fix(falcon): Don't exhaust request body stream

Author: szokeasaurusrex

sentry_sdk/integrations/falcon.py

@@ -43,6 +43,12 @@
     FALCON3 = False
 
 
+_FALCON_UNSET = None  # type: Optional[object]
+if FALCON3:  # falcon.request._UNSET is only available in Falcon 3.0+
+    with capture_internal_exceptions():
+        from falcon.request import _UNSET as _FALCON_UNSET  # type: ignore[import-not-found, no-redef]
+
+
 class FalconRequestExtractor(RequestExtractor):
     def env(self):
         # type: () -> Dict[str, Any]
@@ -73,27 +79,23 @@ def raw_data(self):
         else:
             return None
 
-    if FALCON3:
-
-        def json(self):
-            # type: () -> Optional[Dict[str, Any]]
-            try:
-                return self.request.media
-            except falcon.errors.HTTPBadRequest:
-                return None
-
-    else:
-
-        def json(self):
-            # type: () -> Optional[Dict[str, Any]]
-            try:
-                return self.request.media
-            except falcon.errors.HTTPBadRequest:
-                # NOTE(jmagnusson): We return `falcon.Request._media` here because
-                # falcon 1.4 doesn't do proper type checking in
-                # `falcon.Request.media`. This has been fixed in 2.0.
-                # Relevant code: https://github.com/falconry/falcon/blob/1.4.1/falcon/request.py#L953
-                return self.request._media
+    def json(self):
+        # type: () -> Optional[Dict[str, Any]]
+        # fallback to cached_media = None if self.request._media is not available
+        cached_media = None
+        with capture_internal_exceptions():
+            # self.request._media is the cached self.request.media
+            # value. It is only available if self.request.media
+            # has already been accessed. Therefore, reading
+            # self.request._media will not exhaust the raw request
+            # stream (self.request.bounded_stream) because it has
+            # already been read if self.request._media is set.
+            cached_media = self.request._media
+
+        if cached_media is not _FALCON_UNSET:
+            return cached_media
+
+        return None
 
 
 class SentryFalconMiddleware:

tests/integrations/falcon/test_falcon.py

@@ -211,9 +211,12 @@ def test_falcon_large_json_request(sentry_init, capture_events):
 
     data = {"foo": {"bar": "a" * 2000}}
 
+    assert_passed = False
+
     class Resource:
         def on_post(self, req, resp):
-            assert req.media == data
+            nonlocal assert_passed
+            assert_passed = req.media == data
             sentry_sdk.capture_message("hi")
             resp.media = "ok"
 
@@ -225,7 +228,7 @@ def on_post(self, req, resp):
     client = falcon.testing.TestClient(app)
     response = client.simulate_post("/", json=data)
     assert response.status == falcon.HTTP_200
-
+    assert assert_passed
     (event,) = events
     assert event["_meta"]["request"]["data"]["foo"]["bar"] == {
         "": {"len": 2000, "rem": [["!limit", "x", 1021, 1024]]}
@@ -460,3 +463,48 @@ def test_span_origin(sentry_init, capture_events, make_client):
     (_, event) = events
 
     assert event["contexts"]["trace"]["origin"] == "auto.http.falcon"
+
+
+def test_falcon_request_media(sentry_init):
+    # test_passed stores whether the test has passed.
+    test_passed = False
+
+    # test_failure_reason stores the reason why the test failed
+    # if test_passed is False. The value is meaningless when
+    # test_passed is True.
+    test_failure_reason = "test endpoint did not get called"
+
+    class SentryCaptureMiddleware:
+        def process_request(self, _req, _resp):
+            # This capture message forces Falcon event processors to run
+            # before the request handler runs
+            sentry_sdk.capture_message("Processing request")
+
+    class RequestMediaResource:
+        def on_post(self, req, _):
+            nonlocal test_passed, test_failure_reason
+            raw_data = req.bounded_stream.read()
+
+            # If the raw_data is empty, the request body stream
+            # has been exhausted by the SDK. Test should fail in
+            # this case.
+            test_passed = raw_data != b""
+            test_failure_reason = "request body has been read"
+
+    sentry_init(integrations=[FalconIntegration()])
+
+    try:
+        app_class = falcon.App  # Falcon ≥3.0
+    except AttributeError:
+        app_class = falcon.API  # Falcon <3.0
+
+    app = app_class(middleware=[SentryCaptureMiddleware()])
+    app.add_route("/read_body", RequestMediaResource())
+
+    client = falcon.testing.TestClient(app)
+
+    client.simulate_post("/read_body", json={"foo": "bar"})
+
+    # Check that simulate_post actually calls the resource, and
+    # that the SDK does not exhaust the request body stream.
+    assert test_passed, test_failure_reason