Change behaviour to behavior

docs/agents.md

@@ -444,12 +444,12 @@ user_id=123 message='Hello John, would you be free for coffee sometime next week
 
 ## Model errors
 
-If models behave unexpectedly (e.g., the retry limit is exceeded, or their API returns `503`), agent runs will raise [`UnexpectedModelBehaviour`][pydantic_ai.exceptions.UnexpectedModelBehaviour].
+If models behave unexpectedly (e.g., the retry limit is exceeded, or their API returns `503`), agent runs will raise [`UnexpectedModelBehavior`][pydantic_ai.exceptions.UnexpectedModelBehavior].
 
 In these cases, [`agent.last_run_messages`][pydantic_ai.Agent.last_run_messages] can be used to access the messages exchanged during the run to help diagnose the issue.
 
 ```py
-from pydantic_ai import Agent, ModelRetry, UnexpectedModelBehaviour
+from pydantic_ai import Agent, ModelRetry, UnexpectedModelBehavior
 
 agent = Agent('openai:gpt-4o')
 
@@ -464,7 +464,7 @@ def calc_volume(size: int) -> int:  # (1)!
 
 try:
     result = agent.run_sync('Please get me the volume of a box with size 6.')
-except UnexpectedModelBehaviour as e:
+except UnexpectedModelBehavior as e:
     print('An error occurred:', e)
     #> An error occurred: Retriever exceeded max retries count of 1
     print('cause:', repr(e.__cause__))

docs/index.md

@@ -125,7 +125,7 @@ async def main():
 
 1. This [agent](agents.md) will act as first-tier support in a bank. Agents are generic in the type of dependencies they accept and the type of result they return. In this case, the support agent has type `#!python Agent[SupportDependencies, SupportResult]`.
 2. Here we configure the agent to use [OpenAI's GPT-4o model](api/models/openai.md), you can also set the model when running the agent.
-3. The `SupportDependencies` dataclass is used to pass data, connections, and logic into the model that will be needed when running [system prompt](agents.md#system-prompts) and [retriever](agents.md#retrievers) functions. PydanticAI's system of dependency injection provides a type-safe way to customise the behaviour of your agents, and can be especially useful when running unit tests and evals.
+3. The `SupportDependencies` dataclass is used to pass data, connections, and logic into the model that will be needed when running [system prompt](agents.md#system-prompts) and [retriever](agents.md#retrievers) functions. PydanticAI's system of dependency injection provides a type-safe way to customise the behavior of your agents, and can be especially useful when running unit tests and evals.
 4. Static [system prompts](agents.md#system-prompts) can be registered with the [`system_prompt` keyword argument][pydantic_ai.Agent.__init__] to the agent.
 5. Dynamic [system prompts](agents.md#system-prompts) can be registered with the [`@agent.system_prompt`][pydantic_ai.Agent.system_prompt] decorator, and can make use of dependency injection. Dependencies are carried via the [`CallContext`][pydantic_ai.dependencies.CallContext] argument, which is parameterized with the `deps_type` from above. If the type annotation here is wrong, static type checkers will catch it.
 6. [Retrievers](agents.md#retrievers) let you register "tools" which the LLM may call while responding to a user. Again, dependencies are carried via [`CallContext`][pydantic_ai.dependencies.CallContext], and any other arguments become the tool schema passed to the LLM. Pydantic is used to validate these arguments, and errors are passed back to the LLM so it can retry.

