Location of all stores that sell Orca passes. Scraped from https://www.soundtransit.org/sites/default/files/documents/orca_retail_locations.pdf and put in a more viewable format (GeoJSON)

_stores.json

add_manual_stores.py

from geojson import Point, FeatureCollection
import geojson

def create_points():
    points = []
    f = open('undecoded_coords.txt', 'r')

    for line in f.readlines():
        store_name, address, hours, coords = line.strip().split('|')
        lat, lng = coords.split(',')
        feature = Point((float(lng), float(lat)))
        feature['properties'] = {
            'name': store_name,
            'address': address,
            'hours': hours
        }
        points.append(feature)
    f.close()
    return points

def add_to_geojson(points):
    existing_points = geojson.load(open('stores.json', 'r')).geometries
    all_points = FeatureCollection(existing_points + points)
    return all_points

def main():
    points = create_points()
    collection = add_to_geojson(points)
    with open('new_stores.json', 'w') as f:
        geojson.dump(collection, f)

if __name__ == '__main__':
    main()

parse_pdf.py

import re
from geopy.geocoders import Nominatim
from geojson import Point, FeatureCollection
import geojson
import time

def parse_raw(f_name):
    f = open(f_name, 'r')
    stores = []
    for line in f:
        # Not a location
        if '•' not in line:
            # SE Seattle, NW Seattle etc. are not cities
            if 'Seattle' in line:
                city = 'Seattle'
            else:
                city = line.strip()
            continue

        loc_and_hour_string = line.strip()
        loc_string, hours = loc_and_hour_string.split('•')
        matches = re.match(r"([a-z\s]+)(.+)", loc_string, re.I)
        store_name, loc = matches[1], matches[2]
        address = loc.strip() + ', ' + city + ', WA'

        # Get rid of the trailing comma
        store_name, address, hours = store_name.strip(), address.strip(), hours.strip()
        stores.append((store_name, address, hours))
    return stores

def get_coords(address):
    geolocator = Nominatim(user_agent="orca_scraper")
    return geolocator.geocode(address)

def convert_to_geojson(stores):
    undecoded_stores = []
    decoded_stores = []
    for store in stores:
        store_name, address, hours = store
        location = get_coords(address)
        if location is None:
            undecoded_stores.append(store)
            continue
        lat, lon = location.latitude, location.longitude
        feature = Point((lon, lat))
        feature['properties'] = {
            'name': store_name,
            'address': address,
            'hours': hours
        }
        decoded_stores.append(feature)
        time.sleep(1) # Don't overload the API with calls
    return undecoded_stores, FeatureCollection(decoded_stores)

def main():
    FILENAME = 'orca_locations.txt'
    stores = parse_raw(FILENAME)
    undecoded_stores, decoded_stores = convert_to_geojson(stores)

    with open('stores.json', 'w') as f:
        f.write(geojson.dumps(decoded_stores))

    with open('undecoded_stores.txt', 'w') as f:
        for store, address, hours in undecoded_stores:
            f.write(f'{store}|{address}|{hours}\n')

if __name__ == '__main__':
    main()