test-assignment-parcellab/src/integration/geocoding/osm.ts

import { compact } from 'lodash';
import NodeGeocoder from 'node-geocoder';
import fetch from 'node-fetch';
import PQueue from 'p-queue';
import { mean } from '../../utils/math';
import type { GeocodingProvider } from './types';

export const createOsmClient = (): GeocodingProvider => {
    const geocoder = NodeGeocoder({
        provider: 'openstreetmap',
        fetch: (url, options) => {
            return fetch(url, {
                ...options,
                headers: {
                    ...options?.headers,
                    'user-agent':
                        'test-assignment-parcellab by Inga. This agent is only supposed to send a few requests in two weeks starting on October 23th 2023, and none after that',
                },
            });
        },
    });

    // Rate limit according to https://operations.osmfoundation.org/policies/nominatim/
    const queue = new PQueue({
        interval: 1000,
        intervalCap: 1,
        timeout: 2000,
        throwOnTimeout: true,
    });

    return {
        geocode: async (query) =>
            queue.add(async () => {
                const result = await geocoder.geocode(query);

                if (!result.length) {
                    throw new Error('No results found');
                }

                const meanLatitude = mean(
                    compact(result.map(({ latitude }) => latitude)),
                );
                const meanLongitude = mean(
                    compact(result.map(({ longitude }) => longitude)),
                );
                if (
                    !result.every(
                        ({ latitude, longitude }) =>
                            latitude &&
                            longitude &&
                            Math.abs(latitude - meanLatitude) < 0.01 &&
                            Math.abs(longitude - meanLongitude) < 0.01,
                    )
                ) {
                    throw new Error('Ambiguous address');
                }

                return {
                    latitude: meanLatitude,
                    longitude: meanLongitude,
                };
            }),
    };
};