pydantic_ai/__init__.py

@@ -2,7 +2,7 @@
 
 from .agent import Agent
 from .dependencies import CallContext
-from .exceptions import ModelRetry, UnexpectedModelBehaviour, UserError
+from .exceptions import ModelRetry, UnexpectedModelBehavior, UserError
 
-__all__ = 'Agent', 'CallContext', 'ModelRetry', 'UnexpectedModelBehaviour', 'UserError', '__version__'
+__all__ = 'Agent', 'CallContext', 'ModelRetry', 'UnexpectedModelBehavior', 'UserError', '__version__'
 __version__ = version('pydantic_ai')

pydantic_ai/_retriever.py

@@ -10,7 +10,7 @@
 
 from . import _pydantic, _utils, messages
 from .dependencies import AgentDeps, CallContext, RetrieverContextFunc, RetrieverParams, RetrieverPlainFunc
-from .exceptions import ModelRetry, UnexpectedModelBehaviour
+from .exceptions import ModelRetry, UnexpectedModelBehavior
 
 # Usage `RetrieverEitherFunc[AgentDependencies, P]`
 RetrieverEitherFunc = _utils.Either[
@@ -101,7 +101,7 @@ def _on_error(self, exc: ValidationError | ModelRetry, call_message: messages.To
         self._current_retry += 1
         if self._current_retry > self.max_retries:
             # TODO custom error with details of the retriever
-            raise UnexpectedModelBehaviour(f'Retriever exceeded max retries count of {self.max_retries}') from exc
+            raise UnexpectedModelBehavior(f'Retriever exceeded max retries count of {self.max_retries}') from exc
         else:
             if isinstance(exc, ValidationError):
                 content = exc.errors(include_url=False)

pydantic_ai/agent.py

@@ -660,7 +660,7 @@ async def _handle_model_response(
                         return _MarkFinalResult(result_data)
 
             if not model_response.calls:
-                raise exceptions.UnexpectedModelBehaviour('Received empty tool call message')
+                raise exceptions.UnexpectedModelBehavior('Received empty tool call message')
 
             # otherwise we run all retriever functions in parallel
             messages: list[_messages.Message] = []
@@ -723,7 +723,7 @@ async def _handle_streamed_model_response(
                 pass
             structured_msg = model_response.get()
             if not structured_msg.calls:
-                raise exceptions.UnexpectedModelBehaviour('Received empty tool call message')
+                raise exceptions.UnexpectedModelBehavior('Received empty tool call message')
             messages: list[_messages.Message] = [structured_msg]
 
             # we now run all retriever functions in parallel
@@ -748,7 +748,7 @@ async def _validate_result(
     def _incr_result_retry(self) -> None:
         self._current_result_retry += 1
         if self._current_result_retry > self._max_result_retries:
-            raise exceptions.UnexpectedModelBehaviour(
+            raise exceptions.UnexpectedModelBehavior(
                 f'Exceeded maximum retries ({self._max_result_retries}) for result validation'
             )
 

pydantic_ai/exceptions.py

@@ -2,7 +2,7 @@
 
 import json
 
-__all__ = 'ModelRetry', 'UserError', 'UnexpectedModelBehaviour'
+__all__ = 'ModelRetry', 'UserError', 'UnexpectedModelBehavior'
 
 
 class ModelRetry(Exception):
@@ -30,7 +30,7 @@ def __init__(self, message: str):
         super().__init__(message)
 
 
-class UnexpectedModelBehaviour(RuntimeError):
+class UnexpectedModelBehavior(RuntimeError):
     """Error caused by unexpected Model behavior, e.g. an unexpected response code."""
 
     message: str

pydantic_ai/models/gemini.py

@@ -36,7 +36,7 @@
 from pydantic import Discriminator, Field, Tag
 from typing_extensions import NotRequired, TypedDict, TypeGuard, assert_never
 
-from .. import UnexpectedModelBehaviour, _pydantic, _utils, exceptions, result
+from .. import UnexpectedModelBehavior, _pydantic, _utils, exceptions, result
 from ..messages import (
     ArgsObject,
     Message,
@@ -192,7 +192,7 @@ async def _make_request(self, messages: list[Message], streamed: bool) -> AsyncI
         async with self.http_client.stream('POST', url, content=request_json, headers=headers) as r:
             if r.status_code != 200:
                 await r.aread()
-                raise exceptions.UnexpectedModelBehaviour(f'Unexpected response from gemini {r.status_code}', r.text)
+                raise exceptions.UnexpectedModelBehavior(f'Unexpected response from gemini {r.status_code}', r.text)
             yield r
 
     @staticmethod
@@ -223,7 +223,7 @@ async def _process_streamed_response(http_response: HTTPResponse) -> EitherStrea
                     break
 
         if start_response is None:
-            raise UnexpectedModelBehaviour('Streamed response ended without content or tool calls')
+            raise UnexpectedModelBehavior('Streamed response ended without content or tool calls')
 
         if _extract_response_parts(start_response).is_left():
             return GeminiStreamStructuredResponse(_content=content, _stream=aiter_bytes)
@@ -287,7 +287,7 @@ def get(self, *, final: bool = False) -> Iterable[str]:
                 for part in parts:
                     yield part['text']
             else:
-                raise UnexpectedModelBehaviour(
+                raise UnexpectedModelBehavior(
                     'Streamed response with unexpected content, expected all parts to be text'
                 )
 
@@ -334,7 +334,7 @@ def get(self, *, final: bool = False) -> ModelStructuredResponse:
                 combined_parts.extend(parts)
             elif not candidate.get('finish_reason'):
                 # you can get an empty text part along with the finish_reason, so we ignore that case
-                raise UnexpectedModelBehaviour(
+                raise UnexpectedModelBehavior(
                     'Streamed response with unexpected content, expected all parts to be function calls'
                 )
         return _structured_response_from_parts(combined_parts, timestamp=self._timestamp)
@@ -534,14 +534,14 @@ def _extract_response_parts(
     Returns Either a list of function calls (Either.left) or a list of text parts (Either.right).
