pydantic-ai-slim 1.0.0b1__py3-none-any.whl → 1.0.1__py3-none-any.whl

pydantic_ai/mcp.py

@@ -20,6 +20,7 @@ from typing_extensions import Self, assert_never, deprecated
 
 from pydantic_ai.tools import RunContext, ToolDefinition
 
+from .direct import model_request
 from .toolsets.abstract import AbstractToolset, ToolsetTool
 
 try:
@@ -300,6 +301,8 @@ class MCPServer(AbstractToolset[Any], ABC):
         return self
 
     async def __aexit__(self, *args: Any) -> bool | None:
+        if self._running_count == 0:
+            raise ValueError('MCPServer.__aexit__ called more times than __aenter__')
         async with self._enter_lock:
             self._running_count -= 1
             if self._running_count == 0 and self._exit_stack is not None:
@@ -327,11 +330,7 @@ class MCPServer(AbstractToolset[Any], ABC):
         if stop_sequences := params.stopSequences:  # pragma: no branch
             model_settings['stop_sequences'] = stop_sequences
 
-        model_response = await self.sampling_model.request(
-            pai_messages,
-            model_settings,
-            models.ModelRequestParameters(),
-        )
+        model_response = await model_request(self.sampling_model, pai_messages, model_settings=model_settings)
         return mcp_types.CreateMessageResult(
             role='assistant',
             content=_mcp.map_from_model_response(model_response),
@@ -401,16 +400,7 @@ class MCPServerStdio(MCPServer):
     from pydantic_ai.mcp import MCPServerStdio
 
     server = MCPServerStdio(  # (1)!
-        'deno',
-        args=[
-            'run',
-            '-N',
-            '-R=node_modules',
-            '-W=node_modules',
-            '--node-modules-dir=auto',
-            'jsr:@pydantic/mcp-run-python',
-            'stdio',
-        ]
+        'uv', args=['run', 'mcp-run-python', 'stdio'], timeout=10
     )
     agent = Agent('openai:gpt-4o', toolsets=[server])
 
@@ -419,7 +409,7 @@ class MCPServerStdio(MCPServer):
             ...
     ```
 
-    1. See [MCP Run Python](../mcp/run-python.md) for more information.
+    1. See [MCP Run Python](https://github.com/pydantic/mcp-run-python) for more information.
     2. This will start the server as a subprocess and connect to it.
     """
 
@@ -455,6 +445,7 @@ class MCPServerStdio(MCPServer):
         self,
         command: str,
         args: Sequence[str],
+        *,
         env: dict[str, str] | None = None,
         cwd: str | Path | None = None,
         tool_prefix: str | None = None,
@@ -467,7 +458,6 @@ class MCPServerStdio(MCPServer):
         sampling_model: models.Model | None = None,
         max_retries: int = 1,
         elicitation_callback: ElicitationFnT | None = None,
-        *,
         id: str | None = None,
     ):
         """Build a new MCP server.
@@ -581,8 +571,8 @@ class _MCPServerHTTP(MCPServer):
 
     def __init__(
         self,
-        *,
         url: str,
+        *,
         headers: dict[str, str] | None = None,
         http_client: httpx.AsyncClient | None = None,
         id: str | None = None,
@@ -732,16 +722,15 @@ class MCPServerSSE(_MCPServerHTTP):
     from pydantic_ai import Agent
     from pydantic_ai.mcp import MCPServerSSE
 
-    server = MCPServerSSE('http://localhost:3001/sse')  # (1)!
+    server = MCPServerSSE('http://localhost:3001/sse')
     agent = Agent('openai:gpt-4o', toolsets=[server])
 
     async def main():
-        async with agent:  # (2)!
+        async with agent:  # (1)!
             ...
     ```
 
-    1. E.g. you might be connecting to a server run with [`mcp-run-python`](../mcp/run-python.md).
-    2. This will connect to a server running on `localhost:3001`.
+    1. This will connect to a server running on `localhost:3001`.
     """
 
     @property
@@ -765,7 +754,7 @@ class MCPServerHTTP(MCPServerSSE):
     from pydantic_ai import Agent
     from pydantic_ai.mcp import MCPServerHTTP
 
