korting/src/scraper.js

const cheerio = require('cheerio');

function roundMoney(n) {
  return Math.round(n * 100) / 100;
}

// Turns a human-written amount into a plain JS-parsable number string.
// When both separators occur, the last one is the decimal separator
// ("1.299,00" and "1,299.00" both mean 1299.00); a lone comma is decimal
// only when exactly two digits follow it, otherwise it groups thousands.
function normalizeAmount(raw) {
  if (raw == null) return null;
  const match = String(raw).trim().match(/-?\d[\d.,]*/);
  if (!match) return null;
  const str = match[0].replace(/[.,]+$/, '');
  const lastDot = str.lastIndexOf('.');
  const lastComma = str.lastIndexOf(',');
  if (lastDot !== -1 && lastComma !== -1) {
    return lastComma > lastDot
      ? str.replace(/\./g, '').replace(',', '.')
      : str.replace(/,/g, '');
  }
  if (lastComma !== -1) {
    return /,\d{2}$/.test(str) ? str.replace(',', '.') : str.replace(/,/g, '');
  }
  return str;
}

function parsePriceValue(raw) {
  const normalized = normalizeAmount(raw);
  if (normalized == null) return null;
  const value = parseFloat(normalized);
  return Number.isFinite(value) ? roundMoney(value) : null;
}

function searchJsonLdNode(node) {
  if (Array.isArray(node)) {
    for (const item of node) {
      const price = searchJsonLdNode(item);
      if (price != null) return price;
    }
    return null;
  }
  if (node && typeof node === 'object') {
    const type = node['@type'];
    const isProduct = type === 'Product' || (Array.isArray(type) && type.includes('Product'));
    if (isProduct && node.offers) {
      const offers = Array.isArray(node.offers) ? node.offers[0] : node.offers;
      const price = parsePriceValue(offers && offers.price);
      if (price != null) return price;
    }
    if (Array.isArray(node['@graph'])) {
      const price = searchJsonLdNode(node['@graph']);
      if (price != null) return price;
    }
  }
  return null;
}

function findPriceInJsonLd($) {
  const scripts = $('script[type="application/ld+json"]').toArray();
  for (const el of scripts) {
    let data;
    try {
      data = JSON.parse($(el).contents().text());
    } catch {
      continue;
    }
    const price = searchJsonLdNode(data);
    if (price != null) return price;
  }
  return null;
}

function findPriceInMeta($) {
  const selectors = ['meta[property="product:price:amount"]', 'meta[property="og:price:amount"]'];
  for (const selector of selectors) {
    const price = parsePriceValue($(selector).attr('content'));
    if (price != null) return price;
  }
  return null;
}

// Thousands-grouped amounts first, then a plain amount, so "€1.299,00"
// is not truncated to "€1.29".
const TEXT_PRICE_RE = /€\s?(\d{1,3}(?:[.,]\d{3})+[.,]\d{2}|\d+[.,]\d{2})/;

function findPriceInText($) {
  const text = $('body').clone().find('script, style, noscript').remove().end()
    .text();
  const match = text.match(TEXT_PRICE_RE);
  return match ? parsePriceValue(match[1]) : null;
}

function extractPrice(html) {
  const $ = cheerio.load(html);
  let price = findPriceInJsonLd($);
  if (price != null) return { price, method: 'json-ld' };
  price = findPriceInMeta($);
  if (price != null) return { price, method: 'meta' };
  price = findPriceInText($);
  if (price != null) return { price, method: 'regex' };
  return { price: null, method: null };
}

const FETCH_TIMEOUT_MS = 15000;
const USER_AGENT = 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 '
  + '(KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36';

function defaultFetch(url) {
  return fetch(url, {
    signal: AbortSignal.timeout(FETCH_TIMEOUT_MS),
    headers: { 'User-Agent': USER_AGENT },
  });
}

async function fetchAndExtractPrice(url, { fetchImpl = defaultFetch } = {}) {
  let response;
  try {
    response = await fetchImpl(url);
  } catch (err) {
    return { price: null, method: null, error: err };
  }
  if (!response.ok) {
    return { price: null, method: null, error: new Error(`HTTP ${response.status}`) };
  }
  const html = await response.text();
  const { price, method } = extractPrice(html);
  return { price, method, error: null };
}

module.exports = { extractPrice, fetchAndExtractPrice, parsePriceValue };