donkit-llm 0.1.4__tar.gz → 0.1.5__tar.gz

{donkit_llm-0.1.4 → donkit_llm-0.1.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: donkit-llm
-Version: 0.1.4
+Version: 0.1.5
 Summary: Unified LLM model implementations for Donkit (OpenAI, Azure OpenAI, Claude, Vertex AI, Ollama)
 License: MIT
 Author: Donkit AI

{donkit_llm-0.1.4 → donkit_llm-0.1.5}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "donkit-llm"
-version = "0.1.4"
+version = "0.1.5"
 description = "Unified LLM model implementations for Donkit (OpenAI, Azure OpenAI, Claude, Vertex AI, Ollama)"
 authors = ["Donkit AI <opensource@donkit.ai>"]
 license = "MIT"

{donkit_llm-0.1.4 → donkit_llm-0.1.5}/src/donkit/llm/openai_model.py

@@ -263,7 +263,24 @@ class OpenAIModel(LLMModelAbstract):
                     kwargs["tool_choice"] = request.tool_choice
 
         if request.response_format:
-            kwargs["response_format"] = request.response_format
+            # OpenAI requires specific format for structured output
+            # If response_format is a JSON Schema dict with "type": "object", wrap it
+            if isinstance(request.response_format, dict):
+                if request.response_format.get("type") == "object":
+                    # This is a JSON Schema - wrap it in json_schema format
+                    kwargs["response_format"] = {
+                        "type": "json_schema",
+                        "json_schema": {
+                            "name": "response",
+                            "strict": True,
+                            "schema": request.response_format,
+                        },
+                    }
+                else:
+                    # Already in correct format or simple type
+                    kwargs["response_format"] = request.response_format
+            else:
+                kwargs["response_format"] = request.response_format
 
         return kwargs
 

{donkit_llm-0.1.4 → donkit_llm-0.1.5}/src/donkit/llm/vertex_model.py

@@ -330,6 +330,39 @@ class VertexAIModel(LLMModelAbstract):
 
         return convert(schema)
 
+    def _build_config_kwargs(
+        self, request: GenerateRequest, system_instruction: str | None = None
+    ) -> dict[str, Any]:
+        """Build configuration kwargs for Vertex AI generate/generate_stream."""
+        config_kwargs: dict[str, Any] = {
+            "temperature": request.temperature
+            if request.temperature is not None
+            else 0.2,
+            "top_p": request.top_p if request.top_p is not None else 0.95,
+            "max_output_tokens": request.max_tokens
+            if request.max_tokens is not None
+            else 8192,
+        }
+        if system_instruction:
+            config_kwargs["system_instruction"] = system_instruction
+        if request.stop:
+            config_kwargs["stop_sequences"] = request.stop
+        if request.response_format:
+            config_kwargs["response_mime_type"] = "application/json"
+            # If response_format is a JSON Schema dict with "type": "object", use it directly
+            if isinstance(request.response_format, dict):
+                if request.response_format.get("type") == "object":
+                    # This is a JSON Schema - use it directly
+                    config_kwargs["response_schema"] = self._clean_json_schema(
+                        request.response_format
+                    )
+                elif "schema" in request.response_format:
+                    # Already wrapped in schema key
+                    config_kwargs["response_schema"] = self._clean_json_schema(
+                        request.response_format["schema"]
+                    )
+        return config_kwargs
+
     async def generate(self, request: GenerateRequest) -> GenerateResponse:
         """Generate a response using Vertex AI."""
         await self.validate_request(request)
@@ -410,26 +443,7 @@ class VertexAIModel(LLMModelAbstract):
                 contents.append(user_content)
                 i += 1
 
-        config_kwargs = {
-            "temperature": request.temperature
-            if request.temperature is not None
-            else 0.2,
-            "top_p": request.top_p if request.top_p is not None else 0.95,
-            "max_output_tokens": request.max_tokens
-            if request.max_tokens is not None
-            else 8192,
-        }
-        if system_instruction:
-            config_kwargs["system_instruction"] = system_instruction
-        if request.stop:
-            config_kwargs["stop_sequences"] = request.stop
-        if request.response_format:
-            config_kwargs["response_mime_type"] = "application/json"
-            if "schema" in request.response_format:
-                config_kwargs["response_schema"] = self._clean_json_schema(
-                    request.response_format["schema"]
-                )
-
+        config_kwargs = self._build_config_kwargs(request, system_instruction)
         config = genai.types.GenerateContentConfig(**config_kwargs)
 
         if request.tools:
@@ -584,25 +598,7 @@ class VertexAIModel(LLMModelAbstract):
                 contents.append(user_content)
                 i += 1
 
-        config_kwargs: dict[str, Any] = {
-            "temperature": request.temperature
-            if request.temperature is not None
-            else 0.2,
-            "top_p": request.top_p if request.top_p is not None else 0.95,
-            "max_output_tokens": request.max_tokens
-            if request.max_tokens is not None
-            else 8192,
-        }
-        if system_instruction:
-            config_kwargs["system_instruction"] = system_instruction
-        if request.stop:
-            config_kwargs["stop_sequences"] = request.stop
-        if request.response_format:
-            config_kwargs["response_mime_type"] = "application/json"
-            if "schema" in request.response_format:
-                config_kwargs["response_schema"] = self._clean_json_schema(
-                    request.response_format["schema"]
-                )
+        config_kwargs = self._build_config_kwargs(request, system_instruction)
         config_kwargs["automatic_function_calling"] = (
             genai.types.AutomaticFunctionCallingConfig(maximum_remote_calls=100)
         )