     """
     if len(response['candidates']) != 1:
-        raise UnexpectedModelBehaviour('Expected exactly one candidate in Gemini response')
+        raise UnexpectedModelBehavior('Expected exactly one candidate in Gemini response')
     parts = response['candidates'][0]['content']['parts']
     if _all_function_call_parts(parts):
         return _utils.Either(left=parts)
     elif _all_text_parts(parts):
         return _utils.Either(right=parts)
     else:
-        raise exceptions.UnexpectedModelBehaviour(
+        raise exceptions.UnexpectedModelBehavior(
             f'Unsupported response from Gemini, expected all parts to be function calls or text, got: {parts!r}'
         )
 

pydantic_ai/models/openai.py

@@ -13,7 +13,7 @@
 from openai.types.chat.chat_completion_chunk import ChoiceDeltaToolCall
 from typing_extensions import assert_never
 
-from .. import UnexpectedModelBehaviour, _utils, result
+from .. import UnexpectedModelBehavior, _utils, result
 from ..messages import (
     ArgsJson,
     Message,
@@ -184,7 +184,7 @@ async def _process_streamed_response(response: AsyncStream[ChatCompletionChunk])
         try:
             first_chunk = await response.__anext__()
         except StopAsyncIteration as e:  # pragma: no cover
-            raise UnexpectedModelBehaviour('Streamed response ended without content or tool calls') from e
+            raise UnexpectedModelBehavior('Streamed response ended without content or tool calls') from e
         timestamp = datetime.fromtimestamp(first_chunk.created, tz=timezone.utc)
         delta = first_chunk.choices[0].delta
         start_cost = _map_cost(first_chunk)
@@ -194,7 +194,7 @@ async def _process_streamed_response(response: AsyncStream[ChatCompletionChunk])
             try:
                 next_chunk = await response.__anext__()
             except StopAsyncIteration as e:
-                raise UnexpectedModelBehaviour('Streamed response ended without content or tool calls') from e
+                raise UnexpectedModelBehavior('Streamed response ended without content or tool calls') from e
             delta = next_chunk.choices[0].delta
             start_cost += _map_cost(next_chunk)
 

pydantic_ai/result.py

@@ -279,7 +279,7 @@ async def validate_structured_result(
         assert self._result_schema is not None, 'Expected _result_schema to not be None'
         match = self._result_schema.find_tool(message)
         if match is None:
-            raise exceptions.UnexpectedModelBehaviour(
+            raise exceptions.UnexpectedModelBehavior(
                 f'Invalid message, unable to find tool: {self._result_schema.tool_names()}'
             )
 

tests/models/test_gemini.py

@@ -12,7 +12,7 @@
 from pydantic import BaseModel
 from typing_extensions import Literal, TypeAlias
 
-from pydantic_ai import Agent, ModelRetry, UnexpectedModelBehaviour, UserError, _utils
+from pydantic_ai import Agent, ModelRetry, UnexpectedModelBehavior, UserError, _utils
 from pydantic_ai.messages import (
     ArgsObject,
     ModelStructuredResponse,
@@ -511,7 +511,7 @@ def handler(_: httpx.Request):
     m = GeminiModel('gemini-1.5-flash', http_client=gemini_client)
     agent = Agent(m, system_prompt='this is the system prompt')
 
-    with pytest.raises(UnexpectedModelBehaviour) as exc_info:
+    with pytest.raises(UnexpectedModelBehavior) as exc_info:
         await agent.run('Hello')
 
     assert str(exc_info.value) == snapshot('Unexpected response from gemini 401, body:\ninvalid request')
@@ -535,7 +535,7 @@ async def test_heterogeneous_responses(get_gemini_client: GetGeminiClient):
     gemini_client = get_gemini_client(response)
     m = GeminiModel('gemini-1.5-flash', http_client=gemini_client)
     agent = Agent(m)
-    with pytest.raises(UnexpectedModelBehaviour) as exc_info:
+    with pytest.raises(UnexpectedModelBehavior) as exc_info:
         await agent.run('Hello')
 
     assert str(exc_info.value) == snapshot(
@@ -573,7 +573,7 @@ async def test_stream_text_no_data(get_gemini_client: GetGeminiClient):
     gemini_client = get_gemini_client(stream)
     m = GeminiModel('gemini-1.5-flash', http_client=gemini_client)
     agent = Agent(m)
-    with pytest.raises(UnexpectedModelBehaviour, match='Streamed response ended without con'):
