init.js

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