google-img-scrap 1.0.7 → 1.0.8

package/CHANGELOG.md

@@ -1,27 +1,39 @@
 # Changelog
 
+### 1.0.8
+
+- Fixed "ERROR: Cannot assign to "queryName" because it is a constant" (by GaspardCulis)
+- Removed gstatic url
+- Added average color, id, title and originalUrl
+
 ### 1.0.7
+
 - Readme update
 
 ### 1.0.6
+
 - Fixed types
-- Added ```limit``` to limit the size of the results
+- Added `limit` to limit the size of the results
 
 ### 1.0.5
+
 - Added types (by christophe77)
 
 ### v1.0.4
 
-- New option ```urlMatch```. You now get image when an url match a string (example: "cdn")
-- New option ```filterByTitles```. Filter images by titles
+- New option `urlMatch`. You now get image when an url match a string (example: "cdn")
+- New option `filterByTitles`. Filter images by titles
 
 ### v1.0.3
-- New option ```execute```. allow you to execute a function to remove "gstatic.com" domains for example
+
+- New option `execute`. allow you to execute a function to remove "gstatic.com" domains for example
 
 ### v1.0.2
+
 - Cannot set 'domains' and 'excludeDomains' as same time
 - Fixed some bugs
-- New option ```excludeWords```
+- New option `excludeWords`
 
 ### v1.0.1
-- Added the missing dependencie
+
+- Added the missing dependencie

package/README.md

@@ -1,6 +1,6 @@
-# Google-img-scrap v1.0.7
+# Google-img-scrap
 
-Scrap images from google images with customs pre filled options
+Scrap images from google images with customs pre filled dorking options
 
 ## Update
 
@@ -19,22 +19,22 @@ npm i google-img-scrap
 ## Import
 
 ```js
-const { GOOGLE_IMG_SCRAP , GOOGLE_QUERY } = require('google-img-scrap');
+const { GOOGLE_IMG_SCRAP, GOOGLE_QUERY } = require("google-img-scrap");
 ```
 
 ## Query Params
 
-- "search" (String) what you want to search
-- "execute" (Function) allow you to execute a function to remove "gstatic.com" domains for example
-- "excludeWords" (String[]) exclude some words from the search
-- "domains" (String[]) filter by domains
-- "excludeDomains" (String[]) exclude some domains
-- "safeSearch" (Boolean) active safe search or not for nsfw for example
-- "custom" (String) add extra query
-- "urlMatch" (String[][]) get image when an url match a string (example: "cdn") | ```example below```
-- "filterByTitles" (String[][]) filter images by titles | ```example below```
-- "query" (Object) set a query (can be [TYPE, DATE, COLOR, SIZE, LICENCE, EXTENSION]) (use GOOGLE_QUERY items | ```example below```
-- "limit" (Int) to limit the size of the results
+- "search" `string` what you want to search
+- "execute" `(element: FinalResult) => FinalResult | undefined` allow you to execute a function to filter results
+- "excludeWords" `string[]` exclude some words from the search
+- "domains" `string[]` filter by domains
+- "excludeDomains" `string[]` exclude some domains
+- "safeSearch" `boolean` active safe search or not for nsfw for example
+- "custom" `string` add extra query
+- "urlMatch" `string[][]` get image when an url match a string (example: "cdn") | `example below`
+- "filterByTitles" `string[][]` filter images by titles | `example below`
+- "query" `GoogleQuery` set a query (can be [TYPE, DATE, COLOR, SIZE, LICENCE, EXTENSION]) (use GOOGLE_QUERY items | `example below`
+- "limit" `number` to limit the size of the results
 
