Add support for third party errors

weaviate/collections/grpc/query.py

@@ -83,13 +83,15 @@ def __init__(
         validate_arguments: bool,
         uses_125_api: bool,
         uses_127_api: bool,
+        uses_133_api: bool,
     ):
         super().__init__(weaviate_version, consistency_level, validate_arguments)
         self._name: str = name
         self._tenant = tenant
         self._validate_arguments = validate_arguments
         self.__uses_125_api = uses_125_api
         self.__uses_127_api = uses_127_api
+        self.__uses_133_api = uses_133_api
 
     def __parse_near_options(
         self,
@@ -474,6 +476,7 @@ def __create_request(
             uses_123_api=True,
             uses_125_api=self.__uses_125_api,
             uses_127_api=self.__uses_127_api,
+            uses_133_api=self.__uses_133_api,
             collection=self._name,
             limit=limit,
             offset=offset,

weaviate/collections/queries/base_executor.py

@@ -85,6 +85,7 @@ def __init__(
 
         self.__uses_125_api = connection._weaviate_version.is_at_least(1, 25, 0)
         self.__uses_127_api = connection._weaviate_version.is_at_least(1, 27, 0)
+        self.__uses_133_api = connection._weaviate_version.is_at_least(1, 33, 0)
         self._query = _QueryGRPC(
             connection._weaviate_version,
             self._name,
@@ -93,6 +94,7 @@ def __init__(
             validate_arguments=self._validate_arguments,
             uses_125_api=self.__uses_125_api,
             uses_127_api=self.__uses_127_api,
+            uses_133_api=self.__uses_133_api,
         )
 
     def __retrieve_timestamp(

weaviate/connect/v4.py

@@ -73,6 +73,7 @@
     WeaviateGRPCUnavailableError,
     WeaviateInvalidInputError,
     WeaviateQueryError,
+    WeaviateQueryThirdPartyError,
     WeaviateRetryError,
     WeaviateStartUpError,
     WeaviateTenantGetError,
@@ -187,6 +188,38 @@ def set_integrations(self, integrations_config: List[_IntegrationConfig]) -> Non
             self._headers.update(integration._to_header())
             self.__additional_headers.update(integration._to_header())
 
+    def _parse_search_errors(self, sr: search_get_pb2.SearchReply) -> None:
+        if sr.third_party_error is not None and sr.third_party_error.provider_name != "":
+            raise WeaviateQueryThirdPartyError(
+                sr.third_party_error.full_error,
+                sr.third_party_error.provider_name,
+                sr.third_party_error.error_from_provider,
+                sr.third_party_error.status_code,
+                sr.third_party_error.request_id,
+                "GRPC search",
+            )
+        if sr.internal_error is not None and sr.internal_error != "":
+            raise WeaviateQueryError(sr.internal_error, "GRPC search")
+
+    def _parse_batch_errors(self, res: batch_pb2.BatchObjectsReply) -> Dict[int, str]:
+        objects: Dict[int, str] = {}
+        for err in res.errors:
+            if err.third_party_error is not None:
+                objects[err.index] = f"""
+                Provider ({err.third_party_error.provider_name}) service error {err.third_party_error.status_code}
+
+                    Error details:
+
+                    'Provider:' {err.third_party_error.provider_name}
+                    'Provider error code:' {err.third_party_error.status_code}
+                    'Provider message:' {err.third_party_error.error_from_provider}
+
+                    Weaviate error message: {err.third_party_error.full_error}
+                    """
+                continue
+            objects[err.index] = err.error
+        return objects
+
     @overload
     def _make_client(self, colour: Literal["async"]) -> AsyncClient: ...
 
