derma/scrape.js

const puppeteer = require('puppeteer');
const { pathOr, unionWith, prop, eqBy } = require('ramda');
const parse_args = require('minimist');

const graphql_endpoint = 'https://www.facebook.com/api/graphql/';

const get_upcoming_events = pathOr(
  null,
  'data.page.upcoming_events'.split('.'),
);

const merge_edges = unionWith(eqBy(prop('event_id')));

const load_page = async (page, event_page) => {
  try {
    const graphql_data = new Promise((resolve, reject) => {
      page.on('response', async (response) => {
        if (graphql_endpoint === response.request().url()) {
          const text = await response.json();
          const upcoming_events = get_upcoming_events(text);
          if (upcoming_events !== null) {
            resolve(upcoming_events);
          }
        }
      });
    });
    await page.goto(event_page);
    await page.evaluate(() => window.scrollBy(0, window.innerHeight));
    return await graphql_data;
  } catch (e) {
    console.error(e);
  }
};

const argv = parse_args(process.argv.slice(2));
const page_ids = pathOr('', ['page_ids'], argv)
  .split(',')
  .filter((str) => str.length !== 0)
  .map((page_id) => `https://www.facebook.com/${page_id}/events/`);

const event_ids = pathOr('', ['event_ids'], argv)
  .split(',')
  .filter((str) => str.length !== 0)
  .map((event_id) => `https://www.facebook.com/events/${event_id}`);

const get_city_name = (event) =>
  pathOr('', 'event_place.city.contextual_name'.split('.'), event);

const get_event_host = (event) =>
  pathOr('', 'event_place.contextual_name'.split('.'), event);

const edge_to_node = (edge) => edge.node;

const map_event = (edge) => {
  const event = edge.node;
  const ticket_url = pathOr('', ['event_buy_ticket_url'], event);
  const city = get_city_name(event);
  const host = get_event_host(event);
  return {
    date: event.time_range,
    name: event.name,
    event_id: event.id,
    ticket_url,
    location: {
      host: host,
      location: city,
    },
  };
};

(async () => {
  const browser = await puppeteer.launch({
    headless: true,
    args: ['--disable-dev-shm-usage'],
  });

  let events = [];

  for (let page_id of page_ids) {
    const page = await browser.newPage();
    const data = await load_page(page, page_id);
    const edges = data.edges.map(map_event);
    events = merge_edges(edges, events);
  }

  console.log(JSON.stringify(events));
  process.exit();
})();