mistralai 1.5.2__py3-none-any.whl → 1.6.0__py3-none-any.whl

mistralai/_hooks/types.py

@@ -7,16 +7,19 @@ from typing import Any, Callable, List, Optional, Tuple, Union
 
 
 class HookContext:
+    base_url: str
     operation_id: str
     oauth2_scopes: Optional[List[str]] = None
     security_source: Optional[Union[Any, Callable[[], Any]]] = None
 
     def __init__(
         self,
+        base_url: str,
         operation_id: str,
         oauth2_scopes: Optional[List[str]],
         security_source: Optional[Union[Any, Callable[[], Any]]],
     ):
+        self.base_url = base_url
         self.operation_id = operation_id
         self.oauth2_scopes = oauth2_scopes
         self.security_source = security_source
@@ -25,21 +28,30 @@ class HookContext:
 class BeforeRequestContext(HookContext):
     def __init__(self, hook_ctx: HookContext):
         super().__init__(
-            hook_ctx.operation_id, hook_ctx.oauth2_scopes, hook_ctx.security_source
+            hook_ctx.base_url,
+            hook_ctx.operation_id,
+            hook_ctx.oauth2_scopes,
+            hook_ctx.security_source,
         )
 
 
 class AfterSuccessContext(HookContext):
     def __init__(self, hook_ctx: HookContext):
         super().__init__(
-            hook_ctx.operation_id, hook_ctx.oauth2_scopes, hook_ctx.security_source
+            hook_ctx.base_url,
+            hook_ctx.operation_id,
+            hook_ctx.oauth2_scopes,
+            hook_ctx.security_source,
         )
 
 
 class AfterErrorContext(HookContext):
     def __init__(self, hook_ctx: HookContext):
         super().__init__(
-            hook_ctx.operation_id, hook_ctx.oauth2_scopes, hook_ctx.security_source
+            hook_ctx.base_url,
+            hook_ctx.operation_id,
+            hook_ctx.oauth2_scopes,
+            hook_ctx.security_source,
         )
 
 

mistralai/_version.py

@@ -3,10 +3,10 @@
 import importlib.metadata
 
 __title__: str = "mistralai"
-__version__: str = "1.5.2"
+__version__: str = "1.6.0"
 __openapi_doc_version__: str = "0.0.2"
-__gen_version__: str = "2.497.0"
-__user_agent__: str = "speakeasy-sdk/python 1.5.2 2.497.0 0.0.2 mistralai"
+__gen_version__: str = "2.548.6"
+__user_agent__: str = "speakeasy-sdk/python 1.6.0 2.548.6 0.0.2 mistralai"
 
 try:
     if __package__ is not None:

mistralai/agents.py

@@ -46,6 +46,7 @@ class Agents(BaseSDK):
         prediction: Optional[
             Union[models.Prediction, models.PredictionTypedDict]
         ] = None,
+        parallel_tool_calls: Optional[bool] = None,
         retries: OptionalNullable[utils.RetryConfig] = UNSET,
         server_url: Optional[str] = None,
         timeout_ms: Optional[int] = None,
@@ -66,6 +67,7 @@ class Agents(BaseSDK):
         :param frequency_penalty: frequency_penalty penalizes the repetition of words based on their frequency in the generated text. A higher frequency penalty discourages the model from repeating words that have already appeared frequently in the output, promoting diversity and reducing repetition.
         :param n: Number of completions to return for each request, input tokens are only billed once.
         :param prediction:
+        :param parallel_tool_calls:
         :param retries: Override the default retry configuration for this method
         :param server_url: Override the default server URL for this method
         :param timeout_ms: Override the default request timeout configuration for this method in milliseconds
@@ -78,6 +80,8 @@ class Agents(BaseSDK):
 
         if server_url is not None:
             base_url = server_url
+        else:
+            base_url = self._get_url(base_url, url_variables)
 
         request = models.AgentsCompletionRequest(
             max_tokens=max_tokens,
@@ -100,6 +104,7 @@ class Agents(BaseSDK):
             prediction=utils.get_pydantic_model(
                 prediction, Optional[models.Prediction]
             ),
+            parallel_tool_calls=parallel_tool_calls,
             agent_id=agent_id,
         )
 
