derma/scrape.js

const puppeteer = require('puppeteer');
const { pathOr, unionWith, prop, eqBy, maxBy } = require('ramda');
const parse_args = require('minimist');
const url = require('url');
const path = require('path');
const fs = require('fs').promises;
const gm = require('gm').subClass({ imageMagick: true });

const graphql_endpoint = 'https://www.facebook.com/api/graphql/';
const facebook_event_url = (event_id) =>
  `https://www.facebook.com/events/${event_id}/`;

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 load_event = async (page, event_id) => {
  try {
    const image_data = new Promise((resolve, reject) => {
      const images = [];
      page.on('response', async (response) => {
        const response_url = response.request().url();
        const { pathname } = new URL(response_url);
        const ext = path.extname(pathname);
        if (ext === '.jpg') {
          const image = await response.buffer();
          images.push(image);
        }
      });
      page.on('domcontentloaded', async () => {
        resolve(images);
      });
    });

    await page.goto(facebook_event_url(event_id));
    const images = await image_data;
    const image = images.reduce((res, image) =>
      maxBy((item) => item.length, res, image),
    );
    return { image };
  } catch (e) {
    console.error(e);
  }
};

const write_image = (path, image) =>
  fs.writeFile(path, image, { encoding: null });

const gm_write = (image, path) => {
  return new Promise((resolve, reject) =>
    image.write(path, (err) => (!err ? resolve() : reject(err))),
  );
};

const write_resized = async (image_path, original) => {
  const image = gm(original);
  const size = await new Promise((resolve, reject) => {
    image.size((err, value) => (!err ? resolve(value) : resolve(null)));
  });

  if (size === null) {
    throw new Error('Could not get image.');
  }

  let { height: y, width: x } = size;

  if (y % 2 === 1) {
    y = y + 1;
  }

  if (x % 2 === 1) {
    x = x + 1;
  }

  image.resize(x, y);

  if (y > x) {
    const z = (y - x) / 2;
    image.crop(x, x, 0, z);
  }

  if (y < x) {
    const z = (x - y) / 2;
    image.crop(y, y, z, 0);
  }

  return gm_write(image, image_path);
};

const save_images = async ({ image = null, event_id }) => {
  if (image === null) {
    return [];
  }
  const original_path = `./events/img/${event_id}.jpg`;
  const resized_path = `./events/img/square-${event_id}.jpg`;
  const original = write_image(original_path, image);
  const resized_square = write_resized(resized_path, image);
  try {
    const res = await Promise.all([original, resized_square]);
    return { original: original_path, square: resized_path };
  } catch (err) {
    console.error(err);
    return { original: null };
  }
};

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 () => {
  fs.mkdir('./events/img', { recursive: true }).catch(console.error);
  const browser = await puppeteer.launch({
    headless: false,
    args: ['--disable-dev-shm-usage'],
  });

  let events = [];

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

    events = await Promise.all(
      events.map(async (event) => {
        const event_page = await browser.newPage();
        const event_data = await load_event(event_page, event.event_id);
        return {
          ...event_data,
          ...event,
        };
      }),
    );

    events = await Promise.all(
      events.map(async (event) => {
        const images = await save_images(event);
        delete event.image;
        return {
          images,
          ...event,
        };
      }),
    );
  }
  console.log(events);
})();

/* (async () => {
 *   const image_path = './events/img/439439046887956.jpg';
 *   const image = await fs.readFile(image_path);
 *   write_resized(image_path, image);
 * })(); */