covid19/generate.js

#!/usr/bin/env node

const path = require('path');
const fs = require('fs');
const parseCsv = require('csv-parse/lib/sync');
const pug = require('pug');
const {execSync} = require('child_process');

const publicDir = path.join(__dirname, 'public');
const templatesDir = path.join(__dirname, 'tmpl');
const dataDir = path.join(__dirname, 'data');
const covidGitDir = path.join(__dirname, 'COVID-19');
const covidDataDir = path.join(covidGitDir, 'csse_covid_19_data');
const timeSeriesDir = path.join(covidDataDir, 'csse_covid_19_time_series');

const promiseMe = (fn) => {
	return new Promise((resolve, reject) => {
		fn((err, result) => {
			if (err) {
				reject(err);
				return;
			}

			resolve(result);
		});
	});
};

fs.mkdirSync(path.join(__dirname, 'public', 'countries'), {
	recursive: true,
});

fs.copyFileSync(
	path.join(__dirname, 'node_modules', 'chart.js', 'dist', 'Chart.bundle.min.js'),
	path.join(publicDir, 'Chart.bundle.js'),
);

fs.copyFileSync(
	path.join(__dirname, 'node_modules', 'bootstrap', 'dist', 'css', 'bootstrap.css'),
	path.join(publicDir, 'bootstrap.css'),
);

const deathsGlobalCsv = path.join(timeSeriesDir, 'time_series_covid19_deaths_global.csv');
const confirmedGlobalCsv = path.join(timeSeriesDir, 'time_series_covid19_confirmed_global.csv');
const deathsUSCsv = path.join(timeSeriesDir, 'time_series_covid19_deaths_US.csv');
const confirmedUSCsv = path.join(timeSeriesDir, 'time_series_covid19_confirmed_US.csv');
const populationStatesCsv = path.join(dataDir, 'SCPRC-EST2019-18+POP-RES.csv');
const populationCountriesCsv = path.join(dataDir, 'population-world-wikipedia.tsv');

const lastGlobalDeathsUpdate = execSync(`git -C "${covidGitDir}" log  -n 1 --pretty=format:'%ci' "${deathsGlobalCsv}"`, {
	encoding: 'utf8',
});
const lastUSDeathsUpdate = execSync(`git -C "${covidGitDir}" log -n 1 --pretty=format:'%ci' "${deathsUSCsv}"`, {
	encoding: 'utf8',
});

const lastUpdate = new Date(lastGlobalDeathsUpdate > lastUSDeathsUpdate ?
	lastGlobalDeathsUpdate :
	lastUSDeathsUpdate
);

const zeroPad = value => value < 10 ? `0${value}` : value.toString();
const toSafeName = x => x.replace(/[^A-Za-z]/g, '-').toLowerCase();

const processGlobalDeaths = async () => {
	const globalStart = Date.now();
	let start = Date.now();

	const timeSeriesGlobalRaw = fs.readFileSync(deathsGlobalCsv, {encoding: 'utf8'});
	console.log(`read global deaths CSV in ${Date.now() - start}ms`);

	start = Date.now();
	const timeSeriesUSRaw = fs.readFileSync(deathsUSCsv, { encoding: 'utf8' });
	console.log(`read US deaths CSV in ${Date.now() - start}ms`);

	start = Date.now();
	const populationUSRaw = fs.readFileSync(populationStatesCsv, {encoding: 'utf8'});
	console.log(`read US states population CSV in ${Date.now() - start}ms`);

	start = Date.now();
	const populationCountriesRaw = fs.readFileSync(populationCountriesCsv, {encoding: 'utf8'});
	console.log(`read countries population CSV in ${Date.now() - start}ms`);

	start = Date.now();
	let tsGlobalRecords = parseCsv(timeSeriesGlobalRaw, {
		cast: true,
		columns: true,
	});
	console.log(`parsed global deaths CSV in ${Date.now() - start}ms`);

	start = Date.now();
	let tsUSRecords = parseCsv(timeSeriesUSRaw, {
		cast: true,
		columns: true,
	});
	console.log(`parsed US deaths CSV in ${Date.now() - start}ms`);

