extract-phone-dataset-from-selectra.info.js

https://selectra.info/telecom/guides/comparatifs/telephones-mobiles

function getPhoneData(tableDivEl) {
  try {
    let rawData = Array.from(tableDivEl.querySelectorAll("tr"))
      .map(tr => ({
        key: tr.querySelector("th").textContent,
        value: parseInt(tr
            .querySelector("td")
            .textContent
            .replace(/\s+/g, " ")
            .replace(/.*: ?/, "")
          , 10)
      }))

    let input = {};

    for (let item of rawData) {
      if (item.value)
        input[item.key] = item.value;
    }

    let paragraph = tableDivEl
    let price;
    let euroSymbolFound;
    do {
      if (!euroSymbolFound)
        euroSymbolFound = paragraph.querySelector("img[alt=euro]")

      let priceEl = paragraph.querySelector("strong");
      if (priceEl && euroSymbolFound) {
        let priceText = priceEl.textContent
        let priceDigits = priceText.replace(/[^0-9]/g, "")

        price = parseInt(priceDigits, 10)
      }
      paragraph = paragraph.nextElementSibling
    } while (!price && paragraph)

    return { input, output: { price } }
  } catch (e) {
    return null
  }
}

function normalize(data) {
  let max = {};
  for (key in data[0])
    max[key] = Math.max(...data.map(x => x[key]))

  for (item of data) {
    for (key in item) {
      item[key] /= max[key]
      if (item[key] !== 0 && !item[key])
        delete item[key]
    }
  }    
}


var trainingData = Array
  .from(document.querySelectorAll("div.table-responsive"))
  .map(getPhoneData)
  .filter(x => !!x)

normalize(trainingData.map(d => d.input))
normalize(trainingData.map(d => d.output))

console.log(JSON.stringify(trainingData))
console.log(trainingData.length)