Merge pull request #6 from Countly/lighthouse-useragentdata

Added useragentdata for bot detection too

CHANGELOG.md

@@ -1,3 +1,7 @@
+## 23.12.4
+
+* Enhanced userAgentData detection for bot filtering
+
 ## 23.12.3
 
 * Added bot detection for workers

cypress/e2e/user_agent.cy.js

@@ -1,6 +1,7 @@
 /* eslint-disable cypress/no-unnecessary-waiting */
 /* eslint-disable require-jsdoc */
 var Countly = require("../../Countly.js");
+var Utils = require("../../modules/Utils.js");
 var hp = require("../support/helper");
 
 function initMain() {
@@ -55,11 +56,19 @@ describe("User Agent tests ", () => {
         hp.haltAndClearStorage(() => {
             initMain();
             // setting ua value to strings that can pass the regex test
+            expect(Countly._internals.userAgentSearchBotDetection("")).to.equal(false);
             expect(Countly._internals.userAgentSearchBotDetection("123")).to.equal(false);
             expect(Countly._internals.userAgentSearchBotDetection("Googlebot")).to.equal(true);
             expect(Countly._internals.userAgentSearchBotDetection("Google")).to.equal(false);
             expect(Countly._internals.userAgentSearchBotDetection("HeadlessChrome")).to.equal(true);
             expect(Countly._internals.userAgentSearchBotDetection("Chrome-Lighthouse")).to.equal(true);
+            expect(Countly._internals.userAgentSearchBotDetection("Lighthouse")).to.equal(true);
+        });
+    });
+    // userAgentData is not supported by all browsers yet so it is hard to test with consistency
+    it("Check if currentUserAgentDataString override works", () => {
+        hp.haltAndClearStorage(() => {
+            expect(Utils.currentUserAgentDataString('123')).to.equal("123");
         });
     });
 });

modules/Constants.js

@@ -104,7 +104,7 @@ var healthCheckCounterEnum = Object.freeze({
     errorMessage: "cly_hc_error_message",
 });
 
-var SDK_VERSION = "23.12.3";
+var SDK_VERSION = "23.12.4";
 var SDK_NAME = "javascript_native_web";
 
 // Using this on document.referrer would return an array with 15 elements in it. The 12th element (array[11]) would be the path we are looking for. Others would be things like password and such (use https://regex101.com/ to check more)

modules/CountlyClass.js

@@ -19,6 +19,7 @@ import {
     add_event_listener,
     get_event_target,
     currentUserAgentString,
+    currentUserAgentDataString,
     userAgentDeviceDetection,
     userAgentSearchBotDetection,
     get_page_coord,
@@ -4624,6 +4625,7 @@ class CountlyClass {
             processScrollView: processScrollView,
             processScroll: processScroll,
             currentUserAgentString: currentUserAgentString,
+            currentUserAgentDataString: currentUserAgentDataString,
             userAgentDeviceDetection: userAgentDeviceDetection,
             userAgentSearchBotDetection: userAgentSearchBotDetection,
             getRequestQueue: getRequestQueue,

modules/Utils.js

@@ -319,23 +319,39 @@ function currentUserAgentString(uaOverride) {
     }
 
     var ua_raw = navigator.userAgent;
-    // check if userAgentData is supported and userAgent is not available, use it
+    // check if userAgentData is supported and userAgent is not available, then use it
     if (!ua_raw) {
-        if (navigator.userAgentData) {
-            // turn brands array into string
-            ua_raw = navigator.userAgentData.brands.map(function (e) {
-                return e.brand + ":" + e.version;
-            }).join();
-            // add mobile info
-            ua_raw += (navigator.userAgentData.mobile ? " mobi " : " ");
-            // add platform info
-            ua_raw += navigator.userAgentData.platform;
-        }
+        ua_raw = currentUserAgentDataString();
     }
     // RAW USER AGENT STRING
     return ua_raw;
 }
 
