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 => res.text()).then(data =>
   7                 jsonic( // jsonic is used because JSON.parse() requires strict JSON and eval() allows for remote code execution
   8                         cheerio.load(data, null, false)                            // parse HTML
   9                         ("script")                                                 // find script tags
  10                         .toArray()                                                 // convert cheerio list to array
  11                         .map(script => script.children[0]?.data)                   // map script tags to their inline code
  12                         .find(script => script?.startsWith("AF_initDataCallback")) // find script that contains init data
  13                         .slice("AF_initDataCallback(".length, -");".length)        // remove call to init function
  14                 ).data[31][0][12][2].map(elem => new Object({ // map the parts of the init data we know/care about to something readable
  15                         image: {
  16                                 url: elem[1][3][0],
  17                                 size: {
  18                                         width: elem[1][3][2],
  19                                         height: elem[1][3][1],
  20                                 },
  21                         },
  22                         preview: {
  23                                 url: elem[1][2][0],
  24                                 size: {
  25                                         width: elem[1][2][2],
  26                                         height: elem[1][2][1],
  27                                 },
  28                         },
  29                         color: elem[1][6], // average color of the image, probably (used as placeholder while loading the image)
  30                         link: elem[1][9][2003][2],
  31                         title: elem[1][9][2003][3], // there is some more data in elem[1][9] that could potentially be useful
  32                 }))
  33         )
  34
  35 /*
  36
  37 In case google makes changes, here are some snippets used to reverse engineer the format:
  38
  39 1. Find which script contains the init data (use the query astolfo+images for this)
  40 -----------------------------------------------------------------------------------
  41
  42         scripts.find(script => script.search("https://steamcdn-a.akamaihd.net/steamcommunity/public/images/items/622220/f4d2d4074167411a7e15b9a845cf18b434c02af3.jpg") >= 0)
  43
  44 2. Reverse engineer format of init data passed to AF_initDataCallback
  45 ---------------------------------------------------------------------
  46
  47 const findStrings = (obj, path = "") => {
  48         let found = []
  49
  50         for (k in obj) {
  51                 let v = obj[k]
  52                 let t = typeof v
  53                 let p = path + "." + k
  54
  55                 if (t == "object")
  56                         found = found.concat(findStrings(v, p))
  57                 else if (t == "string")
  58                         found.push([v, p])
  59         }
  60
  61         return found
  62 }
  63
  64         console.log(findStrings(initData))
  65
  66 */