+    with pytest.raises(UnexpectedModelBehavior, match='Streamed response ended without con'):
         async with agent.run_stream('Hello'):
             pass
 
@@ -691,5 +691,5 @@ async def test_stream_text_heterogeneous(get_gemini_client: GetGeminiClient):
 
     msg = 'Streamed response with unexpected content, expected all parts to be text'
     async with agent.run_stream('Hello') as result:
-        with pytest.raises(UnexpectedModelBehaviour, match=msg):
+        with pytest.raises(UnexpectedModelBehavior, match=msg):
             await result.get_data()

tests/models/test_openai.py

@@ -23,7 +23,7 @@
 from openai.types.completion_usage import CompletionUsage, PromptTokensDetails
 from typing_extensions import TypedDict
 
-from pydantic_ai import Agent, ModelRetry, UnexpectedModelBehaviour, _utils
+from pydantic_ai import Agent, ModelRetry, UnexpectedModelBehavior, _utils
 from pydantic_ai.messages import (
     ArgsJson,
     ModelStructuredResponse,
@@ -424,6 +424,6 @@ async def test_no_content(allow_model_requests: None):
     m = OpenAIModel('gpt-4', openai_client=mock_client)
     agent = Agent(m, result_type=MyTypedDict)
 
-    with pytest.raises(UnexpectedModelBehaviour, match='Streamed response ended without con'):
+    with pytest.raises(UnexpectedModelBehavior, match='Streamed response ended without con'):
         async with agent.run_stream(''):
             pass

tests/test_agent.py

@@ -6,7 +6,7 @@
 from inline_snapshot import snapshot
 from pydantic import BaseModel
 
-from pydantic_ai import Agent, CallContext, ModelRetry, UnexpectedModelBehaviour, UserError
+from pydantic_ai import Agent, CallContext, ModelRetry, UnexpectedModelBehavior, UserError
 from pydantic_ai.messages import (
     ArgsJson,
     ArgsObject,
@@ -445,7 +445,7 @@ def empty(_: list[Message], _info: AgentInfo) -> ModelAnyResponse:
 
     agent = Agent(FunctionModel(empty))
 
-    with pytest.raises(UnexpectedModelBehaviour, match='Received empty tool call message'):
+    with pytest.raises(UnexpectedModelBehavior, match='Received empty tool call message'):
         agent.run_sync('Hello')
 
 
@@ -455,7 +455,7 @@ def empty(_: list[Message], _info: AgentInfo) -> ModelAnyResponse:
 
     agent = Agent(FunctionModel(empty))
 
-    with pytest.raises(UnexpectedModelBehaviour, match=r'Exceeded maximum retries \(1\) for result validation'):
+    with pytest.raises(UnexpectedModelBehavior, match=r'Exceeded maximum retries \(1\) for result validation'):
         agent.run_sync('Hello')
     assert agent.last_run_messages == snapshot(
         [

tests/test_streaming.py

@@ -7,7 +7,7 @@
 import pytest
 from inline_snapshot import snapshot
 
-from pydantic_ai import Agent, UnexpectedModelBehaviour, UserError
+from pydantic_ai import Agent, UnexpectedModelBehavior, UserError
 from pydantic_ai.messages import (
     ArgsJson,
     ArgsObject,
@@ -154,7 +154,7 @@ async def text_stream(_messages: list[Message], _: AgentInfo) -> AsyncIterator[s
 
     agent = Agent(FunctionModel(stream_function=text_stream), result_type=tuple[str, str])
 
-    with pytest.raises(UnexpectedModelBehaviour, match=r'Exceeded maximum retries \(1\) for result validation'):
+    with pytest.raises(UnexpectedModelBehavior, match=r'Exceeded maximum retries \(1\) for result validation'):
         async with agent.run_stream(''):
             pass
 
@@ -232,7 +232,7 @@ async def stream_structured_function(_messages: list[Message], _: AgentInfo) ->
 
     agent = Agent(FunctionModel(stream_function=stream_structured_function), result_type=tuple[str, int])
 
-    with pytest.raises(UnexpectedModelBehaviour, match='Received empty tool call message'):
+    with pytest.raises(UnexpectedModelBehavior, match='Received empty tool call message'):
         async with agent.run_stream('hello'):
             pass
 
@@ -247,7 +247,7 @@ async def stream_structured_function(_messages: list[Message], _: AgentInfo) ->
     async def ret_a(x: str) -> str:
         return x
 
-    with pytest.raises(UnexpectedModelBehaviour, match=r'Exceeded maximum retries \(1\) for result validation'):
+    with pytest.raises(UnexpectedModelBehavior, match=r'Exceeded maximum retries \(1\) for result validation'):
         async with agent.run_stream('hello'):
             pass
     assert agent.last_run_messages == snapshot(