scrape-url/dist/index.cjs

Scrape URL with CSS selectors

'use strict';

var cheerio = require('cheerio');
var wretch = require('wretch');

async function scrape(method, options, selectors) {
  let url = options.url;
  if (!/\w:\/\//.test(options.url)) {
    url = `http://${options.url}`;
  }
  let normalizedSelectors = selectors;
  if (typeof selectors === "string") {
    normalizedSelectors = [selectors];
  }
  const body = method === "POST" ? await wretch(url).post(options.body || {}).text() : await wretch(url).get().text();
  const $ = cheerio.load(body);
  const result = [].concat(
    normalizedSelectors.map((selector) => {
      const elements = [];
      const selected = $(selector);
      selected.each((_, element) => {
        elements.push($(element));
      });
      return elements;
    })
  );
  return result.reduce((acc, curr) => {
    if (Array.isArray(curr)) {
      return acc.concat(curr);
    }
    return acc;
  }, []);
}
async function get(options, selectors) {
  return scrape("GET", options, selectors);
}
async function post(options, selectors) {
  return scrape("POST", options, selectors);
}

exports.get = get;
exports.post = post;