-    server = MCPServerHTTP('http://localhost:3001/sse')  # (1)!
+    server = MCPServerHTTP('http://localhost:3001/sse')
     agent = Agent('openai:gpt-4o', toolsets=[server])
 
     async def main():
@@ -773,8 +762,7 @@ class MCPServerHTTP(MCPServerSSE):
             ...
     ```
 
-    1. E.g. you might be connecting to a server run with [`mcp-run-python`](../mcp/run-python.md).
-    2. This will connect to a server running on `localhost:3001`.
+    1. This will connect to a server running on `localhost:3001`.
     """
 
 

pydantic_ai/messages.py

@@ -1,6 +1,7 @@
 from __future__ import annotations as _annotations
 
 import base64
+import hashlib
 from abc import ABC, abstractmethod
 from collections.abc import Sequence
 from dataclasses import KW_ONLY, dataclass, field, replace
@@ -88,6 +89,13 @@ class SystemPromptPart:
     __repr__ = _utils.dataclasses_no_defaults_repr
 
 
+def _multi_modal_content_identifier(identifier: str | bytes) -> str:
+    """Generate stable identifier for multi-modal content to help LLM in finding a specific file in tool call responses."""
+    if isinstance(identifier, str):
+        identifier = identifier.encode('utf-8')
+    return hashlib.sha1(identifier).hexdigest()[:6]
+
+
 @dataclass(init=False, repr=False)
 class FileUrl(ABC):
     """Abstract base class for any URL-based file."""
@@ -115,17 +123,31 @@ class FileUrl(ABC):
         compare=False, default=None
     )
 
+    identifier: str | None = None
+    """The identifier of the file, such as a unique ID. generating one from the url if not explicitly set
+
+    This identifier can be provided to the model in a message to allow it to refer to this file in a tool call argument,
+    and the tool can look up the file in question by iterating over the message history and finding the matching `FileUrl`.
+
+    This identifier is only automatically passed to the model when the `FileUrl` is returned by a tool.
+    If you're passing the `FileUrl` as a user message, it's up to you to include a separate text part with the identifier,
+    e.g. "This is file <identifier>:" preceding the `FileUrl`.
+    """
+
     def __init__(
         self,
         url: str,
+        *,
         force_download: bool = False,
         vendor_metadata: dict[str, Any] | None = None,
         media_type: str | None = None,
+        identifier: str | None = None,
     ) -> None:
         self.url = url
-        self.vendor_metadata = vendor_metadata
         self.force_download = force_download
+        self.vendor_metadata = vendor_metadata
         self._media_type = media_type
+        self.identifier = identifier or _multi_modal_content_identifier(url)
 
     @pydantic.computed_field
     @property
@@ -162,11 +184,12 @@ class VideoUrl(FileUrl):
     def __init__(
         self,
         url: str,
+        *,
         force_download: bool = False,
         vendor_metadata: dict[str, Any] | None = None,
         media_type: str | None = None,
         kind: Literal['video-url'] = 'video-url',
-        *,
+        identifier: str | None = None,
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         _media_type: str | None = None,
     ) -> None:
@@ -175,6 +198,7 @@ class VideoUrl(FileUrl):
             force_download=force_download,
             vendor_metadata=vendor_metadata,
             media_type=media_type or _media_type,
+            identifier=identifier,
         )
         self.kind = kind
 
@@ -235,11 +259,12 @@ class AudioUrl(FileUrl):
     def __init__(
         self,
         url: str,
+        *,
         force_download: bool = False,
         vendor_metadata: dict[str, Any] | None = None,
         media_type: str | None = None,
         kind: Literal['audio-url'] = 'audio-url',
-        *,
+        identifier: str | None = None,
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         _media_type: str | None = None,
     ) -> None:
@@ -248,6 +273,7 @@ class AudioUrl(FileUrl):
             force_download=force_download,
             vendor_metadata=vendor_metadata,
             media_type=media_type or _media_type,
+            identifier=identifier,
         )
         self.kind = kind
 
