init.js

   1 const fetch = require("node-fetch");
   2 const cheerio = require("cheerio");
   3 const jsonic = require("jsonic");
   4
   5 /*
   6 In case google makes minor changes, here are some snippets used to reverse engineer the format:
   7
   8
   9 Find which script to use (use the query astolfo+images for this)
  10 ----------------------------------------------------------------
  11
  12         .then(scripts => scripts.filter(script => script.children[0]?.data?.search("http") >= 0))
  13         .then(scripts => scripts.reduce((a, b, i) => b.children[0]?.data?.search("https://steamcdn-a.akamaihd.net/steamcommunity/public/images/items/622220/f4d2d4074167411a7e15b9a845cf18b434c02af3.jpg") >= 0 ? i : a), -1)
  14
  15 Reverse engineer data passed to AF_initDataCallback
  16 ---------------------------------------------------
  17
  18 const findStrings = (obj, path = "") => {
  19         let found = [];
  20
  21         for (k in obj) {
  22                 let v = obj[k];
  23                 let t = typeof v;
  24                 let p = path + "." + k
  25
  26
  27                 if (t == "object")
  28                         found = found.concat(findStrings(v, p));
  29                 else if (t == "string")
  30                         found.push([v, p]);
  31         }
  32
  33         return found;
  34 };
  35
  36         .then(findStrings)
  37
  38 Dump data
  39 ---------
  40         const util = require("util");
  41
  42         .then(obj => util.inspect(obj, {showHidden: false, depth: 3, colors: true}))
  43         .then(console.log)
  44 */
  45
  46 const makeImage = elem => {
  47         return {
  48                 url: elem[0],
  49                 size: {
  50                         width: elem[1],
  51                         height: elem[2],
  52                 }
  53         }
  54 };
  55
  56 module.exports.search = (query, userAgent = "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:97.0) Gecko/20100101 Firefox/97.0") => {
  57         return fetch("https://www.google.com/search?tbm=isch&q=" + encodeURIComponent(query), {headers: {"User-Agent": userAgent}})
  58                 .then(res => res.text())
  59                 .then(data => cheerio.load(data, null, false))
  60                 .then(content => content("script"))
  61                 .then(scripts => scripts.toArray())
  62                 .then(scripts => scripts.map(script => script.children[0]?.data))
  63                 .then(scripts => scripts.filter(script => script?.search("http") >= 0))
  64                 .then(scripts => scripts[4])
  65                 .then(script => script.slice("AF_initDataCallback(".length, script.length - ");".length))
  66                 .then(jsonic)
  67                 .then(data => data.data[31][0][12][2])
  68                 .then(data => data.map(elem => elem[1]))
  69                 .then(data => data.map(elem => new Object({
  70                         preview: makeImage(elem[2]),
  71                         image: makeImage(elem[3]),
  72                         color: elem[6],
  73                         link: elem[9][2003][2],
  74                         title: elem[9][2003][3],
  75                 })));
  76 }