Added test cases for services: form-detector, groq-service, session-tracker.#24
Conversation
| }); | ||
|
|
||
| test('getIntentContext resets expired session', async () => { | ||
| const oldTime = Date.now() - (3 * 60 * 60 * 1000); |
There was a problem hiding this comment.
The TTL in the actual code is 2 hours, but this test uses 3 hours. It passes today by accident. Change the hardcoded 3 * to read sessionTracker.SESSION_TTL_MS directly so if someone changes the TTL later, the test actually catches it. Also worth adding a second test that checks a session just before expiry is NOT reset.
| apiSpy.mockRestore(); | ||
| }); | ||
|
|
||
| test('calls Groq API and parses valid JSON response', async () => { |
There was a problem hiding this comment.
This test only checks the happy path but never confirms isFormFill is absent. The whole point of the form-fill shortcut is that it sets isFormFill: true and skips the API. Without asserting isFormFill is falsy on the API path, someone could accidentally set it everywhere and this test wouldn't catch it. Add expect(result.isFormFill).toBeFalsy().
| expect(result.suggestions[0].text).toBe('suggestion one'); | ||
| }); | ||
|
|
||
| test('parseResponse returns empty suggestions on invalid payload', () => { |
There was a problem hiding this comment.
The console.error call inside parseResponse fires visibly every time this test runs. It's expected behaviour but pollutes CI output. Wrap the call in jest.spyOn(console, 'error').mockImplementation(() => {}) and restore it after.
There was a problem hiding this comment.
The retry logic on 429 responses in callWithRetry has no test at all. If someone breaks that retry, nothing fails. Worth adding a test that mocks the first fetch as 429 and the second as success, and asserts fetch was called twice.
There was a problem hiding this comment.
There's a bug in the source worth flagging: _detectOS checks for Windows NT 11 but real Windows 11 browsers still report Windows NT 10.0 in the UA string — same as Windows 10. That branch can never match. The tests don't cover this so the bug is invisible. Either fix _detectOS to use navigator.userAgentData instead, or at minimum add a test with a real Windows 11 UA string that documents what actually comes back.
Shantanugupta43
left a comment
Shantanugupta43
left a comment
There was a problem hiding this comment.
Hi thanks for contributing there are few improvements worth addressing.
2 participants
Added some test cases covering functionality for test cases.