@@ -132,6 +137,7 @@ class Agents(BaseSDK):
 
         http_res = self.do_request(
             hook_ctx=HookContext(
+                base_url=base_url or "",
                 operation_id="agents_completion_v1_agents_completions_post",
                 oauth2_scopes=[],
                 security_source=get_security_from_env(
@@ -143,12 +149,14 @@ class Agents(BaseSDK):
             retry_config=retry_config,
         )
 
-        data: Any = None
+        response_data: Any = None
         if utils.match_response(http_res, "200", "application/json"):
             return utils.unmarshal_json(http_res.text, models.ChatCompletionResponse)
         if utils.match_response(http_res, "422", "application/json"):
-            data = utils.unmarshal_json(http_res.text, models.HTTPValidationErrorData)
-            raise models.HTTPValidationError(data=data)
+            response_data = utils.unmarshal_json(
+                http_res.text, models.HTTPValidationErrorData
+            )
+            raise models.HTTPValidationError(data=response_data)
         if utils.match_response(http_res, "4XX", "*"):
             http_res_text = utils.stream_to_text(http_res)
             raise models.SDKError(
@@ -204,6 +212,7 @@ class Agents(BaseSDK):
         prediction: Optional[
             Union[models.Prediction, models.PredictionTypedDict]
         ] = None,
+        parallel_tool_calls: Optional[bool] = None,
         retries: OptionalNullable[utils.RetryConfig] = UNSET,
         server_url: Optional[str] = None,
         timeout_ms: Optional[int] = None,
@@ -224,6 +233,7 @@ class Agents(BaseSDK):
         :param frequency_penalty: frequency_penalty penalizes the repetition of words based on their frequency in the generated text. A higher frequency penalty discourages the model from repeating words that have already appeared frequently in the output, promoting diversity and reducing repetition.
         :param n: Number of completions to return for each request, input tokens are only billed once.
         :param prediction:
+        :param parallel_tool_calls:
         :param retries: Override the default retry configuration for this method
         :param server_url: Override the default server URL for this method
         :param timeout_ms: Override the default request timeout configuration for this method in milliseconds
@@ -236,6 +246,8 @@ class Agents(BaseSDK):
 
         if server_url is not None:
             base_url = server_url
+        else:
+            base_url = self._get_url(base_url, url_variables)
 
         request = models.AgentsCompletionRequest(
             max_tokens=max_tokens,
@@ -258,6 +270,7 @@ class Agents(BaseSDK):
             prediction=utils.get_pydantic_model(
                 prediction, Optional[models.Prediction]
             ),
+            parallel_tool_calls=parallel_tool_calls,
             agent_id=agent_id,
         )
 
@@ -290,6 +303,7 @@ class Agents(BaseSDK):
 
         http_res = await self.do_request_async(
             hook_ctx=HookContext(
+                base_url=base_url or "",
                 operation_id="agents_completion_v1_agents_completions_post",
                 oauth2_scopes=[],
                 security_source=get_security_from_env(
@@ -301,12 +315,14 @@ class Agents(BaseSDK):
             retry_config=retry_config,
         )
 
-        data: Any = None
+        response_data: Any = None
         if utils.match_response(http_res, "200", "application/json"):
             return utils.unmarshal_json(http_res.text, models.ChatCompletionResponse)
         if utils.match_response(http_res, "422", "application/json"):
-            data = utils.unmarshal_json(http_res.text, models.HTTPValidationErrorData)
-            raise models.HTTPValidationError(data=data)
+            response_data = utils.unmarshal_json(
+                http_res.text, models.HTTPValidationErrorData
+            )
+            raise models.HTTPValidationError(data=response_data)
         if utils.match_response(http_res, "4XX", "*"):
             http_res_text = await utils.stream_to_text_async(http_res)
             raise models.SDKError(
@@ -362,6 +378,7 @@ class Agents(BaseSDK):
         prediction: Optional[
             Union[models.Prediction, models.PredictionTypedDict]
         ] = None,
+        parallel_tool_calls: Optional[bool] = None,
         retries: OptionalNullable[utils.RetryConfig] = UNSET,
         server_url: Optional[str] = None,
         timeout_ms: Optional[int] = None,
@@ -384,6 +401,7 @@ class Agents(BaseSDK):
         :param frequency_penalty: frequency_penalty penalizes the repetition of words based on their frequency in the generated text. A higher frequency penalty discourages the model from repeating words that have already appeared frequently in the output, promoting diversity and reducing repetition.
         :param n: Number of completions to return for each request, input tokens are only billed once.
         :param prediction:
+        :param parallel_tool_calls:
         :param retries: Override the default retry configuration for this method
         :param server_url: Override the default server URL for this method
         :param timeout_ms: Override the default request timeout configuration for this method in milliseconds
@@ -396,6 +414,8 @@ class Agents(BaseSDK):
 
