pydantic-ai-slim 0.0.53__tar.gz → 0.0.55__tar.gz

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pydantic-ai-slim
-Version: 0.0.53
+Version: 0.0.55
 Summary: Agent Framework / shim to use Pydantic with LLMs, slim package
 Author-email: Samuel Colvin <samuel@pydantic.dev>
 License-Expression: MIT
@@ -29,7 +29,7 @@ Requires-Dist: exceptiongroup; python_version < '3.11'
 Requires-Dist: griffe>=1.3.2
 Requires-Dist: httpx>=0.27
 Requires-Dist: opentelemetry-api>=1.28.0
-Requires-Dist: pydantic-graph==0.0.53
+Requires-Dist: pydantic-graph==0.0.55
 Requires-Dist: pydantic>=2.10
 Requires-Dist: typing-inspection>=0.4.0
 Provides-Extra: anthropic
@@ -45,7 +45,7 @@ Requires-Dist: cohere>=5.13.11; (platform_system != 'Emscripten') and extra == '
 Provides-Extra: duckduckgo
 Requires-Dist: duckduckgo-search>=7.0.0; extra == 'duckduckgo'
 Provides-Extra: evals
-Requires-Dist: pydantic-evals==0.0.53; extra == 'evals'
+Requires-Dist: pydantic-evals==0.0.55; extra == 'evals'
 Provides-Extra: groq
 Requires-Dist: groq>=0.15.0; extra == 'groq'
 Provides-Extra: logfire

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/_agent_graph.py

@@ -79,7 +79,7 @@ class GraphAgentDeps(Generic[DepsT, ResultDataT]):
 
     user_deps: DepsT
 
-    prompt: str | Sequence[_messages.UserContent]
+    prompt: str | Sequence[_messages.UserContent] | None
     new_message_index: int
 
     model: models.Model
@@ -124,7 +124,7 @@ def is_agent_node(
 
 @dataclasses.dataclass
 class UserPromptNode(AgentNode[DepsT, NodeRunEndT]):
-    user_prompt: str | Sequence[_messages.UserContent]
+    user_prompt: str | Sequence[_messages.UserContent] | None
 
     system_prompts: tuple[str, ...]
     system_prompt_functions: list[_system_prompt.SystemPromptRunner[DepsT]]
@@ -151,7 +151,7 @@ class UserPromptNode(AgentNode[DepsT, NodeRunEndT]):
 
     async def _prepare_messages(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None,
         message_history: list[_messages.ModelMessage] | None,
         run_context: RunContext[DepsT],
     ) -> tuple[list[_messages.ModelMessage], _messages.ModelRequest]:
@@ -166,16 +166,18 @@ class UserPromptNode(AgentNode[DepsT, NodeRunEndT]):
                 messages = ctx_messages.messages
                 ctx_messages.used = True
 
+        parts: list[_messages.ModelRequestPart] = []
         if message_history:
             # Shallow copy messages
             messages.extend(message_history)
             # Reevaluate any dynamic system prompt parts
             await self._reevaluate_dynamic_prompts(messages, run_context)
-            return messages, _messages.ModelRequest([_messages.UserPromptPart(user_prompt)])
         else:
-            parts = await self._sys_parts(run_context)
+            parts.extend(await self._sys_parts(run_context))
+
+        if user_prompt is not None:
             parts.append(_messages.UserPromptPart(user_prompt))
-            return messages, _messages.ModelRequest(parts)
+        return messages, _messages.ModelRequest(parts)
 
     async def _reevaluate_dynamic_prompts(
         self, messages: list[_messages.ModelMessage], run_context: RunContext[DepsT]

