Merge e69577d246 into eb0e3f8c18

litellm/litellm_core_utils/exception_mapping_utils.py

@@ -1167,6 +1167,26 @@ def _map_vertex_exception(
             llm_provider=custom_llm_provider,
             litellm_debug_info=extra_information,
         )
+    elif (
+        "429 Quota exceeded" in error_str
+        or "Quota exceeded for" in error_str
+        or "Resource exhausted" in error_str
+        or "IndexError: list index out of range" in error_str
+        or "429 Unable to submit request because the service is temporarily out of capacity." in error_str
+    ):
+        raise RateLimitError(
+            message=f"litellm.RateLimitError: {custom_llm_provider}Exception - {error_str}",
+            model=model,
+            llm_provider=custom_llm_provider,
+            litellm_debug_info=extra_information,
+            response=httpx.Response(
+                status_code=429,
+                request=httpx.Request(
+                    method="POST",
+                    url=" https://cloud.google.com/vertex-ai/",
+                ),
+            ),
+        )
     elif "403" in error_str:
         raise BadRequestError(
             message=f"{custom_llm_provider.capitalize()}Exception BadRequestError - {error_str}",
@@ -1198,26 +1218,6 @@ def _map_vertex_exception(
                 ),
             ),
         )
-    elif (
-        "429 Quota exceeded" in error_str
-        or "Quota exceeded for" in error_str
-        or "Resource exhausted" in error_str
-        or "IndexError: list index out of range" in error_str
-        or "429 Unable to submit request because the service is temporarily out of capacity." in error_str
-    ):
-        raise RateLimitError(
-            message=f"litellm.RateLimitError: {custom_llm_provider}Exception - {error_str}",
-            model=model,
-            llm_provider=custom_llm_provider,
-            litellm_debug_info=extra_information,
-            response=httpx.Response(
-                status_code=429,
-                request=httpx.Request(
-                    method="POST",
-                    url=" https://cloud.google.com/vertex-ai/",
-                ),
-            ),
-        )
     elif (
         isinstance(getattr(original_exception, "status_code", None), int)
         and 500 <= original_exception.status_code < 600

tests/test_litellm/litellm_core_utils/test_exception_mapping_utils.py

@@ -452,6 +452,55 @@ def test_vertex_ai_rate_limit_error_mapping(error_message, should_raise_rate_lim
             )
 
 
+# Regression tests for https://github.com/BerriAI/litellm/issues/34954
+# A Gemini/Vertex 429 whose RESOURCE_EXHAUSTED body carries a sub-second retry
+# hint such as "Please retry in 18.403470473s." must still map to RateLimitError.
+# The digits "403" in that delay used to hit an unanchored `"403" in error_str`
+# branch that was evaluated before the quota branch, yielding BadRequestError(403)
+# and disabling Router retries for a transient rate limit.
+def _gemini_quota_body(retry_delay: str) -> str:
+    return (
+        '{"error": {"code": 429, "message": "You exceeded your current quota. '
+        "Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, "
+        "limit: 15. Please retry in " + retry_delay + '.", "status": "RESOURCE_EXHAUSTED"}}'
+    )
+
+
+@pytest.mark.parametrize(
+    "retry_delay",
+    [
+        "18.403470473s",  # digits contain "403" - the bug case
+        "18.9s",  # control: no "403" digits, already mapped correctly
+    ],
+)
+def test_gemini_429_quota_maps_to_rate_limit_regardless_of_retry_delay(retry_delay):
+    original_exception = Exception(_gemini_quota_body(retry_delay))
+
+    with pytest.raises(litellm.RateLimitError) as excinfo:
+        exception_type(
+            model="gemini/gemini-2.5-flash",
+            original_exception=original_exception,
+            custom_llm_provider="vertex_ai",
+        )
+    assert excinfo.value.status_code == 429
+
+
+def test_vertex_genuine_403_still_maps_to_bad_request():
+    body = (
+        '{"error": {"code": 403, "message": "Permission denied on resource project foo.", '
+        '"status": "PERMISSION_DENIED"}}'
+    )
+    original_exception = Exception(body)
+
+    with pytest.raises(litellm.BadRequestError) as excinfo:
+        exception_type(
+            model="gemini/gemini-2.5-flash",
+            original_exception=original_exception,
+            custom_llm_provider="vertex_ai",
+        )
+    assert excinfo.value.status_code == 403
+
+
 class TestGetBodyErrorCode:
     """Unit tests for _get_body_error_code helper."""