	start = Date.now();
	let populationUSStateRecords = parseCsv(populationUSRaw, {
		cast: true,
		columns: true,
	});
	console.log(`parsed US states population CSV in ${Date.now() - start}ms`);

	start = Date.now();
	let populationCountriesRecords = parseCsv(populationCountriesRaw, {
		cast: true,
		columns: true,
		delimiter: '\t',
		ltrim: true,
		rtrim: true,
	});
	console.log(`parsed countries population CSV in ${Date.now() - start}ms`);

	const statePopulationMap = populationUSStateRecords
		.sort((a, b) => a.NAME.localeCompare(b.NAME))
		.reduce((map, item) => {
			map[item.NAME] = Number(item.POPESTIMATE2019);
			return map;
		}, {});

	const countryNameMap = {
		'Cape Verde': 'Cabo Verde',
		'DR Congo': 'Congo (Brazzaville)',
		'Congo': 'Congo (Kinshasa)',
		'Ivory Coast': 'Cote d\'Ivoire',
		'Czech Republic': 'Czechia',
		'Vatican City': 'Holy See',
		'South Korea': 'Korea, South',
		'Taiwan': 'Taiwan*',
		'East Timor': 'Timor-Leste',
		'Palestine': 'West Bank and Gaza',
		'Myanmar': 'Burma',
		'São Tomé and Príncipe': 'Sao Tome and Principe',
	};

	const countryPopulationMap = populationCountriesRecords
		.sort((a, b) => a.Country.localeCompare(b.Country))
		.reduce((map, item) => {
			map[item.Country] = Number(item.Pop_2019.toString().replace(/,/g, ''));
			const otherName = countryNameMap[item.Country];
			if (otherName) {
				map[otherName] = map[item.Country];
			}
			return map;
		}, {});

	countryPopulationMap['Diamond Princess'] = statePopulationMap['Diamond Princess'] = 3711;
	countryPopulationMap['MS Zaandam'] = 1829;
	countryPopulationMap['Kosovo'] = 1810463;
	statePopulationMap['Grand Princess'] = 3533;

	// tsGlobalRecords = tsGlobalRecords.filter((record) => {
	// 	return record['Country/Region'] === 'US';
	// });

	const getGrowthRate = (record) => {
		const ts = record.timeSeriesDaily;
		const len = ts.length;
		if (len < 2) {
			return 0;
		}
		const latest = len - 1;
		const earliest = Math.max(len - 14, 0);
		const pow = 1/(latest - earliest + 1);
		const hi = ts[latest].value;
		const lo = Math.max(ts[earliest].value, 0.5);
		if (hi === 0 && lo < 1) {
			return 0;
		}
		return Math.pow((hi / lo), pow) - 1;
	};

	// state/county data is separated for the US and doesn't need to be rolled up
	tsUSRecords.forEach((usRecord) => {
		const newRecord = {
			...usRecord,
			needsRollup: false,
			Long: usRecord.Long_,
			'Province/State': usRecord.Province_State,
			'Country/Region': usRecord.Country_Region,
		};

		delete newRecord.UID;
		delete newRecord.iso2;
		delete newRecord.iso3;
		delete newRecord.code3;
		delete newRecord.FIPS;
		delete newRecord.Combined_Key;
		delete newRecord.Long_;
		delete newRecord.Province_State;
		delete newRecord.Country_Region;

		tsGlobalRecords.push(newRecord);
	});

	start = Date.now();
	tsGlobalRecords.forEach((record) => {
		record.timeSeriesDaily = [];
		record.timeSeriesMonthly = [];
		const dateColumns = Object.keys(record).filter(x => /^\d+\/\d+\/\d+$/.test(x))
			.map(key => {
				return {
					key,
					date: new Date(key),
				};
			})
			.sort((a, b) => {
				if (a.date.getTime() === b.date.getTime()) {
					return 0;
				}

				return a.date.getTime() < b.date.getTime() ? -1 : 1;
			});

		const toSortableDate = date => [
			date.getUTCFullYear(),
			zeroPad(date.getUTCMonth() + 1),
			zeroPad(date.getUTCDate()),
		].join('-');

