fix(Litestar): Apply failed_request_status_codes to exceptions raised in middleware
#4074
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #4074 +/- ##
==========================================
+ Coverage 79.48% 79.49% +0.01%
==========================================
Files 141 141
Lines 15809 15815 +6
Branches 2703 2703
==========================================
+ Hits 12565 12572 +7
Misses 2382 2382
+ Partials 862 861 -1
|
|
Relates to #3134 |
|
@sentrivana @antonpirker @untitaker can you review this please? 🙂 |
sentry_sdk/integrations/asgi.py
Outdated
| @@ -57,17 +57,6 @@ | |||
| TRANSACTION_STYLE_VALUES = ("endpoint", "url") | |||
|
|
|||
|
|
|||
| def _capture_exception(exc, mechanism_type="asgi"): | |||
There was a problem hiding this comment.
I would prefer to keep this function the same, as a standalone function instead of a method. I don't think it has to be modified to implement your changes
sentry_sdk/integrations/asgi.py
Outdated
| def _capture_lifespan_exception(self, exc): | ||
| # type: (Exception) -> None | ||
| return self._capture_exception(exc) | ||
|
|
||
| def _capture_request_exception(self, exc): |
There was a problem hiding this comment.
I am a bit unsure of the difference between these two functions. Please add docstrings to both to clarify the distinction
sentry_sdk/integrations/litestar.py
Outdated
| def _capture_request_exception(self, exc): | ||
| # type: (Exception) -> None | ||
| """Ignore exceptions for requests here: we catch them in Litestar.after_exception handler.""" |
There was a problem hiding this comment.
I am slightly unsure why this function is needed/how it relates to what you are trying to explain in this PR. Could you please explain in a bit more detail?
Also, I think you need a pass here:
| def _capture_request_exception(self, exc): | |
| # type: (Exception) -> None | |
| """Ignore exceptions for requests here: we catch them in Litestar.after_exception handler.""" | |
| def _capture_request_exception(self, exc): | |
| # type: (Exception) -> None | |
| """Ignore exceptions for requests here: we catch them in Litestar.after_exception handler.""" | |
| pass |
There was a problem hiding this comment.
Currently, Litestar integration catches exceptions raised in middleware twice:
- In the custom ASGI handler.
- In the
Litestar.after_exceptionhandler, which is added inpatch_app_init.
The custom ASGI handler doesn't filter exceptions using failed_request_status_codes, so all exceptions (including those not filtered by the expected status code) appear in Sentry.
Interestingly, this behavior doesn't apply to exceptions raised in request handlers because Litestar prepends the built-in exception handling middleware.
This leads us to the conclusion that the ASGI handler only catches exceptions raised in middleware or lifespan handlers. It's still valuable to catch exceptions from lifespan handlers.
In this case, we patch _capture_request_exception to prevent catching exceptions raised in middleware.
…LitestarASGIMiddleware
… SentryAsgiMiddleware for better clarity and support for derived integrations.
This is a fix for #4021: exceptions raised in middleware were sent without taking into account
failed_request_status_codesvalue.See the test case for an example.
Thank you for contributing to
sentry-python! Please add tests to validate your changes, and lint your code usingtox -e linters.Running the test suite on your PR might require maintainer approval. The AWS Lambda tests additionally require a maintainer to add a special label, and they will fail until this label is added.