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/agent.py

@@ -242,7 +242,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
     @overload
     async def run(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         result_type: None = None,
         message_history: list[_messages.ModelMessage] | None = None,
@@ -257,7 +257,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
     @overload
     async def run(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         result_type: type[RunResultDataT],
         message_history: list[_messages.ModelMessage] | None = None,
@@ -271,7 +271,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
 
     async def run(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         result_type: type[RunResultDataT] | None = None,
         message_history: list[_messages.ModelMessage] | None = None,
@@ -335,7 +335,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
     @asynccontextmanager
     async def iter(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         result_type: type[RunResultDataT] | None = None,
         message_history: list[_messages.ModelMessage] | None = None,
@@ -372,6 +372,12 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
             print(nodes)
             '''
             [
+                UserPromptNode(
+                    user_prompt='What is the capital of France?',
+                    system_prompts=(),
+                    system_prompt_functions=[],
+                    system_prompt_dynamic_functions={},
+                ),
                 ModelRequestNode(
                     request=ModelRequest(
                         parts=[
@@ -497,7 +503,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
     @overload
     def run_sync(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         message_history: list[_messages.ModelMessage] | None = None,
         model: models.Model | models.KnownModelName | str | None = None,
@@ -511,7 +517,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
     @overload
     def run_sync(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         result_type: type[RunResultDataT] | None,
         message_history: list[_messages.ModelMessage] | None = None,
@@ -525,7 +531,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
 
     def run_sync(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         result_type: type[RunResultDataT] | None = None,
         message_history: list[_messages.ModelMessage] | None = None,
@@ -586,7 +592,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
     @overload
     def run_stream(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         result_type: None = None,
         message_history: list[_messages.ModelMessage] | None = None,
@@ -601,7 +607,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
     @overload
     def run_stream(
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         result_type: type[RunResultDataT],
         message_history: list[_messages.ModelMessage] | None = None,
@@ -616,7 +622,7 @@ class Agent(Generic[AgentDepsT, ResultDataT]):
     @asynccontextmanager
     async def run_stream(  # noqa C901
         self,
-        user_prompt: str | Sequence[_messages.UserContent],
+        user_prompt: str | Sequence[_messages.UserContent] | None = None,
         *,
         result_type: type[RunResultDataT] | None = None,
         message_history: list[_messages.ModelMessage] | None = None,
@@ -1355,6 +1361,12 @@ class AgentRun(Generic[AgentDepsT, ResultDataT]):
         print(nodes)
         '''
         [
+            UserPromptNode(
+                user_prompt='What is the capital of France?',
+                system_prompts=(),
+                system_prompt_functions=[],
+                system_prompt_dynamic_functions={},
+            ),
             ModelRequestNode(
                 request=ModelRequest(
                     parts=[

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/models/__init__.py

@@ -427,7 +427,7 @@ def infer_model(model: Model | KnownModelName | str) -> Model:
         from .cohere import CohereModel
 
         return CohereModel(model_name, provider=provider)
-    elif provider in ('deepseek', 'openai'):
+    elif provider in ('deepseek', 'openai', 'azure'):
         from .openai import OpenAIModel
 
         return OpenAIModel(model_name, provider=provider)

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/models/anthropic.py

@@ -31,7 +31,14 @@ from ..messages import (
 from ..providers import Provider, infer_provider
 from ..settings import ModelSettings
 from ..tools import ToolDefinition
-from . import Model, ModelRequestParameters, StreamedResponse, cached_async_http_client, check_allow_model_requests
+from . import (
+    Model,
+    ModelRequestParameters,
+    StreamedResponse,
+    cached_async_http_client,
+    check_allow_model_requests,
+    get_user_agent,
+)
 
 try:
     from anthropic import NOT_GIVEN, APIStatusError, AsyncAnthropic, AsyncStream
@@ -226,10 +233,12 @@ class AnthropicModel(Model):
                 tools=tools or NOT_GIVEN,
                 tool_choice=tool_choice or NOT_GIVEN,
                 stream=stream,
+                stop_sequences=model_settings.get('stop_sequences', NOT_GIVEN),
                 temperature=model_settings.get('temperature', NOT_GIVEN),
                 top_p=model_settings.get('top_p', NOT_GIVEN),
                 timeout=model_settings.get('timeout', NOT_GIVEN),
                 metadata=model_settings.get('anthropic_metadata', NOT_GIVEN),
+                extra_headers={'User-Agent': get_user_agent()},
             )
         except APIStatusError as e:
             if (status_code := e.status_code) >= 400:

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/models/bedrock.py

@@ -42,12 +42,14 @@ if TYPE_CHECKING:
     from mypy_boto3_bedrock_runtime.type_defs import (
         ContentBlockOutputTypeDef,
         ContentBlockUnionTypeDef,
+        ConverseRequestTypeDef,
         ConverseResponseTypeDef,
         ConverseStreamMetadataEventTypeDef,
         ConverseStreamOutputTypeDef,
         ImageBlockTypeDef,
         InferenceConfigurationTypeDef,
         MessageUnionTypeDef,
+        SystemContentBlockTypeDef,
         ToolChoiceTypeDef,
         ToolTypeDef,
     )
@@ -258,20 +260,19 @@ class BedrockConverseModel(Model):
         else:
             tool_choice = {'auto': {}}
 
-        system_prompt, bedrock_messages = await self._map_message(messages)
+        system_prompt, bedrock_messages = await self._map_messages(messages)
         inference_config = self._map_inference_config(model_settings)
 
-        params = {
+        params: ConverseRequestTypeDef = {
             'modelId': self.model_name,
             'messages': bedrock_messages,
-            'system': [{'text': system_prompt}],
+            'system': system_prompt,
             'inferenceConfig': inference_config,
-            **(
-                {'toolConfig': {'tools': tools, **({'toolChoice': tool_choice} if tool_choice else {})}}
-                if tools
-                else {}
-            ),
         }
+        if tools:
+            params['toolConfig'] = {'tools': tools}
+            if tool_choice:
+                params['toolConfig']['toolChoice'] = tool_choice
 
         if stream:
             model_response = await anyio.to_thread.run_sync(functools.partial(self.client.converse_stream, **params))
@@ -293,21 +294,22 @@ class BedrockConverseModel(Model):
             inference_config['temperature'] = temperature
         if top_p := model_settings.get('top_p'):
             inference_config['topP'] = top_p
-        # TODO(Marcelo): This is not included in model_settings yet.
-        # if stop_sequences := model_settings.get('stop_sequences'):
-        #     inference_config['stopSequences'] = stop_sequences
+        if stop_sequences := model_settings.get('stop_sequences'):
+            inference_config['stopSequences'] = stop_sequences
 
         return inference_config
 
-    async def _map_message(self, messages: list[ModelMessage]) -> tuple[str, list[MessageUnionTypeDef]]:
+    async def _map_messages(
+        self, messages: list[ModelMessage]
+    ) -> tuple[list[SystemContentBlockTypeDef], list[MessageUnionTypeDef]]:
         """Just maps a `pydantic_ai.Message` to the Bedrock `MessageUnionTypeDef`."""
-        system_prompt: str = ''
+        system_prompt: list[SystemContentBlockTypeDef] = []
         bedrock_messages: list[MessageUnionTypeDef] = []
         for m in messages:
             if isinstance(m, ModelRequest):
                 for part in m.parts:
                     if isinstance(part, SystemPromptPart):
-                        system_prompt += part.content
+                        system_prompt.append({'text': part.content})
                     elif isinstance(part, UserPromptPart):
                         bedrock_messages.extend(await self._map_user_prompt(part))
                     elif isinstance(part, ToolReturnPart):

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/models/cohere.py

@@ -118,7 +118,7 @@ class CohereModel(Model):
                 'cohere' or an instance of `Provider[AsyncClientV2]`. If not provided, a new provider will be
                 created using the other parameters.
         """
-        self._model_name: CohereModelName = model_name
+        self._model_name = model_name
 
         if isinstance(provider, str):
             provider = infer_provider(provider)
@@ -163,6 +163,7 @@ class CohereModel(Model):
                 messages=cohere_messages,
                 tools=tools or OMIT,
                 max_tokens=model_settings.get('max_tokens', OMIT),
+                stop_sequences=model_settings.get('stop_sequences', OMIT),
                 temperature=model_settings.get('temperature', OMIT),
                 p=model_settings.get('top_p', OMIT),
                 seed=model_settings.get('seed', OMIT),