@@ -990,11 +1023,14 @@ def grpc_search(self, request: search_get_pb2.SearchRequest) -> search_get_pb2.S
                 metadata=self.grpc_headers(),
                 timeout=self.timeout_config.query,
             )
-            return cast(search_get_pb2.SearchReply, res)
+            sr = cast(search_get_pb2.SearchReply, res)
+            self._parse_search_errors(sr)
+            return sr
         except RpcError as e:
             error = cast(Call, e)
             if error.code() == StatusCode.PERMISSION_DENIED:
                 raise InsufficientPermissionsError(error)
+
             raise WeaviateQueryError(str(error.details()), "GRPC search")  # pyright: ignore
         except WeaviateRetryError as e:
             raise WeaviateQueryError(str(e), "GRPC search")  # pyright: ignore
@@ -1016,11 +1052,7 @@ def grpc_batch_objects(
                 timeout=timeout,
             )
             res = cast(batch_pb2.BatchObjectsReply, res)
-
-            objects: Dict[int, str] = {}
-            for err in res.errors:
-                objects[err.index] = err.error
-            return objects
+            return self._parse_batch_errors(res)
         except RpcError as e:
             error = cast(Call, e)
             if error.code() == StatusCode.PERMISSION_DENIED:
@@ -1176,7 +1208,9 @@ async def grpc_search(
                 metadata=self.grpc_headers(),
                 timeout=self.timeout_config.query,
             )
-            return cast(search_get_pb2.SearchReply, res)
+            sr = cast(search_get_pb2.SearchReply, res)
+            self._parse_search_errors(sr)
+            return sr
         except AioRpcError as e:
             if e.code().name == PERMISSION_DENIED:
                 raise InsufficientPermissionsError(e)
@@ -1201,11 +1235,8 @@ async def grpc_batch_objects(
                 timeout=timeout,
             )
             res = cast(batch_pb2.BatchObjectsReply, res)
+            return self._parse_batch_errors(res)
 
-            objects: Dict[int, str] = {}
-            for err in res.errors:
-                objects[err.index] = err.error
-            return objects
         except AioRpcError as e:
             if e.code().name == PERMISSION_DENIED:
                 raise InsufficientPermissionsError(e)

weaviate/exceptions.py

@@ -375,3 +375,29 @@ def __init__(self) -> None:
         super().__init__(
             'Weaviate Agents (Alpha) functionality requires additional dependencies. Please install them using: "pip install weaviate-client[agents]"'
         )
+
+
+class WeaviateQueryThirdPartyError(WeaviateQueryError):
+    """Is raised if a query (either gRPC or GraphQL) to Weaviate fails in any way."""
+
+    def __init__(
+        self,
+        full_error: str,
+        provider_name: str,
+        provider_error: str,
+        error_code: int,
+        request_id: str,
+        protocol_type: str,
+    ):
+        request_id_str = f"with request_id {request_id} " if request_id else ""
+
+        msg = f"""Query call with protocol GRPC search failed with message Query with request_id {request_id_str}  failed:
+                Provider ({provider_name}) service error {error_code}.
+                Error details:
+        {"Provider:":>32} {provider_name}
+        {"Provider error code:":>32} {error_code}
+        {"Provider message:":>32} {provider_error}
+                Weaviate error message:
+                {full_error}"""
+        super().__init__(msg, protocol_type)
+        self.message = msg

weaviate/proto/v1/v4/v1/base_pb2.pyi

@@ -237,3 +237,17 @@ class Vectors(_message.Message):
     vector_bytes: bytes
     type: Vectors.VectorType
     def __init__(self, name: _Optional[str] = ..., index: _Optional[int] = ..., vector_bytes: _Optional[bytes] = ..., type: _Optional[_Union[Vectors.VectorType, str]] = ...) -> None: ...
+
+class ThirdPartyError(_message.Message):
+    __slots__ = ["provider_name", "error_from_provider", "full_error", "status_code", "request_id"]
+    PROVIDER_NAME_FIELD_NUMBER: _ClassVar[int]
+    ERROR_FROM_PROVIDER_FIELD_NUMBER: _ClassVar[int]
+    FULL_ERROR_FIELD_NUMBER: _ClassVar[int]
+    STATUS_CODE_FIELD_NUMBER: _ClassVar[int]
+    REQUEST_ID_FIELD_NUMBER: _ClassVar[int]
+    provider_name: str
+    error_from_provider: str
+    full_error: str
+    status_code: int
+    request_id: str
+    def __init__(self, provider_name: _Optional[str] = ..., error_from_provider: _Optional[str] = ..., full_error: _Optional[str] = ..., status_code: _Optional[int] = ..., request_id: _Optional[str] = ...) -> None: ...