		dateColumns.forEach((obj) => {
			const value = Number(record[obj.key]) || 0;
			const date = obj.date;
			delete record[obj.key];
			const sortableKey = toSortableDate(date);
			const lastItem = record.timeSeriesDaily[record.timeSeriesDaily.length - 1];
			record.timeSeriesDaily.push({
				key: sortableKey,
				value,
				delta: lastItem ? value - lastItem.value : 0,
			});
		});

		const monthlyTotals = [];
		const monthlyMaxes = {};
		record.timeSeriesDaily.forEach((item) => {
			const key = item.key.replace(/-\d+$/, '');
			if (!(key in monthlyMaxes)) {
				monthlyMaxes[key] = {
					date: item.key,
					value: item.value,
				};
			} else if (item.key > monthlyMaxes[key].date) {
				monthlyMaxes[key] = {
					date: item.key,
					value: item.value,
				}
			}
		});

		Object.keys(monthlyMaxes).forEach((key) => {
			monthlyTotals.push({
				key,
				value: monthlyMaxes[key].value,
			});
		});

		monthlyTotals.sort((a, b) => a.key.localeCompare(b.key));
		monthlyTotals.forEach((item, i) => {
			const prev = monthlyTotals[i - 1];
			item.delta = prev ? item.value - prev.value : 0;
		});

		record.total = record.timeSeriesDaily.length ?
			record.timeSeriesDaily[record.timeSeriesDaily.length - 1].value :
			0;
		record.timeSeriesMonthly = monthlyTotals;

		record.state = record['Province/State'];
		record.country = record['Country/Region'];
		record.lat = record.Lat;
		record.long = record.Long;
		record.county = record.Admin2 || '';
		record.population = record.Population || null;

		if (record.country === 'US') {
			record.country = 'United States';
		}

		record.countrySafeName = toSafeName(record.country);
		record.stateSafeName = toSafeName(record.state);
		record.countySafeName = toSafeName(record.county);

		delete record['Province/State'];
		delete record['Country/Region'];
		delete record.Lat;
		delete record.Long;
		delete record.Admin2;
		delete record.Population;

		if (!record.population && !record.state && !record.county) {
			const mappedPop = countryPopulationMap[record.country];
			if (!mappedPop) {
				console.log(require('util').inspect(countryPopulationMap, false, null, true));
				throw new Error('found no population for ' + record.country);
			}
			record.population = mappedPop;
		} else if (!record.population && record.state && !record.county && record.country === 'United States') {
			// US territories and cruise ships
			const mappedPop = statePopulationMap[record.state];
			if (!mappedPop) {
				console.log(require('util').inspect(statePopulationMap, false, null, true));
				throw new Error('found no population for ' + record.state);
			}
			// console.log(`setting population to ${mappedPop} for ${record.state}`);
			record.population = mappedPop;
		}

		record.deathsPerMillion = !!record.population ? record.total / record.population * 1000000 : 0;
		record.deathGrowthRate = getGrowthRate(record);
	});

	tsGlobalRecords.sort((a, b) => {
		if (a.country === b.country) {
			return a.state.localeCompare(b.state);
		}

		return a.country.localeCompare(b.country);
	});

	const perCountryTotals = {};
	const perStateTotals = {};
	tsGlobalRecords.forEach((record) => {
		perCountryTotals[record.country] = perCountryTotals[record.country] || {
			total: 0,
			population: 0,
			timeSeriesDaily: {},
			timeSeriesMonthly: {},
			states: [],
			safeName: record.countrySafeName,
		};

		const item = perCountryTotals[record.country];
		if (!record.state && !record.county) {
			// country population
			item.population = record.population;
		}

		if (record.county) {
			record.name = record.county;
			record.safeName = toSafeName(record.county);
		}

