Fix problem with zeros at assistant options (#26)

src/yandex_cloud_ml_sdk/_assistants/assistant.py

@@ -41,10 +41,11 @@ def _kwargs_from_message(cls, proto: ProtoAssistant, sdk: BaseSDK) -> dict[str,
         kwargs = super()._kwargs_from_message(proto, sdk=sdk)
 
         model = sdk.models.completions(proto.model_uri)
-        if max_tokens := proto.completion_options.max_tokens.value:
-            model = model.configure(max_tokens=max_tokens)
-        if temperature := proto.completion_options.temperature.value:
-            model = model.configure(temperature=temperature)
+        completion_options = proto.completion_options
+        if completion_options.HasField('max_tokens'):
+            model = model.configure(max_tokens=completion_options.max_tokens.value)
+        if completion_options.HasField('temperature'):
+            model = model.configure(temperature=completion_options.temperature.value)
         kwargs['model'] = model
 
         kwargs['tools'] = tuple(
@@ -80,23 +81,20 @@ async def _update(
             expiration_policy=expiration_policy
         )
 
-        model_uri: str | None = None
-        model_temperature: float | None = self.model.config.temperature
-        model_max_tokens: int | None = self.model.config.max_tokens
+        model_uri: UndefinedOr[str] | None = UNDEFINED
 
         if is_defined(model):
             if isinstance(model, str):
                 model_uri = self._sdk.models.completions(model).uri
             elif isinstance(model, BaseGPTModel):
                 model_uri = model.uri
-                model_temperature = model.config.temperature
-                model_max_tokens = model.config.max_tokens
+                if not is_defined(temperature) and model.config.temperature is not None:
+                    temperature = model.config.temperature
+                if not is_defined(max_tokens) and model.config.max_tokens is not None:
+                    max_tokens = model.config.max_tokens
             else:
                 raise TypeError('model argument must be str, GPTModel object either undefined')
 
-        model_temperature = get_defined_value(temperature, model_temperature)
-        model_max_tokens = get_defined_value(max_tokens, model_max_tokens)
-
         request = UpdateAssistantRequest(
             assistant_id=self.id,
             name=get_defined_value(name, ''),
@@ -108,11 +106,11 @@ async def _update(
                 max_prompt_tokens=get_defined_value(max_prompt_tokens, None)
             ),
             completion_options=get_completion_options(
-                temperature=model_temperature,
-                max_tokens=model_max_tokens
+                temperature=temperature,
+                max_tokens=max_tokens,
             )
         )
-        if model_uri:
+        if model_uri and is_defined(model_uri):
             request.model_uri = model_uri
 
         self._fill_update_mask(
@@ -125,8 +123,8 @@ async def _update(
                 'expiration_config.expiration_policy': expiration_policy,
                 'instruction': instruction,
                 'model_uri': model_uri,
-                'completion_options.temperature': model_temperature,
-                'completion_options.max_tokens': model_max_tokens,
+                'completion_options.temperature': temperature,
+                'completion_options.max_tokens': max_tokens,
                 'prompt_truncation_options.max_prompt_tokens': max_prompt_tokens,
             }
         )

src/yandex_cloud_ml_sdk/_assistants/domain.py

@@ -48,20 +48,17 @@ async def _create(
         expiration_config = ExpirationConfig.coerce(ttl_days=ttl_days, expiration_policy=expiration_policy)
 
         model_uri: str = ''
-        model_temperature: float | None = None
-        model_max_tokens: int | None = None
         if isinstance(model, str):
             model_uri = self._sdk.models.completions(model).uri
         elif isinstance(model, BaseGPTModel):
             model_uri = model.uri
-            model_temperature = model.config.temperature
-            model_max_tokens = model.config.max_tokens
+            if not is_defined(temperature) and model.config.temperature is not None:
+                temperature = model.config.temperature
+            if not is_defined(max_tokens) and model.config.max_tokens is not None:
+                max_tokens = model.config.max_tokens
         else:
             raise TypeError('model argument must be str, GPTModel object either undefined')
 
-        model_temperature = get_defined_value(temperature, model_temperature)
-        model_max_tokens = get_defined_value(max_tokens, model_max_tokens)
-
         tools_: tuple[BaseTool, ...] = ()
         if is_defined(tools):
             # NB: mypy doesn't love abstract class used as TypeVar substitution here
@@ -79,8 +76,8 @@ async def _create(
             ),
             model_uri=model_uri,
             completion_options=get_completion_options(
-                temperature=model_temperature,
-                max_tokens=model_max_tokens
+                temperature=temperature,
+                max_tokens=max_tokens
             ),
             tools=[tool._to_proto() for tool in tools_]
         )

src/yandex_cloud_ml_sdk/_assistants/utils.py

@@ -3,27 +3,29 @@
 
 from yandex.cloud.ai.assistants.v1.common_pb2 import CompletionOptions, PromptTruncationOptions
 
+from yandex_cloud_ml_sdk._types.misc import UndefinedOr, is_defined
+
 
 def get_completion_options(
     *,
-    temperature: float | None,
-    max_tokens: int | None,
+    temperature: UndefinedOr[float] | None,
+    max_tokens: UndefinedOr[int] | None,
 ) -> CompletionOptions:
     options = CompletionOptions()
-    if temperature is not None:
+    if temperature is not None and is_defined(temperature):
         options.temperature.value = temperature
-    if max_tokens is not None:
+    if max_tokens is not None and is_defined(max_tokens):
         options.max_tokens.value = max_tokens
 
     return options
 
 
 def get_prompt_trunctation_options(
     *,
-    max_prompt_tokens: int | None
+    max_prompt_tokens: UndefinedOr[int] | None
 ) -> PromptTruncationOptions:
     options = PromptTruncationOptions()
-    if max_prompt_tokens is not None:
+    if max_prompt_tokens is not None and is_defined(max_prompt_tokens):
         options.max_prompt_tokens.value = max_prompt_tokens
 
     return options

src/yandex_cloud_ml_sdk/_types/resource.py

@@ -60,7 +60,7 @@ def _update_from_proto(self, proto: Message) -> Self:
 
     def _fill_update_mask(self, mask: FieldMask, fields: dict[str, Any]) -> None:
         for key, value in fields.items():
-            if is_defined(value) and value is not None:
+            if is_defined(value):
                 mask.paths.append(key)