 ## Result
 
@@ -43,60 +43,60 @@ const { GOOGLE_IMG_SCRAP , GOOGLE_QUERY } = require('google-img-scrap');
   url: 'https://images.google.com/search?tbm=isch&tbs=itp:clipart,qdr:y,ic:gray,isz:l,il:ol,ift:jpg&q=cats%20%20%20-%22black%22%20-%22white%22&name=content&name2=content2',
   result: [
     {
+      id: "HA6fW6faerBfPM",
+      title: "CAT eating a fish",
+      originalUrl: "https://media.gettyimages.com/vectors/cat-article.html",
       url: 'https://media.gettyimages.com/vectors/cat-eating-fish-vector-id1216628506',
-      height: '1024',
-      width: '1024'
+      averageColor: "rgb(241, 25, 60)",
+      averageColorObject: { r: 241, g: 25, b: 60},
+      height: 1024,
+      width: 1024
     },
     {
+      id: "OPSfyUtrsrYUI",
+      title: "Cat",
+      originalUrl: "https://www.ariatrade.gr/images/products/2021/10/article.html",
       url: 'https://www.ariatrade.gr/images/products/2021/10/110294_1.jpg',
-      height: '768',
-      width: '1024'
+      averageColor: "rgb(201, 250, 65)",
+      averageColorObject: { r: 201, g: 250, b: 65},
+      height: 768,
+      width: 1024
     },
-    {
-      url: 'https://media.gettyimages.com/illustrations/panther-leaping-illustration-id152406879?s=2048x2048',
-      height: '2048',
-      width: '2048'
-    },
-    {
-      url: 'https://media.gettyimages.com/illustrations/botany-plants-antique-engraving-illustration-erythrina-variegata-illustration-id970781520',
-      height: '1024',
-      width: '828'
-    }
+    ...
   ]
-  ...
 }
 ```
 
 ## How to use ?
 
-- **For the query parameter you need to set the name in upper case !**
+**NOTE**: For the query parameter you need to set the name in upper case !
 
 ## Simple example
 
 Search cats images
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+  });
 
-    console.log(test);
+  console.log(test);
 })();
 ```
 
-## Removing gstatic.com
+## Filtering
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "demon slayer background hd",
-        execute: function(element){
-            if(!element.url.match('gstatic.com')) return element;
-        }
-    });
-
-    console.log(test);
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "demon slayer background hd",
+    execute: function (element) {
+      if (element.url.length < 20) return element;
+    },
+  });
+
+  console.log(test);
 })();
 ```
 
@@ -105,33 +105,33 @@ Search cats images
 All query options are optional (see below for all the options)
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        query: {
-            TYPE: GOOGLE_QUERY.TYPE.CLIPART,
-            DATE: GOOGLE_QUERY.DATE.YEAR,
-            COLOR: GOOGLE_QUERY.COLOR.BLACK_AND_WHITE,
-            SIZE: GOOGLE_QUERY.SIZE.LARGE,
-            LICENCE: GOOGLE_QUERY.LICENCE.COMMERCIAL_AND_OTHER,
-            EXTENSION: GOOGLE_QUERY.EXTENSION.JPG
-        },
-    });
-
-    console.log(test);
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    query: {
+      TYPE: GOOGLE_QUERY.TYPE.CLIPART,
+      DATE: GOOGLE_QUERY.DATE.YEAR,
+      COLOR: GOOGLE_QUERY.COLOR.BLACK_AND_WHITE,
+      SIZE: GOOGLE_QUERY.SIZE.LARGE,
+      LICENCE: GOOGLE_QUERY.LICENCE.COMMERCIAL_AND_OTHER,
+      EXTENSION: GOOGLE_QUERY.EXTENSION.JPG,
+    },
+  });
+
+  console.log(test);
 })();
 ```
 
 ## Limit result size
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        limit: 5,
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    limit: 5,
+  });
 
-    console.log(test);
+  console.log(test);
 })();
 ```
 
@@ -140,26 +140,26 @@ All query options are optional (see below for all the options)
 Only scrap from a specific domain
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        domains: ["alamy.com", "istockphoto.com", "vecteezy.com"],
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    domains: ["alamy.com", "istockphoto.com", "vecteezy.com"],
+  });
 
-    console.log(test);
+  console.log(test);
 })();
 ```
 
 ## Exclude domains
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        excludeDomains: ["istockphoto.com", "alamy.com"]
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    excludeDomains: ["istockphoto.com", "alamy.com"],
+  });
 
-    console.log(test);
+  console.log(test);
 })();
 ```
 
@@ -168,39 +168,39 @@ Only scrap from a specific domain
 If you don' like black cats and white cats
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        excludeWords: ["black", "white"], //If you don't like black cats and white cats
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    excludeWords: ["black", "white"], //If you don't like black cats and white cats
+  });
 
-    console.log(test, test.result.length);
+  console.log(test);
 })();
 ```
 
 ## Safe search (no nsfw)
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        safeSearch: false,
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    safeSearch: false,
+  });
 
-    console.log(test);
+  console.log(test);
 })();
 ```
 
 ## Custom query params
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        custom: "name=content&name2=content2",
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    custom: "name=content&name2=content2",
+  });
 
-    console.log(test);
+  console.log(test);
 })();
 ```
 
@@ -209,17 +209,17 @@ If you don' like black cats and white cats
 - urlMatch work like filterByTiles
 
 ```js
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        //will build something like this "(draw and white) or (albino and white)"
-        filterByTitles: [
-            ["draw", "white"],
-            ["albino", "white"]
-        ],
-    });
-
-    console.log(test);
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    //will build something like this "(draw and white) or (albino and white)"
+    filterByTitles: [
+      ["draw", "white"],
+      ["albino", "white"],
+    ],
+  });
+
+  console.log(test);
 })();
 ```
 