         if server_url is not None:
             base_url = server_url
+        else:
+            base_url = self._get_url(base_url, url_variables)
 
         request = models.AgentsCompletionStreamRequest(
             max_tokens=max_tokens,
@@ -418,6 +438,7 @@ class Agents(BaseSDK):
             prediction=utils.get_pydantic_model(
                 prediction, Optional[models.Prediction]
             ),
+            parallel_tool_calls=parallel_tool_calls,
             agent_id=agent_id,
         )
 
@@ -450,6 +471,7 @@ class Agents(BaseSDK):
 
         http_res = self.do_request(
             hook_ctx=HookContext(
+                base_url=base_url or "",
                 operation_id="stream_agents",
                 oauth2_scopes=[],
                 security_source=get_security_from_env(
@@ -462,7 +484,7 @@ class Agents(BaseSDK):
             retry_config=retry_config,
         )
 
-        data: Any = None
+        response_data: Any = None
         if utils.match_response(http_res, "200", "text/event-stream"):
             return eventstreaming.EventStream(
                 http_res,
@@ -471,8 +493,10 @@ class Agents(BaseSDK):
             )
         if utils.match_response(http_res, "422", "application/json"):
             http_res_text = utils.stream_to_text(http_res)
-            data = utils.unmarshal_json(http_res_text, models.HTTPValidationErrorData)
-            raise models.HTTPValidationError(data=data)
+            response_data = utils.unmarshal_json(
+                http_res_text, models.HTTPValidationErrorData
+            )
+            raise models.HTTPValidationError(data=response_data)
         if utils.match_response(http_res, "4XX", "*"):
             http_res_text = utils.stream_to_text(http_res)
             raise models.SDKError(
@@ -528,6 +552,7 @@ class Agents(BaseSDK):
         prediction: Optional[
             Union[models.Prediction, models.PredictionTypedDict]
         ] = None,
+        parallel_tool_calls: Optional[bool] = None,
         retries: OptionalNullable[utils.RetryConfig] = UNSET,
         server_url: Optional[str] = None,
         timeout_ms: Optional[int] = None,
@@ -550,6 +575,7 @@ class Agents(BaseSDK):
         :param frequency_penalty: frequency_penalty penalizes the repetition of words based on their frequency in the generated text. A higher frequency penalty discourages the model from repeating words that have already appeared frequently in the output, promoting diversity and reducing repetition.
         :param n: Number of completions to return for each request, input tokens are only billed once.
         :param prediction:
+        :param parallel_tool_calls:
         :param retries: Override the default retry configuration for this method
         :param server_url: Override the default server URL for this method
         :param timeout_ms: Override the default request timeout configuration for this method in milliseconds
@@ -562,6 +588,8 @@ class Agents(BaseSDK):
 
         if server_url is not None:
             base_url = server_url
+        else:
+            base_url = self._get_url(base_url, url_variables)
 
         request = models.AgentsCompletionStreamRequest(
             max_tokens=max_tokens,
@@ -584,6 +612,7 @@ class Agents(BaseSDK):
             prediction=utils.get_pydantic_model(
                 prediction, Optional[models.Prediction]
             ),
+            parallel_tool_calls=parallel_tool_calls,
             agent_id=agent_id,
         )
 
@@ -616,6 +645,7 @@ class Agents(BaseSDK):
 
         http_res = await self.do_request_async(
             hook_ctx=HookContext(
+                base_url=base_url or "",
                 operation_id="stream_agents",
                 oauth2_scopes=[],
                 security_source=get_security_from_env(
@@ -628,7 +658,7 @@ class Agents(BaseSDK):
             retry_config=retry_config,
         )
 
