processHTML.js

const cheerio = require('cheerio')

const processHTML = (html, filterValues, maxDaysBack) => {
  const $ = cheerio.load(html)
  const data = []
  let nextPage = 'https://news.ycombinator.com/' + $('.morelink').attr('href')

  //.each to loop over all class athing elements
  $('.athing').each((index, element) => {
    const title = $(element).find('.titleline > a').text()
    const link = $(element).find('.titleline > a').attr('href')

    // .age carries the exact post time in its title attribute,
    // e.g. title="2026-07-31T09:12:04 1754... "
    const ageTitle = $(element).next().find('.age').attr('title')
    const postedISO = ageTitle ? ageTitle.split(' ')[0] : null

    const titleLower = title.toLowerCase()
    const hasFilterItem = filterValues.some((item) => titleLower.includes(item))

    const daysAgo = postedISO
      ? (Date.now() - new Date(postedISO).getTime()) / (1000 * 60 * 60 * 24)
      : Infinity

    // rows are in reverse-chronological order, so the first row past the
    // cutoff means every later page is older too. Stop paginating there.
    if (daysAgo <= maxDaysBack) {
      data.push({ link, title, postedISO, hasFilterItem })
    } else {
      nextPage = null
    }
  })

  return { data, nextPage }
}

module.exports = processHTML