@@ -227,10 +227,10 @@ If you don' like black cats and white cats
 
 ```js
 {
-  SIZE: { 
-    LARGE, 
-    MEDIUM, 
-    ICON 
+  SIZE: {
+    LARGE,
+    MEDIUM,
+    ICON
   },
   COLOR: {
     BLACK_AND_WHITE,
@@ -248,10 +248,10 @@ If you don' like black cats and white cats
     BLACK,
     BROWN
   },
-  TYPE: { 
-    CLIPART, 
-    DRAW, 
-    GIF 
+  TYPE: {
+    CLIPART,
+    DRAW,
+    GIF
   },
   EXTENSION: {
     JPG,
@@ -263,15 +263,15 @@ If you don' like black cats and white cats
     ICO,
     RAW
   },
-  DATE: { 
-    DAY, 
-    WEEK, 
-    MONTH, 
-    YEAR 
+  DATE: {
+    DAY,
+    WEEK,
+    MONTH,
+    YEAR
   },
-  LICENCE: { 
-    CREATIVE_COMMONS, 
-    COMMERCIAL_AND_OTHER 
+  LICENCE: {
+    CREATIVE_COMMONS,
+    COMMERCIAL_AND_OTHER
   }
 }
 ```

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "google-img-scrap",
-  "version": "1.0.7",
-  "description": "Scrap images from google images with customs pre filled options",
+  "version": "1.0.8",
+  "description": "Scrap images from google images with customs pre filled dorking options",
   "main": "./src/google-img-scrap.js",
   "types": "./types/index.d.ts",
   "directories": {
@@ -34,6 +34,10 @@
     {
       "name": "christophe77",
       "url": "https://github.com/christophe77"
+    },
+    {
+      "name": "GaspardCulis",
+      "url": "https://github.com/GaspardCulis"
     }
   ],
   "license": "MIT",

