cities.js

const express = require("express");
const cheerio = require("cheerio");
const httpism = require("httpism");

if (process.env.PORT) {
  startApp({ port: process.env.PORT, fetchCities })
} else {
  module.exports = startApp
}

async function startApp({ port, fetchCities }) {
  const app = express();
  app.get("/", (_, res) => fetchCities().then(html => res.json(parseCities(html))))
  await new Promise(function(resolve, reject) {
    app.listen(port, err => err ? reject(err) : resolve());
  })
}

function fetchCities() {
  return httpism.get("https://en.wikipedia.org/wiki/List_of_metropolitan_areas_in_Asia")
}

function parseCities(html) {
  const $ = cheerio.load(html);
  const table = $("h2:contains('List') + table")

  const headings = table.find("th").toArray()
    .map((th, index) => ({ text: $(th).text().trim(), index }))

  const columns = [
    {
      heading: /Area/, parse: td => ({
        name: td.text().trim(),
        url: `https://en.wikipedia.org${td.find('a').attr('href')}`
      })
    },
    { heading: /Country/,     parse: td => ({ country: td.text().trim() }) },
    { heading: /Population/,  parse: parsePopulation },
    { heading: /Demographia/, parse: parsePopulation }
  ].map(parser => ({
    heading: headings.find(h => parser.heading.test(h.text)).index,
    parse: parser.parse
  }))

  return table.find('tr:not(:has(th))').toArray()
    .map(tr => columns.reduce((city, column) => Object.assign(
      column.parse($(tr).find("td").eq(column.heading)), city
    ), {}))
}

function parsePopulation(td) {
  const population = Number(td.text().trim().replace(/\[.+\]+/g, '').replace(/,/g, ''))
  return population > 0 ? { population } : {}
};