Get status report for Facebook user's own liked pages based on page's latest post time.

fb-pages-status.js

/**
 * Script for generating Facebook user's own liked pages status report as JSON. The report consists of active, inactive,
 * unavailable page Ids validated based on page's latest post time.
 *
 * Unfortunately Facebook doesn't support "unlike" a page via API. So, these IDs should be used with a userscript.
 */

const got = require("got");
const cheerio = require("cheerio");
const fs = require("fs");

/**
 * Temporary Facebook Graph API access token acquired from
 * [Graph API Explorer]{@link https://developers.facebook.com/tools/explorer}. "user_likes" permission is needed.
 *
 * @const {string}
 */
const accessToken = "Paste access token here";

/**
 * Main function for this script.
 *
 * @async
 * @param {boolean} [readFromCachedPages = false] - If true; pages.json will be read from filesystem if there is any.
 */
(async function (readFromCachedPages = false) {
  let pageInfos;

  if (readFromCachedPages) pageInfos = require("./pages.json");

  pageInfos = pageInfos || await getLikedPageInfos();
  writeToFile(pageInfos, "pages.json");

  const report = await reportPageStatus(pageInfos);
  writeToFile(report, "report.json");
})();

/**
 * Uses Facebook Graph API to get all the /me liked pages id and name.
 *
 * @async
 * @param {object} [pageInfos={}] - An object populated with {id: name} of all the liked pages.
 * @returns {Promise<object>} - Promise object represents pageInfos object.
 */
async function getLikedPageInfos(pageInfos = {}) {
  let url = `https://graph.facebook.com/me?fields=likes.summary(true)&access_token=${accessToken}&pretty=0&limit=100`;
  let totalPageCount;

  do {
    const [error, response] = await forThis(got(url, {json: true}));

    if (error) {
      console.log(error);
      break;
    }

    const likes = response.body;
    const data = likes["data"];

    for (const {name, id} of data) pageInfos[id] = name;

    if (totalPageCount === undefined) totalPageCount = likes["summery"]["total_count"];
    url = likes["paging"]["next"];
  } while (url);

  console.log(`Total (${Object.keys(pageInfos).length} / ${totalPageCount}) liked page info. acquired.`);

  return pageInfos;
}

/**
 * Scrape FB for determining every pages status.
 *
 * @param {object} pageInfos - pageInfo object with page Ids.
 * @param {Date.} [date=] - UTC time for an acceptable date for page's latest post. If the time of latest post is
 * before the acceptable date, then the page count as inactive. Default value is first day of current year.
 * @returns {Promise<Object>} - Promise object of report object.
 */
async function reportPageStatus(pageInfos, date = Date.UTC(new Date().getFullYear(), 0)) {
  const unavailable = [];
  const inactive = [];
  const active = [];

  const ids = Object.keys(pageInfos);

  for (const id of ids) {
    console.log(`Visiting ${pageInfos[id]} ...`);

    const pageState = await getPageState(id, date);
    const page = {[pageInfos[id]]: `https://fb.com/pg/${id}/posts`};

    if (pageState === null) {
      console.log("|- Unavailable!");
      unavailable.push(page);
    } else if (pageState) {
      console.log("|- Inactive.");
      inactive.push(page);
    } else {
      console.log("|- Active.");
      active.push(page);
    }
  }

  return {
    [`${unavailable.length} Unavailable`]: unavailable,
    [`${inactive.length} Inactive`]: inactive,
    [`${active.length} Active`]: active,
  };
}

async function getPageState(id, date) {
  const [error, response] = await forThis(got(`https://www.facebook.com/pg/${id}/posts`, {
    headers: {"user-agent": "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:15.0) Gecko/20100101 Firefox/15.0.1"}
  }));

  if (error) {
    console.log(`Skipped - $id scraping failed!`);
    return null;
  }

  const postTimes = extractPostTimes(response.body);

  return isInactive(postTimes, date);
}

function extractPostTimes(html) {
  const $ = cheerio.load(html);

  return $(".userContentWrapper")
      .map((index, element) => $(element))
      .get()
      .map(post => post.find("abbr").attr("data-utime"));
}

function isInactive(postTimes, date) {
  let latestPost = postTimes[0];
  if (!latestPost) return true;

  // First post can be an older featured post. In this case, the second post is the latest post.
  if (postTimes.length > 1 && latestPost < postTimes[1]) latestPost = postTimes[1];

  return latestPost <= date / 1000;
}

/**
 * Write file to local storage.
 *
 * @param {object} data - JSON object that will be written in a file.
 * @param {string} fileName - File name for the storage.
 */
function writeToFile(data, fileName) {
  fs.writeFile(fileName, JSON.stringify(data, null, 4), "utf8", error => {
    console.log(error || `./${fileName} saved!`);
  });
}

/**
 * Functional style Promise error handling.
 *
 * @param {Promise} promise
 * @returns {Promise<*[]>} - Either successful response or error.
 */
async function forThis(promise) {
  try {
    const response = await promise;
    return [null, response];
  } catch (error) {
    return [error, null];
  }
}

package.json

{
  "name": "fb-pages-status",
  "version": "1.0.0",
  "description": "Get status report for Facebook user's own liked pages based on page's latest post time.",
  "main": "fb-pages-status.js",
  "scripts": {
    "test": "echo \"Error: no test specified\" && exit 1"
  },
  "keywords": ["facebook", "pages", "graph api", "scrape", "likes"],
  "author": "Nissan Ahmed",
  "license": "ISC",
  "dependencies": {
    "cheerio": "^1.0.0-rc.3",
    "got": "^9.6.0"
  }
}

userscript.js

// ==UserScript==
// @name One Click Multiple Tab Open
// @description Open 20 tabs from JSON pageIDs in facebook by pressing "x".
// @namespace ni554n.github.io
// @include https://*.facebook.com/*
// @grant GM_getValue
// @grant GM_setValue
// @grant GM_openInTab
// ==/UserScript==

// Copy paste an active / inactive pageInfo array from report.json.
const pageInfos = [];

document.addEventListener('keydown', (event) => {
  if (event.key === 'x') {
    event.preventDefault();

    let count = GM_getValue("count", 0);
    
    for (let i = 0; i < 20; i++) {
      GM_openInTab(`${Object.values(pageInfos[count++])[0]}`, true);
    }

    GM_setValue("count", count);
  }
});