@@ -295,11 +321,12 @@ class ImageUrl(FileUrl):
     def __init__(
         self,
         url: str,
+        *,
         force_download: bool = False,
         vendor_metadata: dict[str, Any] | None = None,
         media_type: str | None = None,
         kind: Literal['image-url'] = 'image-url',
-        *,
+        identifier: str | None = None,
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         _media_type: str | None = None,
     ) -> None:
@@ -308,6 +335,7 @@ class ImageUrl(FileUrl):
             force_download=force_download,
             vendor_metadata=vendor_metadata,
             media_type=media_type or _media_type,
+            identifier=identifier,
         )
         self.kind = kind
 
@@ -350,11 +378,12 @@ class DocumentUrl(FileUrl):
     def __init__(
         self,
         url: str,
+        *,
         force_download: bool = False,
         vendor_metadata: dict[str, Any] | None = None,
         media_type: str | None = None,
         kind: Literal['document-url'] = 'document-url',
-        *,
+        identifier: str | None = None,
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         _media_type: str | None = None,
     ) -> None:
@@ -363,6 +392,7 @@ class DocumentUrl(FileUrl):
             force_download=force_download,
             vendor_metadata=vendor_metadata,
             media_type=media_type or _media_type,
+            identifier=identifier,
         )
         self.kind = kind
 
@@ -405,24 +435,26 @@ class DocumentUrl(FileUrl):
             raise ValueError(f'Unknown document media type: {media_type}') from e
 
 
-@dataclass(repr=False)
+@dataclass(init=False, repr=False)
 class BinaryContent:
     """Binary content, e.g. an audio or image file."""
 
     data: bytes
     """The binary data."""
 
-    media_type: AudioMediaType | ImageMediaType | DocumentMediaType | str
-    """The media type of the binary data."""
-
     _: KW_ONLY
 
-    identifier: str | None = None
-    """Identifier for the binary content, such as a URL or unique ID.
+    media_type: AudioMediaType | ImageMediaType | DocumentMediaType | str
+    """The media type of the binary data."""
 
-    This identifier can be provided to the model in a message to allow it to refer to this file in a tool call argument, and the tool can look up the file in question by iterating over the message history and finding the matching `BinaryContent`.
+    identifier: str
+    """Identifier for the binary content, such as a unique ID. generating one from the data if not explicitly set
+    This identifier can be provided to the model in a message to allow it to refer to this file in a tool call argument,
+    and the tool can look up the file in question by iterating over the message history and finding the matching `BinaryContent`.
 
-    This identifier is only automatically passed to the model when the `BinaryContent` is returned by a tool. If you're passing the `BinaryContent` as a user message, it's up to you to include a separate text part with the identifier, e.g. "This is file <identifier>:" preceding the `BinaryContent`.
+    This identifier is only automatically passed to the model when the `BinaryContent` is returned by a tool.
+    If you're passing the `BinaryContent` as a user message, it's up to you to include a separate text part with the identifier,
+    e.g. "This is file <identifier>:" preceding the `BinaryContent`.
     """
 
     vendor_metadata: dict[str, Any] | None = None
@@ -435,6 +467,21 @@ class BinaryContent:
     kind: Literal['binary'] = 'binary'
     """Type identifier, this is available on all parts as a discriminator."""
 
+    def __init__(
+        self,
+        data: bytes,
+        *,
+        media_type: AudioMediaType | ImageMediaType | DocumentMediaType | str,
+        identifier: str | None = None,
+        vendor_metadata: dict[str, Any] | None = None,
+        kind: Literal['binary'] = 'binary',
+    ) -> None:
+        self.data = data
+        self.media_type = media_type
+        self.identifier = identifier or _multi_modal_content_identifier(data)
+        self.vendor_metadata = vendor_metadata
+        self.kind = kind
+
     @property
     def is_audio(self) -> bool:
         """Return `True` if the media type is an audio type."""
@@ -786,7 +833,7 @@ ModelRequestPart = Annotated[
 class ModelRequest:
     """A request generated by Pydantic AI and sent to a model, e.g. a message from the Pydantic AI app to the model."""
 
-    parts: list[ModelRequestPart]
+    parts: Sequence[ModelRequestPart]
     """The parts of the user message."""
 
     _: KW_ONLY
@@ -941,7 +988,7 @@ ModelResponsePart = Annotated[
 class ModelResponse:
     """A response from a model, e.g. a message from the model to the Pydantic AI app."""
 
