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                 cheerio.load(data, null, false)                                               // parse HTML
   8                         ("script")                                                                // find script tags
   9                         .toArray()                                                                // convert cheerio list to array
  10                         .map(script => script.children[0]?.data)                                  // map script tags to their inline code
  11                         .filter(script => script?.startsWith("AF_initDataCallback"))              // find script that contains init data
  12                         .map(script => script.slice("AF_initDataCallback(".length, -");".length)) // remove call to init function
  13                         .map(jsonic)                                                              // jsonic is used because JSON.parse() requires strict JSON and eval() allows remote code execution
  14                         .find(data => data.key == "ds:1")                                         // for some reason there are two init datas, one is empty tho
  15                         .data[31][0][12][2].map(elem => elem[1] && new Object({                   // map the parts of the init data we know/care about to something readable
  16                                 image: {
  17                                         url: elem[1][3][0],
  18                                         size: {
  19                                                 width: elem[1][3][2],
  20                                                 height: elem[1][3][1],
  21                                         },
  22                                 },
  23                                 preview: {
  24                                         url: elem[1][2][0],
  25                                         size: {
  26                                                 width: elem[1][2][2],
  27                                                 height: elem[1][2][1],
  28                                         },
  29                                 },
  30                                 color: elem[1][6],          // probably average color of the image (used as placeholder while loading the image)
  31                                 link: elem[1][9][2003][2],
  32                                 title: elem[1][9][2003][3], // there is some more data in elem[1][9] that could potentially be useful
  33                         }))
  34                         .filter(elem => elem)
  35         )
  36
  37 module.exports.searchRandom = (query, userAgent) => module.exports.search(query, userAgent)
  38         .then(results => results[Math.floor(Math.random() * results.length))
  39
  40 /*
  41
  42 In case google makes changes, here are some snippets used to reverse engineer the format:
  43
  44 1. Find which script contains the init data (use the query astolfo+images for this)
  45 -----------------------------------------------------------------------------------
  46
  47         scripts.find(script => script.search("https://steamcdn-a.akamaihd.net/steamcommunity/public/images/items/622220/f4d2d4074167411a7e15b9a845cf18b434c02af3.jpg") >= 0)
  48
  49 2. Reverse engineer format of init data passed to AF_initDataCallback
  50 ---------------------------------------------------------------------
  51
  52 const findStrings = (obj, path = "") => {
  53         let found = []
  54
  55         for (k in obj) {
  56                 let v = obj[k]
  57                 let t = typeof v
  58                 let p = path + "." + k
  59
  60                 if (t == "object")
  61                         found = found.concat(findStrings(v, p))
  62                 else if (t == "string")
  63                         found.push([v, p])
  64         }
  65
  66         return found
  67 }
  68
  69         console.log(findStrings(initData))
  70
  71 */