parserWB2

/*
Наш канал про Таблицы и скрипты в Telegram: https://t.me/google_sheets
Оглавление канала (конечно, в Таблице): https://docs.google.com/spreadsheets/d/1u0aRlwb0rQHbjE9avbgCqIJSQ3EvPfTxT-h54jTQh8I/edit#gid=0&range=A1:B1
Чат: https://t.me/google_spreadsheets_chat
Донаты: https://t.me/google_spreadsheets_chat/160663
Написать: https://t.me/namokonov
*/

const regExps = {
  name: /<meta itemprop="name" content="(.+?)">/,
  price: /<meta itemprop="price" content="(.+?)">/,
  curr: /<meta itemprop="priceCurrency" content="(.+?)">/,
  orders: /ordersCount":(\d+)/,
  q: /quantity":(\d+)/,
  brand: /<meta itemprop="brand" content="(.+?)">/,
  rating: /<meta itemprop="ratingValue" content="(.+?)">/,
  reviewCount: /<meta itemprop="reviewCount" content="(.+?)">/,
  image: /<meta itemprop="image" content="(.+?)"/
}

const sheetNames = {
  shMain: 'URLS',
  shPaste: 'RESULT',
  shLog: 'LOG'
}

const step = 1000;

function onOpen(e) {
  SpreadsheetApp.getUi()
    .createMenu(">–]> В ЭТОМ МЕНЮ СКРИПТЫ")
    .addItem("Парсим!", "parserWb2")
    .addToUi();
}

function parserWb2() {
  const ss = SpreadsheetApp.getActive();
  const sh = ss.getSheetByName(sheetNames.shMain);
  const shPaste = ss.getSheetByName(sheetNames.shPaste);
  const shLog = ss.getSheetByName(sheetNames.shLog);
  const d = sh.getDataRange().getValues();
  const urls = d.slice(1).filter(h => h[0]).map(h => h[0]);
  let requests = requestPrepare(urls);

  shPaste.getRange("A2:L").clearContent();

  while (requests.length) {
    const startTime = new Date();
    const partRequests = requests.splice(0, step);
    const r = UrlFetchApp.fetchAll(partRequests);
    let arr = [];

    for (n in r) {
      const response = r[n].getContentText();
      const row = responseParse(response);
      const url = urls[n];
      const date = new Date();
      row.unshift(date, url);
      arr = arr.concat([row]);
    }

    shLog.appendRow([new Date(), partRequests.length, arr.filter(r => r[2]).length, secondsBetweenDates(startTime)]);
    shPaste.getRange(shPaste.getLastRow() + 1, 1, arr.length, arr[0].length).setValues(arr);
    SpreadsheetApp.flush();
  }
}

function responseParse(r) {
  let row = [];
  for (y in regExps) {
    const regExp = regExps[y];
    let result = m(r, regExp);
    result = y == 'price' ? result * 1 : result;
    row.push(result);
  }
  return row;
}


function requestPrepare(urls) {
  return urls.map(url => ({ 'url': url, 'muteHttpExceptions': true }));
}

function createUrlImage(y) {
  return y ? `https://img1.wbstatic.net/big/new/${Math.floor(y / 10000) * 10000}/${y}-1.jpg` : 0;
}

function m(response, regExp) {
  regExp = new RegExp(regExp);
  return response.match(regExp) ? response.match(regExp)[1] : 0;
}

function secondsBetweenDates(st) {
  const et = new Date();
  var dif = st.getTime() - et.getTime();
  var Seconds_from_T1_to_T2 = dif / 1000;
  return Math.abs(Seconds_from_T1_to_T2);
}