		// roll up up state/county data
		if (record.state && record.county) {
			const stateItem = perStateTotals[record.state] = perStateTotals[record.state] || {
				name: record.state,
				safeName: record.stateSafeName,
				country: record.country,
				countrySafeName: record.countrySafeName,
				total: 0,
				population: 0,
				deathsPerMillion: 0,
				timeSeriesDaily: {},
				timeSeriesMonthly: {},
				counties: [],
			};

			if (!stateItem.population && record.country === 'United States') {
				const population = statePopulationMap[record.state];
				if (!population) {
					console.log(require('util').inspect(statePopulationMap, false, null, true));
					throw new Error('no population for ' + record.state);
				}

				stateItem.population = population;
			}

			stateItem.total += record.total;
			record.timeSeriesDaily.forEach((ts) => {
				stateItem.timeSeriesDaily[ts.key] = stateItem.timeSeriesDaily[ts.key] || {
					value: 0,
					delta: 0,
				};
				stateItem.timeSeriesDaily[ts.key].value += ts.value;
				stateItem.timeSeriesDaily[ts.key].delta += ts.delta;
			});

			record.timeSeriesMonthly.forEach((ts) => {
				stateItem.timeSeriesMonthly[ts.key] = stateItem.timeSeriesMonthly[ts.key] || {
					value: 0,
					delta: 0,
				};
				stateItem.timeSeriesMonthly[ts.key].value += ts.value;
				stateItem.timeSeriesMonthly[ts.key].delta += ts.delta;
			});

			stateItem.counties.push(record);
		} else {
			item.states.push(record);
		}

		if (record.needsRollup === false) {
			return;
		}

		item.total += record.total;
		record.timeSeriesDaily.forEach((ts) => {
			item.timeSeriesDaily[ts.key] = item.timeSeriesDaily[ts.key] || {
				value: 0,
				delta: 0,
			};
			item.timeSeriesDaily[ts.key].value += ts.value;
			item.timeSeriesDaily[ts.key].delta += ts.delta;
		});

		record.timeSeriesMonthly.forEach((ts) => {
			item.timeSeriesMonthly[ts.key] = item.timeSeriesMonthly[ts.key] || {
				value: 0,
				delta: 0,
			};
			item.timeSeriesMonthly[ts.key].value += ts.value;
			item.timeSeriesMonthly[ts.key].delta += ts.delta;
		});
	});

	Object.keys(perStateTotals).forEach((stateName) => {
		const item = perStateTotals[stateName];
		const stateItem = {
			name: stateName,
			safeName: item.safeName,
			country: item.country,
			countrySafeName: item.countrySafeName,
			total: item.total,
			counties: item.counties,
			population: item.population,
			deathsPerMillion: item.population > 0 ? item.total / item.population * 1000000 : 0,
			timeSeriesDaily: Object.keys(item.timeSeriesDaily).sort().map((date) => {
				return {
					key: date,
					value: item.timeSeriesDaily[date].value,
					delta: item.timeSeriesDaily[date].delta,
				};
			}),
			timeSeriesMonthly: Object.keys(item.timeSeriesMonthly).sort().map((date) => {
				return {
					key: date,
					value: item.timeSeriesMonthly[date].value,
					delta: item.timeSeriesMonthly[date].delta,
				};
			}),
		};

		stateItem.deathGrowthRate = getGrowthRate(stateItem);

		// insert into states array for the country
		perCountryTotals[item.country].states.push(stateItem);
	});

	const countryArr = Object.keys(perCountryTotals).map((countryName) => {
		const item = perCountryTotals[countryName];
		if (!item.population) {
			// some countries don't have roll up (e.g. Canada, China, Australia)
			// and only contain province data instead of data for the whole country.
			// population doesn't get set in those cases.
			item.population = countryPopulationMap[countryName];
		}

		const countryItem = {
			name: countryName,
			safeName: item.safeName,
			total: item.total,
			states: item.states,
			population: item.population,
			deathsPerMillion: item.population > 0 ? item.total / item.population * 1000000 : 0,
			timeSeriesDaily: Object.keys(item.timeSeriesDaily).sort().map((date) => {
				return {
					key: date,
					value: item.timeSeriesDaily[date].value,
					delta: item.timeSeriesDaily[date].delta,
				};
			}),
			timeSeriesMonthly: Object.keys(item.timeSeriesMonthly).sort().map((date) => {
				return {
					key: date,
					value: item.timeSeriesMonthly[date].value,
					delta: item.timeSeriesMonthly[date].delta,
				};
			}),
		};