-    parts: list[ModelResponsePart]
+    parts: Sequence[ModelResponsePart]
     """The parts of the model message."""
 
     _: KW_ONLY
@@ -967,18 +1014,30 @@ class ModelResponse:
     provider_name: str | None = None
     """The name of the LLM provider that generated the response."""
 
-    provider_details: dict[str, Any] | None = field(default=None)
+    provider_details: Annotated[
+        dict[str, Any] | None,
+        # `vendor_details` is deprecated, but we still want to support deserializing model responses stored in a DB before the name was changed
+        pydantic.Field(validation_alias=pydantic.AliasChoices('provider_details', 'vendor_details')),
+    ] = None
     """Additional provider-specific details in a serializable format.
 
     This allows storing selected vendor-specific data that isn't mapped to standard ModelResponse fields.
     For OpenAI models, this may include 'logprobs', 'finish_reason', etc.
     """
 
-    provider_response_id: str | None = None
+    provider_response_id: Annotated[
+        str | None,
+        # `vendor_id` is deprecated, but we still want to support deserializing model responses stored in a DB before the name was changed
+        pydantic.Field(validation_alias=pydantic.AliasChoices('provider_response_id', 'vendor_id')),
+    ] = None
     """request ID as specified by the model provider. This can be used to track the specific request to the model."""
 
-    def price(self) -> genai_types.PriceCalculation:
-        """Calculate the price of the usage.
+    @deprecated('`price` is deprecated, use `cost` instead')
+    def price(self) -> genai_types.PriceCalculation:  # pragma: no cover
+        return self.cost()
+
+    def cost(self) -> genai_types.PriceCalculation:
+        """Calculate the cost of the usage.
 
         Uses [`genai-prices`](https://github.com/pydantic/genai-prices).
         """

pydantic_ai/models/__init__.py

@@ -728,6 +728,7 @@ def infer_model(model: Model | KnownModelName | str) -> Model:  # noqa: C901
         'openrouter',
         'together',
         'vercel',
+        'litellm',
     ):
         from .openai import OpenAIChatModel
 

pydantic_ai/models/anthropic.py

@@ -536,7 +536,7 @@ class AnthropicModel(Model):
         }
 
 
-def _map_usage(message: BetaMessage | BetaRawMessageStreamEvent) -> usage.RequestUsage:
+def _map_usage(message: BetaMessage | BetaRawMessageStartEvent | BetaRawMessageDeltaEvent) -> usage.RequestUsage:
     if isinstance(message, BetaMessage):
         response_usage = message.usage
     elif isinstance(message, BetaRawMessageStartEvent):
@@ -544,12 +544,7 @@ def _map_usage(message: BetaMessage | BetaRawMessageStreamEvent) -> usage.Reques
     elif isinstance(message, BetaRawMessageDeltaEvent):
         response_usage = message.usage
     else:
-        # No usage information provided in:
-        # - RawMessageStopEvent
-        # - RawContentBlockStartEvent
-        # - RawContentBlockDeltaEvent
-        # - RawContentBlockStopEvent
-        return usage.RequestUsage()
+        assert_never(message)
 
     # Store all integer-typed usage values in the details, except 'output_tokens' which is represented exactly by
     # `response_tokens`
@@ -586,10 +581,8 @@ class AnthropicStreamedResponse(StreamedResponse):
         current_block: BetaContentBlock | None = None
 
         async for event in self._response:
-            self._usage += _map_usage(event)
-
             if isinstance(event, BetaRawMessageStartEvent):
-                pass
+                self._usage = _map_usage(event)
 
             elif isinstance(event, BetaRawContentBlockStartEvent):
                 current_block = event.content_block
@@ -652,7 +645,7 @@ class AnthropicStreamedResponse(StreamedResponse):
                     pass
 
             elif isinstance(event, BetaRawMessageDeltaEvent):
-                pass
+                self._usage = _map_usage(event)
 
             elif isinstance(event, BetaRawContentBlockStopEvent | BetaRawMessageStopEvent):  # pragma: no branch
                 current_block = None

pydantic_ai/models/bedrock.py

@@ -2,7 +2,6 @@ from __future__ import annotations
 
 import functools
 import typing
-import warnings
 from collections.abc import AsyncIterator, Iterable, Iterator, Mapping
 from contextlib import asynccontextmanager
 from dataclasses import dataclass, field
@@ -601,7 +600,7 @@ class BedrockStreamedResponse(StreamedResponse):
     _provider_name: str
     _timestamp: datetime = field(default_factory=_utils.now_utc)
 