-        data: Any = None
+        response_data: Any = None
         if utils.match_response(http_res, "200", "text/event-stream"):
             return eventstreaming.EventStreamAsync(
                 http_res,
@@ -637,8 +667,10 @@ class Agents(BaseSDK):
             )
         if utils.match_response(http_res, "422", "application/json"):
             http_res_text = await utils.stream_to_text_async(http_res)
-            data = utils.unmarshal_json(http_res_text, models.HTTPValidationErrorData)
-            raise models.HTTPValidationError(data=data)
+            response_data = utils.unmarshal_json(
+                http_res_text, models.HTTPValidationErrorData
+            )
+            raise models.HTTPValidationError(data=response_data)
         if utils.match_response(http_res, "4XX", "*"):
             http_res_text = await utils.stream_to_text_async(http_res)
             raise models.SDKError(

mistralai/basesdk.py

@@ -231,6 +231,10 @@ class BaseSDK:
                     req.headers,
                     get_body_content(req),
                 )
+
+                if client is None:
+                    raise ValueError("client is required")
+
                 http_res = client.send(req, stream=stream)
             except Exception as e:
                 _, e = self.sdk_configuration.get_hooks().after_error(
@@ -303,6 +307,10 @@ class BaseSDK:
                     req.headers,
                     get_body_content(req),
                 )
+
+                if client is None:
+                    raise ValueError("client is required")
+
                 http_res = await client.send(req, stream=stream)
             except Exception as e:
                 _, e = self.sdk_configuration.get_hooks().after_error(

mistralai/chat.py

@@ -122,6 +122,7 @@ class Chat(BaseSDK):
         prediction: Optional[
             Union[models.Prediction, models.PredictionTypedDict]
         ] = None,
+        parallel_tool_calls: Optional[bool] = None,
         safe_prompt: Optional[bool] = None,
         retries: OptionalNullable[utils.RetryConfig] = UNSET,
         server_url: Optional[str] = None,
@@ -145,6 +146,7 @@ class Chat(BaseSDK):
         :param frequency_penalty: frequency_penalty penalizes the repetition of words based on their frequency in the generated text. A higher frequency penalty discourages the model from repeating words that have already appeared frequently in the output, promoting diversity and reducing repetition.
         :param n: Number of completions to return for each request, input tokens are only billed once.
         :param prediction:
+        :param parallel_tool_calls:
         :param safe_prompt: Whether to inject a safety prompt before all conversations.
         :param retries: Override the default retry configuration for this method
         :param server_url: Override the default server URL for this method
@@ -158,6 +160,8 @@ class Chat(BaseSDK):
 
         if server_url is not None:
             base_url = server_url
+        else:
+            base_url = self._get_url(base_url, url_variables)
 
         request = models.ChatCompletionRequest(
             model=model,
@@ -181,6 +185,7 @@ class Chat(BaseSDK):
             prediction=utils.get_pydantic_model(
                 prediction, Optional[models.Prediction]
             ),
+            parallel_tool_calls=parallel_tool_calls,
             safe_prompt=safe_prompt,
         )
 
@@ -213,6 +218,7 @@ class Chat(BaseSDK):
 
         http_res = self.do_request(
             hook_ctx=HookContext(
+                base_url=base_url or "",
                 operation_id="chat_completion_v1_chat_completions_post",
                 oauth2_scopes=[],
                 security_source=get_security_from_env(
@@ -224,12 +230,14 @@ class Chat(BaseSDK):
             retry_config=retry_config,
         )
 