package/src/back/google-img-scrap-1.0.7-.js

@@ -0,0 +1,184 @@
+const got = require("got");
+const { FastHTMLParser } = require("fast-html-dom-parser");
+
+const { GOOGLE_CONSTANT } = require("../constant/GOOGLE_CONSTANT");
+const { GOOGLE_QUERY } = require("../constant/query/GOOGLE_QUERY");
+const EXTENSIONS = require("../constant/extensions/IMAGES_EXTENSIONS.json");
+
+const { buildQuery, unicodeToChar } = require("../utils/UTILS");
+
+/**
+ * Validation of the arguments passed
+ * @param {import("../../types").Config} config
+ */
+function verify(config) {
+  if (config.excludeDomains && config.domains)
+    throw "Can not set 'excludeDomains' and 'domains' as same times";
+
+  if (!config.search || config.search.trim() == "")
+    throw "'search' can not be empty";
+
+  if (config.query) {
+    const queryToVerify = Object.keys(GOOGLE_QUERY);
+
+    for (const key of Object.keys(config.query)) {
+      if (!queryToVerify.includes(key)) throw `Invalide query name '${key}'`;
+
+      const VALUES = Object.values(GOOGLE_QUERY[key]);
+      const ACTUAL_VALUE = config.query[key];
+      if (!VALUES.includes(ACTUAL_VALUE))
+        throw `'${ACTUAL_VALUE}' is not a valide argument for the query : '${key}'`;
+    }
+  }
+}
+
+/**
+ * Verifify the url is an image
+ * @param {string} content
+ * @returns {boolean}
+ */
+function containImage(content = "") {
+  return EXTENSIONS.some((extension) => content.includes(extension));
+}
+
+/**
+ *Parse the html from google image to get the images links
+ * @param {string} url
+ * @returns {import("../../types").FinalResult[]}
+ */
+async function parse(url) {
+  const result = [];
+
+  const response = await got(url, {
+    headers: GOOGLE_CONSTANT.headers,
+  });
+  const parser = new FastHTMLParser(response.body);
+
+  const scripts = parser.getElementsByTagName("script");
+
+  if (!scripts) return result;
+
+  for (const script of scripts) {
+    const body = script.innerHTML;
+
+    const valide = containImage(body);
+
+    if (valide) {
+      const regex = /\["(http.+?)",(\d+),(\d+)\]/gi;
+
+      let res = null;
+
+      while ((res = regex.exec(body)) != null) {
+        if (res.length >= 4 && res[1].match(/http/gi).length < 2)
+          result.push({
+            url: unicodeToChar(res[1]),
+            height: res[2],
+            width: res[3],
+          });
+      }
+    }
+  }
+
+  return result;
+}
+
+/**
+ * Main function to build google image dork URL
+ * @param {import("../../types").Config} config
+ * @returns {import("../../types").Results}
+ */
+async function GOOGLE_IMG_SCRAP(config = {}) {
+  verify(config);
+
+  //exclude domains
+  const EXCLUDE_DOMAINS = [];
+  if (config.excludeDomains)
+    config.excludeDomains.forEach((domain) =>
+      EXCLUDE_DOMAINS.push(`-site:"${domain}"`)
+    );
+
+  //domains
+  const DOMAINS = [];
+  if (config.domains)
+    config.domains.forEach((domain) => DOMAINS.push(`site:"${domain}"`));
+
+  //exclude words
+  const EXCLUDE_WORDS = [];
+  if (config.excludeWords)
+    config.excludeWords.forEach((word) => EXCLUDE_WORDS.push(`-"${word}"`));
+
+  //filter by titles
+  const FILTER_TITLE = [];
+  if (config.filterByTitles)
+    config.filterByTitles.forEach((titleFilter) => {
+      const value = titleFilter.map((title) => {
+        return `intitle:"${title}"`;
+      });
+
+      FILTER_TITLE.push(`(${value.join(" AND ")})`);
+    });
+
+  //url match words
+  const URL_MATCH = [];
+  if (config.urlMatch)
+    config.urlMatch.forEach((urlMatch) => {
+      const value = urlMatch.map((content) => {
+        return `inurl:${content}`;
+      });
+
+      URL_MATCH.push(`(${value.join(" AND ")})`);
+    });
+
+  //building url
+  const SEARCH_TERM =
+    config.search +
+    " " +
+    URL_MATCH.join(" OR ") +
+    " " +
+    FILTER_TITLE.join(" OR ") +
+    " " +
+    EXCLUDE_WORDS.join(" ") +
+    " " +
+    EXCLUDE_DOMAINS.join(" ") +
+    " " +
+    DOMAINS.join(" OR ");
+
+  const SEARCH = encodeURIComponent(SEARCH_TERM.trim());
+  const QUERY = Object.assign(GOOGLE_CONSTANT.forceGoogleImage, {
+    [GOOGLE_CONSTANT.queryParam]: Object.values(config.query || {}).join(","),
+    q: SEARCH,
+  });
+
+  const CUSTOM_PARAM = config.custom ? `&${config.custom}` : "";
+  const SAFE_SEARCH = config.safeSearch ? `&safe=active` : "";
+
+  const URL =
+    GOOGLE_CONSTANT.url + buildQuery(QUERY) + CUSTOM_PARAM + SAFE_SEARCH;
+
+  //parsing
+  const result = await parse(URL);
+
+  //excute function
+  let finalResult = [];
+  if (config.execute)
+    result.forEach((element) => {
+      const value = config.execute(element);
+      if (value) finalResult.push(value);
+    });
+  else finalResult = result;
+
+  //limit result
+  let slicedResult = [];
+  const { limit } = config;
+
+  if (limit && limit > 0 && finalResult.length > limit) {
+    slicedResult = finalResult.slice(0, limit);
+  }
+  //result
+  return {
+    url: URL,
+    result: slicedResult.length > 0 ? slicedResult : finalResult,
+  };
+}
+
+module.exports = { GOOGLE_IMG_SCRAP, GOOGLE_QUERY };

package/src/constant/GOOGLE_CONSTANT.js

@@ -1,13 +1,13 @@
 const GOOGLE_CONSTANT = {
-    url: "https://images.google.com/search",
-    queryParam: "tbs",
-    forceGoogleImage: {
-        tbm: "isch" //needed to search on google image instead of google
-    },
-    headers: {
-        'User-Agent':
-        'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_11_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/61.0.3163.100 Safari/537.36'
-    }
+  url: "https://images.google.com/search",
+  queryParam: "tbs",
+  forceGoogleImage: {
+    tbm: "isch", //needed to search on google image instead of google
+  },
+  headers: {
+    "User-Agent":
+      "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_11_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/61.0.3163.100 Safari/537.36",
+  },
 };
 
