fix(KUI-1369): noIndexMiddleware and noIndexMiddleware.test has been added to make sure that URL with aap*. is not indexed by google

jest.config.js

@@ -1,3 +1,4 @@
 module.exports = {
+  setupFilesAfterEnv: ['jest-extended/all'],
   testEnvironment: 'jsdom',
 }

package.json

@@ -85,8 +85,10 @@
     "husky": "^9.0.11",
     "jest": "^29.7.0",
     "jest-environment-jsdom": "29.7.0",
+    "jest-extended": "^4.0.2",
     "lint-staged": "^15.2.5",
     "mini-css-extract-plugin": "^2.9.0",
+    "node-mocks-http": "^1.14.1",
     "nodemon": "^3.1.3",
     "null-loader": "^4.0.1",
     "path": "^0.12.7",

server/server.js

@@ -160,6 +160,12 @@ server.use(
   })
 )
 
+/* ********************************
+ * ******* No index middleware ****
+ * ********************************
+ */
+server.use(require('./utils/noIndexMiddleware.js'))
+
 /* **********************************
  * ******* APPLICATION ROUTES *******
  * **********************************

server/utils/noIndexMiddleware.js

@@ -0,0 +1,28 @@
+const { server: serverConfig } = require('../configuration')
+
+/**
+ * Middleware that adds "noindex" robots header if "server host" and the request
+ * host ("forwarded host") doesn't match. This is done to prevent Google from
+ * indexing app.kth.se/* when we want the user to use www.kth.se/*
+ *
+ * The "server host" (SERVER_HOST_URL) is the primary host we expected the
+ * app to be hosted on (ie https://www.kth.se).
+ *
+ * The orginal host of the request is replaced with an Azure host
+ * (*.azurewebsites.net) before reaching our app so we look at the "forwarded
+ * host" ("x-forwarded-host" header) .
+ *
+ * Default, if x-forwarded-host is missing, is to do nothing.
+ */
+const noIndexMiddleware = function (req, res, next) {
+  const forwardedHost = req.header('x-forwarded-host')
+  if (forwardedHost) {
+    const serverHostUrl = new URL(serverConfig.hostUrl)
+    const serverHost = serverHostUrl.host
+    if (serverHost !== forwardedHost) {
+      res.set('x-robots-tag', 'noindex')
+    }
+  }
+  next()
+}
+module.exports = noIndexMiddleware

server/utils/noIndexMiddleware.test.js

@@ -0,0 +1,50 @@
+const httpMocks = require('node-mocks-http')
+
+const noIndexMiddleware = require('./noIndexMiddleware')
+
+const mockedServerHost = 'mockedserverhost.example.com'
+jest.mock('../configuration', () => ({
+  server: { hostUrl: 'https://mockedserverhost.example.com' },
+}))
+
+describe('noIndexMiddleware', () => {
+  it('should not set "X-Robots-Tag" header if "X-Forwarded-Host" is missing', () => {
+    const next = jest.fn()
+    const { req, res } = httpMocks.createMocks({
+      headers: {},
+    })
+
+    noIndexMiddleware(req, res, next)
+
+    expect(res.getHeader('X-Robots-Tag')).toBeUndefined()
+    expect(next).toHaveBeenCalledOnce()
+  })
+
+  it('should not set "X-Robots-Tag" header if "X-Forwarded-Host" match SERVER_HOST_URL', () => {
+    const next = jest.fn()
+    const { req, res } = httpMocks.createMocks({
+      headers: {
+        'X-Forwarded-Host': mockedServerHost,
+      },
+    })
+
+    noIndexMiddleware(req, res, next)
+
+    expect(res.getHeader('X-Robots-Tag')).toBeUndefined()
+    expect(next).toHaveBeenCalledOnce()
+  })
+
+  it('should set "X-Robots-Tag" header if "X-Forwarded-Host" does not match SERVER_HOST_URL', () => {
+    const next = jest.fn()
+    const { req, res } = httpMocks.createMocks({
+      headers: {
+        'X-Forwarded-Host': 'anotherhost.example.com',
+      },
+    })
+
+    noIndexMiddleware(req, res, next)
+
+    expect(res.getHeader('X-Robots-Tag')).toBe('noindex')
+    expect(next).toHaveBeenCalledOnce()
+  })
+})