-        data: Any = None
+        response_data: Any = None
         if utils.match_response(http_res, "200", "application/json"):
             return utils.unmarshal_json(http_res.text, models.ChatCompletionResponse)
         if utils.match_response(http_res, "422", "application/json"):
-            data = utils.unmarshal_json(http_res.text, models.HTTPValidationErrorData)
-            raise models.HTTPValidationError(data=data)
+            response_data = utils.unmarshal_json(
+                http_res.text, models.HTTPValidationErrorData
+            )
+            raise models.HTTPValidationError(data=response_data)
         if utils.match_response(http_res, "4XX", "*"):
             http_res_text = utils.stream_to_text(http_res)
             raise models.SDKError(
@@ -279,6 +287,7 @@ class Chat(BaseSDK):
         prediction: Optional[
             Union[models.Prediction, models.PredictionTypedDict]
         ] = None,
+        parallel_tool_calls: Optional[bool] = None,
         safe_prompt: Optional[bool] = None,
         retries: OptionalNullable[utils.RetryConfig] = UNSET,
         server_url: Optional[str] = None,
@@ -302,6 +311,7 @@ class Chat(BaseSDK):
         :param frequency_penalty: frequency_penalty penalizes the repetition of words based on their frequency in the generated text. A higher frequency penalty discourages the model from repeating words that have already appeared frequently in the output, promoting diversity and reducing repetition.
         :param n: Number of completions to return for each request, input tokens are only billed once.
         :param prediction:
+        :param parallel_tool_calls:
         :param safe_prompt: Whether to inject a safety prompt before all conversations.
         :param retries: Override the default retry configuration for this method
         :param server_url: Override the default server URL for this method
@@ -315,6 +325,8 @@ class Chat(BaseSDK):
 
         if server_url is not None:
             base_url = server_url
+        else:
+            base_url = self._get_url(base_url, url_variables)
 
         request = models.ChatCompletionRequest(
             model=model,
@@ -338,6 +350,7 @@ class Chat(BaseSDK):
             prediction=utils.get_pydantic_model(
                 prediction, Optional[models.Prediction]
             ),
+            parallel_tool_calls=parallel_tool_calls,
             safe_prompt=safe_prompt,
         )
 
@@ -370,6 +383,7 @@ class Chat(BaseSDK):
 
         http_res = await self.do_request_async(
             hook_ctx=HookContext(
+                base_url=base_url or "",
                 operation_id="chat_completion_v1_chat_completions_post",
                 oauth2_scopes=[],
                 security_source=get_security_from_env(
@@ -381,12 +395,14 @@ class Chat(BaseSDK):
             retry_config=retry_config,
         )
 
-        data: Any = None
+        response_data: Any = None
         if utils.match_response(http_res, "200", "application/json"):
             return utils.unmarshal_json(http_res.text, models.ChatCompletionResponse)
         if utils.match_response(http_res, "422", "application/json"):
-            data = utils.unmarshal_json(http_res.text, models.HTTPValidationErrorData)
-            raise models.HTTPValidationError(data=data)
+            response_data = utils.unmarshal_json(
+                http_res.text, models.HTTPValidationErrorData
+            )
+            raise models.HTTPValidationError(data=response_data)
         if utils.match_response(http_res, "4XX", "*"):
             http_res_text = await utils.stream_to_text_async(http_res)
             raise models.SDKError(
@@ -444,6 +460,7 @@ class Chat(BaseSDK):
         prediction: Optional[
             Union[models.Prediction, models.PredictionTypedDict]
         ] = None,
+        parallel_tool_calls: Optional[bool] = None,
         safe_prompt: Optional[bool] = None,
         retries: OptionalNullable[utils.RetryConfig] = UNSET,
         server_url: Optional[str] = None,
@@ -469,6 +486,7 @@ class Chat(BaseSDK):
         :param frequency_penalty: frequency_penalty penalizes the repetition of words based on their frequency in the generated text. A higher frequency penalty discourages the model from repeating words that have already appeared frequently in the output, promoting diversity and reducing repetition.
         :param n: Number of completions to return for each request, input tokens are only billed once.
         :param prediction:
+        :param parallel_tool_calls:
         :param safe_prompt: Whether to inject a safety prompt before all conversations.
         :param retries: Override the default retry configuration for this method
         :param server_url: Override the default server URL for this method
@@ -482,6 +500,8 @@ class Chat(BaseSDK):
 
         if server_url is not None:
             base_url = server_url
+        else:
+            base_url = self._get_url(base_url, url_variables)
 
         request = models.ChatCompletionStreamRequest(
             model=model,
@@ -507,6 +527,7 @@ class Chat(BaseSDK):
             prediction=utils.get_pydantic_model(
                 prediction, Optional[models.Prediction]
             ),
+            parallel_tool_calls=parallel_tool_calls,
             safe_prompt=safe_prompt,
         )
 