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/models/gemini.py

@@ -506,6 +506,7 @@ class _GeminiGenerationConfig(TypedDict, total=False):
     top_p: float
     presence_penalty: float
     frequency_penalty: float
+    stop_sequences: list[str]
 
 
 class _GeminiContent(TypedDict):

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/models/groq.py

@@ -31,7 +31,7 @@ from ..messages import (
 from ..providers import Provider, infer_provider
 from ..settings import ModelSettings
 from ..tools import ToolDefinition
-from . import Model, ModelRequestParameters, StreamedResponse, check_allow_model_requests
+from . import Model, ModelRequestParameters, StreamedResponse, check_allow_model_requests, get_user_agent
 
 try:
     from groq import NOT_GIVEN, APIStatusError, AsyncGroq, AsyncStream
@@ -208,6 +208,7 @@ class GroqModel(Model):
                 parallel_tool_calls=model_settings.get('parallel_tool_calls', NOT_GIVEN),
                 tools=tools or NOT_GIVEN,
                 tool_choice=tool_choice or NOT_GIVEN,
+                stop=model_settings.get('stop_sequences', NOT_GIVEN),
                 stream=stream,
                 max_tokens=model_settings.get('max_tokens', NOT_GIVEN),
                 temperature=model_settings.get('temperature', NOT_GIVEN),
@@ -217,6 +218,7 @@ class GroqModel(Model):
                 presence_penalty=model_settings.get('presence_penalty', NOT_GIVEN),
                 frequency_penalty=model_settings.get('frequency_penalty', NOT_GIVEN),
                 logit_bias=model_settings.get('logit_bias', NOT_GIVEN),
+                extra_headers={'User-Agent': get_user_agent()},
             )
         except APIStatusError as e:
             if (status_code := e.status_code) >= 400:

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/models/mistral.py

@@ -39,6 +39,7 @@ from . import (
     ModelRequestParameters,
     StreamedResponse,
     check_allow_model_requests,
+    get_user_agent,
 )
 
 try:
@@ -199,6 +200,8 @@ class MistralModel(Model):
                 top_p=model_settings.get('top_p', 1),
                 timeout_ms=self._get_timeout_ms(model_settings.get('timeout')),
                 random_seed=model_settings.get('seed', UNSET),
+                stop=model_settings.get('stop_sequences', None),
+                http_headers={'User-Agent': get_user_agent()},
             )
         except SDKError as e:
             if (status_code := e.status_code) >= 400:
@@ -236,6 +239,8 @@ class MistralModel(Model):
                 timeout_ms=self._get_timeout_ms(model_settings.get('timeout')),
                 presence_penalty=model_settings.get('presence_penalty'),
                 frequency_penalty=model_settings.get('frequency_penalty'),
+                stop=model_settings.get('stop_sequences', None),
+                http_headers={'User-Agent': get_user_agent()},
             )
 
         elif model_request_parameters.result_tools:
@@ -249,6 +254,7 @@ class MistralModel(Model):
                 messages=mistral_messages,
                 response_format={'type': 'json_object'},
                 stream=True,
+                http_headers={'User-Agent': get_user_agent()},
             )
 
         else:
@@ -257,6 +263,7 @@ class MistralModel(Model):
                 model=str(self._model_name),
                 messages=mistral_messages,
                 stream=True,
+                http_headers={'User-Agent': get_user_agent()},
             )
         assert response, 'A unexpected empty response from Mistral.'
         return response

