scrape.js

'use strict';
var fs = require('fs');
var request = require('request');
var Q = require('q');
var cheerio = require('cheerio');

function parseListen(txt) {
  return parseInt(txt.listens.replace(new RegExp(',', 'g'), ''), 10);
}

function readJson(name) {
  var f = fs.readFileSync(name, 'utf-8');

  return JSON.parse(f);
}

function createUrl(text) {
  var textArr = text.split(' ');

  return 'http://www.last.fm/search?q=' + textArr.join('+');
}

function getHtml(url) {
  var deferred = Q.defer();

  request(url, function(err, res, body) {
    if (!err && res.statusCode === 200) {
      deferred.resolve(body);
    }
  });

  return deferred.promise;
}

function scrapeListens(html) {
  var $ = cheerio.load(html);
  var selected = $('p.grid-items-item-aux-text');

  if (selected[0] &&
      selected[0].children[0] &&
      selected[0].children[0].data) {
    return selected[0].children[0].data.trim();
  }
  return '0';
}

function sortByListens(arr) {
  return arr.sort(function(a, b) {
    var la = parseListen(a);
    var lb = parseListen(b);

     if (la < lb) {
       return 1;
     }
      if (la > lb) {
        return -1;
      }
      return 0;
    });
}

function makeTable(arr) {
  var str = arr.map(function(obj) {
    return '|' + obj.name + '|' + obj.listens + '|\n';
  }).join('');

  return '|Band Name | Unique Listeners on last.fm|\n' +
    '|---------|--------|\n' + str;
}

function writeTableFor(json) {
  var outName = json.replace('json', 'md');
  var names = readJson(json);
  var urls = names.map(createUrl);
  var promisesArr = urls.map(function(url) {
    return getHtml(url)
    .then(scrapeListens);
  });

  Q.all(promisesArr)
  .then(function(listens) {
    return listens.map(function(num, i) {
      return {
        name: names[i],
        listens: num
      };
    });
  })
  .then(sortByListens)
  .then(makeTable)
  .then(function(txt) {
    fs.writeFileSync(outName, txt);
    console.log('done writing ', outName);
  })
  .fail(function(err) {
    console.log(err);
  });
}

writeTableFor('ec.json');
writeTableFor('untold.json');