-module.exports = { GOOGLE_CONSTANT };
+module.exports = { GOOGLE_CONSTANT };

package/src/constant/extensions/IMAGES_EXTENSIONS.json

@@ -1 +1 @@
-["jpg", "gif", "bmp", "png", "svg", "webp", "ico", "raw"]
+["jpg", "gif", "bmp", "png", "svg", "webp", "ico", "raw"]

package/src/constant/query/GOOGLE_COLORS.js

@@ -1,14 +1,16 @@
-const COLORS = ["red",
-                "blue", 
-                "purple", 
-                "orange", 
-                "yellow", 
-                "green", 
-                "teal", 
-                "pink", 
-                "white", 
-                "gray", 
-                "black", 
-                "brown"];
+const COLORS = [
+  "red",
+  "blue",
+  "purple",
+  "orange",
+  "yellow",
+  "green",
+  "teal",
+  "pink",
+  "white",
+  "gray",
+  "black",
+  "brown",
+];
 
-module.exports = { COLORS };
+module.exports = { COLORS };

package/src/constant/query/GOOGLE_PARAMS.js

@@ -1,9 +1,17 @@
 const SIZE_PARAM = "isz",
-      COLOR_PARAM = "ic",
-      SPECIFIC_COLOR_PARAM = "Cisc",
-      TYPE_PARAM = "itp",
-      DATE_PARAM = "qdr",
-      LICENCE_PARAM = "il",
-      IMAGE_EXTENSION_PARAM = "ift";
+  COLOR_PARAM = "ic",
+  SPECIFIC_COLOR_PARAM = "Cisc",
+  TYPE_PARAM = "itp",
+  DATE_PARAM = "qdr",
+  LICENCE_PARAM = "il",
+  IMAGE_EXTENSION_PARAM = "ift";
 
-module.exports = { SIZE_PARAM, COLOR_PARAM, SPECIFIC_COLOR_PARAM, TYPE_PARAM, DATE_PARAM, LICENCE_PARAM, IMAGE_EXTENSION_PARAM };
+module.exports = {
+  SIZE_PARAM,
+  COLOR_PARAM,
+  SPECIFIC_COLOR_PARAM,
+  TYPE_PARAM,
+  DATE_PARAM,
+  LICENCE_PARAM,
+  IMAGE_EXTENSION_PARAM,
+};

package/src/constant/query/GOOGLE_QUERY.js

@@ -1,50 +1,59 @@
-const { SIZE_PARAM, COLOR_PARAM, SPECIFIC_COLOR_PARAM, TYPE_PARAM, DATE_PARAM, LICENCE_PARAM, IMAGE_EXTENSION_PARAM } = require('./GOOGLE_PARAMS');
-const { COLORS } = require('./GOOGLE_COLORS');
+const {
+  SIZE_PARAM,
+  COLOR_PARAM,
+  SPECIFIC_COLOR_PARAM,
+  TYPE_PARAM,
+  DATE_PARAM,
+  LICENCE_PARAM,
+  IMAGE_EXTENSION_PARAM,
+} = require("./GOOGLE_PARAMS");
+const { COLORS } = require("./GOOGLE_COLORS");
 const EXTENSIONS = require("../extensions/IMAGES_EXTENSIONS.json");
 
 const GOOGLE_QUERY = {
-    SIZE: {
-        LARGE: SIZE_PARAM + ":l",
-        MEDIUM: SIZE_PARAM + ":m",
-        ICON: SIZE_PARAM + ":i",
-    },
-
-    COLOR: {
-        BLACK_AND_WHITE: COLOR_PARAM + ":gray",
-        TRANSPARENT: COLOR_PARAM + ":trans",
-    },
-
-    TYPE: {
-        CLIPART: TYPE_PARAM + ":clipart",
-        DRAW: TYPE_PARAM + ":lineart",
-        GIF: TYPE_PARAM + ":animated",
-    },
-
-    EXTENSION: {},
-
-    DATE: {
-        DAY: DATE_PARAM + ":d",
-        WEEK: DATE_PARAM + ":w",
-        MONTH: DATE_PARAM + ":m",
-        YEAR: DATE_PARAM + ":y",
-    },
-
-    LICENCE: {
-        CREATIVE_COMMONS: LICENCE_PARAM + ":cl",
-        COMMERCIAL_AND_OTHER: LICENCE_PARAM + ":ol",
-    },
+  SIZE: {
+    LARGE: SIZE_PARAM + ":l",
+    MEDIUM: SIZE_PARAM + ":m",
+    ICON: SIZE_PARAM + ":i",
+  },
+
+  COLOR: {
+    BLACK_AND_WHITE: COLOR_PARAM + ":gray",
+    TRANSPARENT: COLOR_PARAM + ":trans",
+  },
+
+  TYPE: {
+    CLIPART: TYPE_PARAM + ":clipart",
+    DRAW: TYPE_PARAM + ":lineart",
+    GIF: TYPE_PARAM + ":animated",
+  },
+
+  EXTENSION: {},
+
+  DATE: {
+    DAY: DATE_PARAM + ":d",
+    WEEK: DATE_PARAM + ":w",
+    MONTH: DATE_PARAM + ":m",
+    YEAR: DATE_PARAM + ":y",
+  },
+
+  LICENCE: {
+    CREATIVE_COMMONS: LICENCE_PARAM + ":cl",
+    COMMERCIAL_AND_OTHER: LICENCE_PARAM + ":ol",
+  },
 };
 
