test: explicitly assert for Log.error calls in `registerForPush`

[sherwinski]

Description

1 Line Summary

Explicitly tests for an error in the registerForPush test.

Details

When running current tests, the following error is emitted:

  console.error
    PushPermissionNotGrantedError: The user dismissed the permission prompt.
        at SubscriptionManager.subscribeFcmFromPage (/Website-SDK/src/shared/managers/SubscriptionManager.ts:461:17)
        at SubscriptionManager.subscribe (/Website-SDK/src/shared/managers/SubscriptionManager.ts:162:13)
        at Function.internalRegisterForPush (/Website-SDK/src/shared/helpers/SubscriptionHelper.ts:30:35)
        at Function.registerForPush (/Website-SDK/src/shared/helpers/SubscriptionHelper.ts:20:12)
        at Function.registerForPushNotifications (/Website-SDK/src/shared/helpers/InitHelper.ts:104:5)
        at PromptsManager.internalShowNativePrompt (/Website-SDK/src/page/managers/PromptsManager.ts:188:5)
        at NotificationsNamespace.requestPermission (/Website-SDK/src/onesignal/NotificationsNamespace.ts:142:5)
        at PushSubscriptionNamespace.optIn (/Website-SDK/src/onesignal/PushSubscriptionNamespace.ts:96:7)
        at Object.<anonymous> (/Website-SDK/__test__/unit/push/registerForPush.test.ts:33:5) {
      reason: 1
    }

This change adds an assertion since the error is part of the test's expected behavior.

Even though the error occurs, the tests still pass. For that reason, I assumed it's not part of the test's expected behavior. Please feel free to correct me if my understanding there is wrong.

Note: If needed we can add something like expect(Log.error).toHaveBeenCalled() to ensure that we are testing for the existence of the error.

Systems Affected

• [x] WebSDK
• [ ] Backend
• [ ] Dashboard

Validation

Tests

Info

Checklist

• [x] All the automated tests pass or I explained why that is not possible
• [x] I have personally tested this on my machine or explained why that is not possible
• [x] I have included test coverage for these changes or explained why they are not needed

Programming Checklist Interfaces:

• [x] Don't use default export
• [x] New interfaces are in model files

Functions:

• [x] Don't use default export
• [x] All function signatures have return types
• [x] Helpers should not access any data but rather be given the data to operate on.

Typescript:

• [x] No Typescript warnings
• [x] Avoid silencing null/undefined warnings with the exclamation point

Other:

• [x] Iteration: refrain from using elem of array syntax. Prefer forEach or use map
• [x] Avoid using global OneSignal accessor for context if possible. Instead, we can pass it to function/constructor so that we don't call OneSignal.context

Screenshots

Info

Checklist

• [ ] I have included screenshots/recordings of the intended results or explained why they are not needed

---

Related Tickets

---

---

This change is 

[shepherd-l]

Even though the error occurs, the tests still pass. For that reason, I assumed it's not part of the test's expected behavior. Please feel free to correct me if my understanding there is wrong.

I was also unsure at first but I think it's part of the test's expected behaviour as discussed here: https://github.com/OneSignal/OneSignal-Website-SDK/pull/1192#pullrequestreview-2298504262

Note: If needed we can add something like expect(Log.error).toHaveBeenCalled() to ensure that we are testing for the existence of the error.

Good idea, this makes sense to do inorder to avoid confusion