garminconnect/client.py (2)

894-898: ⚠️ Potential issue | 🟡 Minor

Stale docstring references diauth.garmin.com.

Now that the endpoint is domain-aware via self._di_token_url, the docstring claim "POST to diauth.garmin.com" is misleading for CN users — it's the one place a future reader would look to confirm the host is dynamic.

📝 Suggested doc tweak

         """Exchange a CAS service ticket for native DI + IT Bearer tokens.
 
-        POST to diauth.garmin.com to get a DI OAuth2 token, then exchange
-        for an IT token via services.garmin.com.
+        POST to ``self._di_token_url`` (``diauth.{domain}``) to get a DI
+        OAuth2 token, then exchange for an IT token via services.{domain}.
         """

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@garminconnect/client.py` around lines 894 - 898, The docstring for the method
that exchanges a CAS service ticket (the function which posts to obtain DI/IT
tokens and uses self._di_token_url) incorrectly hardcodes "POST to
diauth.garmin.com"; update the docstring to remove the static hostname and
instead describe that the code POSTs to the domain-aware DI token endpoint
referenced by self._di_token_url (or simply "the DI token endpoint"), and
mention that the host is determined by self._di_token_url so it works for CN or
other domains.

---

956-980: 🧹 Nitpick | 🔵 Trivial

Refresh path correctly uses the instance URL — verify CN refresh under load.

_refresh_di_token now POSTs to self._di_token_url, which directly addresses the reported invalid_grant "User doesn't exist in DB for GarminGUID" failure on CN. The PR notes manual verification with a garmin.cn account.

One thing worth noting (not blocking): on a refresh failure, _refresh_session (line 1028) only logs at debug level and swallows the exception, so a regression here would be silent — the next API call would just continue with the (possibly stale) di_token until it 401s and triggers a re-refresh loop. Consider whether a CN-domain integration test or at least a unit test asserting self._di_token_url is the POST target on domain="garmin.cn" would be worth adding to lock this down.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@garminconnect/client.py` around lines 956 - 980, Add a unit test that
verifies _refresh_di_token posts to the instance-specific URL by instantiating
the client with domain="garmin.cn" (or mocking instance resolution), mocking the
_http_post method and asserting it is called with self._di_token_url as the URL
and with the expected headers/data; specifically target the _refresh_di_token
method and mock _http_post so the test fails if code still posts to a hardcoded
CN endpoint. Optionally, add a small test that exercises _refresh_session to
confirm failures are at least logged (mocking _refresh_di_token to raise) so
regressions that silently swallow DI refresh errors are detected.