 //build extension
 EXTENSIONS.forEach((EXTENSION) => {
-    const queryName = EXTENSION.toUpperCase();
-    GOOGLE_QUERY.EXTENSION[queryName] = IMAGE_EXTENSION_PARAM + ":" + EXTENSION;
+  const queryName = EXTENSION.toUpperCase();
+  GOOGLE_QUERY.EXTENSION[queryName] = IMAGE_EXTENSION_PARAM + ":" + EXTENSION;
 });
 
 //build colors
 COLORS.forEach((COLOR) => {
-    const queryName = COLOR.toUpperCase();
-    GOOGLE_QUERY.COLOR[queryName] = COLOR_PARAM + ":specific," + SPECIFIC_COLOR_PARAM + ":" + COLOR;
+  const queryName = COLOR.toUpperCase();
+  GOOGLE_QUERY.COLOR[queryName] =
+    COLOR_PARAM + ":specific," + SPECIFIC_COLOR_PARAM + ":" + COLOR;
 });
 
-module.exports = { GOOGLE_QUERY };
+module.exports = { GOOGLE_QUERY };

package/src/google-img-scrap.js

@@ -3,12 +3,14 @@ const { FastHTMLParser } = require("fast-html-dom-parser");
 
 const { GOOGLE_CONSTANT } = require("./constant/GOOGLE_CONSTANT");
 const { GOOGLE_QUERY } = require("./constant/query/GOOGLE_QUERY");
-const { TRANSLATOR } = require("./constant/translator/TRANSLATOR");
 const EXTENSIONS = require("./constant/extensions/IMAGES_EXTENSIONS.json");
 
 const { buildQuery, unicodeToChar } = require("./utils/UTILS");
 
-//verify good configuration
+/**
+ * Validation of the arguments passed
+ * @param {import("../types").Config} config
+ */
 function verify(config) {
   if (config.excludeDomains && config.domains)
     throw "Can not set 'excludeDomains' and 'domains' as same times";
@@ -30,16 +32,20 @@ function verify(config) {
   }
 }
 
-//verify imag extension
+/**
+ * Verifify the url is an image
+ * @param {string} content
+ * @returns {boolean}
+ */
 function containImage(content = "") {
-  for (const EXTENSION of EXTENSIONS) {
-    if (content.includes(EXTENSION)) return true;
-  }
-
-  return false;
+  return EXTENSIONS.some((extension) => content.includes(extension));
 }
 
-//parse HTML
+/**
+ *Parse the html from google image to get the images links
+ * @param {string} url
+ * @returns {import("../types").FinalResult[]}
+ */
 async function parse(url) {
   const result = [];
 
@@ -58,17 +64,45 @@ async function parse(url) {
     const valide = containImage(body);
 
     if (valide) {
-      const regex = /\["(http.+?)",(\d+),(\d+)\]/gi;
+      //getting image url, height, width, average
+      const regex =
+        /\["(http[^"]+?)",(\d+),(\d+)\],[\w\d]+?,[\w\d]+?,"rgb\((\d+),(\d+),(\d+)\)"/gi;
+
+      //getting originalUrl, title, id
+      const secondRegex = /\[[\w\d]+?,"([^"]+?)","(http[^"]+?)","([^"]+?)"/gi;
 
       let res = null;
+      let secondRes = null;
+
+      while (
+        (res = regex.exec(body)) != null &&
+        (secondRes = secondRegex.exec(body)) != null
+      ) {
+        if (
+          res.length >= 4 &&
+          res[1].match(/http/gi).length < 2 &&
+          secondRes.length === 4 &&
+          secondRes[2].match(/http/gi).length < 2
+        ) {
+          const [r, g, b] = [res[4], res[5], res[6]].map((e) =>
+            parseInt(e, 10)
+          );
 
-      while ((res = regex.exec(body)) != null) {
-        if (res.length >= 4 && res[1].match(/http/gi).length < 2)
           result.push({
+            id: secondRes[1],
+            title: secondRes[3],
             url: unicodeToChar(res[1]),
-            height: res[2],
-            width: res[3],
+            originalUrl: unicodeToChar(secondRes[2]),
+            averageColor: `rgb(${r}, ${g}, ${b})`,
+            averageColorObject: {
+              r,
+              g,
+              b,
+            },
+            height: parseInt(res[2], 10),
+            width: parseInt(res[3], 10),
           });
+        }
       }
     }
   }
@@ -76,9 +110,12 @@ async function parse(url) {
   return result;
 }
 