+/**
+ *  Forms user agent string from userAgentData by concatenating brand, version, mobile and platform
+ *  @memberof Countly._internals
+ *  @param {string} uaOverride - a string value to pass instead of ua value
+ *  @returns {string} currentUserAgentString - user agent string from userAgentData
+ */
+function currentUserAgentDataString(uaOverride) {
+    if (uaOverride) {
+        return uaOverride;
+    }
+
+    var ua = "";
+    if (navigator.userAgentData) {
+        // turn brands array into string
+        ua = navigator.userAgentData.brands.map(function (e) {
+            return e.brand + ":" + e.version;
+        }).join();
+        // add mobile info
+        ua += (navigator.userAgentData.mobile ? " mobi " : " ");
+        // add platform info
+        ua += navigator.userAgentData.platform;
+    }
+    return ua;
+}
+
 /**
  *  Returns device type information according to user agent string
  *  @memberof Countly._internals
@@ -348,7 +364,7 @@ function userAgentDeviceDetection(uaOverride) {
     if (uaOverride) {
         userAgent = uaOverride;
     }
-    else if (navigator.userAgentData.mobile) {
+    else if (navigator.userAgentData && navigator.userAgentData.mobile) {
         return "phone";
     }
     else {
@@ -384,9 +400,18 @@ function userAgentDeviceDetection(uaOverride) {
  */
 function userAgentSearchBotDetection(uaOverride) {
     // search bot regexp
-    var searchBotRE = /(CountlySiteBot|nuhk|Googlebot|GoogleSecurityScanner|Yammybot|Openbot|Slurp|MSNBot|Ask Jeeves\/Teoma|ia_archiver|bingbot|Google Web Preview|Mediapartners-Google|AdsBot-Google|Baiduspider|Ezooms|YahooSeeker|AltaVista|AVSearch|Mercator|Scooter|InfoSeek|Ultraseek|Lycos|Wget|YandexBot|Yandex|YaDirectFetcher|SiteBot|Exabot|AhrefsBot|MJ12bot|TurnitinBot|magpie-crawler|Nutch Crawler|CMS Crawler|rogerbot|Domnutch|ssearch_bot|XoviBot|netseer|digincore|fr-crawler|wesee|AliasIO|contxbot|PingdomBot|BingPreview|HeadlessChrome|Chrome-Lighthouse)/;
-    // true if the user agent string contains a search bot string pattern
-    return searchBotRE.test(uaOverride || currentUserAgentString());
+    const searchBotRE = /(CountlySiteBot|nuhk|Googlebot|GoogleSecurityScanner|Yammybot|Openbot|Slurp|MSNBot|Ask Jeeves\/Teoma|ia_archiver|bingbot|Google Web Preview|Mediapartners-Google|AdsBot-Google|Baiduspider|Ezooms|YahooSeeker|AltaVista|AVSearch|Mercator|Scooter|InfoSeek|Ultraseek|Lycos|Wget|YandexBot|Yandex|YaDirectFetcher|SiteBot|Exabot|AhrefsBot|MJ12bot|TurnitinBot|magpie-crawler|Nutch Crawler|CMS Crawler|rogerbot|Domnutch|ssearch_bot|XoviBot|netseer|digincore|fr-crawler|wesee|AliasIO|contxbot|PingdomBot|BingPreview|HeadlessChrome|Lighthouse)/;
+
+    // check override first
+    if (uaOverride) {
+        return searchBotRE.test(uaOverride);
+    }
+
+    // check both userAgent and userAgentData, as one of them might be containing the information we are looking for
+    const ua_bot = searchBotRE.test(currentUserAgentString());
+    const uaData_bot = searchBotRE.test(currentUserAgentDataString());
+
+    return ua_bot || uaData_bot;
 }
 
 /**
@@ -614,5 +639,6 @@ export {
     loadCSS,
     showLoader,
     checkIfLoggingIsOn,
-    hideLoader
+    hideLoader,
+    currentUserAgentDataString
 }; 

package.json

@@ -1,6 +1,6 @@
 {
   "name": "countly-sdk-js",
-  "version": "23.12.3",
+  "version": "23.12.4",
   "description": "Countly JavaScript SDK",
   "type": "module",
   "main": "Countly.js",