@@ -539,6 +560,7 @@ class Chat(BaseSDK):
 
         http_res = self.do_request(
             hook_ctx=HookContext(
+                base_url=base_url or "",
                 operation_id="stream_chat",
                 oauth2_scopes=[],
                 security_source=get_security_from_env(
@@ -551,7 +573,7 @@ class Chat(BaseSDK):
             retry_config=retry_config,
         )
 
-        data: Any = None
+        response_data: Any = None
         if utils.match_response(http_res, "200", "text/event-stream"):
             return eventstreaming.EventStream(
                 http_res,
@@ -560,8 +582,10 @@ class Chat(BaseSDK):
             )
         if utils.match_response(http_res, "422", "application/json"):
             http_res_text = utils.stream_to_text(http_res)
-            data = utils.unmarshal_json(http_res_text, models.HTTPValidationErrorData)
-            raise models.HTTPValidationError(data=data)
+            response_data = utils.unmarshal_json(
+                http_res_text, models.HTTPValidationErrorData
+            )
+            raise models.HTTPValidationError(data=response_data)
         if utils.match_response(http_res, "4XX", "*"):
             http_res_text = utils.stream_to_text(http_res)
             raise models.SDKError(
@@ -619,6 +643,7 @@ class Chat(BaseSDK):
         prediction: Optional[
             Union[models.Prediction, models.PredictionTypedDict]
         ] = None,
+        parallel_tool_calls: Optional[bool] = None,
         safe_prompt: Optional[bool] = None,
         retries: OptionalNullable[utils.RetryConfig] = UNSET,
         server_url: Optional[str] = None,
@@ -644,6 +669,7 @@ class Chat(BaseSDK):
         :param frequency_penalty: frequency_penalty penalizes the repetition of words based on their frequency in the generated text. A higher frequency penalty discourages the model from repeating words that have already appeared frequently in the output, promoting diversity and reducing repetition.
         :param n: Number of completions to return for each request, input tokens are only billed once.
         :param prediction:
+        :param parallel_tool_calls:
         :param safe_prompt: Whether to inject a safety prompt before all conversations.
         :param retries: Override the default retry configuration for this method
         :param server_url: Override the default server URL for this method
@@ -657,6 +683,8 @@ class Chat(BaseSDK):
 
         if server_url is not None:
             base_url = server_url
+        else:
+            base_url = self._get_url(base_url, url_variables)
 
         request = models.ChatCompletionStreamRequest(
             model=model,
@@ -682,6 +710,7 @@ class Chat(BaseSDK):
             prediction=utils.get_pydantic_model(
                 prediction, Optional[models.Prediction]
             ),
+            parallel_tool_calls=parallel_tool_calls,
             safe_prompt=safe_prompt,
         )
 
@@ -714,6 +743,7 @@ class Chat(BaseSDK):
 
         http_res = await self.do_request_async(
             hook_ctx=HookContext(
+                base_url=base_url or "",
                 operation_id="stream_chat",
                 oauth2_scopes=[],
                 security_source=get_security_from_env(
@@ -726,7 +756,7 @@ class Chat(BaseSDK):
             retry_config=retry_config,
         )
 
-        data: Any = None
+        response_data: Any = None
         if utils.match_response(http_res, "200", "text/event-stream"):
             return eventstreaming.EventStreamAsync(
                 http_res,
@@ -735,8 +765,10 @@ class Chat(BaseSDK):
             )
         if utils.match_response(http_res, "422", "application/json"):
             http_res_text = await utils.stream_to_text_async(http_res)
-            data = utils.unmarshal_json(http_res_text, models.HTTPValidationErrorData)
-            raise models.HTTPValidationError(data=data)
+            response_data = utils.unmarshal_json(
+                http_res_text, models.HTTPValidationErrorData
+            )
+            raise models.HTTPValidationError(data=response_data)
         if utils.match_response(http_res, "4XX", "*"):
             http_res_text = await utils.stream_to_text_async(http_res)
             raise models.SDKError(