-//main
+/**
+ * Main function to build google image dork URL
+ * @param {import("../types").Config} config
+ * @returns {import("../types").Results}
+ */
 async function GOOGLE_IMG_SCRAP(config = {}) {
-  //verify config
   verify(config);
 
   //exclude domains
@@ -144,10 +181,7 @@ async function GOOGLE_IMG_SCRAP(config = {}) {
   const SAFE_SEARCH = config.safeSearch ? `&safe=active` : "";
 
   const URL =
-    GOOGLE_CONSTANT.url +
-    buildQuery(QUERY, TRANSLATOR) +
-    CUSTOM_PARAM +
-    SAFE_SEARCH;
+    GOOGLE_CONSTANT.url + buildQuery(QUERY) + CUSTOM_PARAM + SAFE_SEARCH;
 
   //parsing
   const result = await parse(URL);

package/src/utils/UTILS.js

@@ -1,24 +1,20 @@
-function buildQuery(query, translator){
-    const result = [];
+function buildQuery(query) {
+  const result = [];
 
-    const params = Object.keys(query);
-    const toTranslate = Object.keys(translator);
+  const params = Object.keys(query);
 
-    for(const param of params){
-        const queryName = param;
-        if(toTranslate.includes(param)) queryName = toTranslate[param];
+  for (const param of params) {
+    const queryName = param;
+    result.push(`${queryName}=${query[param]}`);
+  }
 
-        result.push(`${queryName}=${query[param]}`);
-    }
-
-    return "?" + result.join('&');
+  return "?" + result.join("&");
 }
 
 function unicodeToChar(text) {
-    return text.replace(/\\u[\dA-F]{4}/gi, 
-           function (match) {
-                return String.fromCharCode(parseInt(match.replace(/\\u/g, ''), 16));
-           });
- }
+  return text.replace(/\\u[\dA-F]{4}/gi, function (match) {
+    return String.fromCharCode(parseInt(match.replace(/\\u/g, ""), 16));
+  });
+}
 
-module.exports = { buildQuery, unicodeToChar };
+module.exports = { buildQuery, unicodeToChar };

package/test/test-filter-titles.js

@@ -1,16 +1,16 @@
-const { GOOGLE_IMG_SCRAP } = require('../src/google-img-scrap');
+const { GOOGLE_IMG_SCRAP } = require("../src/google-img-scrap");
 
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        filterByTitles: [
-            ["draw", "white"],
-            ["albino", "white"]
-        ],
-        execute: function(element){
-            if(!element.url.match('gstatic.com')) return element;
-        }
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    filterByTitles: [
+      ["draw", "white"],
+      ["albino", "white"],
+    ],
+    execute: function (element) {
+      if (!element.url.match("gstatic.com")) return element;
+    },
+  });
 
-    console.log(test, test.result.length);
-})();
+  console.log(test, test.result.length);
+})();

package/test/test-last-version.js

@@ -0,0 +1,19 @@
+const { GOOGLE_IMG_SCRAP } = require("../src/google-img-scrap");
+const V107 = require("../src/back/google-img-scrap-1.0.7-.js");
+
+// console.log(GOOGLE_QUERY);
+
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+  });
+
+  const test2 = await V107.GOOGLE_IMG_SCRAP({
+    search: "cats",
+    execute: function (element) {
+      if (!element.url.match("gstatic.com")) return element;
+    },
+  });
+
+  console.log(test.result.length, test2.result.length);
+})();

package/test/test-simple.js

@@ -0,0 +1,11 @@
+const { GOOGLE_IMG_SCRAP, GOOGLE_QUERY } = require("../src/google-img-scrap");
+
+// console.log(GOOGLE_QUERY);
+
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+  });
+
+  console.log(test, test.result.length);
+})();

package/test/test-url-match.js

@@ -1,16 +1,13 @@
-const { GOOGLE_IMG_SCRAP } = require('../src/google-img-scrap');
+const { GOOGLE_IMG_SCRAP } = require("../src/google-img-scrap");
 
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        urlMatch: [
-            ["cdn"],
-            ["istockphoto"]
-        ],
-        execute: function(element){
-            if(!element.url.match('gstatic.com')) return element;
-        }
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    urlMatch: [["cdn"], ["istockphoto"]],
+    execute: function (element) {
+      if (!element.url.match("gstatic.com")) return element;
+    },
+  });
 
