convert.js

// Following file is licensed under WTFPL public licence. It's open licence.
//npm i xlsx readline-sync

const fs = require('fs');
const xlsx = require('xlsx');
const readlineSync = require('readline-sync');

async function read() {
	const gymns = String(await require('fs').promises.readFile('./secondary-schools.geojson'))
	const c_list = xlsx.readFile('registery.xlsx');
	const c_sheets = c_list.SheetNames;

	return [
		JSON.parse(gymns)['features'],
		xlsx.utils.sheet_to_json(c_list.Sheets[c_sheets[0]])
	]
}

function printWebsite(website) {
	if(!website)
		return undefined
	if(website.startsWith('http')) {
		return website;
	} else {
		return 'http://' + website;
	}
}

function printName(str) {
	if(!str)
		return undefined
	return str.toLowerCase().replace(/(?:^|\s|["'([{])+\S/g, match => match.toUpperCase());
}

function printPhone(tel) {
	if(!tel)
		return undefined;
	return tel.replace(/ /g, '').replace('+48', '').replace(/^/, '+48 ').replace(/(\d{3})(\d{3})(\d{3})/, '$1 $2 $3')
}

console.log('reading');
(async () => {
	read().then(async ([gymnasiums, correct]) => {
		console.log(gymnasiums.length)
		const toFix = gymnasiums.filter(it => it['properties']['addr:city'] && it['properties']['addr:street'] && it['properties']['addr:housenumber']);
		console.log('read');
		
		let i=1;
		const result = await Promise.all(toFix.map(async it => {
			const matches = correct.filter(school => (school['MiejscowośćGUS'] === it['properties']['addr:city'] || school['Miejscowość'] === it['properties']['addr:city']) &&
					school['Ulica'].match(it['properties']['addr:street']) &&
					school['Numer domu'] === it['properties']['addr:housenumber']);
					
			console.log(`${i++} / ${toFix.length}   : ${matches.length}`)

			if(matches.length === 1) {
				return {d: it, m: matches[0]};
			}
			// uncomment to manually choose from multiple matches
			//  else if(matches.length > 1) {
			// 	console.log(it['properties']['name'], it['properties']['addr:street'], it['properties']['addr:housenumber'], it['properties']['website'], 'WYBIERZ:');

			// 	let j=0;
			// 	matches.map(school => {
			// 		console.log(`${j++}:  ${school['Nazwa placówki']} ${school['Ulica']} ${school['Numer domu']},  ${school['Adres www']}    y or n:?`)
			// 	})
			// 	return {d: it, m: matches[readlineSync.question('Zaakceptuj wybór numer:')]};
			// }
			return {};
		}));
		
		const nonEmpty = await Promise.all(result.filter(n => !(Object.keys(n).length === 0 && n.constructor === Object)));
		
		console.log(`Fixed ${nonEmpty.length} items`);
		
		let download = [];
		const features = await Promise.all(nonEmpty.map(async it => {
			download.push(`${it.d.id[0]}${it.d.id.split('/')[1]}`)
			return {
				type: "Feature",
				properties: {...it.d.properties, ...{
					//'@id': undefined,
					'name': printName(it.m['Nazwa placówki']),
					'operator': printName(it.m['Organ prowadzący']),
					'phone': printPhone(it.m['Numer telefonu'] || it.m['Drugi numer telefonu']),
					'email': it.d.properties['addr:email'] || it.m['Adres email'],
					'website': printWebsite(it.m['Adres www']),
					'isced:level': it.m['Typ podmiotu'] === 'Przedszkole' ? 0 :
						it.m['Typ podmiotu'] === 'Szkoła podstawowa' ? '1;2' :
						it.m['Typ podmiotu'] === 'Liceum ogólnokształcące' ? 3 :
						it.m['Typ podmiotu'] === 'Technikum' ? 3 :
						it.m['Typ podmiotu'] === 'Szkoła policealna' ? 4 : undefined
				}},
				geometry: it.d.geometry,
				id: it.d.id
			}
		}));

		const geojson = {
			type: "FeatureCollection",
			generator: "custom script",
			copyright: "The data included in this document is from www.openstreetmap.org and dane.gov.pl The data is made available under ODbL.",
			timestamp: "2020-08-12T14:11:03Z",
			features: features
		};

		fs.writeFileSync('./result.geojson', JSON.stringify(geojson));
		fs.writeFileSync('./down', JSON.stringify(download));
	});
})();