weather-scraper/scrapers/3Bmeteo.js

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

    //TODO not 100% sure that this would work as it was never tested
    return parseFloat(text)
}

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

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

    const tablePrevisioni = await page
        .locator('.table-previsioni-ora')
        .waitHandle()
    const rows = await tablePrevisioni.$$('.row-table.noPad')

    const startTime = isToday ? new Date().getHours() : 0
    const endTime = 23

    for (let i = startTime; i <= endTime; i++) {
        try {
            const row = rows[rows.length - 1 - (endTime - i)]
            // console.log(await row.evaluate(el => el.children))
            const [rowLeft, rowRight] = await row.$$(':scope > div')

            const codeDiv = (await (await rowLeft.$('.row-table')).$$('div'))[2]
            const [tempDiv, precDiv] = await (
                await rowRight.$('.row-table')
            ).$$('div')

            const code = await codeDiv.evaluate(el => el.textContent.trim())
            const temp = await (await tempDiv.$('span')).evaluate(parseTemp)
            const prec = await (await precDiv.$('span')).evaluate(parsePrec)
            result[i] = { temp, prec, code }
        } catch (error) {
            result[i] = null
            console.log(error)
        }
    }

    await page.close()

    return result
}

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.3bmeteo.com/meteo/pisa/${text}`

    const results = await Promise.all([
        scrapePage(browser, url(''), true),
        scrapePage(browser, url('1'), false),
        scrapePage(browser, url('2'), 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),
    }
}