-    console.log(test, test.result.length);
-})();
+  console.log(test, test.result.length);
+})();

package/test/test-wallpaper.js

@@ -1,17 +1,21 @@
-const { GOOGLE_IMG_SCRAP , GOOGLE_QUERY } = require('../src/google-img-scrap');
+const { GOOGLE_IMG_SCRAP, GOOGLE_QUERY } = require("../src/google-img-scrap");
 
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "demon slayer background hd",
-        query: {
-            SIZE: GOOGLE_QUERY.SIZE.LARGE,
-        },
-        domains: ["alphacoders.com"],
-        safeSearch: false,
-        execute: function(element){
-            if(!element.url.match('gstatic.com')) return element;
-        }
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "demon slayer background hd",
+    query: {
+      SIZE: GOOGLE_QUERY.SIZE.LARGE,
+    },
+    domains: ["alphacoders.com"],
+    safeSearch: false,
+    execute: function (element) {
+      if (!element.url.match("gstatic.com")) return element;
+    },
+  });
 
-    console.log(test, test.result[test.result.length-1].url, test.result.length);
-})();
+  console.log(
+    test,
+    test.result[test.result.length - 1].url,
+    test.result.length
+  );
+})();

package/test/test.js

@@ -1,26 +1,26 @@
-const { GOOGLE_IMG_SCRAP , GOOGLE_QUERY } = require('../src/google-img-scrap');
+const { GOOGLE_IMG_SCRAP, GOOGLE_QUERY } = require("../src/google-img-scrap");
 
 // console.log(GOOGLE_QUERY);
 
-(async function(){
-    const test = await GOOGLE_IMG_SCRAP({
-        search: "cats",
-        query: {
-            TYPE: GOOGLE_QUERY.TYPE.CLIPART,
-            DATE: GOOGLE_QUERY.DATE.YEAR,
-            COLOR: GOOGLE_QUERY.COLOR.BLACK_AND_WHITE,
-            SIZE: GOOGLE_QUERY.SIZE.LARGE,
-            LICENCE: GOOGLE_QUERY.LICENCE.COMMERCIAL_AND_OTHER,
-            EXTENSION: GOOGLE_QUERY.EXTENSION.JPG
-        },
-        excludeWords: ["black", "white"], //If you don't like black and white cats
-        custom: "name=content&name2=content2",
-        safeSearch: false,
-        execute: function(element){
-            if(!element.url.match('gstatic.com')) return element;
-        }
-        // excludeDomains: ["istockphoto.com", "alamy.com"]
-    });
+(async function () {
+  const test = await GOOGLE_IMG_SCRAP({
+    search: "cats",
+    query: {
+      TYPE: GOOGLE_QUERY.TYPE.CLIPART,
+      DATE: GOOGLE_QUERY.DATE.YEAR,
+      COLOR: GOOGLE_QUERY.COLOR.BLACK_AND_WHITE,
+      SIZE: GOOGLE_QUERY.SIZE.LARGE,
+      LICENCE: GOOGLE_QUERY.LICENCE.COMMERCIAL_AND_OTHER,
+      EXTENSION: GOOGLE_QUERY.EXTENSION.JPG,
+    },
+    excludeWords: ["black", "white"], //If you don't like black and white cats
+    custom: "name=content&name2=content2",
+    safeSearch: false,
+    execute: function (element) {
+      if (!element.url.match("gstatic.com")) return element;
+    },
+    // excludeDomains: ["istockphoto.com", "alamy.com"]
+  });
 
-    console.log(test, test.result.length);
-})();
+  console.log(test, test.result.length);
+})();

package/types/index.d.ts

@@ -1,6 +1,6 @@
 type Config = {
   search: string;
-  limit?:number;
+  limit?: number;
   query?: {
     TYPE?: string;
     DATE?: string;
@@ -17,15 +17,27 @@ type Config = {
   execute?: (element: FinalResult) => FinalResult | undefined;
   filterByTitles?: [string[]];
 };
+
 type FinalResult = {
+  id: string;
+  title: string;
+  originalUrl: string;
   url: string;
-  height: string;
-  width: string;
+  averageColor: string;
+  averageColorObject: {
+    r: number;
+    g: number;
+    b: number;
+  };
+  height: number;
+  width: number;
 };
+
 type Results = {
   url: string;
   result: FinalResult[];
 };
+
 type GoogleQuery = {
   SIZE: {
     LARGE: string;

package/src/constant/translator/TRANSLATOR.js

@@ -1,3 +0,0 @@
-const TRANSLATOR = {};
-
-module.exports = { TRANSLATOR };