-    async def _get_event_iterator(self) -> AsyncIterator[ModelResponseStreamEvent]:  # noqa: C901
+    async def _get_event_iterator(self) -> AsyncIterator[ModelResponseStreamEvent]:
         """Return an async iterator of [`ModelResponseStreamEvent`][pydantic_ai.messages.ModelResponseStreamEvent]s.
 
         This method should be implemented by subclasses to translate the vendor-specific stream of events into
@@ -638,18 +637,11 @@ class BedrockStreamedResponse(StreamedResponse):
                     index = content_block_delta['contentBlockIndex']
                     delta = content_block_delta['delta']
                     if 'reasoningContent' in delta:
-                        if text := delta['reasoningContent'].get('text'):
-                            yield self._parts_manager.handle_thinking_delta(
-                                vendor_part_id=index,
-                                content=text,
-                                signature=delta['reasoningContent'].get('signature'),
-                            )
-                        else:  # pragma: no cover
-                            warnings.warn(
-                                f'Only text reasoning content is supported yet, but you got {delta["reasoningContent"]}. '
-                                'Please report this to the maintainers.',
-                                UserWarning,
-                            )
+                        yield self._parts_manager.handle_thinking_delta(
+                            vendor_part_id=index,
+                            content=delta['reasoningContent'].get('text'),
+                            signature=delta['reasoningContent'].get('signature'),
+                        )
                     if 'text' in delta:
                         maybe_event = self._parts_manager.handle_text_delta(vendor_part_id=index, content=delta['text'])
                         if maybe_event is not None:  # pragma: no branch

pydantic_ai/models/gemini.py

@@ -211,7 +211,9 @@ class GeminiModel(Model):
         generation_config = _settings_to_generation_config(model_settings)
         if model_request_parameters.output_mode == 'native':
             if tools:
-                raise UserError('Gemini does not support structured output and tools at the same time.')
+                raise UserError(
+                    'Gemini does not support `NativeOutput` and tools at the same time. Use `output_type=ToolOutput(...)` instead.'
+                )
 
             generation_config['response_mime_type'] = 'application/json'
 

pydantic_ai/models/google.py

@@ -264,6 +264,14 @@ class GoogleModel(Model):
         yield await self._process_streamed_response(response, model_request_parameters)  # type: ignore
 
     def _get_tools(self, model_request_parameters: ModelRequestParameters) -> list[ToolDict] | None:
+        if model_request_parameters.builtin_tools:
+            if model_request_parameters.output_tools:
+                raise UserError(
+                    'Gemini does not support output tools and built-in tools at the same time. Use `output_type=PromptedOutput(...)` instead.'
+                )
+            if model_request_parameters.function_tools:
+                raise UserError('Gemini does not support user tools and built-in tools at the same time.')
+
         tools: list[ToolDict] = [
             ToolDict(function_declarations=[_function_declaration_from_tool(t)])
             for t in model_request_parameters.tool_defs.values()
@@ -334,7 +342,9 @@ class GoogleModel(Model):
         response_schema = None
         if model_request_parameters.output_mode == 'native':
             if tools:
-                raise UserError('Gemini does not support structured output and tools at the same time.')
+                raise UserError(
+                    'Gemini does not support `NativeOutput` and tools at the same time. Use `output_type=ToolOutput(...)` instead.'
+                )
             response_mime_type = 'application/json'
             output_object = model_request_parameters.output_object
             assert output_object is not None
@@ -559,6 +569,10 @@ class GeminiStreamedResponse(StreamedResponse):
                     )
                     if maybe_event is not None:  # pragma: no branch
                         yield maybe_event
+                elif part.executable_code is not None:
+                    pass
+                elif part.code_execution_result is not None:
+                    pass
                 else:
                     assert part.function_response is not None, f'Unexpected part: {part}'  # pragma: no cover