init.js

   1 const fetch = require("node-fetch")
   2 const cheerio = require("cheerio")
   3 const jsonic = require("jsonic")
   4
   5 const debug = arg => {
   6         console.log(arg)
   7         return arg
   8 }
   9
  10 module.exports.search = (query, userAgent = "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:97.0) Gecko/20100101 Firefox/97.0") =>
  11         fetch("https://www.google.com/search?tbm=isch&q=" + encodeURIComponent(debug(query)), {headers: {"User-Agent": userAgent}}).then(res => res.text()).then(data =>
  12                 cheerio.load(data, null, false)                                               // parse HTML
  13                         ("script")                                                                // find script tags
  14                         .toArray()                                                                // convert cheerio list to array
  15                         .map(script => script.children[0]?.data)                                  // map script tags to their inline code
  16                         .filter(script => script?.startsWith("AF_initDataCallback"))              // find script that contains init data
  17                         .map(script => script.slice("AF_initDataCallback(".length, -");".length)) // remove call to init function
  18                         .map(jsonic)                                                              // jsonic is used because JSON.parse() requires strict JSON and eval() allows remote code execution
  19                         .find(data => data.key == "ds:1")                                         // for some reason there are two init datas, one is empty tho
  20                         .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
  21                                 image: {
  22                                         url: elem[1][3][0],
  23                                         size: {
  24                                                 width: elem[1][3][2],
  25                                                 height: elem[1][3][1],
  26                                         },
  27                                 },
  28                                 preview: {
  29                                         url: elem[1][2][0],
  30                                         size: {
  31                                                 width: elem[1][2][2],
  32                                                 height: elem[1][2][1],
  33                                         },
  34                                 },
  35                                 color: elem[1][6],          // probably average color of the image (used as placeholder while loading the image)
  36                                 link: elem[1][9][2003][2],
  37                                 title: elem[1][9][2003][3], // there is some more data in elem[1][9] that could potentially be useful
  38                         }))
  39                         .filter(elem => elem)
  40         )
  41 /*
  42
  43 In case google makes changes, here are some snippets used to reverse engineer the format:
  44
  45 1. Find which script contains the init data (use the query astolfo+images for this)
  46 -----------------------------------------------------------------------------------
  47
  48         scripts.find(script => script.search("https://steamcdn-a.akamaihd.net/steamcommunity/public/images/items/622220/f4d2d4074167411a7e15b9a845cf18b434c02af3.jpg") >= 0)
  49
  50 2. Reverse engineer format of init data passed to AF_initDataCallback
  51 ---------------------------------------------------------------------
  52
  53 const findStrings = (obj, path = "") => {
  54         let found = []
  55
  56         for (k in obj) {
  57                 let v = obj[k]
  58                 let t = typeof v
  59                 let p = path + "." + k
  60
  61                 if (t == "object")
  62                         found = found.concat(findStrings(v, p))
  63                 else if (t == "string")
  64                         found.push([v, p])
  65         }
  66
  67         return found
  68 }
  69
  70         console.log(findStrings(initData))
  71
  72 */