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/models/openai.py

@@ -39,6 +39,7 @@ from . import (
     StreamedResponse,
     cached_async_http_client,
     check_allow_model_requests,
+    get_user_agent,
 )
 
 try:
@@ -271,6 +272,7 @@ class OpenAIModel(Model):
                 tool_choice=tool_choice or NOT_GIVEN,
                 stream=stream,
                 stream_options={'include_usage': True} if stream else NOT_GIVEN,
+                stop=model_settings.get('stop_sequences', NOT_GIVEN),
                 max_completion_tokens=model_settings.get('max_tokens', NOT_GIVEN),
                 temperature=model_settings.get('temperature', NOT_GIVEN),
                 top_p=model_settings.get('top_p', NOT_GIVEN),
@@ -281,6 +283,7 @@ class OpenAIModel(Model):
                 logit_bias=model_settings.get('logit_bias', NOT_GIVEN),
                 reasoning_effort=model_settings.get('openai_reasoning_effort', NOT_GIVEN),
                 user=model_settings.get('openai_user', NOT_GIVEN),
+                extra_headers={'User-Agent': get_user_agent()},
             )
         except APIStatusError as e:
             if (status_code := e.status_code) >= 400:
@@ -611,7 +614,8 @@ class OpenAIResponsesModel(Model):
                 truncation=model_settings.get('openai_truncation', NOT_GIVEN),
                 timeout=model_settings.get('timeout', NOT_GIVEN),
                 reasoning=reasoning,
-                user=model_settings.get('user', NOT_GIVEN),
+                user=model_settings.get('openai_user', NOT_GIVEN),
+                extra_headers={'User-Agent': get_user_agent()},
             )
         except APIStatusError as e:
             if (status_code := e.status_code) >= 400:

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/models/wrapper.py

@@ -37,6 +37,9 @@ class WrapperModel(Model):
         async with self.wrapped.request_stream(messages, model_settings, model_request_parameters) as response_stream:
             yield response_stream
 
+    def customize_request_parameters(self, model_request_parameters: ModelRequestParameters) -> ModelRequestParameters:
+        return self.wrapped.customize_request_parameters(model_request_parameters)
+
     @property
     def model_name(self) -> str:
         return self.wrapped.model_name

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/providers/__init__.py

@@ -52,6 +52,10 @@ def infer_provider(provider: str) -> Provider[Any]:
         from .deepseek import DeepSeekProvider
 
         return DeepSeekProvider()
+    elif provider == 'azure':
+        from .azure import AzureProvider
+
+        return AzureProvider()
     elif provider == 'google-vertex':
         from .google_vertex import GoogleVertexProvider
 

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/providers/azure.py

@@ -87,9 +87,9 @@ class AzureProvider(Provider[AsyncOpenAI]):
                     'Must provide one of the `azure_endpoint` argument or the `AZURE_OPENAI_ENDPOINT` environment variable'
                 )
 
-            if not api_key and 'OPENAI_API_KEY' not in os.environ:  # pragma: no cover
+            if not api_key and 'AZURE_OPENAI_API_KEY' not in os.environ:  # pragma: no cover
                 raise UserError(
-                    'Must provide one of the `api_key` argument or the `OPENAI_API_KEY` environment variable'
+                    'Must provide one of the `api_key` argument or the `AZURE_OPENAI_API_KEY` environment variable'
                 )
 
             if not api_version and 'OPENAI_API_VERSION' not in os.environ:  # pragma: no cover

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/settings.py

@@ -128,6 +128,19 @@ class ModelSettings(TypedDict, total=False):
     * Groq
     """
 
+    stop_sequences: list[str]
+    """Sequences that will cause the model to stop generating.
+
+    Supported by:
+
+    * OpenAI
+    * Anthropic
+    * Bedrock
+    * Mistral
+    * Groq
+    * Cohere
+    """
+
 
 def merge_model_settings(base: ModelSettings | None, overrides: ModelSettings | None) -> ModelSettings | None:
     """Merge two sets of model settings, preferring the overrides.

{pydantic_ai_slim-0.0.53 → pydantic_ai_slim-0.0.55}/pydantic_ai/tools.py

@@ -48,7 +48,7 @@ class RunContext(Generic[AgentDepsT]):
     """The model used in this run."""
     usage: Usage
     """LLM usage associated with the run."""
-    prompt: str | Sequence[_messages.UserContent]
+    prompt: str | Sequence[_messages.UserContent] | None
     """The original user prompt passed to the run."""
     messages: list[_messages.ModelMessage] = field(default_factory=list)
     """Messages exchanged in the conversation so far."""