tusubtitulo.com scrapper module

index.js

var got = require("got");
var cheerio = require("cheerio");
var Fuzzy = require("fuzzy-search");

var base_url = "https://www.tusubtitulo.com";

var links = [];

module.exports = {
  index:  index,
  detail: detail,
  search: search,
  episodes: episodes
}

function index(){
  var subs_index_url = base_url+'/series.php';
  return got(subs_index_url).then(onIndexLoaded);

  function onIndexLoaded(res){
    var html = res.body;
    $ = cheerio.load(html);

    links = $("#showindex td:not([colspan]) a");
    var data = links.map((i, el) => {
      var link = $(el).attr("href");
      var name = $(el).text();
      var id   = link.split("/");
      id = id[id.length-1];

      return {name, link: base_url+link, id}
    }).get();

    return data;
  }
}

function detail(show_id){
  var detail_url = base_url+"/show/"+show_id;
  return got(detail_url).then(onDetailLoaded);

  function onDetailLoaded(res){
    var html = res.body;
    var $    = cheerio.load(html);

    var show_el       = $("#contenido");
    var title         = show_el.find("p.titulo").text();
    var numbers_label = show_el.find("td.SectionTitle").text();
    var numbers       = numbers_label.match(/\d+/g);
    var episodes      = numbers[1];
    var seasons       = numbers[0];

    return {title, seasons, episodes};
  }
}

function search(text){
  return index().then(data => {
    var fuzzy = new Fuzzy(data, ["name"], {
      caseSensitive: false,
      sort: true
    });

    var search_res = fuzzy.search(text);
    return detail(search_res);
  })
}

function episodes(show_id, season){
  var url = base_url+'/ajax_loadShow.php?show='+show_id+'&season='+season;
  return got(url).then(onSeasonLoaded);

  function onSeasonLoaded(res){
    var html = res.body;
    var $ = cheerio.load(html);

    var eps = [];


    $("table").each((index, el) => {
      // cada tabla es un episodio
      var links = [];
      var trs = $(el).find("tr");

      var title = $(trs[0]).text().trim();

      trs.each((index, el) => {
        if(index < 2){
          return;
        }
        var trs    = $(el);
        var lang   = trs.find("td:nth-last-child(3)").text().trim();
        var link   = trs.find("td:nth-last-child(1) a").attr("href");

        if(link){
          links.push({lang, link});
        }
      });

      eps.push({title, links});
    });

    return eps;
  }
}

package.json

{
  "name": "tusubtitulo-scrapper",
  "version": "0.1.0",
  "description": "Scrapper module for tusubtitulo.com",
  "main": "index.js",
  "scripts": {
    "test": "echo \"Error: no test specified\" && exit 1"
  },
  "author": "fuken",
  "license": "MIT",
  "dependencies": {
    "cheerio": "^0.20.0",
    "fuzzy-search": "^1.3.5",
    "got": "^6.3.0"
  }
}