		countryItem.deathGrowthRate = getGrowthRate(countryItem);
		return countryItem;
	});

	const worldData = {
		name: 'The World',
		safeName: 'the-world',
		total: 0,
		countries: countryArr,
		timeSeriesDaily: {},
		timeSeriesMonthly: {},
	};

	countryArr.forEach((countryData) => {
		worldData.total += countryData.total;

		countryData.timeSeriesDaily.forEach((ts) => {
			worldData.timeSeriesDaily[ts.key] = worldData.timeSeriesDaily[ts.key] || {
				value: 0,
				delta: 0,
			};
			worldData.timeSeriesDaily[ts.key].value += ts.value;
			worldData.timeSeriesDaily[ts.key].delta += ts.delta;
		});

		countryData.timeSeriesMonthly.forEach((ts) => {
			worldData.timeSeriesMonthly[ts.key] = worldData.timeSeriesMonthly[ts.key] || {
				value: 0,
				delta: 0,
			};
			worldData.timeSeriesMonthly[ts.key].value += ts.value;
			worldData.timeSeriesMonthly[ts.key].delta += ts.delta;
		});
	});

	worldData.timeSeriesDaily = Object.keys(worldData.timeSeriesDaily).sort().map((date) => {
		return {
			key: date,
			value: worldData.timeSeriesDaily[date].value,
			delta: worldData.timeSeriesDaily[date].delta,
		};
	});
	worldData.timeSeriesMonthly = Object.keys(worldData.timeSeriesMonthly).sort().map((date) => {
		return {
			key: date,
			value: worldData.timeSeriesMonthly[date].value,
			delta: worldData.timeSeriesMonthly[date].delta,
		};
	});

	worldData.deathGrowthRate = getGrowthRate(worldData);
	worldData.population = 7781841000;
	worldData.deathsPerMillion = worldData.total / worldData.population * 1000000;

	console.log(`transformed data in ${Date.now() - start}ms`);

	start = Date.now();
	const worldTmpl = path.join(templatesDir, 'world.pug');
	const worldHtml = pug.renderFile(worldTmpl, {
		data: worldData,
		$title: 'The World',
		lastUpdate,
	});

	const targetFile = path.join(publicDir, 'index.html');
	fs.writeFileSync(targetFile, worldHtml);
	console.log(`wrote to ${targetFile} in ${Date.now() - start}ms`);
	// fs.writeFileSync(path.join(publicDir, 'countries.json'), JSON.stringify(countryArr, null, '  '));

	const singleCountryTmpl = path.join(templatesDir, 'country.pug');
	const singleStateTmpl = path.join(templatesDir, 'state.pug');

	const countryFn = pug.compileFile(singleCountryTmpl);
	const stateFn = pug.compileFile(singleStateTmpl);

	await Promise.all(countryArr.map(async (countryData) => {
		const start = Date.now();
		const targetFile = path.join(publicDir, 'countries', countryData.safeName + '.html');
		const countryHtml = countryFn({
			data: countryData,
			$title: countryData.name,
			lastUpdate,
		});
		console.log(`writing to ${targetFile}`);
		await promiseMe(callback => fs.writeFile(targetFile, countryHtml, callback));
		console.log(`wrote to ${targetFile} in ${Date.now() - start}ms`);

		if (countryData.states.length) {
			await Promise.all(countryData.states.map(async (stateData) => {
				if (!stateData.name || !stateData.counties || !stateData.counties.length) {
					return;
				}
				const start = Date.now();
				const targetFile = path.join(publicDir, 'countries', countryData.safeName + '-' + stateData.safeName + '.html');
				const stateHtml = stateFn({
					data: stateData,
					$title: stateData.name + ' - ' + countryData.name,
					lastUpdate,
				});
				await promiseMe(callback => fs.writeFile(targetFile, stateHtml, callback));
				console.log(`wrote to ${targetFile} in ${Date.now() - start}ms`);
			}));
		}
	}));

	console.log(`finished in ${((Date.now() - globalStart) / 1000).toFixed(2)}s`);
};

processGlobalDeaths()
	.then(() => {
		console.log('all done');
	})
	.catch((err) => {
		console.log(err);
		process.exit(1);
	});