|
# Turns out alltheplaces has done all the hard work here; we can use their
|
|
# (CC0-licensed) data rather than trying to replicate the scraper ourselves.
|
|
#
|
|
# Unfortunately, many of the stores' individual location searches, including the
|
|
# parent TJX's list at https://www.tjx.com/stores, don't provide a list of
|
|
# stores, and only a search result. Some do, e.g. Sierra, with a chunk of
|
|
# javascript containing a list of JS objects, but this isn't consistent across
|
|
# stores, and I'm too lazy to reimplement something for every store. So, instead
|
|
# we take advantage of the hard work of those who have gone before us!
|
|
|
|
import requests
|
|
import json
|
|
|
|
data = requests.get('https://alltheplaces-data.openaddresses.io/runs/2024-04-20-13-31-46/output/tjx.geojson')
|
|
|
|
chains = {}
|
|
|
|
for store in data.json()['features']:
|
|
# store = {
|
|
# "type": "Feature",
|
|
# "id": "iaLJnlhrRR8daHXO0SGtTHQ2aYM=",
|
|
# "properties": {
|
|
# "ref": "93743",
|
|
# "@spider": "tjx",
|
|
# "shop": "department_store",
|
|
# "addr:full": "655 Sydney Ave",
|
|
# "addr:city": "Windsor",
|
|
# "addr:state": "ON",
|
|
# "addr:postcode": "N8X 5C4",
|
|
# "addr:country": "CA",
|
|
# "name": "Windsor",
|
|
# "phone": "+1 519-250-0494",
|
|
# "opening_hours": "Mo-Fr 09:30-21:00; Sa 09:00-21:00; Su 10:00-18:00",
|
|
# "brand": "Marshalls",
|
|
# "brand:wikidata": "Q15903261",
|
|
# "nsi_id": "marshalls-53f9e5"
|
|
# },
|
|
# "geometry": {
|
|
# "type": "Point",
|
|
# "coordinates": [
|
|
# -82.9981994628906,
|
|
# 42.2717170715332
|
|
# ]
|
|
# }
|
|
# },
|
|
if not store['properties']['brand'] in chains:
|
|
chains[store['properties']['brand']] = []
|
|
chains[store['properties']['brand']].append({
|
|
"type": "Feature",
|
|
"geometry": store['geometry'],
|
|
"properties": {
|
|
"name": store['properties']['name'],
|
|
"addr": store['properties']['addr:full'],
|
|
"city": store['properties']['addr:city'],
|
|
"state": store['properties']['addr:state'],
|
|
"postcode": store['properties']['addr:postcode'],
|
|
"country": store['properties']['addr:country'],
|
|
},
|
|
})
|
|
|
|
safe_name = lambda s: ''.join([c.lower() for c in s if c.isalpha()])
|
|
|
|
for chain, features in chains.items():
|
|
geojson = {
|
|
"type": "FeatureCollection",
|
|
"features": features,
|
|
}
|
|
|
|
with open(f"data/{safe_name(chain)}.geojson", "w") as f:
|
|
f.write(json.dumps(geojson))
|
|
|
|
print(f"{len(features)} {chain} locations found")
|
|
|
|
with open('data/chains.js', 'w') as f:
|
|
for chain in chains:
|
|
f.write(f"import {safe_name(chain)} from './{safe_name(chain)}.geojson?url';\n")
|
|
f.write('\nexport default {\n')
|
|
for chain in chains:
|
|
f.write(f" \"{chain}\": {safe_name(chain)},\n")
|
|
f.write("};\n")
|