weather-scraper/scrapers/iLMeteo.js

const parseTemp = el => {
    return parseFloat(el.textContent)
}
const parsePrec = el => {
    const text = el.textContent
    if (text === '- assenti -') return 0
    return parseFloat(text)
}

const scrapePage = async (browser, url, isToday) => {
    let result = {}

    const page = await browser.newPage()
    await page.goto(url)

    const weatherTable = await page.locator('.weather_table').waitHandle()

    const startTime = isToday ? new Date().getHours() + 1 : 1
    const endTime = 24

    if (isToday) {
        try {
            const row = await weatherTable.$$('.latest_detection')
            const fields = await row[1]?.$$('td')
            const temp = await fields[2].evaluate(parseTemp)
            const prec = await fields[3].evaluate(parsePrec)
            result[startTime - 1] = { temp, prec }
        } catch (error) {
            retult[startTime - 1] = null
            // console.log(error)
        }
    }
    for (let i = startTime; i <= endTime; i++) {
        try {
            const selector = isToday
                ? `.forecast_1h[data-dialogid="${i}"] td`
                : `.forecast_1h[data-dialogid="${1000 + i - 1}"] td`

            const fields = await weatherTable.$$(selector)

            const temp = await fields[2].evaluate(parseTemp)
            const prec = await fields[3].evaluate(parsePrec)
            result[i] = { temp, prec }
        } catch (error) {
            retult[i] = null
            // console.log(error)
        }
    }

    await page.close()

    return result
}

const rearrangeResults = results => {
    for (let i = 0; i < 6; i++) {
        const midnight = results[i][24]
        delete results[i][24]
        results[i + 1][0] = midnight
    }
    delete results[6][24]

    return results
}

const getDaySummary = day => {
    let minTemp = Number.MAX_VALUE
    let maxTemp = Number.MIN_VALUE
    let totPrec = 0
    for (const h in day) {
        minTemp = Math.min(minTemp, day[h].temp)
        maxTemp = Math.max(maxTemp, day[h].temp)
        totPrec += day[h].prec
    }
    return {
        minTemp,
        maxTemp,
        totPrec,
    }
}

export default async browser => {
    const url = text => `https://www.ilmeteo.it/meteo/Pisa/${text}`

    const results = rearrangeResults(
        await Promise.all([
            scrapePage(browser, url(''), true),
            scrapePage(browser, url('domani'), false),
            scrapePage(browser, url('dopodomani'), false),
            scrapePage(browser, url('3'), false),
            scrapePage(browser, url('4'), false),
            scrapePage(browser, url('5'), false),
            scrapePage(browser, url('6'), false),
        ]),
    )

    return {
        today: results[0],
        tomorrow: results[1],
        dayAfterTomorrow: results[2],
        week: results.map(getDaySummary),
    }
}