xml-to-yaml.js

/**
 * Transform RSS-style XML with geolocation data to YAML
 *
 * {@link https://yaml.org}
 *
 * ```
 * mkdir -p data
 * yarn add hard-rejection cheerio he fast-xml-parser js-yaml
 * # ...
 * node xml-to-yaml.js
 * ```
 */
require('hard-rejection/register')

const cheerio = require('cheerio')
const he = require('he')
const path = require('path')
const xml = require('fast-xml-parser')
const yaml = require('js-yaml')
const { readFile, writeFile } = require('fs').promises

;(async () => {
  const data = await readFile(path.join(__dirname, 'data.xml'), 'utf8')

  if (!xml.validate(data)) {
    throw new Error('data is invalid XML')
  }

  const addressTag = /<\/?address>/g
  const brTag = /<br\s*\/?>/g
  const zipPattern = /(\d{5})(-\d+)?$/

  const items = xml.parse(data).rss.channel.item.reduce((memo, item) => {
    const name = he.decode(item.title)
    if (!(name in memo)) {
      memo[name] = {
        name,
        links: [item.link],
        locations: [],
      }
    }

    if (
      typeof item['geo:lat'] !== 'number' ||
      typeof item['geo:long'] !== 'number'
    ) {
      throw new Error(`${item.title} lacks coordinates`)
    }

    const address = item['content:encoded']
      .replace(addressTag, '')
      .split(brTag)
      .map(s => s.trim())
      .filter(Boolean)
    let zip = address[address.length - 1].match(zipPattern)[1]

    zip = !Number.isNaN(parseInt(zip, 10)) ? parseInt(zip, 10) : zip

    memo[name].locations.push({
      name: item.description ? he.decode(item.description) : name,
      address: {
        street: address[address.length === 3 ? 1 : 0],
        zip,
      },
      geo: {
        latitude: item['geo:lat'],
        longitude: item['geo:long'],
      },
    })

    return memo
  }, {})

  const writes = Object.keys(items).reduce((memo, name) => {
    const filename = path.join(
      __dirname,
      'data',
      name
        .toLowerCase()
        .replace(/\.|&/g, ' ')
        .replace(/’/g, '')
        .trim()
        .replace(/ +/g, '-') + '.yml'
    )

    return memo.concat([
      [filename, writeFile(filename, yaml.safeDump(items[name]))],
    ])
  }, [])

  await Promise.all(writes.map(([, x]) => x))

  console.log(writes.map(([filename]) => filename).join('\n'))
})()